190
ACTIVE ANTENNAS
ACTIVE ANTENNAS
The present article is an introduction to the topic of active
antennas. The first section is a description of the field suitable
for reading by almost any undergraduate science major. The
next section is an in-depth reexamination of the subject, including equations and some derivations. Its basic idea is to
provide the reader with enough tools that he or she can evaluate whether it is an active antenna that he or she might need
for a specific application. The final section is a discussion of
where active antennas are finding and will find application.
We should mention here that, if one really needs to design
active antennas, one will need to go further than this article.
The set of references to the primary research literature given
in this article is by no means complete, nor is it meant to be.
A good way to get started on the current literature on this
topic would be a reading of the overview monograph of Navarro and Chang (1). We will not cover active amplifiers in
this article. However, this topic is treated in the book edited
by York and Popovic (2).
AN INTRODUCTION TO ACTIVE ANTENNAS
An antenna is a structure that converts electromagnetic energy propagating in free space into voltage and current in an
electrical circuit and/or vice versa. In a transceiver system,
the antenna is used both to receive and to transmit free-space
waves. At minimum, a transceiver then must consist of a signal source that serves to drive the antenna as well as a receiver circuit that reads out the signal from the antenna. Until recently, practically all antenna systems operating in the
microwave frequency regime (operation frequencies greater
than 1 billion cycles per second, or 1 GHz) were mostly designed to isolate the antenna from the circuitsthat is, to
find ways to make system operation independent of the antennas electrical characteristics. In contradistinction, an active antenna is one in which the antenna actually serves as a
circuit element of either the driver or the readout circuit. To
understand why this is different from conventional antenna
driving or readout will require us to take a bit of a historical
trip through the last century or so.
Actually, the first antenna was an active one. Heinrich
Hertz, back in 1884 (2a), was the first to demonstrate that
one could generate radio waves and that they would propagate from a transmitter to a receiver at the speed of light.
The apparatus used is schematically depicted in Fig. 1. The
idea of the transmitter is that, by discharging an induction
coil (a wire looped about a magnetic core such that the composite device can store significant amounts of magnetic energy) into a spark gap, one can generate a current in the 5
mm diameter wire. The voltage in the spark gap induces a
current in the wires, which in turn induces a voltage in the
wires, and this voltage in turn induces current, so that the
voltage and current propagate along the two pieces of the wire
J. Webster (ed.), Wiley Encyclopedia of Electrical and Electronics Engineering. Copyright # 1999 John Wiley & Sons, Inc.
Induction
coil
5 mm
3/4-cm
spark gap
1m
35 cm
30 cm
2 mm
(a)
(b)
Figure 1. Hertz apparatus for (a) transmitting and (b) receiving radio waves, where the transmitting antenna serves to choose a specific
frequency of the spark gap voltage to transmit to the receiving antenna, which also serves to pick out this special frequency from the
free-space waveform and turn this electromagnetic disturbance into
a voltage across the receiver antenna gap.
to either side of the gap as waves, appearing much like a onedimensional slice through a water wave propagating away
from the point where a pebble has struck the waters surface
(the spark gap). A wave will propagate rectilinearly until it
encounters an obstruction, at which point it can suffer reflection from or transmission into the barrier that the obstruction
presents. There will be reflections then off the metal spheres on
the ends of the wire. The spark will generate a broad spectrum
of frequencies or wavelengths. The reflections off the two ends,
though, will tend to cancel each other except at certain special
frequencies. The effect at these wrong frequencies is much like
the effect of throwing a handful of pebbles into the pond and noting that, in between the points where the pebbles struck, the
waves are much more indistinct than they are far from where
the handful struck the surface. The special frequencies are ones
which just fit into the region between the spheres. The current
needs to be zero at the two ends in order to fit, whereas the voltage needs to be maximum at the ends. The current and voltage
waves at the right frequencies may appear as depicted in Fig. 2.
The Hertz transmitter is the archetypical active antenna.
The source is the spark gap, which is actually placed in the
antenna. The antenna then acts as a filter to pick the right
frequency out of a large number of frequencies that could be
launched from the gap. The receiver is picked to be of a length
to also select this primary frequency.
Hertz-style spark gap transmitters, after further development and popularization by Marconi, were in use for fifty
years after Hertz. However, such transmitters exhibit some
rather severe drawbacks. The main problem is that the simple resonant dipole antenna (that is, a straight wire antenna
with a gap or a feeder cable used to feed in current) is a filter
with a poor frequency selection. Namely, if one increases the
I
V
Figure 2. Current and voltage waveforms for the lowest-order (least
number of zeros) waveform for the Hertz transmitter of Fig. 1(a). The
current must go to zero at the points where the wire ends, whereas
the potential will be highest there.
Power transmitted
ACTIVE ANTENNAS
191
100%
75%
f0
2f0
3f0
4f0
Figure 3. A sketch of what the transmission as a function of frequency might look like for the Hertzian dipole antenna of Figs. 1
and 2.
frequency by 50%, there is 75% as much power transmitted
at this frequency as at the first resonance, which is called
the fundamental. There is a second resonance at twice the
frequency of the first resonance, and another at each integer
multiple of the fundamental. With increasing frequency, the
transmitted power decreases a little and then flattens out
around the second resonance, decreases a little, flattens out
at the third resonance, etc., as is illustrated in Fig. 3. If the
spark discharge is really broadband (that is, if it generates a
large number of frequencies where the highest frequency may
be many times the lowest), then what is transmitted by the
antenna will also be broadband, although with somewhat
higher transmission at the fundamental frequency and its
harmonics than in between. In the very early days of radio,
this was somewhat acceptable, although any information impressed on such a broadband carrier would be rather severely
degraded upon reception. However, the demise of the spark
gap transmitter was really instigated by the early success of
radio, which caused the available frequency bands to begin to
fill up rapidly. This band filling led to the formation of the
Federal Communications Commission (FCC) in 1934, which
was charged with allocation of frequency bands. The allocation by nature led to a ban on spark gap transmitters, which
were needlessly wasting bandwidth.
In a later experiment, Hertz noticed that the waves he was
generating would tend to have a component that hugged the
ground and could therefore travel over the horizon and, in
fact, across the Atlantic Ocean, skimming along the surface
of the water. Other researchers noticed that the effect became
more pronounced at wavelengths longer than the roughly 2 m
wavelength that Hertz originally used. (For the frequencies
and wavelengths of some important frequency bands, see Table 1.) In order for wave transmission to be useful, however,
the transmitted signal needs to carry information. Impressing
information on the wave is called modulating the carrier. One
can modulate the height (amplitude), the frequency, and so
on. The discovery of a technique to amplitude-modulate the
waves coming off an antenna (in 1906) then led to the inception of AM radio in bands with wavelengths greater than 300
m, which corresponds to roughly 1 MHz. AM radio became
commercial in 1920. By the 1930s, other researchers noted
that waves with frequencies around 10 MHz, corresponding
to a wavelength around 30 m, could be quite efficiently propagated over the horizon by bouncing the wave off the ionosphere. This led to the radio bands known as short-wave. In
1939, a researcher realized a technique to modulate the frequency of the wave. This realization led in the 1950s to FM
radio, which was allocated the band around 100 MHz with
192
ACTIVE ANTENNAS
Table 1. A Listing of the Allocated Microwave and
Millimeter-Wave Bands as Defined by the Frequency and
Wavelength Range Within Each Band
L
Battery
Band
Designation
Frequency
(GHz)
Wavelength
L
S
C
X
Ku
K
Ka
Q
U
V
E
W
D
G
Y
12
24
48
812
1218
1826
2640
3350
4060
5075
6080
75110
110170
140220
220325
1530 cm
7.515 cm
3.757.5 cm
2.53.75 cm
1.672.5 cm
1.151.67 cm
0.751.15 cm
69 mm
57.5 mm
46 mm
3.755 mm
2.74 mm
1.82.7 mm
1.42.1 mm
0.91.4 mm
a corresponding wavelength around 3 m. However, the FM
technique was used first during World War II as a radar modulation technique. Radars today are at frequencies above
roughly 1 GHz or wavelengths below 30 cm.
There is a fundamental difference between circuits that operate at frequencies whose corresponding wavelengths are
less than the maximum circuit dimension and those that are
large compared to the carrier wavelength. The effect is closely
related to the concept of impedance. As was mentioned above,
in the wire antenna, the voltage and current reinforce each
other and thereby travel on the antenna as waves. The same
effect takes place in a circuit. At any point along the path
(line) in a circuit, one defines the ratio of voltage at one frequency to the current at the same frequency as the impedance
at that frequency. For a sinusoidal waveform, if the impedance tends to preserve the phase relationship (where the
wave peaks lie, relatively), then we say that the impedance is
resistive. If the impedance tends to drive the voltage peaks
forward with respect to the current peaks, we say that the
impedance is capacitive; in the opposite case we say that the
impedance is inductive. In a small circuit (small compared to
a wavelength), one generally tries to carefully design passive
componentsresistors, capacitors, and inductorsso that
Ground
Figure 5. A circuit with lumped elements connected by wire segments.
they exhibit large local impedance, that is, large impedance
within their physical dimensions. When the circuit is small,
one would like to control the phase and amplitude of the wave
at discrete points by using lumped elements and thereby minimizing line effects. The lines (wires) between the components
have little or no effect on the electromagnetic disturbances
passing through the circuit, then, as the impedances in the
wires are small and reasonably independent of their lengths.
When the circuit is large, the lines themselves effectively become circuit elements, and they themselves must be carefully
designed in order to exhibit the proper impedances. To illustrate, consider the parallel plate capacitor of Fig. 4. The capacitance is maximized by maximizing the permittivity (a
material parameter equal to the ratio of electrial displacement to applied electric field) and area A while minimizing
the plate spacing d. However, the fact that the capacitance
depends on the plate spacing d is the important point here.
Consider the circuit of Fig. 5 as an example. The only ground
in the figure is the one on the battery, but the wires connecting the circuit elements together in essence form at each point
a capacitor, with a point on the wire that is carrying charge
as the upper plate and the ground as the lower. This capacitance changes as a function of position along the wire. For a
small enough circuit (relative to the wavelength of the highest
frequency carried by the circuit), the effect is not too important, as the wireground pair has small capacitance and the
position-varying effect is small. For a large circuit, the effect
is disastrous, as we shall consider below. The effect is identical to the effect of Fresnel coefficients in optics.
Consider the circuit of Fig. 6. We will now discuss what
happens when impedances are not carefully controlled. This
leads to the concept of impedance matching. Let us first say
that the circuit is short (compared to a wavelength). If the
load resistor, RL, is not matched to (that is, is not equal to, or,
one could say, not impedance matched to) the resistance of the
source, RS, some amount of reflection will occur at RL, propagate back to RS, be reflected with a reversal of sign at RL,
Area A
++++++++++++++
RS
d
RL
Figure 4. Schematic depiction of a parallel plate capacitor in which
the flow of a current will tend to change the upper plate, causing a
voltage difference between upper and lower plates. The capacitance
is defined as the ratio of the amount of change of the upper plate to
the magnitude of the voltage this change induces between the plates.
Figure 6. A circuit in which one is trying to supply power from a
source with internal resistance RS to a load with resistance RL. The
power transfer is maximized only when RS and RL are equal, in which
case half the power supplied by the source is supplied to the load, the
other half being dissipated in the source and causing it to heat.
ACTIVE ANTENNAS
Inner
conductor
Outer conductor
(ground potential)
Figure 7. A coaxial cable in which signals are carried on an inner
conductor and in which the grounded outer conductor serves to carry
the ground plane along with the signal in order to give a constant
impedance along the line.
propagate back to RL, etc. The reflections add up perfectly out
of phase (that is, simply subtract from one another) at the
source and load, and the amount of power supplied to the load
is less than optimal. In this limit of a small circuit, it is as if
the load will not allow the source to supply as much power as
it is capable of. Let us now say that the line is well-designed
but long compared to the wavelength used. Then the same
argument applies to the reflections, but in this case the source
does not know that the load is there until several wave periods have passed (several maxima and minima of the waveform have left the source), so the source supplies all the power
it can. The power, though, is not allowed to be fully absorbed
by the load, and some of it will rattle around the line until it
is radiated or absorbed. As we mentioned above, in a long
enough circuit the wire itself becomes a distributed element
that is, one with an impedance of its own. If the distance to
the nearest ground is not kept fixed along the line, the inductance and capacitance become dependent on the position. In
this case, we have distributed reflections all along the line
and the circuit will probably not work at all. This spatial variability of the line impedance is remediable, though, as illustrated by the drawing of a coaxial cable in Fig. 7. The idea is
that, if the line brings along its own ground plane in the form
of a grounded outer conductor, the characteristic impedance
of the line can be kept constant with distance. Such a line,
which carries its own ground plane, is called a transmission
line. The problem becomes the connection of the line to the
source and load (i.e., impedance matching).
Before going on to discuss the conventional solution versus
the new active antenna solution, perhaps we should summarize a bit. In AM, short-wave, and FM applications, the wavelengths are of order greater than meters. If one considers typical receivers, the whole circuit will generally be small
compared to the carrier wavelength. This is also to say that
Oscillator
Antenna
Matching
network
Matching
network
Transmission
line
Transmission
line
193
in all of these cases, the antennas will be active in the sense
that the antenna presents an impedance to the circuit. (Recall
that an active antenna is any antenna in which an active element lies within a wavelength of the antenna and is used as
an element to match the antenna impedance to the decoder
impedance.) To passively match an antenna to the receiver
circuit, one needs pieces of line comparable to a wavelength.
However, from here on we shall not be interested in the lowfrequency case but rather in the well-above-1-GHz case, as
AM, FM, and TV technologies are mature technologies. During World War II, radar was the application that drove the
frequencies above 1 GHz (wavelength less than 30 cm). In a
radar, one sends out a pulse and, from the returned, scattered
wave, tries to infer as much as possible about the target. Target resolution is inversely proportional to wavelength. There
has been a constant drive to shorten wavelength. Therefore,
as is indicated by Table 1, bands have been allocated out to
hundreds of gigahertz. Presently, however, there are a plethora of nonmilitary drivers for pushing to higher-frequency
communication systems that are compact and have lower
power dissipation. However, the conventional solution, which
was developed originally for radars, is really not conducive to
compactness nor to the pressures of cost minimization of the
commercial market.
A typical conventional transmitter is schematically depicted in Fig. 8. A main concept here is that the transmission
lines and matching networks are being used to isolate the
oscillator from the amplifier and the amplifier from the antenna, in contrast to the situation in an active antenna. There
were a number of reasons why the conventional solution took
on the form it did. Among them was the urgency of World
War II. Radar was developed rapidly in both Great Britain
and the United States in the 1930s and 1940s. Rapid development required numerous researchers working in parallel.
When operating frequencies exceeded 1 GHz (corresponding
to 30 cm wavelengths), passive matching networks, whose
main requirement is that they must consist of lines of lengths
comparable to a wavelength, became convenient to construct
(in terms of size) for ground-based radar. In this case, then,
the oscillators could be optimized independently of the amplifiers, which in turn could be optimized independently of the
antennas and the receiver elements. The impedances of the
individual pieces didnt matter, as the matching networks
could be used to effectively transform the effective impedances looking into an element into something completely different for purposes of matching pieces of the network to each
other. There are costs associated with such a solution, though,
such as total system size as well as the tolerances that components must satisfy. However, once the technique was in place,
the industry standardized on the conventional solution and
perfected it to the point where it was hard to challenge. The
reemergence of the active solution owes itself to two indepen-
Matching
network
Matching
network
Amplifier
Figure 8. Schematic of a conventional RF microwave transmitter in which each individual element of the transmitter is
matched to each other element.
194
ACTIVE ANTENNAS
Dc
power
in
Input 1
Input 2
Metal
Dielectric
Output
a
b
Figure 9. Schematic depiction of a feedback system that can operate
as an oscillator when G is greater than 1, the feedback is positive,
and there is a delay in feeding back the output to the input.
dent technologies, the emergence of high-frequency solid-state
devices and the development of planar circuit and planar antenna technology.
A single frequency of electromagnetic energy must be generated in a so-called oscillatorthat is, a circuit that converts
dc electrical power to ac electromagnetic power at the proper
frequency. The basic operation of an oscillator can be described with respect to Fig. 9. What is shown here schematically is an amplifier in which a portion b (1) of the output
is fed back to the input with either a plus or a minus sign.
When the feedback is off (b 0), then the signal out will be
just G times the input. When the feedback is negative, the
output will be less than G times the input. However, in the
negative feedback mode, the stability to noise increases, since
fluctuations will be damped. That is, if the output fluctuates
up, this lowers the effective input, whereas if the output fluctuates down, the output is driven up. The opposite is true in
the positive feedback case. In the positive feedback case, if
there were no fluctuations, any input would cause the output
to increase until all of the dc power in as well as all of the
input signal showed up at the output. (This is all of the power
that can show up at the output. Such behavior is typical of
unstable operation.) This would not be such an interesting
case; however, there are always fluctuations of the input, and
the positive feedback will cause these to grow. If there is a
delay from output to input, then fluctuations with a period
corresponding to this delay will be favored, as a rise in the
input will show up as a rise in the output one period later,
and rapidly all of the dc power in will be converted to power
out at this magic frequency.
A real circuit operates a bit more interestingly than our
ideal one. In a real circuit, as the fluctuations build up, the
gain is affected and some elements absorb power, but the oscillations still take place, although perhaps with a different
frequency and amplitude from what one would have predicted
from nondynamic measurements.
The transistor was first demonstrated in 1947, with publication in 1948 (3), and the diode followed shortly (4). Although the field effect transistor (FET) was proposed in 1952
(5), it was not until the mid 1960s that the technology had
come far enough that it could be demonstrated (6). The FET
(and variations thereof) is presently the workhorse microwave
three-terminal device. Two-terminal transfer electron devices
(a)
(b)
Figure 10. Views of (a) a microstrip and (b) a coplanar waveguide
line. In the microstrip, the ground plane is the lower electrode,
whereas in the coplanar waveguide the ground plane is placed on the
surface of the dielectric substrate.
(TEDs) were used before the FET for microwave applications
and are still in use, but tend to have a much lower wall plug
efficiency (dc to ac conversion), especially as the amplifying
device of an oscillator. Radar systems, however, were already
in use in the late 1930s. Essentially all of the microwave
sources in radars up until the 1970s operated on principles
that required that the source have physical dimensions larger
than a wavelength, and perhaps many wavelengths. This fact
almost required the conventional solution to be used. Transistors, though, can have active areas with dimensions of micrometers; even packaged hybrid devices can have complete
packages of dimensions smaller than a millimeter. The transistor can therefore act as an amplifier with dimensions much
smaller than a wavelength and does not, therefore, need to be
placed in a conventional (passive) solution design.
The last piece of our story of the new active antenna era
involves the development of printed circuit technology, along
with slot and patch antennas. The two most common planar
open waveguide designs are microstrip line and coplanar
waveguide (CPW). Depictions of these waveguide lines are
given in Fig. 10. The idea behind the microstrip line is to
propagate electromagnetic energy along the lines by confining
the electric field between the upper signal line and a lower
ground plane. As the upper line carries current, a magnetic
field encircles the upper line. As power flow takes place in a
direction perpendicular to the electric and magnetic fields, the
power flow is mostly between the signal line and the ground
Output strip
Gap
Drain
Figure 11. A simple transistor oscillator implemented in CPW technology.
ACTIVE ANTENNAS
(a)
195
(b)
(a)
;
;
;;
(b)
Figure 12. A depiction of (a) a patch antenna in a microstrip line
and (b) a slot antenna in a CPW line.
Input line
line in the dielectric. On a low-frequency wire (a line whose
transverse dimensions are small compared to a wavelength),
the voltage and current waveforms reinforce each other. The
coupling of the electric and magnetic fields in the microstrip
is analogous to the coupling of voltage and current on the
Hertz antenna wire, except that the microstrip line can be
electrically long in the sense that the distance from the signal
line to the ground plane is kept constant so that the impedance can be kept constant, as with the earlier-discussed coaxial cable. Lines that carry along their ground planes are generally referred to as transmission lines. Components (i.e.
capacitors and inductors) can be built into the line by changing the width, cutting gaps into the upper line, or putting
slits in the ground plane. In this sense, we can still describe
transmission line circuits by conventional circuit theory if we
use a special circuit model for the line itself. The CPW line is
quite similar to the microstrip line except that there the
ground planes are on top of the dielectric slab. Either of these
line types is reasonably easy to fabricate, as one needs only
to buy a metal-coated dielectric plate and then pattern the
needed shapes by photographically defining the patterns using a technique known as photolithography, a process common to all present-day circuit fabrication. These planar structures are quite compatible with transistor technology, as is
indicated by the simple transistor oscillator circuit depicted
in Fig. 11. The gap in the line on the drain side is there in
order to provide the proper feedback for oscillation. In this
case, the total oscillator linear dimension can be less than
a wavelength.
In order to have an active antenna, one needs to have a
radiating elementthat is, a passive antenna element in the
active antenna. There are antenna technologies which are
compatible with microstrip and CPW technologies, and the
resulting antenna types are illustrated in Fig. 12. The idea
behind either of these antenna types is that the patch (slit) is
designed to have a transverse length that matches the operating wavelength (as we discussed in conjunction with
Hertz dipole antennas). In the case of the patch, the electric
field points primarily from the patch to the ground plane, as
is illustrated in Fig. 13. The edges of the transverse (to the
input line) dimension will then have a field pattern as
sketched in Fig. 13(a), and the longitudinal edges will have a
field pattern as sketched in Fig. 13(b), with a composite
sketch given in Fig. 13(c). The important part of the sketches,
however, is really the so-called fringing fields in Fig. 13(a)
(c)
Figure 13. Illustration of the electric field directions along (a) the
nonradiating edge and (b) the radiating edge, and (c) a schematic
depiction of the edge fields around the patch.
that is, the fields that point neither up nor down but rather
across. Beyond the longitudinal edges of the patch are fields,
in phase for the two edges, that are normal to the surface. It
is these fields (when combined with transverse magnetic
fringe fields in the same strips) that give rise to the upward
radiation. Similar arguments describe the operation of the slit
antenna if one exchanges the electric and magnetic fields in
the argument.
We have now introduced all of the pieces necessary to describe the new resurgence in active antenna research. A possible active antenna design could appear as in Fig. 14 (7),
where the transistor is actually mounted right into the patch
antenna element, and therefore the design can be quite compact. That is, the source plus oscillator plus antenna can all
be fitted into less than a wavelength. The design of Fig. 14,
which comes from R. Comptons group at Cornell (31,32), will
be discussed further in the next section.
/4
short
Gate
Source
Source
Slit
Drain
Drain
bias
Bypass
capacitor
Source leads,
grounded
Figure 14. Depiction of the upper surface metallization of a microstrip active patch antenna discussed in Ref. 7. The short on the gate
together with the slit between gate and drain provides the proper
feedback delay to cause oscillation.
196
ACTIVE ANTENNAS
There are a number of advantages to the use of active antennas. One is that an active antenna can be made compact.
Compactness in itself is advantageous, as throughout the history of microelectronics, miniaturization has led to lowered
costs. There are two more advantages, though, which relate
to compactness. One is that the power-handling capabilities
of a device go down with increasing frequency. We would
therefore like to find ways to combine the power from several
devices. One can try to add together outputs from various oscillators in the circuit before feeding them to the elements,
but this goes back to the conventional solution. A more advantageous design is to make an array of antennas, with proper
spacing relative to the wavelength and antenna sizes, and
add the power of the locked oscillators in the array quasioptically in free space. (In other words, optical radiation tends
to radiate into free space, whereas radio frequency in microwave radiation needs to be kept in guiding waveguides until
encroachment on radiating elements. Quasi-optics uses the
principle of the optical interferometer to combine multiple coherent microwave fields in free space.) The locking requires
that the oscillators talk to each other so that the phases of all
of the array elements stay in a given relation. As will be discussed in more detail in the next section, however, an important problem at present in the active antenna field relates to
keeping elements locked yet still being able to modulate the
output as well as steer the beam in order to be able to electronically determine on output direction. These issues will be
discussed in the next section and taken up in more detail in
the last section.
where E is the electric field vector, B is the magnetic induction vector, H is the magnetic field vector, D is the electric
displacement vector, J is the current density vector, and is
the volume density of charge. An additional important quantity is S, the Poynting vector, defined by
SOME QUANTITATIVE DISCUSSION
OF ASPECTS OF ACTIVE ANTENNAS
Integrating this equation throughout a volume V and using
Gausss theorem,
S dV = S dA
In order to be able to make calculations on active antennas,
it is important to know what level of approximation is necessary in order to obtain results. An interesting point is that,
although the operating frequency of active antennas is high,
the circuit tends to be small in total extent relative to the
operating wavelength, and therefore the primary design tool
is circuit theory mixed with transmission line theory. These
techniques are approximate, and a most important point in
working with high frequencies is to know where a given technique is applicable. Exact treatments of all effects, however,
prove to be impossible to carry out analytically. Numerical
approaches tend to be hard to interpret unless one has a
framework to use. The combined circuit transmission-line
framework is the one generally applied. When it begins to
break down, one tends to use numerical techniques to bootstrap it back to reality. We will presently try to uncover the
basic approximations of transmission line and circuit theory.
Maxwells equations are the basic defining equations for
all electromagnetic phenomena, and they are expressible in
MKSA units as (8)
B
B
t
D
D
H = J +
t
D =
E =
B = 0
S = E H
If one takes the divergence of S, one finds
E H)
S = (E
If one assumes a free-space region,
D = 0E
B = 0 H
which is therefore lossless,
J =0
and charge-free,
=0
(where 0 is the permittivity of free space and 0 is the permeability of free space), one can use vector identities and Maxwells equations to obtain
S =
0
E E ) 0 (H
H H)
(E
2 t
2 t
where dA is the differential area times the unit normal pointing out of the surface of the volume V, one finds that
S dA =
We Wm
t
t
where We is the electric energy density
We =
0
E E dV
2
and Wm is the magnetic energy density
Wm =
0
H H dV
2
The interpretation of the above is that the amount of S flowing out of V is the amount of change of the energy within.
One therefore associates energy flow with S E H. This is
important in describing energy flow in wires as well as transmission lines and waveguides of all types. As was first described by Heaviside (9), the energy flow in a wire occurs not
inside the wire but around it. That is, as the wire is highly
conductive, there is essentially no field inside it except at the
surface, where the outer layer of oscillating charges have no
outer shell to cancel their effect. There is therefore a radial
electric field emanating from the surface of the wire, which
ACTIVE ANTENNAS
combines with an azimuthal magnetic field that rings the current flow to yield an E H surrounding the wire and pointing
down its axis.
It was Pocklington in 1897 (10) who made the formal structure of the fields around a wire a bit more explicit and, in the
effort, also formed the basis for the approximation upon
which most of circuit and transmission line theory rests, the
quasi-static approximation. A simplified version of his argument is as follows. Assume an xyz Cartesian coordinate
system where the axis of the wire is the z axis. One then
assumes that all of the field quantities f(x, y, z, t) vary as
f (x, y, z, t) = f (x, y) cos(z t + )
If one assumes that the velocity of propagation of the abovedefined wave is c (00)1/2, the speed of light, then one can
write that
=
The assumption in the above that f(x, y) is independent of z,
by substitution of the above into Maxwells equations, can be
shown to be equivalent to the assumption that the transverse
field components Ex, Ey, Bx, and By all satisfy relations of the
form
Ex
z |Ex |
which is the crux of the quasistatic approximation. With the
above approximation, one finds that
t Et =
t Ht = J
where
+ e y
t = e x
x
y
which is just the transverse, and therefore two-dimensional,
gradient operator. These equations are just the electro- and
magnetostatic equations for the transverse fields, whereas
the propagation equation above shows that these static transverse field configurations are propagated forward as if they
corresponded to a plane wave field configuration. If the magnetic field is caused by the current in the wire, it rings the
wire, whereas if the electric field is static, it must appear to
emanate from charges in the wire and point outward at right
angles to the magnetic field. If this is true, then the Poynting
vector S will point along the direction of propagation and the
theory is self-consistent, if approximate.
If we wish to guide power, then the quasistatic picture
must come close to holding, as the Poynting vector is in the
right direction for guidance. The more general approximate
theory that comes from Pocklingtons quasistatic approximation is generally called transmission line theory. To derive this
theory, first consider the two-wire transmission line of Fig.
197
Ground plane
Equiphase
lines
C1
z axis
J lines
V2
Conductor 1
Figure 15. A sketch of a two-conductor transmission line where
some equipotentials and some current lines are drawn in, as well as
a volume V1 with outward-pointing normal dA1. There is also an outward-pointing normal dA2 associated with the area bounded by contour C2.
15. If we are to have something that we can actually call a
transmission line, then we would hope that we can find
equiphase fronts of the electromagnetic disturbance propagating in the gap crossing the gap conductor and that we can find
lines along which the current flows on the current-carrying
conductor. Otherwise (if the equiphases closed on themselves
and/or we had eddies in the current), it would be hard to
think of the structure as any form of guiding structure. Let
us say we form an area in the gap with two walls of the foursided contour C1 surrounding this area following equiphases
an infinitesimal distance dz from each other. We can then
write
E dA 1 =
B
B
dA 1
t
where dA1 corresponds to an upward-pointing normal from
the enclosed area. One generally defines the integral as
B dA 1 =
where is the magnetic flux. We often further define the flux
as the inductance of the structure times the current:
= Li
The integral with the curl in it can be rewritten by Stokes
theorem as
E dl
E dA 1 =
C1
where C1 is the contour enclosing the area. If we define
v = E dl
on the two equiphase lines of the contour C1, where v is an ac
voltage (this is the main approximation in the above, as it is
only strictly true for truly static fields), then, noting that v
does not change along two of the boundaries of the contour
(because they are the infinitesimal walls on constant-voltage
plates) and making the other two connecting lines infinitesimal, we note that the relation between the curl of E and the
magnetic field reduces to
v(z + dz) v(z) =
(Li)
t
198
ACTIVE ANTENNAS
where it has been tacitly assumed that geometric deviations
from rectilinearity are small enough that one can approximately use Cartesian coordinates, which can be rewritten in
the form
i
v
=l
z
t
i(z)
v(z)
l
vi
(1)
=0
t
J dV2 =
Q=
dV2
Q = Cv
Noting that
J dV2 =
J dA 2
where dA2 is the outward-pointing normal to the boundary of
the volume V2 and where one usually defines
i=
J dA 2
and letting the volume V have infinitesimal thickness, one
finds that
J dA 2 = i(z + dz) i(z)
vo
g vo
tation, as is schematically depicted in Fig. 16(a). One can verify this by writing Kirchhoff s laws for the nodes with v(z
dz) and v(z) using the relations
v=l
i
t
i=c
v
t
Figure 16(b) illustrates the circuit equivalent for a lossy (and
therefore dispersive) transmission line, where r represents
the resistance encountered by the current in the metallization
and where g represents any conductance of the substrate material that might allow leakage to ground. A major point of
the diagram is that the structure need not be uniform in order
to have a transmission line representation, although one may
find that irregularities in the structure will lead to longitudinally varying inductances and capacitances.
The solution to the circuit equations will have a wave nature and will exhibit propagation characteristics, which we
discussed previously. In a region with constant l and c, one
can take a z derivative of Eq. (1) and a t derivative of Eq. (2)
and substitute to obtain
2v
2v
lc
=0
z2
t 2
which is a wave equation with solutions
Putting this together with the above, we find
v
i
=c
z
t
and
is also sometimes defined in terms of capacitance C and voltage v by
Figure 16. A circuit equivalent for (a) a lossless and (b) a lossy transmission line. The actual stages should be infinitesimally long, and
the ls and cs can vary with distance down the line. In reality, one
can find closed-form solutions for the waves in nominally constant l
and c segments and put them together with boundary conditions.
dV2
t
where the total change Q, given by
(b)
which is the equation of charge conservation. Integrating this
equation over a volume V2 that encloses the current-carrying
conductor whose walls lie perpendicular to the current lines
gives
vi
and noting that the divergence of a curl is zero, substituting
for D, we find
J+
(a)
where l is an inductance per unit length, which may vary
with longitudinal coordinate z if the line has longitudinal
variation of geometry. A similar manipulation can be done
with the second and third of Maxwells equations. Taking
H) = J + D
(
t
v(z + dz)
l
(2)
where c is the capacitance per length of the structure, and
where longitudinal variations in line geometry will lead to a
longitudinal variation of c. The system of partial differential
equations for the voltage and current have a circuit represen-
v(z, t) = vf cos(t z + f ) + vb cos(t + z + b )
(3)
where vf is the amplitude of a forward-going voltage wave, vb
is the amplitude of a backward-going voltage wave, and
= lc
;;;
;;;
ACTIVE ANTENNAS
Similarly, taking a t derivative of Eq. (1) and a z derivative
of Eq. (2) and substituting gives
2i
2i
lc 2 = 0
2
z
t
l
i
c f
l
i
vb =
c b
vf =
allowing us to write that
= lc
and the line impedance Z0 is given by
Z0 =
Figure 17. Schematic depiction of a top view of the metallized surface of an FET, where G denotes gate, D drain, and S source.
which indicates that we can make the identification that the
line phase velocity vp is given by
which will have a solution analogous to the one in Eq. (3)
above, but with
vp =
l/c
Oftentimes, we assume that we can write (the sinusoidal
steady-state representation)
Z(z ) =
Z + jZ0 tan (z l)
v(z l)
= Z0 l
i(z l)
Z0 + jZl tan (z l)
R=
i(z, t) = Re[i(z)e jt ]
so that we can write
v
= jli
z
i
= jcv
z
with solutions
v(z) = vf e j z + vb e j z
i(z) = if e j z ib e j z
Z0 i(l) = vf e j l vb e j l
hold, and from them we can find
vf =
vb =
Zl Z0
Zl + Z0
and has the meaning of a Fresnel coefficient (8). This is the
reflection we discussed in the last section, which causes the
difference between large and small circuit dimensions.
One could ask what the use was of going at some length
into Poynting vectors and transmission lines when the discussion is about active antennas. The answer is that any antenna
system, at whatever frequency or of whatever design, is a system for directing power from one place to another. To direct
power from one place to another requires constantly keeping
the Poynting vector pointed in the right direction. As we can
surmise from the transmission line derivation, line irregularities may cause the Poynting vector to wobble (with attendant
reflections down the line due to attendant variations in the l
and c), but the picture must stay close to correct for power to
get from one end of the system to another. For this reason,
active antennas, even at very high frequencies (hundreds of
gigahertz), can still be discussed in terms of transmission
lines, impedances, and circuit equivalents, although ever
greater care must be used in applying these concepts at increasingly higher frequencies.
;;;;;;
;;;;
Let us say now that we terminate the line with a lumped
impedance Zl at location l. At the coordinate l, then, the relations
Zl i(l) = vf e j l + vb e j l
+ Z0 )i(l) e
j l
Z0 )i(l) e
j l
which gives
Schottky barrier
S
n+
Ohmic contact
i(l)
[(Zl + Z0 ) e j (lz) + (Zl Z0 ) e j (lz) ]
v(z) =
2
i(l)
[(Zl + Z0 ) e j (lz) (Zl Z0 ) e j (lz) ]
i(z) =
2Z0
(4)
This equation allows us to, in essence, move the load from the
plane l to any other plane. This transformation can be used
to eliminate line segments and thereby use circuits on them
directly. However, note that line lengths at least comparable
to a wavelength are necessary in order to significantly alter
the impedance. At the plane z l, then, we can further note
that the ratio of the reflected voltage coefficient vb and the
forward-going vf , which is the voltage reflection coefficient, is
given by
v(z, t) = Re[v(z)e jt ]
1
(Zl
2
1
2 (Zl
199
Ga
As
Ohmic contact
Figure 18. Schematic depiction of the cross section of the active region of a GaAs FET. Specific designs can vary significantly in the
field-effect family.
200
ACTIVE ANTENNAS
+Vds
Vgs = 2 V
G
Ids
5V
(a)
Rd
Cgs
vin ZT
V
g mV
Rds Cds
ZL vout
Vds
10 V
(b)
Figure 19. (a) Circuit element diagram with voltages and currents
labeled for (b), where a typical IV curve is depicted.
The next piece of an active antenna that needs to be discussed is the active element. Without too much loss of generality, we will take our device to be a field effect transistor
(FET). The FET as such was first described by Shockley in
1952 (5), but the MESFET (metalsemiconductor FET),
which is todays workhorse active device for microwave circuitry, was not realized until 1965 (6), when gallium arsenide
(GaAs) fabrication techniques became workable albeit only as
a laboratory demonstration. [Although we will discuss the
MESFET in this section, it should be pointed out that the
silicon MOSFET (metaloxidesemiconductor FET) is the
workhorse device of digital electronics and therefore the most
common of all electronic devices presently in existence by a
very large margin.] A top view of an FET might appear as in
Fig. 17. As is shown clearly in the figure, an FET is a threeterminal device with gate, drain, and source regions. A cross
section of the active region (that is, where the gate is very
narrow) might appear as in Fig. 18. The basic idea is that the
saturation-doped n region causes current to flow through the
ohmic contacts from drain to source (that is, electrons flow
from source to drain), but the current is controlled in magnitude by the electric field generated by the reverse bias voltage
applied to the gate electrode. The situation is described in a
bit more detail in Fig. 19, where bias voltages are defined and
a typical IV curve for dc operation is given. Typically the
bias is supplied by a circuit such as that of Fig. 20. In what
follows, we will simply assume that the biases are properly
applied and isolated, and we will consider the ac operation.
An ac circuit model is given in Fig. 21. If one uses the proper
number of circuit values, these models can be quite accurate,
but the values do vary from device to device, even when the
Figure 21. Intrinsic model for a common-source FET with external
load and termination impedances and including gate and drain resistive parasitics, where ZT is the gate termination impedance, Rg is
the gate (metallization) resistance, Cgs is the gate-to-source capacitance, Cgd is the gate-to-drain capacitance, gm is the channel transconductance, Rds is the channel (drain-to-source) resistance, Cds is the
channel capacitance, Rd is the drain (metallization) resistance, and
ZL is the load impedance.
devices were fabricated at the same time and on the same
substrate. Usually, the data sheet with a device, instead of
specifying the circuit parameters, will specify the parameters
of the device S, which are defined as in Fig. 22 and which can
be measured in a straightforward manner by a network analyzer. The S parameters are defined by the equation
V1
S11 S12
V1+
=
(5)
V2
S21 S22
V2+
An important parameter of the circuit design is the transfer function of the transistor circuit, which can be defined as
the ratio of vo to vi as defined in Fig. 21. To simplify further
analysis, we will ignore the package parasitics Rg and Rd in
comparison with other circuit parameters, and thereby we
will carry out further analysis on the circuit depicted in Fig.
23. The circuit can be solved by writing a simultaneous system of equations for the two nodal voltages vi and vo. These
sinusoidal steady-state equations become
vi = v
jCgd (vo vi ) + gm vi + jCds vo +
vo
vo
+
=0
Rds
ZL
The system can be rewritten in the form
1
1
vo j(Cgd + Cds ) +
+
= vi (gm + jCgd )
Rds
ZL
which gives us our transfer function T in the form
+Vds
T=
Vgs
Zt
Intrinsic ac model
Vgs = 3 V
vi
Cgd
+
Vgs = 1 V
D
Vgs
Rg
Vgs = 0
Idss
ZL
vo
=
vi
gm + jCgd
1
1
j (Cgd + Cgs ) +
+
Rds
ZL
vo
Figure 20. Typical FET circuit including the bias voltages vgs and
vds as well as the ac voltages vi and vo, where the conductors represent
ac blocks and the capacitors dc blocks.
V 1+ej z
V1 e j z
Two-port network,
biased FET
S
V 2+ e j z
V 2 e j z
Figure 22. Schematic depiction of an FET as a two-port device that
defines the quantities used in the S matrix of Eq. (5).
ACTIVE ANTENNAS
ii
+
v ZT
vi
Cgs
io
vo
Cgd
RT
+
v
Cds
g mv
Rds
ZL
Zi
ZT
Ri
jXT
Figure 23. Simplified transistor circuit used for analyzing rather
general amplifier and oscillator circuits, where the circuit parameter
definitions are as in Fig. 22.
Oftentimes we are interested in open-circuit parametersfor
example, the circuit transfer function when ZL is large compared to other parameters. We often call this parameter G
the open-circuit gain. We can write this open-circuit gain in
the form
G=
201
gm Rds + jCgd Rds
vo
=
vi oc
j (Cgd + Cgs ) Rds + 1
It is useful to look at approximate forms. It is generally
true that
Cgd Cds , Cgs
and for usual operating frequencies it is also generally true
that
jXi
Figure 24. Diagram depicting the transistor and its load as a oneport device that, when matched to its termination so that there is
no real or imaginary part to the total circuit impedance, will allow
for oscillations.
Let us now consider an oscillator circuit. The basic idea is
illustrated in the one-port diagram of Fig. 24. The transistors
gain, together with feedback to the input loop through the
capacitor Cgd, can give the transistor an effective negative input impedance, which can lead to oscillation if the real and
imaginary parts of the total impedance (that is, ZT in parallel
with the Zi of the transistor plus load) cancel. The idea is
much like that illustrated in Fig. 25 for a feedback network.
One sees that the output of the feedback network can be expressed as
vo = G( j)[vi H( j)vo ]
or, on rearranging terms,
1
Rds
Cds
vo
G( j)
=
vi
1 + G( j)H( j)
Using both of the above in our equations for T and G, we find
which clearly will exhibit oscillationthat is, have an output
voltage without an applied input voltagewhen
gm Rds
R
1+
ZL
G = gm Rds
T=
H( j ) =
Clearly, from the above, one sees that the loaded gain will be
lower than the unloaded gain, as we would expect. Making
only the first of our two above approximations, we can write
the above equations as
T=
gm Rds
1 + jds +
G=
Rds
ZL
gm Rds
1 + jds
where ds is a time constant given by
ds =
What we need to do to see if we can achieve oscillation is to
investigate the input impedance of our transistor and load
seen as a one-port network. Clearly we can write the input
current of Fig. 23 as
ii = jCgs vi + jCgd (vi vo )
and then, using the full expression for T to express vo as a
function of vi, one finds
Zi =
ii
= jCgs + jCgd
1 +
vi
1
Cds Rds
We see that, in this limit, the high-frequency gain is damped.
Also, an interesting observation is that, at some frequency ,
an inductive load could be used to cancel the damping and
obtain a purely real transfer function at that frequency. This
effect is the one that allows us to use the transistor in an oscillator.
1
G( j)
gm jCgd
1
1
j (Cgd + Cds ) +
+
Rds
ZL
+
vi
G( j)
vo
H( j )
Figure 25. Depiction of a simple feedback network.
202
ACTIVE ANTENNAS
which can be somewhat simplified to yield
gm Rds + 1 + jds +
Zi = jCgs + jCgd
Rds
ZL
R
1 + jds + d
ZL
The condition for oscillation in such a system can be expressed in either of the forms
i T = 1
or
o L = 1
We can again invoke a limit in which ds 1 and then write
Zi = jCgs + jCgd
ZL (1 + gm Rds + Rds )
Rds + ZL
where the s are defined in the caption of Fig. 26. If both ZT
and ZL were passive loadsthat is, loads consisting of resistance, inductance, and capacitance, then we would have that
Perhaps the most interesting thing about this expression is
that if
|T | < 1
|L | < 1
ZL = j L
and the conditions for unconditional stability (nonoscillation
at any frequency) would be that
and
gm Rds 1
|i | < 1
|o | < 1
then clearly
Clearly, we can express i and o as series of reflections such
that
Ri < 0
Whether or not Xi can be made to match any termination is
another question, which we will take up in the next paragraph.
As was mentioned earlier, generally the data sheet one obtains with an FET has plots of the frequency dependence of
the S parameters rather than values for the equivalent circuit
parameters. Oscillator analysis is therefore usually carried
out using a model of the circuit such as that depicted in Fig.
26, where the transistor is represented by its measured S matrix. The S matrix is defined as the matrix of reflection and
transmission coefficients. That is to say, with referrence to
the figure, S11 would be the complex ratio of the field reflected
from the device divided by the field incident on the device.
S21 would be the field transmitted from the device divided by
the field incident on the device. S12 would be the field incident
from the load side of the device divided by the power incident
on the device, and S22 would be the power reflected from the
load side of the device divided by the power incident on the
device. For example, if there is only an input from ZT, then
i = S11 + S12 L S21 + S12 L S22 L S21
+ S12 L S22 L S22 L S21 +
o = S22 + S21 T S12 + S21 T S11 T S12
+ S21 T S11 T S11 T S12 +
Using the fact that
xn =
n=0
1
1x
we can reexpress the s as
S12 S21 L
1 S22 L
S S
o = S22 + 12 21 T
1 S22 T
i = S11 +
If we denote the determinant of the S matrix by
i = S11
= S11 S22 S12 S21
If there is only an input from ZL, then
and define a transistor parameter by
o = S22
=
S
ZT
T i
ZL
o L
Figure 26. Schematic depiction of an oscillator circuit in which the
transistor is represented by its S matrix and calculation is done in
terms of reflection coefficients T looking into the gate termination,
i looking into the gate source port of the transistor, o looking into
its drain source port, and L looking into the load impedance.
1 |S11 |2 |S22 |2 + ||2
2|S12 S21 |
then some tedious algebra leads to the result that stability
requires
>1
<1
At frequencies where the above are not satisfied, oscillation
can occur if the load and termination impedances, ZL and ZT
respectively, are chosen properly. Oscillator design is discussed in various texts (1114). Generally, though, oscillator
ACTIVE ANTENNAS
design involves finding instability points and not predicting
the dynamics once oscillation is achieved. Here we are discussing only oscillators which are self-damping. External circuits can be used to damp the behavior of an oscillator, but
here we are discussing only those that damp themselves independent of an external circuit. The next paragraph will discuss these dynamics.
If a transistor circuit is designed to be unstable, then as
soon as the dc bias is raised to a level where the circuit
achieves the set of unstable values, the circuits output within
the range of unstable frequencies rises rapidly and dramatically. The values that we took in the equivalent ac circuit,
though, were small-signal parameters. As the circuit output
increases, the signal will eventually no longer be small. The
major thing that changes in this limit is that the input resistance to the transistor saturates, so that (14)
Ri = Ri + mv2
i
1
+ (Ri + RT )i +
i dt = 0
t
C
Recalling the equivalent circuit of Fig. 23 and recalling that
Cgs Cgd
we see that, approximately at any rate, we should have a relation between vi and ii of the form
ii = Cgs
vi
t
Using this iv relation in the above, we find that
2 v Ri RT
t 2
L
mv2
Ri RT
a
t
b
Dielectric
constant
x
Figure 28. A patch antenna and Cartesian coordinate system.
which we can rewrite in terms of other parameters as
where the plus sign on the nonlinearity is necessary, for if it
were negative the transistor would burn up or else burn up
the power supply. Generally, m has to be determined empirically, as nonlinear circuit models have parameters that vary
significantly from device to device. For definiteness, let us assume that the ZT is resistive and the ZL is purely inductive.
At the oscillation frequency, the internal capacitance of the
transistor then should cancel the load inductance, but to consider dynamics we need to put in both C and L, as dynamics
take place in the time domain. The dynamic circuit to consider is then as depicted in Fig. 27. The loop equation for this
circuit in the time domain is
L
v
v
+
=0
t
LC
v
2v
+ 02 v = 0
(1 2 v2 )
t 2
t
which is the form of Van der Pols equation (15,16), which
describes the behavior of essentially any oscillator.
Now that we have discussed planar circuits and dynamical
elements that we can put into theory, the time has arrived to
discuss planar antenna structures. Perhaps the best way to
gain understanding of the operation of a patch antenna is by
considering a cavity resonator model of one. A good review of
microstrip antennas is given in Carver and Mink (17) and is
reprinted in Pozar and Schaubert (18). Let us consider a
patch antenna and coordinate system as is illustrated in Fig.
28. The basic idea behind the cavity model is to consider the
region between the patch and ground plane as a resonator. To
do this, we need to apply some moderately crude approximate
boundary conditions. We will assume that there is only a zdirected electric field underneath the patch and that this field
achieves maxima on the edges (open-circuit boundary condition). The magnetic field H will be assumed to have both x
and y components, and its tangential components on the
edges will be zero. (This boundary condition is the one consistent with the open-circuit condition on the electric field and
becomes exact as the thickness of the layer approaches zero,
as there can be no component of current normal to the edge
at the edge, and it is the normal component of the current
that generates the transverse H field.) The electric field satisfying the open-circuit condition can be seen to be given by the
modes
mn
cos kn x cos km y
e mn = e z
abt
ii
where
+
RT
Vi
203
Ri
kn = n/a
L
Figure 27. Circuit used to determine the dynamical behavior of a
transistor oscillator.
km = m/b
m=0
1,
mn =
2, m = 0
2,
m = 0
and n = 0
or n = 0
and n = 0
204
ACTIVE ANTENNAS
b
vi
G1 + jB1
G2 + jB2
vi
G1 + jB1
(a)
G2 jB2
Feedback
patch
(b)
Figure 29. (a) A transmission line model for a patch antenna, and
(b) its circuital equivalent as resonance.
The H field corresponding to the E field then will consist of
modes
h mn =
1 mn
(ex km cos kn x sin km y ey kn sin kn x cos km y)
j abt
As can be gathered from Fig. 13, the primary radiation mode
is the mode with m 1 and n 0.
The basic operation is described by the fact that the boundary conditions are not quite exact. Recall from the earlier argument that accompanied Fig. 13 that the z-directed field
gives rise to a fringe field at the edges y 0 and y b such
that there are strips of y-directed electric field around y 0
and y b. Because the boundary conditions are not quite
correct on H, there will also be strips of x-directed magnetic
fields in these regions. As the Poynting vector is given by
E H, we note that these strips will give rise to a z-directed
Poynting vector. Similar arguments can be applied to the
edges at x 0 and x a. However, the x-directed field at
x 0 has a change of sign at the center of the edge and is
pointwise oppositely directed to the x-directed electric field at
x 0. These fields, therefore, only give rise to very weak radiation, as there is significant cancellation. Analysis of the slot
antenna requires only that we interchange the E and H fields.
The picture of the patch antenna as two radiating strips
allows us to represent it with a transmission line as well as a
circuit model. The original idea is due to Munson (19). The
transmission line model is depicted in Fig. 29. The idea is
that one feeds onto an edge with an admittance (inverse impedance) G1 jB1 and then propagates to a second edge with
admittance G2 jB2. When the circuit is resonant, then the
length of transmission line will simply complex-conjugate the
given load [see Eq. (4)], leading to the circuit representation
of Fig. 29(b). The slot admittance used by Munson (19) was
just that derived for radiation from a slit in a waveguide (20)
as
G1 + jB1 =
a
(1 j 0.636 ln k0 t)
0 Z0
where Z0 is the impedance of free space (0 / 0 377 ), 0
is the free-space wavelength, and k0 is the free-space propagation vector, and where a and t are defined as in Fig. 28. When
the edges are identical (as for a rectangular patch), one can
write
G2 + jB2 = G1 + jB1
FET
Gate
Drain
Figure 30. A design of a microstrip active radiating element.
We have now considered all of the pieces, and therefore it
is time to consider a couple of actual active antenna designs.
Figure 30 depicts one of the early designs from Kai Changs
group at Texas A&M (21). Essentially, the patch here is being
used precisely as the feedback element of an amplifier circuit
(as was described in connection with Fig. 9). A more compact
design is that of Fig. 14 (7). There, the transistor is actually
mounted directly into the patch antenna. The slit between the
gate and drain yields a capacitive feedback element such that
the effective ac circuit equivalent of this antenna may appear
as depicted in Fig. 31. The capacitorinductor pair attached
to the gate lead forms what is often referred to as a tank
circuit, which (if the load were purely real) defines a natural
frequency through the relation
=
1
LC
As was discussed at some length in the last section of this
article, a major argument for the use of active antennas is
that they are sufficiently compact that they can be arrayed
together. Arraying is an important method for free-space
power combining, which is necessary because as the frequency increases, the power-handling capability of active devices decreases. However, element size also decreases with
increasing frequency so that use of multiple coherently combined elements can allow one to fix the total array size and
power more or less independently of frequency, even though
the number of active elements to combine increases. In the
next paragraph, we shall consider some of the basics of
arrays.
Cslit
D
G
ZL
LT
to obtain the input impedance in the form
Zi =
1
1
=
Yi
2G1
Figure 31. Ac circuit equivalent of the active antenna of Fig. 14.
ACTIVE ANTENNAS
which in turn can be written as
kd
sin
sin2 N
2
AF =
2 kd
sin
sin
2
x=0
x=d
x = 2d x = 3d
x = 4d
x = 5d
N 1
x = Nd
Figure 32. Depiction of a linear array of N identical radiating elements.
Consider a linear array such as is depicted in Fig. 32. Now
let us say that the elements are nominally identical apart
from phases that are set by the array operator at each of the
elements. The complex electric field far from the nth element
due to only the nth element is then given by
E n = E e e i n
where Ee is the electric field of a single element. To find out
what is radiated in the direction due to the whole array, we
need to sum the fields from all of the radiators, giving each
radiator the proper phase delay. Each element will get a progressive phase shift kd sin due to its position (see Fig. 32),
where k is the free-space propagation factor, given by
k=
where is the free-space wavelength. With this, we can write
for the total field radiated into the direction due to all n
elements
E t ( ) = E e
N1
e inkd sin e i n
n=0
The sum is generally referred to as the array factor. The intensity, then, in the direction is
where Ri is the input resistance of the transistor circuit as
seen looking into the gate source port and RT is the external
termination resistor placed between the gate and common
source. In the absence of the locking term, one can see that
oscillation will take place with a primary frequency (and some
harmonics) at angular frequency 0 with amplitude
Ri0 RT /m such that
v(t)
Ri0 RT
cos 0t
m
Without being too quantitative, one can say that, if i is close
enough to 0 and A is large enough, the oscillation will lock
AF
AF
kd = /2
kd = 2
kd =
One notes immediately that, if one sets the phases n to
then the intensity in the direction is N2 times the intensity
due to a single element. This is the effect of coherent addition.
One gets a power increase of N plus a directivity increase of
N. To illustrate, let us consider the broadside case where we
take all the n to be zero. In this case, we can write the array
factor in the form
2
N1
1 e iNkd sin 2
ind sin
e
=
n=0
1 e ikd sin
(6)
which is plotted in Fig. 33. Several interesting things can be
noted from the expression and plots. For kd less than , there
is only one central lobe in the pattern. Also, the pattern becomes ever more directed with increasing N. This is called the
directivity effect. If the array has a power-combining efficiency
of 100% (which we have built into our equations by ignoring
actual couplings, etc.), then the total power radiated can only
be N times that of a single element. However, it is radiated
into a lobe that is only 1/N times as wide as that of a single element.
If we are to realize array gain, however, we need to be
certain that the array elements are identical in frequency and
have fixed phase relations in time. This can only take place if
the elements are locked together. The idea of locking is probably best understood in relation to the Van der Pol equation
(16), with an injected term, such that
2 v Ri RT
v
mv2
1
+ 02 v = A cos i t
t 2
L
Ri RT t
2
N1
iknd sin i n
It ( ) = Ie
e
e
n=0
n = nkd sin
205
(a)
(b)
AF
(c)
Figure 33. Plots of the array factor of Eq. (6), where (a) N 1, (b)
N 5 and kd /2, , and 2, and (c) N 10 and kd .
206
ACTIVE ANTENNAS
to i in frequency and phase. If i is not quite close enough
and A not quite big enough (how big A needs to be is a function of how close i is), then the oscillation frequency 0 will
be shifted so that
v(t) = A0 cos[(0 + )t + ]
where and are functions of i and A. These ideas are
discussed in a number of places including Refs. 1, 15, 16, 22,
23, and 24. In order for our array to operate in a coherent
mode, the elements must be truly locked. This locking can
occur through mutual coupling or through the injection of an
external signal to each of the elements.
Ideally, we would like to be able to steer the locked beam.
A number of techniques for doing this are presently under
investigation. Much of the thinking stems from the work Stephan (2528) and Vaughan and Compton (28a). One of the
ideas brought out in these works was that, if the array were
mutually locked and one were to try to inject one of the elements with a given phase, all of the elements would lock to
that phase. However, if one were to inject two elements at
the locked frequency but with different phases, then the other
elements would have to adjust themselves to these phases. In
particular, if one had a locked linear array and one were to
inject the two end elements with phases differing by , then
the other elements would share the phase shift equally so
that there would be a linear phase taper of magnitude uniformly distributed along the array.
A different technique was developed by York (29,30), based
on work he began when working with Compton (31,32). In
this technique, instead of injecting the end elements with the
locked frequency and different phase, one injects with wrong
frequencies. If the amplitudes of these injected frequencies
are set to values that are not strong enough to lock the elements to this wrong frequency, then the elements will retain
their locked frequencies but will undergo phase shifts from
the injected signal. If the elements of the array are locked due
to mutual feedback, trying to inject either end of the array
with wrong frequencies will then tend to give the elements a
linear taperthat is, one in which the phase varies linearly
with distance down the arraywith much the same result as
in the technique of Stephan. This will just linearly steer the
main lobe of the array off broadside and to a new direction.
Such linear scanning is what is needed for many commercial
applications such as tracking or transmitting with minimum
power to a given location.
Another technique, which again uses locking-type ideas, is
that of changing the biases on each of the arrays active devices (3335). Changing the bias of a transistor will alter the
0 at which the active antenna wants to oscillate. For an element locked to another frequency, then, changing the bias
will just change the phase. In this way one can individually
set the phase on each element. There are still a couple of
problems with this approach (as with all the others so far,
which is why this area is still one of active research). One is
that addressing each bias line represents a great increase in
the complexity that we were trying to minimize by using an
active antenna. The other is that the maximum phase shift
obtainable with this technique is from one end of the
array to the other (a limitation that is shared by the phaseshifts-at-the-ends technique). In many phased-array applications, of which electronic warfare is a typical one, one wants
to have true time delay, which means that one would like to
have as much as a phase shift between adjacent elements.
I do not think that the frequency-shifting technique can
achieve this either. Work, however, continues in this exciting area.
APPLICATIONS OF AND PROSPECTS FOR ACTIVE ANTENNAS
Perhaps the earliest application of the active antenna concept
(following that of Hertz) was aimed at solving the smallantenna problem. As we recall, an antenna can be modeled
(roughly) by a series RLC network with the R representing
the radiation resistance. The input impedance of such a combination is given by
Zi =
1 2 /02 + jRC
jC
and so we see that, when the operation frequency is well
below the resonant frequency
0 =
1
LC
and the reciprocal of the RC time constant
= RC
then the antenna appears as a capacitor and radiates quite
inefficiently. The problem of reception is similar. Apparently,
already in 1928 Westinghouse had a mobile antenna receiver
that used a pentode as an inductive loading element in order
to boost the amount of low-frequency radiation that could be
converted to circuit current. In 1974, two works discussed
transistor-based solutions to the short-aerial problem (36,37).
In Ref. 37, the load circuit appeared as in Fig. 34. The idea
was to generate an inductive load whose impedance varied
with frequency, unlike a regular inductor, but so as to increase the antenna bandwidth. The circuits operation is not
intuitively obvious. I think that it is possible that most AM,
short-wave, and FM receivers employ some short-antenna solution whether or not the actual circuit designers were aware
that they were employing active antenna techniques.
Another set of applications where active devices are essentially used as loading elements is in the greater-than-100-
Antenna
R
Q1
Amplifier
Figure 34. A circuit taken from Ref. 37 in which a transistor circuit
is used to load a short antenna. Analysis shows that, in the frequency
regime of interest, the loading circuit appears, when looking toward
the antenna from the amplifier terminals, to cancel the strongly capacitive load of the short antenna.
ACTIVE ANTENNAS
GHz regime. Reviews of progress in this regime are given in
Refs. 1 and 38. To date, most work at frequencies greater
than 100 GHz has involved radio-astronomical receivers. A
problem at such frequencies is a lack of components, including circuit elements so basic as waveguides. Microstrip guides
already start having extra-mode problems at Ku band. Coplanar waveguides can go higher, although to date, rectangular
metallic waveguides are the preferred guiding structures past
about 60 GHz. In W band (normally narrowband, about 94
GHzsee Table 1), there are components, as around 94 GHz
there is an atmospheric window of low propagation loss. However, waveguide tolerances, which must be a small percentage
of the wavelength, are already severe in W band, where the
wavelength is roughly 3 mm. Higher frequencies have to be
handled in free space or, as one says, quasi-optically. Receivers must therefore by nature be downconverting in this 100
GHz regime. Indeed, these types of solutions are the ones being demonstrated by the group at Michigan (38), where receivers will contain multipliers and downconverting mixers
right in the antenna elements in order that CPW can be used
to carry the downconverted signals to the processing electronics. Millimeter-waveterahertz radio astronomy seems to be
a prime niche for quasioptical active antenna solutions.
The first applications of active antennas where solid-state
components were used as gain elements were primarily for
power boosting (3944). Power combining (see reviews in
Refs. 45 and 46) can be hard to achieve. There is a theorem
that grew out of the early days of radiometry and radiative
transfer (in the 1800s), known variously as the brightness
theorem, the Lagrange invariant, or (later) the second law
of thermodynamics. (See, for example, Ref. 8, Chap. 5.) The
theorem essentially states that one cannot increase the
brightness of a source by passive means. This theorem practically means that, if one tries to combine two nominally identical sources by taking their outputs, launching them into
waveguides, and then bringing the two waveguides together
in a Y junction into a single waveguide, the power in the output guide, if the output guide is no larger than either of the
input guides, can be no greater than that of either of the nominally identical sources. This seems to preclude any form of
power combining. There is a bit of a trick here, though. At the
time the brightness theorem was first formulated, there were
no coherent radiation sources. If one takes the output of a
coherent radiation source, splits it in two, and adds it back
together in phase, then the brightness, which was halved, can
be restored. If two sources are locked, they are essentially one
source. (As P. A. M. Dirac said, a photon only interferes with
itself. Indeed, the quantum mechanical meaning of locking is
that the locked sources are sharing a wave function.) Therefore, locked sources can be coherently added if they are properly phased. We will take this up again in a following paragraph.
An alternative to power combining that obviates the need
for locking and precise phase control is amplification of the
signal from a single source at each element. By 1960, solidstate technology had come far enough that antennas integrated with diodes and transistors could be demonstrated.
The technology was to remain a laboratory curiosity until the
1980s, when further improvements in microwave devices
were to render it more practical. Recent research, however,
has been more concentrated on the coherent power combining
of self-oscillator elements. This is not to say that the element-
207
mounted amplifier may not still be of practical use. The main
research issue at present, though, is the limited power available from a single active element at millimeter-wave frequencies.
Another application area is that of proximity detection
(47). The idea is that an oscillator in an antenna element can
be very sensitive to its nearby (several wavelengths) environment. As was discussed previously, variation in distances to
ground planes changes impedances. The proximity of any
metal object will, to some extent, cause the oscillator to be
aware of another ground plane in parallel with the one in the
circuit. This will change the impedance that the oscillator
sees and thereby steer the oscillator frequency. The active antenna of Ref. 47 operated as a self-oscillating mixer. That is,
the active element used the antenna as a load, whereas the
antenna also used a diode mixer between itself and a lowfrequency external circuit. The antenna acted as both a transmitting and a receiving antenna. If there were something
moving near the antenna, the signal reflected off the object
and rereceived might well be at a different frequency than
the shifting oscillator frequency. These two frequencies would
then beat in the mixer, be downconverted, and show up as a
low-frequency beat note in the external circuit. If such a composite device were to be used in a controlled environment,
one could calibrate the output to determine what is occurring.
Navarro and Chang (1, p. 130) mention such applications as
automatic door openers and burglar alarms. The original paper (47) seemed to have a different application in mind, as
the term Doppler sensor was in the title. If one were to carefully control the immediate environment of the self-oscillating
mixer, then reflections off more distant objects that were received by the antenna would beat with the stable frequency
of the oscillator. The resulting beat note of the signals would
then be the Doppler shift of the outgoing signal upon reflection off the surface of the moving object, and from it one could
determine the normal component of the objects velocity. It is
my understanding that some low-cost radars operate on such
a principle. As with other applications, though, the active antenna principle, if only due to size constraints, becomes even
more appealing at millimeter-wave frequencies, and at such
frequencies power constraints favor use of arrays.
An older antenna field that seems to be going through an
active renaissance is that of retroreflection. A retroreflector is
a device that, when illuminated from any arbitrary direction,
will return a signal directly back to the source. Clearly, retroreflectors are useful for return calibration as well as for various tracking purposes. An archetypical passive retroreflector
is a corner cube. Another form of passive reflector is a Van
Atta array (48). Such an array uses wires to interconnect the
array elements so that the phase progression of the incident
signal is conjugated and thereby returned in the direction of
the source. As was pointed out by Friis already in the 1930s,
though, phase conjugation is carried out in any mixer in
which the local oscillator frequency exceeds the signal frequency (49). (A phase conjugate signal is one that takes on
negative values at each phase point on the incoming wave.)
This principle was already being exploited in 1963 for implementing retroreflection (50). This work did not catch on, perhaps for technical reasons. A review in 1994 (51) and designs
for such arrays were demonstrated and presented at the 1995
International Microwave Symposium (52,53). Although both
demonstrations used transistors and patch-type elements,
208
ACTIVE ANTENNAS
Active device
(a)
G
Source bias
Drain bias
Source bias
Gate bias
Source bias
(b)
Figure 35. Schematic depiction of (a) the active surface of a grid
oscillator and (b) a breakout of an internal region of the grid showing
the active device placement relative to the bias lines.
both also employed circulators for isolation and therefore
were not actually active array demonstrations. It would seem
that retroreflection should motivate an active self-oscillating
mixer solution, which will perhaps appear in the future.
As was mentioned earlier in this article, a quite important
application area for active antennas is free-space power combining. As was pointed out then, a number of groups are
working on developing compact elements such as those of Fig.
14 (7) and Fig. 30 (21). As was also previously mentioned, in
order to do coherent power combining, the elements must be
locked. In designs where the elements are spatially packed
tightly enough, proximity can lead to strong enough nearestneighbor coupling so that the array will lock to a common
frequency and phase. Closeness of elements is also desirable
in that arrays with less than /2 spacing will have no sidelobes sapping power from the central array beam. In designs
that do not self-lock, one can inject a locking signal either on
bias lines or spatially from a horn to try to lock to all elements
simultaneously. Of course, the ultimate application would be
for a high-bandwidth, steerable, low-cost transceiver.
Another method of carrying out power combining is to use
the so-called grid oscillator (54,55). The actual structure of a
grid appears in Fig. 35. The operating principle of the grid is
quite a bit different from that of the arrays of weakly coupled
individual elements. Note that there is no ground plane at all
on the back, and there is no ground plane either, per se, on
the front side. Direct optical measurements of the potentials
on the various lines of the grid (56), however, show that the
source bias lines act somewhat like ac grounds. In this sense,
either a drain bias line together with the two closest source
biases, or a gate bias line together with the two horizontally
adjacent bias lines, appears somewhat like CPW. The CPW
lines, however, are periodically loaded ones with periodic active elements alternated with structures that appear like slot
antennas. The radiating edges of the slots are, for the drain
bias lines, the vertical ac connection lines between drain and
drain or, for the gate bias CPW, the horizontal ac gate-to-gate
connection lines. Indeed, the grid is known to lock strongly
between the rows and more weakly between columns. As adjacent row elements are sharing a patch radiator, this behavior should be expected.
In a sense, this strong locking behavior of the grid is both
an advantage and a disadvantage. It is advantageous that the
grid is compact (element spacing can be /6) and further
that it is easy to get the rows to lock to each other. However,
the compactness is also a disadvantage in that it is quite hard
to get any more functionality on the grid. Much effort has
been made in this area to generate functionality by stacking
various grid-based active surfaces such as amplifying surfaces, varactor surfaces for frequency shifting and modulation, doubling surfaces, etc. A problem with stacking is, of
course, diffraction as well as alignment. Alignment tolerance
adds to complexity. Diffraction tends to ease alignment tolerance, but in an inelegant manner. A 100-transistor array with
/6 spacing will have an extent of roughly 1.5 per side. As
the diffraction angle is something like the wavelength divided
by the array diameter, the diffraction angle for such an array
is a good fraction of a radian. One can say that grids are
quasi-optical, but in optics one generally doesnt use apertures much smaller than a millimeter (center optical wavelength of micrometers), for which the diffraction angle would
be roughly a thousandth of a radian. As far as pure combining
efficiency goes, grids are probably the optimal solution. However, more functionality may well be hard to obtain with
this solution.
As we have mentioned, there are a number of techniques
for steering being investigated. There seems to be less work
on modulation, and I do not know of any simultaneous steering of modulated beams to date. Although the field of active
antennas began with the field of radio frequency, it still
seems to be in its infancy. However, as I hope this article has
brought across, there is a significant amount of work ongoing,
and the field of active antennas will grow in the future.
BIBLIOGRAPHY
1. J. A. Navarro and K. Chang, Integrated Active Antennas and Spatial Power Combining, New York: Wiley, 1995.
2. R. A. York and Z. B. Popovic (eds.), Active and Quasi-Optical
Arrays for Solid-State Power Combining, New York: Wiley, 1997.
2a. H. Hertz, Electric Waves, New York: Macmillan, 1983. (This is
a book of reprints of Hertz work in the 1890s.)
3. J. Bardeen and W. Bratain, The transistor: A semiconductor triode, Phys. Rev., 74: 435, 1948.
4. W. Shockley, The theory of pn junctions in semiconductors and
pn junction transistors, Bell Syst. Tech. J., 28: 435, 1949.
5. W. Shockley, A unipolar field-effect transistor, Proc. IEEE, 40:
13651376, 1952.
6. C. A. Mead, Schottky-barrier gate field-effect transistor, Proc.
IEEE, 54: 307308, 1966.
7. R. A. York, R. D. Martinez, and R. C. Compton, Active patch antenna element for array applications, Electron. Lett., 26: 494495,
March 1990.
8. A. R. Mickelson, Physical Optics, New York: Van Nostrand Reinholt, 1992, chap. 2.
9. B. J. Hunt, The Maxwellians, Ithaca, NY: Cornell University
Press, 1991, chap. 3.
ACTIVE FILTERS
10. H. C. Pocklington, Electrical oscillations in wires, Proc. Cambridge Phil. Soc., 324333, 1897.
11. D. M. Pozar, Microwave Engineering, Reading, MA: Addison-Wesley, 1990.
12. P. E. Gray and C. L. Searle, Electronic Principles, Physics, Models
and Circuits, New York: Wiley, 1967.
13. R. E. Collin, Foundations for Microwave Engineering, 2nd ed.,
New York: McGraw-Hill, 1992.
14. K. Chang, Microwave Solid-State Circuits and Applications, New
York: Wiley, 1994.
15. K. Y. Chen et al., Analysis of an experimental technique for determining Van der Pol parameters of a transistor oscillator, IEEE
Trans. Microw. Theory Tech., 46: 914922, 1998.
209
tive antennas, IEEE MTT-S Int. Microw. Symp. Digest, 1994,
pp. 12271230.
35. A. Zarrang, P. S. Hall, and M. Cryan, Active antenna phase control using subharmonic locking, Electron. Lett., 31: 842843,
1995.
36. T. S. M. Maclean and P. A. Ransdale, Short active aerials for
transmission, Int. J. Electron., 36: 261269, 1974.
37. P. K. Rangole and S. S. Midha, Short antenna with active inductace, Electron. Lett., 10: 462463, 1974.
38. G. M. Rebeiz, Millimeter-wave and terahertz integrated circuit
antennas, Proc. IEEE, 80: 17481770, 1996.
39. A. D. Frost, Parametric amplifier antennas, Proc. IRE, 48: 1163
1164, 1960.
16. B. Van der Pol, Forced oscillations in a circuit with a nonlinear
resistance, Phil. Mag., 3: 6580, 1927.
40. J. R. Copeland and W. J. Robertson, Antenna-verters and antennafiers, Electronics, pp. 6871, 1961.
17. K. R. Carver and J. W. Mink, Microstrip antenna technology,
IEEE Trans. Antennas Propag., AP-29: 224, 1981.
41. M. E. Pedinoff, The negative conductance slot amplifier, IRE
Trans. Microw. Theory Tech., 9: 557566, 1961.
18. D. M. Pozar and D. H. Schauber (eds.), Microstrip Antennas, Piscataway, NJ: IEEE Press, 1995.
42. W. J. Robertson, J. R. Copeland, and R. G. Verstraete, Antennafier arrays, IEEE Trans. Antennas Propag., 2: 227233, 1964.
19. R. E. Munson, Conformal microstrip antennas and microstrip
phased arrays, IEEE Trans. Antennas Propag., AP-22: 7478,
1974.
43. K. Fujimoto, Active antennas: Tunnel-diode-loaded dipole, Proc.
IEEE, 53: 174, 1964.
20. R. F. Harrington, Time Harmonic Electromagnetic Fields, New
York: McGraw-Hill, 1961, p. 276.
21. K. Chang, K. A. Hammer, and G. K. Gopalakrishnan, Active radiating element using FET source integrated with microstrip patch
antenna, Electron. Lett., 24: 13471348, 1988.
22. R. Adler, A study of locking phenomena in oscillators, Proc. IRE,
34: 351357, 1946.
23. R. Adler, A study of locking phenomena in oscillators, Proc. IEEE,
61: 13801385, 1973. This is a reprint of (22).
24. K. Kurokawa, Injection locking of microwave solid-state oscillators, Proc. IEEE, 61: 13861410, 1973.
25. K. D. Stephan, Inter-injection-locked oscillators for power combining and phased arrays, IEEE Trans. Microw. Theory Tech., 34:
10171025, 1986.
26. K. D. Stephan and W. A. Morgan, Analysis of inter-injectionlocked oscillators for integrated phased arrays, IEEE Trans. Antennas Propag., 35: 771781, 1987.
27. K. D. Stephan and S. L. Young, Mode stability of radiation-coupled inter-injection-locked oscillators for integrated phased
arrays, IEEE Trans. Microw. Theory Tech., 36: 921924, 1988.
28. W. A. Morgan and K. D. Stephan, An x-band experimental model
of a millimeter-wave inter-injection-locked phase array system,
IEEE Trans. Antennas Propag., 36: 16411645, 1988.
28a. M. J. Vaughan and R. C. Compton, 28 GHz omnidirectional quasioptical transmitter array, IEEE Trans. Microw. Theory Tech.,
MTT-43: 25072509, 1995.
29. R. A. York, Nonlinear analysis of phase relationships in quasioptical oscillator arrays, IEEE Trans. Microw. Theory Tech., 41:
17991809, 1993.
30. P. Liao and R. A. York, A new phase-shifterless beam scanning
technique using arrays of coupled oscillators, IEEE Trans. Microw. Theory Tech., 41: 18101815, 1993.
31. R. A. York and R. C. Compton, Quasi-optical power combining
using mutual synchronized oscillator arrays, IEEE Trans. Microw. Theory Tech., 39: 10001009, 1991.
44. H. H. Meinke, Tunnel diodes integrated with microwave antenna
systems, Radio Electron. Eng., 31: 7680, 1966.
45. K. J. Russell, Microwave power combining techniques, IEEE
Trans. Microw. Theory Tech., 27: 472478, 1979.
46. K. Chang and C. Sun, Millimeter-wave power-combining techniques, IEEE Trans. Microw. Theory Tech., 31: 91107, 1983.
47. B. M. Armstrong et al., Use of microstrip impedance-measurement technique in the design of BARITT piplex Doppler sensor,
IEEE Trans. Microw. Theory Tech., 28: 14371442, 1980.
48. E. D. Sharp and M. A. Diab, Van Atta reflector array, IRE Trans.
Antennas Propag., 8: 436438, 1960.
49. H. T. Friis and C. Feldman, A multiple-unit steerable antenna
for short-wave reception, Bell Syst. Tech. J., 16: 337419, 1937.
50. C. Y. Pon, Retrodirective array using the heterodyne technique,
IEEE Trans. Antennas Propag., 12: 176180, 1964.
51. B. S. Hewitt, The evolution of radar technology into commercial
systems, IEEE MTT-S Int. Microw. Symp. Digest, 1994, pp. 1271
1274.
52. C. W. Poblanz and T. Itoh, A conformal retrodirective array for
radar applications using a heterodyne phase scattering element,
IEEE MTT-S Int. Microw. Symp. Digest, 1995, pp. 905908.
53. Y. Chang, D. C. Scott, and H. R. Fetterman, Microwave phase
conjugation using antenna coupled nonlinear optically pumped
surface, IEEE MTT-S Int. Microw. Symp. Digest, 1995, pp. 1303
1306.
54. Z. B. Popovic, M. Kim, and D. B. Rutledge, Grid oscillators, Int.
J. Infrared Millimeter Waves, 9: 647654, 1988.
55. Z. B. Popovic et al., A 100-MESFET planar grid oscillator, IEEE
Trans. Microw. Theory Tech., 39: 193200, 1991.
56. K. Y. Chen et al., Noninvasive experimental determination of
charge and current distributions on an active surface, IEEE
Trans. Microw. Theory Tech., 44: 10001009, 1996.
ALAN R. MICKELSON
University of Colorado
32. R. A. York and R. C. Compton, Coupled-oscillator arrays for millimeter-wave power-combining and mode-locking, IEEE MTT-S
Int. Microw. Symp. Digest, 1992, pp. 429432.
33. P. S. Hall and P. M. Haskins, Microstrip active patch array with
beam scanning, Electron. Lett., 28: 20562057, 1992.
34. P. S. Hall et al., Phase control in injection locked microstrip ac-
ACTIVE ANTENNAS. See ANTENNA ARRAYS.
ACTIVE ARRAYS. See QUASI-OPTICAL CIRCUITS.
ACTIVE FILTERING. See POWER SYSTEM HARMONICS.
ANTENNA ACCESSORIES
531
Attenuator
R2
R1
R3
Zi
Zo
Figure 1. Resistor configuration for the T attenuator. Zi and Zo are
the resistive impedances presented to the attenuator by external circuits.
ANTENNA ACCESSORIES
Accessories in antennas are devices for radiating or receiving
electromagnetic waves. Their main purpose is to create
means for the perfect operation of antennas. The accessories
interconnect these antennas to the transmission and reception systems efficiently and safely. An electromagnetic coupling generally exists between the antennas and the accessories, and one can explore them to improve antenna
performance. In this way, the accessories are as fundamental
as the antenna itself, whose behavior depends on appropriate connections.
Among several kinds of accessories, the following are most
commonly used: attenuators, baluns, chokes, coax lines, insulators, lightning arresters, planar structures, phase shifters,
twin-lead. These components are described in the next sections.
Design of an Attenuator
The designer has two concerns in designing an attenuator:
1. attenuation in dB
2. input and output impedances
The attenuation is given by Eq. (1):
ATT = 10 log10 N
(1)
where ATT is attenuation in decibels and N is the ratio between the power absorbed by the circuit (from the generator)
and the power delivered to the load.
The design starts from the desired input and output impedances and from the value of the ATT obtained in Eq. (1).
Equations (2), (3), and (4) give the resistances of the T circuit:
ATTENUATORS
Attenuators are circuits designed to introduce a known loss
between input and output ports (1). The power ratio, expressed in decibels, between the input and output represents
the loss in these circuits. The main use of attenuators is in
measuring standing wave ratio (SWR) in antennas and the
transmission coefficient. In SWR measurement (2), one adjusts the attenuation value to maintain equal outputs in the
stationary wave detector at the maximum and minimum
points. Then the SWR in dB is equal to the difference between
the readings of the attenuator.
The external circuits connected to the input and output
ports of the attenuator should present purely resistive impedances. These are always matched to the input and output impedances of the component. Therefore, the resistors always
constitute the attenuator circuit. In general, these resistive
circuits use the T and circuit topologies. Figure 1 shows
the T topology, where R1, R2, and R3 form the circuit. The
external circuitry presents purely resistive impedances (Zi,
Zo) at the input and output of the attenuator. Figure 2 shows
the configuration of the section. If the input and output
impedances are the same (Zi Zo), the circuits T and
become symmetrical, that is, R1 R2.
R3 = 2(NZi Zo )1/2 /(N 1)
(2)
R1 = Zi [(N + 1)/(N 1)] R3
(3)
R2 = Zo [(N + 1)/(N 1)] R3
(4)
Attenuator
R3
Zi
R1
R2
Zo
Figure 2. Resistor configuration for the attenuator. Zi and Zo are
resistive impedances presented to the attenuator by external circuits.
J. Webster (ed.), Wiley Encyclopedia of Electrical and Electronics Engineering. Copyright # 1999 John Wiley & Sons, Inc.
532
ANTENNA ACCESSORIES
very thin, tapered, resistive card. The attenuator device introduces this card in a section of the slotted guide. The adjustment of the penetration depth in the slot allows controlling
the dissipation of power and the desired attenuation. This
type of attenuator has a complex attenuation varying with
penetration depth and frequency.
Attenuator
R1/2
R2/2
R3
Zi
Zo
BALUNS
R1/2
R2/2
Figure 3. Balanced attenuator of the H type. The R1 and R2 resistors from the T circuit are distributed in the lower arms.
Equations (5), (6), and (7) give the resistances for a type
circuit:
R3 = 1/2(N 1)(Zi Zo /N)1/2
(5)
1/R1 = (1/Zi )[(N + 1)/(N 1)] (1/R3 )
(6)
1/R2 = (1/Zo )[(N + 1)/(N 1)] (1/R3 )
(7)
Equations (2), (3), (4) and Eqs. (5), (6), (7) are also valid for
symmetrical circuits (it is sufficient that Zi Zo).
Attenuators T and are applied to unbalanced systems, such as coaxial cables. Bifilar lines feed many antennas
that are intrinsically balanced, such as dipoles. In this case,
the circuits T and change into their balanced versions
and they are called H and O sections, respectively. Figure
3 shows the H circuit resistor configuration, and Fig. 4 illustrates the O circuit configuration. With respect to the circuits T and , there is a distribution of the resistors R3
and R1, R2 in the lower branch. All equations shown previously are valid for this case. However, it is necessary to take
the half of the values of the respective resistors and distribute
them in the upper and lower branches of their circuits.
Attenuators in Waveguides
Waveguides usually feed high-frequency antennas. In this
case, (p. 262, Ref. 2) the attenuators are built starting from a
Attenuator
R3/2
Zi
Ri
R2
Zo
R3/2
Figure 4. Balanced attenuator of the O type. The R3 resistor from
the circuit is distributed in the lower arm.
Although it is important, knowledge of the impedance value
between two terminals is not enough to connect this impedance correctly to a transmission line because of couplings between the terminals and ground.
The equivalent circuit of Fig. 5 represents these couplings
to ground. There are two special cases considered in detail
(3). The first is when Z2 Z3 in magnitude and phase. In this
case, the impedance AB is balanced. Therefore, the voltages
between A to ground and B to ground have the same magnitude and opposite phase.
The second case is when either Z2 or Z3 is zero. In this
case, one of the sides is at ground potential, and the impedance is unbalanced. An example of a balanced line is a parallel wire line (twin-lead). The unbalanced lines are generally
coaxial. Other types of lines exist, where the conductors present different couplings to the ground. An example of this is
the twin lead which has conductors of different thickness.
However, such a line is not common.
Because the conductors have different potentials with respect to the ground, the capacitance of the each conductor to
the ground also differs. Therefore, the current in the two conductors can be different.
Antennas are generally designed for balanced or unbalanced input impedances. This simplifies the connection of the
lines with the antenna. The connection to a symmetrical antenna requires a balanced line, whereas an antenna which
has one of its feeding points at ground potential requires an
unbalanced input.
A simple example of a balanced antenna is the dipole antenna shown in Fig. 6(a), and Fig. 6(b) illustrates a monopole,
which is an unbalanced antenna. Figure 6(c) shows a dipole
antenna with the feeding point between the center and the
end (this is a procedure for achieving different input impedance values). Such an antenna has a current imbalance,
which makes it impossible to feed it satisfactorily either by a
balanced or unbalanced line.
Most frequently it is necessary to feed a balanced antenna
with a coaxial cable and to feed (although less frequent) an
unbalanced antenna with a balanced line. Such connections
require special components to avoid problems with operation
Z1
A
;;;
Z2
Z3
Figure 5. Equivalent circuit of an impedance AB.
(a)
;;;;
(b)
of the system (4). These devices are balanced-to-unbalanced
converters, also known as baluns. The balun makes the voltages and/or currents in the two lines similar in magnitude.
If a coax cable is connected directly in a balanced antenna,
currents are induced in the external part of the external mesh
of the coax. This causes radiation of electromagnetic fields in
unwanted directions. These currents cause an imbalance in
the current distribution of the antenna. This affects the radiation diagram by altering the main lobe (sometimes drastically) and the gain of the antenna. For reception, interference
signals can be induced in the external part of the coaxial cable
and coupled inside the cable feeding the receiver.
The difficulties associated with the connection between an
unbalanced and a balanced system can be understood by considering the coax line below a ground plane. Figure 7 illustrates the connection between a system and a parallel wire
line. In this figure, ZL is the load impedance and ZS is the
impedance associated with the support structures. The resulting currents in this system are equivalent to those created
by ideal generator of Fig. 8.
The currents in lines A and B are not necessarily the same.
The total current leaving point b flows into the line. However,
the total current in b comes from line B and the connection
with the ground. The main purpose of baluns is to ensure that
the currents in lines A and B (Fig. 7) are similar.
Figure 9 illustrates a device capable of introducing symmetry in a line with respect to ground. Symmetry is essential in
all kinds of baluns or balanced systems. Figure 10 shows the
equivalent representation of this kind of balun. In this case
IA IB. Even so, if the length FC is small, the coax cable is
almost short-circuited, and very little power is delivered to
the load. To achieve satisfactory operation, the length FC
should be of the order of a quarter wavelength ( /4). Because
of this restriction, the type of balun shown in Fig. 9 is inherently narrowband.
A possible solution to increasing the bandwidth is to wind
the coaxial into a coil of length FC. This introduces a high
impedance into the system. However, there are design limitations. One of them is in the lower frequency range. In this
case, the impedance of the winding is small compared to the
ANTENNA ACCESSORIES
533
Figure 6. (a) Dipole antenna with symmetrical feeding. (b) Monopole antenna
fed in the base. (c) Dipole antenna with
asymmetrical feeding.
(c)
load (seen at the input of the twin-lead line). The upper limit
is at the resonance point of the space of the windings. A technique used for increasing the inductive effect is to wind the
cable in ferrite cores, which present a high impedance level
on a wider frequency band. Careful designs of structures of
this type allow their operation in frequency band ratios up to
10 : 1 (5).
The baluns most used are those based on quarter-wavelength short-circuited line sections. Such an arrangement is
shown in Fig. 11. In this way, high impedance is obtained
from point B to the external side of the outer mesh of the
coaxial line. This prevents current flow into the outer surface
of the external mesh of the cable.
In practice, the characteristic impedance of the transmission line is usually real, whereas that of the antenna element
is complex. It is frequently desirable to operate the balun at
a length other than a quarter-wavelength, in order to take
advantage of the shunt reactance presented by the balun to
the load, for impedance matching purposes.
Among many coupling matching networks that can be used
to connect the transmission line to the antenna, we can introduce a quarter-wavelength transformer (see Fig. 23). If the
impedance of the antenna is real, the transformer is attached
directly to the load. However, if the antenna impedance is
complex, the transformer is placed at a distance l away from
the antenna. The distance l is chosen so that the impedance
towards the load, at l, is real.
If the transformer is a 2-wire line section, and the line is
a coaxial cable, obviously the coaxial line cannot be connected
directly to the 2-wire line. In this case, a balun may be used
to connect the transformer to the coaxial line.
RADIO-FREQUENCY CHOKES
A radio-frequency choke coil is an inductance designed to offer a high impedance to alternating currents over the frequency range for which the coil is to be used. This result is
obtained by making the inductance of the coil high and the
distributed capacitance low. The result is that the inductance
;;;;
;
;;;;
A
A
B
ZL
ZL
ZS
Figure 7. Unbalanced line connected to a balanced line.
Zg
Figure 8. Idealized generator of the circuit of the Fig. 7.
ZS
534
ANTENNA ACCESSORIES
ZL
;;;;
B
/4
ZS
/4
Figure 9. Connection by a balun.
is in parallel resonance with the distributed capacitance
somewhere in the desired operating frequency range.
A typical radio-frequency choke coil consists of one or more
universally wound coils mounted on an insulating rod or of a
series of pies wound in deep narrow slots in a slotted bobbin.
A long single-layer solenoid is sometimes used.
Proper use of slug-type magnetic cores improves the performance of radio-frequency chokes. These cores increase the
inductance and hence the impedance of the coil without materially affecting the distributed capacitance.
(a)
(b)
Figure 11. (a) Balun bazooka. (b) Cross section of the balun.
d = 27.3 Z0/ ln(b/d) dB/m
Breakdown occurs in an air-filled coaxial line at atmospheric pressure when the maximum electric field Em reaches
a value of approximately 2.9 106 V/m. The average power
P that is transmitted in a matched coaxial line under these
conditions is given by
COAXIAL CABLE
2
P = (Em
/480)b2[ln(b/d)/(b/d)2] W
Coaxial cable is often used in antenna engineering to connect
a transmitter to an antenna, particularly at high frequencies
largely because of convenient construction and practically
perfect shielding between fields inside and outside the line.
The range of impedances obtained most conveniently by coaxial lines is from 30 to 100 . However, because the cable is
an unbalanced structure, its connection to a dipole needs a
transformer called a balun. Figure 12 shows its configuration with the dimensions of interest. d is the internal conductor diameter, and b the outer conductor diameter. Its
characteristic impedance is given by
;
Z0 = 60/ r ln(b/d)
(8)
where r is the relative permittivity of the medium.
The attenuation c of a coaxial line due to ohmic losses in
the conductor is given by
c = (4.343Rs/Z0 )(1/b + 1/d) dB/m
IA
c /2(b + d)
IB
(12)
INSULATORS
A material is an insulator if this is its most dominant characteristic. This does not mean that other electric effects, such
as conductivity and magnetism, are not present, but they are
just less significant. From a macroscopic point of view, the
constitutive relationship of the material, specifically in the
case of insulators (dielectrics), is given by
E
D = E
(13)
where E is the electric field intensity vector and D is the electric flux density vector.
ZL
;;;;;
F
(11)
The first high-order transversal electric (TE) mode that
propagates in a coaxial cable has its cutoff when the average
circumference is about equal to its wavelength, so
(9)
where Rs f/ where f is measured in Hz, in H/m,
and is the conductivity in S/m.
The attenuation d of a coaxial line caused by losses in the
dielectric is given by
(10)
ZS
Figure 10. Equivalent circuit of Fig. 9.
Figure 12. Geometry of a coaxial cable. A dielectric of relative permittivity r fills the space between the inner and outer conductors.
ANTENNA ACCESSORIES
From the microscopic point of view, it is better to use the
relationship introduced by Lorentz (6), where two terms contribute to the vector density of electric flow. The first is related to the vector electric field by the vacuum permittivity,
and the second is called vector polarization P. Mathematically,
D = 0 E + P
(14)
where 0 8.856 1012 F/m.
The contributions of the material to the behavior of vector
P are different for solids, liquids, and gases. These are very
complicated to summarized here [for more details, see Ref.
(6)]. For linear materials, however, P is directly proportional
to E. Therefore
P = 0 e E
(15)
where e is the electric susceptibility to the medium.
Substituting Eq. (15) in Eq. (14), one returns to Eq. (13),
because
= 0 (1 + e )
(16)
Therefore there is no difference between this point of view
and that of the constitutive relationship.
The term that defines the loss of energy in Poyntings theorem has the vector current density (which considers the conduction current) in phase with the intensity of the vector electric field. In the general characterization of dielectrics with
losses, the complex permittivity is introduced and is given by
= j
(17)
This is introduced in Maxwells equations as the current
j( j)E, where the second term is in phase with the electric field. Both and are, generally functions of frequency.
Table 1 lists values of / 0 and / for some representative
535
materials used in antenna engineering. It also lists the breakdown field for some materials.
LIGHTNING ARRESTERS
Lightning is an atmospheric phenomenon with potentially
harmful consequences. It is caused by the accumulation of
electric charges in a cloud and the consequent discharge to
the terrestrial surface. This effect occurs mainly on structures
that offer favorable conditions to discharge. The best protection against atmospheric discharges is the enclosure by a
grounded conductive structure. Even so, such protection when
not expensive, sometimes becomes impracticable.
Antennas must be immune to atmospheric discharges and
maintain integrity as lightning current flows to the ground.
Protection against the direct and induced effects of lightning
has to be designed to avoid damage to operators, equipment,
and structures. For antennas, several techniques can be used,
such as
protective conductors on a radome (7) (without degrading
the performance of the antenna)
grounding the towers (always positioning the antennas
below the tower summit)
lightning arresters (mainly in HF operation, where the
antennas are relatively large structures)
Common types of lightning arresters are varistors, gas discharge devices, and semiconductors. No one type is suitable
for all applications, and each may be combined with another
into a hybrid device. A simple lightning arrester can be assembled by creating a gap between the structure to be protected and the ground. These gaps are adjusted to minimum
width such that no arcing occurs when the transmitter is operating.
PHASE SHIFTERS
Table 1. Characteristics of Insulating Materials
Material
Composition
/0 at
108 Hz
/ at
108 Hz
Breakdown
field at
25C (V/m)
Ceramics
Aluminum oxide
(alumina)
Porcelain
(dry process)
8.8
0.00030
5.04
0.0078
0.00020
0.00020
47.2 106
39.4 106
0.034
15.9 106
Plastics
Polyethylene
Polytetrafluorethylene
(Teflon)
2.26
2.1
Adhesives
Epoxy resin
(Araldite CN-501)
3.35
Types of Phase Shifters
Glass
Fused quartz
a
Data from Ref. (1).
A phase shifter is a two-port component, which provides a
fixed or variable change in the phase of the traveling wave.
The shift is with respect to reference (the line without the
component) and test (the line and the component) lengths.
Therefore it is always understood as the phase difference between the two. The shift may be fixed or variable. The variable phase shift uses mechanical or electronic techniques to
change the phase dynamically. The main uses of these devices
are as testing systems, measurement systems, modulation devices, and phased array antennas. The use of phase shifters
in antenna systems provides controllable steering of the main
beam of the radiation pattern without moving the antenna.
There are several ways to implement these components for
communication systems. The type of phase shifter used in
this process depends on the kind of antenna used, its costs,
and power requirements.
3.78
0.00020
16.1 106
Shifters can provide fixed or variable phase shifts. Fixedphase shifters are usually extra transmission line sections of
certain lengths to shift the phase with respect to the reference
line. Variable shifters use mechanical or electronic means to
536
ANTENNA ACCESSORIES
achieve a dynamical range of phase difference. In antenna
systems, the main use of variable phase shifters is in beam
steering, whereas fixed shifters are used more for beam forming (pattern synthesis).
A mechanically tuned phase shifter usually consists of
variable shorts used with hybrids or, in the case of waveguide
components, dielectric slabs with variable positions in the
guide. A step motor moves the slab across the guide (from its
center toward outer walls), therefore accomplishing maximum or minimum phase shift. Another way of obtaining the
desired mechanically tuned phase shift is by combining variable shorts and hybrids. The movement of the short circuit
along a transmission line results in the phase shift, therefore
making it appear shorter or longer.
There are two classes of electronically tuned phase shifters
depending on component cost, weight, and power handling capability: ferrite and switching circuit shifters. The ferrite
shifter uses slabs of this magnetic material inside waveguides. Because the electrical properties of ferrite change depending on an applied bias magnetic field, this process yields
the desired phase shift. Ferrite phase shifters are essentially
low-cost, bulky, devices with high power handling capability.
Switching-circuit shifters are essentially low-power pin diodes or FETs used as switching (ONOFF states) elements.
Depending on the applied bias current, these devices work
either as short or open circuits. The phase shift is the result
of an extra section of transmission line added to the switching
element. Therefore, depending on the bias current, the wave
traveling along the transmission line has an additional traveling path. Because these devices are binary switches, only
discrete phase shifts are possible. The construction of a continuously variable shifter demands the use of several of these
switching elements. The design of variable phase shifters depends on the expression of the variable shift with respect to
a specific position or applied bias magnetic field or current. In
all cases, the procedures depend on the topology of the device
and its mathematical model.
Ferrite Phase Shifters
Ferrite phase shifters are common in power applications in
waveguide technology. The ferrite phase shifter is the result
of combining two properties of this medium: the dependency
of the velocity of the propagating waves on the direction of
propagation and the application of a bias magnetic field. Ferrite is an anisotropic medium. Therefore, the propagation constant (and the velocity of propagation) depends on the bias
applied to the sample. The anisotropic properties of ferrites
also depend on whether a magnetic field is applied. This is a
macroscopic result of the microscopic interaction of the fields
with atomic nuclei.
The phase shift in these devices is the difference between
forward and backward traveling waves. This is a nonreciprocal phase shift, which depends on an applied external magnetic field. Accurate design of these devices is not straightforward because of the complexity of the mathematical models
of ferrites. A detailed procedure is given by Pozar (8) and
Bornemann (9).
Switching-Circuit Phase Shifters. The switching-circuit phase
shifter uses solid state technology (pin diodes). In these kinds
of applications, pin diodes behave as ONOFF elements. The
2
Figure 13. PIN switching-circuit phase-shifter representation. The
application of the bias current results in a change of the electric
length of the line. The resulting shift is the difference between the
electrical angle in the direct and reverse polarized cases.
switching-circuit shifters are common to several manufacturing technologies, such as microstrip, coplanar guides and
lines, slotline, waveguides, and finlines. The diodes act as binary switches along the line (Fig. 13). With the bias applied,
the diode behaves as a short circuit for the incoming wave.
Therefore, the phase-transmitted signal has two possible
states, determined by which diode is polarized. A variable
phase shift results from mounting several diodes on the line.
The returning wave enters a hybrid and then the antenna.
PLANAR TECHNOLOGY
Planar transmission lines are the basic transmission media
for microwave integrated circuits when hybrid or monolithic
technology is used. Figure 14 shows the cross sections of three
basic planar transmission lines: stripline, microstrip, and
slotline. Several other configurations are possible, such as
listed in Ref. (10). All configurations are variations of the
three main types. Each offers certain advantages, depending
on the circuit type and its application. Some of these configurations are the suspended stripline, the suspended microstrip,
the inverted microstrip, the coplanar guide, and the coplanar
stripline. Planar designs using special substrates have been
applied at frequencies up to 100 GHz.
A designer can produce several microwaves circuits and
electric functions with planar transmission lines. Therefore,
planar technology constitutes a very important technology for
developing antenna accessories in the microwave range.
These components, called distributed circuits, are generally of
the order of a wavelength or more (in the propagation medium). Therefore, in the lower microwave range, these circuits
can have large dimensions. In this case, one uses lumped components.
Several lumped components can be obtained starting from
planar technology. These components are called microwave
integrated circuits (MIC). Figure 15 illustrates lumped and
distributed circuits in planar technology.
Following are the most important design parameters for
planar transmission lines:
effective dielectric constant
characteristic impedance
dispersion
;;
;;
;;
ANTENNA ACCESSORIES
537
Strip conductor
Ground
plane
Dielectric
substrate
Stripline-TEM mode
Ground
plane
(a)
(a)
Strip conductor
Dielectric
substrate
Ground
plane
Microstrip line- quasi-TEM mode
(b)
Ground plane
Ground plane
Slot
Figure 15. Lumped and distributed MICs. (a) Spiral inductor
(lumped element). (b) Band-pass interdigital filter (distributed MIC)
(courtesy of the Department of Electrical Engineering of the University of Braslia).
Dielectric
substrate
w
b
Slotline- non-TEM mode
(c)
Figure 14. Basic planar transmission lines. (a) StriplingTEM
mode. (b) Microstrip linequasi-TEM mode. (c) Slotlinenon-TEM
mode.
losses
working frequency limitations
Several antenna accessories are manufactured for planar
technology. Examples are directional couplers, hybrid rings,
power dividers, and impedance matchers (11,12). These components are used as matching, feeding, and sampling networks in antenna systems.
The effect of the electromagnetic coupling between two
parallel transmission lines (sufficiently near each other) constitutes the design framework of several microwave devices
(especially for antennas). An instance is the directional coupler, which has wide application in circuits and telecommunication systems. It performs several electric functions, such as
separating and combining signal power (received or delivered
from/to antennas). The directional coupler also allows sampling power and separating incident and reflected waves of a
system. This separation is used for power control or performance measurements.
Couplers are either conductively or electromagnetically
coupled transmission line circuits with three, four, or more
ports. In general, a directional coupler is a four-port device.
Two of its ports are mutually decoupled with respect to the
other two. Figure 16 shows the block diagram of a directional
coupler. From a theoretical point of view, if port 1 receives a
microwave signal, this power is delivered to ports 2 and 3,
but no power appears in port 4. Port 4 is decoupled from port
1. On the other hand, if port 2 receives a microwave signal,
then this signal is divided between ports 1 and 4, and port 3
is decoupled relative to port 2.
In the special case where power is equally divided between
ports 2 and 3, the coupler is called a hybrid or 3 dB directional coupler. If the phase difference between ports 2 and 3
is 90, the coupler is called a 90 hybrid or quadrature hybrid.
Another hybrid design provides 180 between the output
ports.
Different topologies of directional couplers exist (11,12).
Figure 17 shows a directional coupler in the topology of parallel lines of /4 length. As in any directional coupler, a fraction
of the incident power in the primary branch is coupled to the
secondary branch. In fact, when a microwave signal is incident in port 1, port 2 receives part of its power. In this topology, another part of the power is coupled to port 3. This coupling happens through the gap, and practically no power is
coupled to port 4, which is terminated by the load. The load
has the same value as the characteristic impedance of the
coupler lines. In parallel line couplers, the smaller the gap
between branches of the coupler, the greater the coupling.
Coupled
output 3
port
Input
port
a
d
Isolated
port
Direct
2 output
port
Figure 16. Block diagram of a directional coupler.
538
ANTENNA ACCESSORIES
Figure 17. Directional couplers in /4 parallel lines in microstrip
technology (courtesy of the Department of Electrical Engineering of
the University of Braslia).
The spacing between conductors can compromise high coupling designs.
The main parameters of a directional coupler are the coupling, directivity, isolation, and transmission factors. The coupling factor is defined by the relationship between the power
in the coupled (port 3) and input (port 1) ports. This factor is
a measurement of the coupling. The transmission factor is
defined by the relationship between the power in ports 2 and
1. The directivity is the relationship between the power in the
isolated (port 4) and coupled (port 3) ports. This measures the
undesirable coupling. The isolation is the relationship between power in the isolated (port 4) and input (port 1) ports.
Isolated
4
port
Coupled
port
Direct
port
Y3
Y1
Input
port
3
Y2
/4
Y4
1
/4
(a)
Figure 19. Hybrid ring (courtesy of the Department of Electrical Engineering of the University of Braslia).
Figure 18(a) shows a coupler in branch line topology
(11,12). A larger coupling factor can be obtained in this directional coupler. It also allows keeping dc continuity between
the ports, which is an interesting feature for phase-shifter design [as shown in Fig. 18(b)]. It allows high power levels. The
branch line coupler uses branching transmission lines to couple 3 dB to 9 dB. The lengths of all branches of this coupler
are /4. However, the widths and consequently the impedances of these branches are different: Y1, Y2, Y3, Y4. The operating principle of this coupler can be presented in the following way: when the input port (port 1) receives a signal,
the energy is divided and it propagates clockwise and counterclockwise through the branches. The power in each output
(ports 2, 3, and 4) depends on the phase relationships between
the propagating signals. In a 3 dB coupler, power is divided
equally between the coupled (port 2) and matched (port 3)
ports. Port 4 is the isolated port.
A special version of the branch line coupler is the hybrid
ring (Fig. 19). The ring has a 1.5 circumference and all sections of the same impedance. It can be assembled on several
technologies, such as coaxial line, stripline, microstrip line, or
even waveguide. The operating principle is same as that of
the branch line coupler. The signal injected in the input port
is divided into two propagating waves with opposite directions. In the coupled output port, the two signals arrive in
phase and reinforce each other, providing an output signal.
In the direct output port, the signal that propagates counterclockwise travels a length of 1.25 , and the signal propagating clockwise travels 0.25 . The two signals arrive at this
port in phase, providing an output signal at this port. For the
same reason, there is an isolated port, which has no output
Bridging
wire
connection
w2
Isolated
Input
s
w1
(b)
Figure 18. (a) Branch line coupler structure. (b) Branch line coupler
in a phase-shifter circuit (courtesy of the Department of Electrical
Engineering of the University of Braslia).
Coupled
/8
/8
Figure 20. Lange coupler.
Direct
ANTENNA ACCESSORIES
From
source
Zl
Zt
To
load
Zl
539
Patch
Feed
Z sZ l
Zt =
/4
(a)
(b)
/4
Patch
Figure 21. Lange coupler in tandem topology (courtesy of the Department of Electrical Engineering of the University of Braslia).
signal. The output signals in the direct and coupled ports are
180 out of phase.
Another topology of the directional coupler is the Lange
coupler (Fig. 20) which is well adapted to microstrip technology (13). It is an interdigital structure with superior performance compared with the parallel line coupler. Its bandwidth
can be greater than one octave. The Lange coupler associated
with tandem topology (Fig. 21) allows obtaining power coupling near 3 dB with a combination of two lower order couplers, whose design dimensions are feasible.
Power dividers are used in feeds and performance measurements of antennas. Directional couplers and hybrid rings
can be used as power dividers. The branch line coupler output
signals are in phase quadrature (90 out of phase). Output
signals of hybrid rings are in opposite phases. However, in
several microwave applications, such as parallel feeds for
phased array antennas, the input has to be divided into an
arbitrary number of signals with equal power and in phase. A
symmetrical power divider of n branches provides successive
division of the input signal into n signals with the same
power and phase at all frequencies. This power divider can
also be used as a signal combiner by inverting the input and
output ports.
There are several kinds of power dividers/combiners. One
of them is the two-branch power divider, shown in Fig. 22. It
Output 1
2 Zc
Zc
Zc
Input
2 Zc
Zc
Output 2
(a)
Output 1
2 Zc
Zc
Zc
R = 2 Zc
Input
2 Zc
Zc
Output 2
(b)
Figure 22. Unmatched (a) and matched (b) two-way power dividers.
Zs
Zl
k3
Zl
(c)
l3
Z5
Z3
l2
Z2
l1
Z1
Stripline
Strip
Microstrip
Substrate
(d)
Figure 23. (a) Quarter-wave impedance transformer, (b) top view of
a patch antenna, (c) microstrip edge feed with quarter-wave transformer, (d) multisection quarter-wave transformer.
is also possible to obtain output signals in phase but with
different power levels.
Impedance matching circuits are another kind of antenna
accessory. They are also manufactured in planar technology
(2). These circuits are commonly used in antenna feeds,
avoiding power losses caused by mismatch. In most cases,
the load has an impedance different from the transmission
line. This is the case in telecommunications systems, where
the output impedance of the generator is usually different
from the input impedance of the antenna. The problem is
how to reduce or eliminate the resulting reflections and
high SWRs.
One method is to cascade transmission line sections (planar structures) or lumped components with different impedances between the mismatched transmission line and its terminations (the system output and the antenna input). Then
the antenna input impedance is transformed. This new impedance value seen at the output point (the end of the transmission line) can be matched to the system for maximum
power transfer.
Different topologies and technologies are used in impedance matching circuits. Some of these are stub association
and quarter-wave transformers. Figure 23(a) shows a schematic of a quarter-wave impedance transformer. The impedance Zl can be matched to a transmission line of characteristic
impedance Zs with a section of transmission line that is a
quarter-wave long based on the wavelength in the transmission line. This characteristic impedance of the matching section is shown as Zt ZsZl. The microstrip feed on Fig. 23(b)
is planar, allowing the patch and the feed to be printed on a
single metallization layer. The impedance of the edge-fed
patch can be transformed by using a quarter-wave matching
540
ANTENNA ACCESSORIES
section of microstrip transmission line, as shown in Figs.
23(a) and 23(c). Another kind of quarter-wave transformer is
the multisection quarter-wave transformer, as shown in Fig.
23(d). This kind of transformer is commonly used in wideband systems.
2r
TWIN LEAD
The bifilar line (twin-lead) is a two-wire parallel conductor
line, which carries power from the generator to the antenna.
Each wire carries equal and opposite currents (180 out of
phase). Because the wires are spaced by a certain distance,
however, their radiated fields do not cancel out completely. If
the two wires were in the same place in space, there would
be no radiated field. However, because this is not possible,
there will be a certain amount of radiation loss. Keeping the
distance between the parallel wires small (typically on the
order of 1% of the wavelength of the radio wave) reduces this
loss. The spacing between conductors is a constraint of the
physical limitations of line construction. Parallel conductor
lines, open-wire lines, two-wire lines, two-wire cable and twowire ribbon cables are also bifilar lines.
Types of Bifilar Lines
The bifilar lines are usually supported a fixed distance apart.
Insulating rods, called spacers, or molded plastics, such as
polystyrene, provide the support. The spacers have to be
placed at short intervals to prevent the wires from moving
apart from each other. The lines are air-insulated, because
this is the medium between them. The spacers have little effect on the impedance behavior of the line. Flexible dielectrically separated (molded) lines have several advantages
over air-insulated types. In this type, a plastic coating similar
to a ribbon surrounds the conductors. These lines maintain
uniform spacing between conductors. They are also less
bulky, weigh less, and are easier to install. One common example of a flexible dielectric bifilar lines is the television receiving cable (called ribbon cable or two-wire cable).
d
Figure 24. The geometry of bifilar lines. In the molded plastic casing, the dielectric surrounding the lines has the shape of a ribbon.
Because most of the field is inside the ribbon, the line is less susceptible to external influences.
Design Equations of Bifilar Lines
The analysis of bifilar transmission lines uses Maxwell equations subject to the appropriate boundary conditions. Following are the main parameters of bifilar transmission lines:
Characteristic impedance
1
Z0 = cosh
d
2r
120
1
= cosh
r
d
2r
(18)
where r is the radius of the conductors, d is the spacing
between them, and r is the relative dielectric of the surrounding medium (Fig. 24).
Attenuation
A(dB/m) = 8.686
1
r
d/2r
(d/2r)2 1
+
(19)
where is the permeability of the surrounding medium,
is the operating frequency, is the loss in the conductors, and / is the relative loss of the dielectric constant.
Propagation constant
Uses of Bifilar Lines
In television reception applications, polyethylene molded ribbon bifilar lines are available with spaces of 2.54 mm (1 in.)
and 1.27 mm (1/2 in.) and conductor sizes of AWG 18 (diameter of 2.053 mm). These lines have characteristic impedance
of 450 and 300 , respectively. The attenuation is quite low
for these receiving applications (typically under 0.03 dB/m for
applications under 20 m). There is also a 75 bifilar line,
which has AWG 12 conductors (diameter of 1.024 mm) and
close spacing between them (see Ref. 14 for a detailed description). The spacing keeps most of the field confined to the solid
dielectric instead of the surrounding air.
Bifilar lines are also used in amateur radio applications
and LF to VHF antennas, although for higher frequency applications, the coaxial line is used more because of its superior
characteristics. The lines are simple to assemble. However,
the designer should avoid sharp bends. Bends change the
characteristic impedance of the lines, which causes reflections
and power mismatch at each bend.
= /v
(20)
where v is the phase velocity in the medium surrounding the conductors. In a ribbon transmission line,
r is 2.56 with (/ ) 0.7. If the wire is made of copper,
then is 58 MS/m. In this case, because most of the
field is inside the dielectric, the relative dielectric constant r is approximately 2.56.
BIBLIOGRAPHY
1. Reference Data for Radio Engineers, 7th ed., Indianapolis, IN:
H.W. Sams, 1985.
2. R. E. Collin, Foundations for Microwave Engineering, 2nd ed.,
New York: McGraw-Hill, 1992.
3. A. W. Rudge et al., The Handbook of Antenna Design, London:
Peter Peregrinus, 1986, Vols. 1 and 2.
4. H. Jasik, Antenna Engineering Handbook, New York: McGrawHill, 1961.
ANTENNA ARRAYS
5. W. L. Weeks, Antenna Engineering, New York: McGraw-Hill,
1968.
6. S. Ramo, J. R. Whinnery, and T. Van Duzer, Fields and Waves
in Communication Electronics, 3rd ed., New York: Wiley, 1994,
pp. 678682.
7. R. H. Golde, Lightning, New York: Academic Press, 1981, Vols. 1
and 2.
8. D. M. Pozar, Microwave Enginering, Reading, MA: Addison-Wesley, 1990.
9. J. Uher, J. Bornemann, and U. Rosenberg, Waveguide Components for Antenna Feed Systems. Theory and CAD, Norwood, MA:
Artech House, 1993.
10. B. Bhat and S. Koul, Stripline-Like Transmission Lines for Microwave Integrated Circuits, New York: Wiley, 1989.
11. E. A. Wolff and R. Kaul, Microwave Engineering and Systems
Applications, New York: Wiley, 1988.
12. L. A. Trinogga, G. Kaizhou, and I. C. Hunter, Practical Microstrip
Circuit Design, London: Ellis Horwood, 1991.
13. J. Lange, Interdigited stripline quadrature hybrid, IEEE Tran.
Microw. Theory Tech., MTT-17: 11501151, 1969.
14. G. Hall (ed.), The ARRL Antenna Book, 15th ed., Newington, NJ:
The American Radio League, 1988, pp. 24.1524.25.
PLINIO R. G. ALVES
LEONARDO R. A. X. MENEZES
PAULO H. P. CARVALHO
FRANKLIN C. SILVA
University of Brasilia
ANTENNA, APERTURE. See APERTURE ANTENNAS.
541
ANTENNA ARRAYS
Consisting of two to thousands of antenna elements, an array antenna presents the ultimate in exible antenna pattern control. This capability includes electronic scanning,
planar or conformable apertures, the ability to transmit or
receive multiple shaped patterns, and adaptive control for
jammers, clutter, and multipath suppression and for the
reduction of cosite interference. Arrays benet from technological advances, and they are the key technology drivers
for solid state T/R (transmit/receive) modules, advanced
signal processing, and photonic technology.
Antenna arrays are often simply called phased arrays,
a term that refers to the progressive phase shift introduced to scan the beam. We will use this term throughout this section, although the beam can be scanned by either time delay devices or phase shifters, depending on the
required systems bandwidth. The distinction will be addressed later.
The basic principle behind the operation of the phased
array, as illustrated in Fig. 1, is that the RF power be divided among a number of elements by a power divider, with
each element signal shifted appropriately in phase or in
time. Relative time delay is portrayed in the sketch by circular phase fronts emanating from each element of the array, with signals either radiated at the same time [Fig. 1(a)]
or delayed in time by an increasing amount from left to
right so that the rightmost signal radiates last. The gure
shows how radiation from each element adds in space so as
to create an outgoing wave with an appropriate scan angle.
Although not shown in the gure, the array power divider
usually provides equal line lengths from the source to each
element because this leads to optimum system bandwidth.
The power divider can also be used to control the signal
amplitude at each element. This unequal power division
is called amplitude tapering, and it provides for sidelobe
control. At broadside, with element spacing chosen properly, the array directivity is that of the broadside aperture,
where A is the aperture area and a is the aperture efciency, which depends on the arrays taper design. The
scanned beam of Fig. 1(b) will have reduced directivity and
(usually) additional losses that further reduce the array
gain.
ARRAY ARCHITECTURE AND COMPONENTS
Figure 1 also indicates the basic components that make up
a phased array. They consist of a power divider/combiner,
phase shifters, and radiating elements. This simple sketch
understates the tremendous number and variety of components that make up array technology, and it also understates the importance of an overriding array architecture
aimed at satisfying the system needs. The following paragraphs will illustrate some of these issues.
Architecture is the character and style of building of a
structure or system. The term array architecture implies
that there are signicant choices to be made in the design
of an antenna array system; these choices can completely
change the design, packaging, cost, and function of the system. The architecture of the system needs to be determined
by the antenna system engineer working with the radar,
communications, and avionics system engineers.
One type of an array may be an ideal solution for a
requirement at, say, 12 GHz but altogether inappropriate at 120 GHz. The 12 GHz design may be a waveguide
array with ferrite phase shifters, but at 120 GHz active
solid state circuitry may have replaced the ferrite phase
shifters, microstrip replaced the waveguide, patch radiators replaced the waveguide elements, and monolithic fabrication replaced the machined part fabrication. These two
arrays will therefore have completely different architectures.
This section will describe some of these components and
illustrate the interdependence of system architecture and
component selection.
Power Dividers
Selecting the way RF power is combined or distributed
within the array is a major architectural choice. Constrained (or often termed corporate) power divider networks can be fabricated using waveguide, coaxial line,
stripline, or any of the various printed circuit transmission
media like microstrip, slot line, and others. Aside from the
choice of transmission medium, the most important choice
is the selection of an organization for the aperture. Many
arrays use a power divider network with as many ports as
antenna elements. This is a very simple organization, with
one array element excited from each port of the power divider. Amplitude illumination (taper) is provided by the
power divider or occasionally by adding loss at the element level. The array is then assembled one element at
a time, with elements supported by the power divider itself. Cost reductions have come from grouping elements of
the array into subarrays; that are readily mass-produced
and that incorporate necessary bias and logic circuitry as
well as RF power division. Figure 2 shows two basic ways
of constructing the aperture. Figure 2(a) shows an array
of column subarrays assembled using brick construction
in which the power divider network, in addition to all the
phase shifters and sometimes the antenna element, are
constructed on a single panel. The distinguishing feature
of brick construction is that the array elements are at the
panels edge. The panel is then inserted into the array as
a brick into a wall. Figure 2(b) shows an array made up
of area subarrays called tiles. Tiles tend to be multilayer
geometries with the power divider network and elements
occupying different layers in the same tile. The tiles are
placed side by side to build the array. These fundamentally
different organizations dictate (and in turn are dictated by)
the selection of solid-state components, circuitry, and the
array elements.
In addition to such constrained networks, a number of
arrays are built using space feeds, wherein a horn antenna illuminates the back or front face of an array to provide power division. Space-fed arrays are more bulky then
corporate-fed arrays, but they are usually far less expen-
J. Webster (ed.), Wiley Encyclopedia of Electrical and Electronics Engineering. Copyright 2007 John Wiley & Sons, Inc.
Antenna Arrays
Figure 1. Radiation from antenna arrays. (a) Array radiating
broadside. (b) Array radiating to angle .
sive and often lighter. Figure 3(a) shows that with one such
space feed, the array is used as an active lens. Array elements at the back face of the array receive the RF power
from the horn feed. Phase shifters installed in the path
between array back and front faces correct for the nearly
spherical wavefront of the incident wave and then add the
progressive phase distribution to scan the beam. Figure
3(b) shows the reectarray concept in which the array is
space-fed from the front. Again the elements at the front
face of the array receive the distributed power, and phase
shifters provide the necessary phase correction and scanning control. The back face of the array is short-circuited,
so the RF signal goes through the phase shifter twice before it radiates. This tends to make reectarrays somewhat
more lossy and the phase shifter more tolerance sensitive
than lens arrays. This feature, and undesirable feed blockage as indicated in Fig. 3(b), for low-angle radiation have
made space-fed lens arrays the more popular choice for
most space-fed systems.
Phase Shifters and Time Delay Units
Phase shifters are devices that change the phase of a signal traveling down the feed transmission line. Usually the
devices have discrete phase states, so one measure of the
quality of the phase shifter is the number of bits, or digital
states, by which it approximates a continuous phase dis-
tribution from zero to 360 degrees. For example, an n-bit
phase shifter has n different phase states that are used in
cascade; it can produce 2n different output phases. A threebit phase shifter consists of a 180-degree state, a 90-degree
state, and a 45-degree state. Starting at zero phase and
switching in various combinations produces 23 or 8 output
phases from 0 to 315 degrees at 45-degree intervals. The
360-degree output phase is the same as 0-degree phase, so
clearly all the required phases are produced with the three
cascaded bits.
Current phase shifters use either ferrite material in the
transmission lines, with propagation constants changed by
electrical currents, or diode-loaded circuits, which are embedded in the transmission line network. Figure 4(a) shows
a ferrite torroid phase shifter in a waveguide conguration.
The phase shift state is selected by passing a current pulse
of calibrated duration and amplitude through the ferrite
core. This is only one of a number of useful ferrite phase
shifter types, and a more complete listing can be found (1)
in the literature. Typically ferrite phase shifters can provide extremely precise phase shift, insertion loss less than
1 dB, and can handle hundreds of watts of average power.
Figure 4(b) shows one section of a switched line phase
shifter that inserts an effective path length l1 or l2 , depending upon the settings of switches s1 and s2 . Diodes can also
be used to switch much longer lengths of line and so can
Antenna Arrays
Figure 2. Array and subarray congurations. (a) Column subarrays with brick construction. (b) Area subarrays with tile construction.
Figure 3. Space fed arrays. (a) Lens array. (b) Reectarray.
be used to control time delay devices. In a phase shifter,
the added sections of line are all a fraction of a wavelength
long, so that the longest electrical path difference through
the device is, as described earlier, one bit less than 360 degrees longer than the shortest path. A time delay device,
however, needs to represent many wavelengths of delay,
and it in fact needs to insert lengths of line comparable
to the physical length of the array. Such devices are often
lossy, complex, bulky, and costly but may be essential for
wideband array performance depending on array size and
scan requirements.
Figure 4. Phase control devices. (a) Ferrite phase shifter. (b)
Diode switched line phase shifter.
Antenna Arrays
The switched line phase shifter is only one of many
solid-state phase shifter designs. Some phase shifters use
a switched line section for one or several of the smaller bits
but hybrid circuits or loaded line circuits for the larger bits.
These and other circuits are sometimes less frequency dependent, require fewer diodes, or occupy less space than
the switched line type. Sharma (2) gives an accounting of
the state of the art in solid-state phase control devices.
Array Elements
The nal, and most important, array component is the radiating element itself. The element determines the ultimate bandwidth, polarization, weight, and cost of the entire radiating system in addition to fundamental architectural issues like integration with the transmission medium
and the choice of distributed solid-state power modules
or central tube source. Practical radiating elements range
from open-ended waveguides to dipoles or monopoles and
a variety of printed elements coupled to the transmission
medium. Figure 5(ac) shows elements used for tile fabrication, while Fig. 5(de) shows elements for brick construction. The microstrip patch-fed tile element of Fig. 5(a)
is simple to fabricate and can be fed on-line or by using
probe feeds, or aperture coupled from a lower-layer circuit, but has only a few percent signal bandwidth in the
array environment. The proximity coupled patch element
of Fig. 5(b) can have bandwidth exceeding 10%. Strip-linefed cavity-backed slots as shown in Fig. 5(c), can be much
wider bandwidth, but they are difcult to fabricate because
of the need to dene the cavities using plated holes. The elements for brick construction are typically wider band but
do protrude from the aperture. Over modest scan sectors,
stripline dipoles [Fig. 5(d)] can have up to 40% bandwidth,
while ared notch elements [Fig. 5(e)] can have up to 4:1
bandwidth.
Figure 5. Array elements. (a) Microstrip patch element. (b) Proximity coupled microstrip element. (c) Stripline fed cavity backed
slot element. (d) Strip line fed dipole element. (e) Flared notch
element fed by stripline center conductor.
ARRAY THEORY
Array Antenna Patterns
The far-eld pattern of an array of N elements in the generalized geometry of Fig. 6 can be written in terms of the
incident transmission line voltage signals an and the element locations r n as
for
where kx = ku, ky = kv, kz = k cos , and k = 2/.
The parameters u = sin cos ; v = sin sin are called
direction cosines of the angular location (, ). The use of
the direction cosine parameters u and v greatly simplies
the evaluation of beamwidth, bandwidth, and scanning parameters, so they are routinely used in array analysis.
In these expressions, k is the vector wave number or
propagation constant of a plane wave radiating in the di-
Figure 6. Generalized array conguration.
rection of (, ) and k = 2/ at wavelength . fn (, ) is the
far-eld pattern of that nth element in the array environment.
The multiplying factor exp(jkR0 )/R0 is common to all
expressions, and it will be suppressed hereafter. In this expression we have assumed each element to have a single
input port, but an obvious generalization would include additional ports by including additional an fn terms. Finally,
the above summation applies for generalized array of any
dimensionality and conforming to any surface.
Antenna Arrays
In general, the element patterns fn (, ) are different
for each array element even though all the elements are
the same. These differences occur because each element
pattern includes the scattered radiation from all other elements, from the array edges, and any other scattering from
the mounting structure. Element pattern differences due
to element interaction will be discussed later. In the special
case of array elements mounted conformal to some nonplanar body, like a rocket or aircraft nose radome, the elements
do not even point in the same direction, so the element patterns are fundamentally different in this case.
Before discussing the general cases, let us consider a
linear array of elements located in a plane at xn with yn =
zn = 0 but otherwise arbitrarily spaced in the x direction:
If all of the element patterns fn (, ) are the same [and
given by f(, )], then the expression becomes
is reduced to half at the angle 0 results in the fractional
bandwidth
Arrays with narrow beamwidths thus have less bandwidth,
and inverse proportionally less bandwidth as the scan angle is increased.
If true time delay devices were used to collimate the
beam, the available excitation would take the form
In this expression the wave number k is the same as that
used in Eq. (5), and it varies linearly with frequency. With
this excitation the far-eld radiation will always have its
peak at u0 for all frequencies, and the array bandwidth is
only limited by device operation, not pattern squint. This
state is highly desirable, but true time delay devices are
costly, lossy, and can constitute a major architectural issue
in the design of arrays systems.
Periodic Arrays and Element Spacing
When this separation can be made, the pattern is expressed
as the product of an element pattern f(, ) and an array
factor (the indicated summation).
NARROWBAND AND WIDEBAND SCANNING
Consider the plane = 0 for convenience. This expression
[Eq. (5)] can have a maximum at different angles depending on the choice of an . To move a principal maximum to
some angle 0 at a single frequency ( =0 ), the required
phased element excitations are given by:
Most phased arrays consist of elements placed in a one- or
two-dimensional planar grid with equal element spacings.
There are several advantages in doing this. First, the periodic grid allows very precise pattern formation. Second,
both the fabrication and array excitation are simplied by
using the periodic arrangement of elements. However, the
periodicity also imposes constraints on element spacing in
order to avoid the formation of unwanted radiation peaks,
called grating lobes.
The grating lobe phenomenon is apparent from an inspection of Eq. (11) below. Consider the one-dimensional
array with elements at the locations xn = ndx and operating at the single frequency 0 scanned by phase shifters.
In the plane = 0 the pattern given by Eq. (7) becomes
for k0 = 2/0 and u0 = sin 0 . Substituting Eq. (6) into Eq.
(5) results in:
The pattern has a maximum value at u = u0 ( = 0 in the
plane = 0) when = 0 . However, for a signal at some other
frequency, since phase shifters produce a phase change that
is nearly independent of frequency, Eq. (7) shows a peak at
the angle whose sine is
The summation has its maximum when the exponent is
zero for all n. This is the peak of the main beam of the
array, but it is also maximum when the exponent is any
multiple p times 2n. At these peaks, or grating lobes, the
radiation is as large as it is at the main beam location (
= 0 ). These grating lobe angles are given by the angles p
for which
for
This beam angle moves, or squints, as a function of frequency as indicated in Fig. 7, with the beam peak farthest from broadside at the lowest frequency and nearest
to broadside for the highest frequency.
This squint angle change can be interpreted in terms of
a fractional bandwidth by assuming an array beamwidth
of u and assuming that the beam is placed exactly at the
angle 0 at center frequency f0 . Then dening the upper
and lower usable frequencies to be those at which the gain
for p all positive and negative integers. Angles described
by sin| p | > 1 do not correspond to real angles, so they are
often referred to as being in imaginary space.
Figure 8 shows the array factor of two linear arrays
scanned to 45 (u0 = 0.707). The rst array, with pattern
shown solid, has 64 elements with 0 /2 spacing, and the
second has 16 elements with 20 spacing. Both patterns
Antenna Arrays
Figure 7. Array beam squint for phase steered array.
Figure 8. Array factors illustrating grating lobe phenomenon. Solid curve: column array of 64 half
wavelength spaced elements scanned to 45 degrees. Dashed curve: column array of 16 elements
with 2 wavelength spacing.
peak at the scan angle, but the envelope of the pattern of
the array with half-wave-spaced elements falls monotonically away from that angle until it reaches a minimum at
u = 0.293 that is half way to a suppressed grating lobe at
u = 1.293 (in imaginary space). The dashed pattern additionally has grating lobes at multiples of 0.5 from the peak,
and a number of them exist and radiate in real space.
Equation (12) also leads to an expression for the largest
spacing allowable for any array that scans to some given
angle. Since only values of sin between zero and 1 represent real angles, the array spacing must be such that for
any scan angle u0 none of the grating lobes enters the region |sin | 1. This results in the constraint on spacing of
for u3 , the beam width between nulls of the array factor.
Typically the beam width u3 is on the order of 20 /L for an
array of length L.
Similar grating lobes are present in two-dimensional
scanning arrays. A planar two-dimensional array with M
N elements on a rectangular lattice, xm = mdx and yn = ndy ,
scanned to angles 0 and 0 represented by the direction
Antenna Arrays
cosines u0 and v0 has an array factor
and has grating lobes up , vq ,
subject to
This equation places similar constraints on dx and dy as in
the one-dimensional array.
Array Mutual Coupling
The previous section assumed that every element in the
array radiated with the same element pattern. In fact an
element in an array does not radiate the same pattern as
if it were radiating in an isolated environment, nor does it
present the same impedance or support the same current
or eld distribution. These effects are all evidence of a phenomenon called mutual coupling, depicted in Fig. 9(a), by
which every element of the array interacts or couples with
every other element. In the general case even the shape
of the current distribution on each element changes with
scan angle, and one must use a higher-order current approximation to evaluate the total radiation for each angle
in space. Between these two limits, one assuming no interaction and one assuming ne scale changes with the scan
angle, is the case where one can speak of single-mode elements and single-mode mutual coupling.
The solution of the general array mutual coupling problem involves a complex integral equation formulation and
numerical solution. The solution is expressed in terms of
a series of basis functions (possibly the harmonics in a
Fourier series distribution) used to approximate the total
current or aperture eld. For the purpose of this discussion, it is convenient to think of these basis functions as
modes and to consider the case where a single mode is a
good representation for the current distribution on each
antenna. This is often a good approximation because most
array elements are small compared to a wavelength, and
all element currents or elds are often nearly the same.
This one-mode assumption makes it simpler to explain
the two complementary views of array mutual coupling.
We will call these the mutual impedance viewpoint and
the element pattern viewpoint.
From the mutual impedance perspective, we assume the
single mode radiates with a pattern f(u, v), which might
be a vector quantity. Each transmission line excites an
element that radiates into all other transmission lines
through their elements, as indicated in Fig. 9(a). For an
N element array on some nth element, the radiating eld
or current, here called In , is related to an input voltage matrix for the whole array by the square N N impedance
matrix Z (V = ZI). The In are unknown, so to compute the
Figure 9. Array mutual coupling. (a) Coupling between array elements. (b) Element pattern P() and reection coefcient magnitude R of center element of unloaded waveguide array (B/ = a/
= 0.4) after Wu (3).
array radiation, one needs to invert the impedance matrix
that relates the applied signals Vm to the produced In :
In summary, from this perspective one can nd the array
radiation from the applied sources by solving for the actual currents (or elds) that result. It turns out that the
common problem of synthesizing a desired radiation pattern is handled by solving for the desired current and then
using the impedance matrix to nd the necessary applied
sources.
The alternative point of view is focused on the array
element patterns that radiate when each element is excited separately, with all other elements terminated in a
matched load. When only one transmission line is excited,
the total pattern is generated as the sum of contributions
from all the element radiation. Consider a small aperture
element that supports a single mode of eld with radiation
pattern f(u, v). If that nth element alone were to radiate
when excited by an incident signal an from the nth transmission line, then the radiation pattern of that element
x x n . The exponential term is due to
would be an e(u, v)ej k
the location of the element in the array. However, as indicated in Fig. 9(a), that element scatters its radiation into
every other element, inducing a eld on any mth element
that is given mnth term of the scattering matrix S. The
total radiation from the array with one element excited is
Antenna Arrays
thus:
where
This expression shows the radiation to consist of a primary
radiation from the excited element, plus a scattered term
given by Snn times the primary term, plus terms Smn times
the primary term but radiating from the location of the
other elements (xm ). From this perspective each element
radiates with a different element pattern fn (u, v)because
each element pattern contains radiation from every element of the array. The total radiation is the sum of all these
element patterns weighted by the incident transmission
line signals. Figure 9(b) shows an example (3) of the element pattern of an isolated antenna element and the same
element in an array. This gure shows the element pattern
of the center element of an array of N waveguide elements
for N = 5, 9, 11 and an innite array. The gure shows the
effect of mutual coupling on element pattern as resulting
in periodic ripples with higher periodicity for longer arrays
and that the end-re gain ( 0 = 90 degrees) reduces because of coupling until it is zero for the innite array case.
One can show that the optimum gain varies like cos for
the innite case.
Beam Broadening and Directivity Loss with Scan
The expressions for the scanned array pattern indicate that
for constant frequency f0 , either phase shifters or time delay units form a beam with peak at the scan parameters (u0 ,
v0 ) and that the shape of that pattern only depends on the
displacement (u u0 , v v0 ) and not the scan parameters.
The pattern is displaced with scan and otherwise remains
unchanged. The array beam width is therefore constant in
(u, v) space for any given azimuth angle , but in terms of
the dependence, the beam width broadens as the array
is scanned from zenith ( = 0) to the horizon. For a large
array and scan angle 0 not too near the horizon, this beam
width is given in terms of the beam width B at broadside
as
The beam width along the scan plane thus broadens like
sec 0 as the array is scanned in .
Accompanying this beam width increase is a decrease
in array directivity so that the directivity is given in terms
of the broadside directivity DB as
ARRAY SYNTHESIS
Many useful pattern synthesis techniques for planar or linear antenna arrays follow directly from existing methods
developed for aperture or continuous one-dimensional antennas. This is so for several reasons. First, as long as the
elements are closely spaced and grating lobes well out of
the radiating region, the array periodicity does not signicantly alter the pattern structure. Second, the distinctions
that do exist come from the mutual coupling and are evident in array edge effects, or equivalently from the observed different element patterns across the array. As long
as the elements support only the single-mode elds, these
issues do not alter the synthesis procedure, since one can
synthesize in terms of the currents and aperture elds that
create radiation, or in terms of the measured or computed
array element patterns, and then include mutual coupling
to evaluate the necessary applied excitation.
The basis for most aperture synthesis is the Fourier
transform relationship between aperture eld and far eld
for a continuous aperture. If the arrays are large and the
elements closely spaced, this procedure is not sensitive to
the discretization or edge effects, and the method is quite
accurate. The transform method is also especially convenient because of its application to arrays that are not periodic and for arrays conformal to gently curved geometries.
Arrays periodic in one or two dimensions have far-eld
patterns describable by discrete Fourier transform pairs.
In one dimension the array factor at wavelength is written
where the sum is taken symmetrically about the array center. The coefcients an are the array element excitation and
are given from orthogonality as
In this expression the integral is taken over the periodic
distance in u space, namely half way to the two nearest
grating lobes for a broadside beam. Used in this way, the
technique gives the best mean square approximation to the
desired pattern. This feature is lost if spacings are less than
half wavelength, although the technique is still useful.
A second technique that has found extensive application is the Woodward synthesis method (4). This approach
uses an orthogonal set of pencil beams to synthesize the desired pattern. The technique has important practical utility because the constituant orthogonal beams are naturally
formed by a Butler (5) matrix or other multiple-beam system.
Other techniques for periodic arrays are based on the
polynomial structure of the far-eld patterns. These include the method of Schelkunov (6), the Dolph-Chebyshev
method (7), and others. Among the most successful and
used methods are the pencil beam synthesis technique
of Taylor (8) and the associated monopulse syntheses
technique of Bayliss (9). These techniques are derived
Antenna Arrays
Figure 10. Low sidelobe sum and difference pattern synthesis. (a) Taylor sum pattern with 30 dB (n = 6) pattern.
(b) Bayliss difference pattern with 30 dB (n = 6) pattern.
as improvements to the equal ripple method of DolphChebyshev, and result in more realizable aperture distributions, improved gain and other advantages. Figure 10(ab)
shows the array factors for 32 element arrays with 30 dB
Taylor and Bayliss distributions. Note that the rst sidelobe in both cases is very close to 30 dB with respect to
the pattern maximum. In general, the discretizing of continuous distributions introduces errors in the synthesized
pattern, and these are more signicant for small arrays or
for arrays that are forced to have very low sidelobes. Space
precludes giving a detailed description of these procedures,
but they are described in detail in a number of references.
Usually discretizing the continuous distribution is not a
problem, but when it is, there are a number of iterative
techniques to converge to the original desired pattern. Notable among these is the work of Elliott (10).
Finally, in addition to these classic synthesis procedures,
there have been many iterative numerical solutions to the
synthesis problem. These have, in general, been shown to
be efcient and useful. One successful iterative procedure
was introduced by Orchard (11) that allows for complete
power pattern design, even to the extent of controlling each
pattern ripple or sidelobe level. Other recently developed
methods have used simulated annealing or genetic algorithms (12).
Aperiodic and Conformal Arrays
Aperiodic Arrays. A periodic array that fully occupies an
entire aperture has several advantages. The directivity of
such a uniform array at broadside is that of the lled aperture, namely 4A/2 for an array with aperture area A.
In addition the pattern (neglecting errors) can have very
low sidelobes as long as spacings are chosen small enough
10
Antenna Arrays
to eliminate grating lobes. However, there are times when
(usually for economy) one chooses to populate an aperture
with fewer elements. This procedure, called thinning, results in an array with nearly the full aperture beam width
but using only a fraction of the elements in the lled array. If this thinning were done by simply increasing the
element spacing of a periodic array, then the resulting pattern would have many undesired grating lobes. In the limit
it would take on an interferometerlike pattern.
Instead, in many thinned arrays, elements are placed at
randomized locations, whether on a rectangular grid or not,
and often excited with uniform illumination. With thinned
arrays the structured sidelobes can be lowered by tapering the density of excited antenna elements, instead of the
aperture power, as done for a lled aperture. This is done
(13) by selecting element locations statistically and choosing element weights as unity or zero with probabilities either equal to or proportional to the lled-array taper. The
proportionality constant K is unity if the one/zero probability is chosen equal to the array taper ratio. At K = 1 the
array is fully populated near the center, where the array
taper is nearly unity. If K is chosen less than unity, then
the array is not fully populated at the center, but the thinning is still proportional to taper. Since the algorithm is
a statistical process, the resulting aperture illumination
and pattern are not unique, but one can describe the average of the ensemble of arrays constructed from the algorithm. For this ensemble one can show that the resulting
average pattern is the sum of two patterns, one of which
is the ideal pattern of the lled, tapered array, multiplied
by the number K. The second pattern is the average sidelobe level, a constant value with no angle dependence. For a
large, highly thinned array the average sidelobe level is approximately 1/Nr , normalized relative to the pattern peak,
where Nr is the number of remaining elements. The average directivity is approximately Nr times the directivity of
an element pattern. The reason for this result is that all
signals add linearly at the beam peak, but elsewhere in the
pattern they combine like the average of a random process.
For this reason the normalized average sidelobe level is at
the level of isotropic radiation, or the factor Nr below the
peak directivity. Figure 11 shows a thinned array resulting
from using the above algorithm directly. The dashes shown
in the gure indicate elements left out of the square, half
wavelength, lattice of a 25 wavelength radius aperture. The
selected ideal pattern for the lled array is a 50 dB Taylor pattern [Fig. 11(b)] and it is approximated with 7845
elements excited. Given this number of elements, the average sidelobe level is about 39 dB, and this is about what is
indicated in the Fig. 11(c). Nearly one hundred thousand
elements would be needed to produce the 50 dB pattern
desired, so clearly thinned arrays do not satisfy most low
sidelobe array requirements. They do, however, present the
least expensive way to provide very narrow beam width
wide-angle-scanned patterns with moderate sidelobes.
Conformal Arrays. Conformal arrays are a special class
of antennas that are built to conform to the surface of some
vehicle, like an aircraft, spacecraft, missile, ship, or even an
automobile. Depending on the array size and the local radius of curvature, this can pose a signicant problem to the
Figure 11. Circular array with elements removed). (a) Geometry (dashes show elements removed. (b) Desired Taylor pattern of
lled array. (c) Pattern of thinned array.
Antenna Arrays
11
Figure 13. Array with phased elements and time-delay applied
at the input to contiguous subarrays.
Figure 12. Generalized array conformed to a body of revolution.
practical realization of any desired pattern. The primary
technical challenge is that the elements point in different
directions, so the control network needs to provide variable
amplitude and phase weighting to scan the array. An extreme but very important example is the use of an array on
a cylindrical surface like that of the vehicle shown in Fig.
12 for scanning over a hemisphere. When a beam radiates
in one direction, it is necessary to commute the amplitude
distribution around the cylinder in order to avoid radiating
energy into undesired regions. Many elaborate networks
have been designed for performing this commutation, but
it remains an expensive process requiring sophisticated design and packaging concepts.
Not all conformal arrays are mounted in such severely
curved shapes as to require signal commutation. Most conformal applications are for ush mounted or very low prole arrays on gently curved surfaces where the challenges
are far less severe. Future applications include airborne
arrays for satcom and aircraft to earth coverage, missile
antennas, and a whole variety of commercial vehicle applications. Conformal arrays will continue to be a major
growth area for array antennas.
been built in laboratory models, and will play a role in providing ultrawideband multifunctional arrays.
Conformality and exible control are the two features
unique to array antennas, and the number of new applications continues to increase to ll demands in the two areas. Conformal scanning arrays provide high-gain antenna
coverage from complex platforms, and when combined with
elements of the increased exibility, arrays will become especially good candidates for commercial applications such
as in wireless communications and automotive radar. As
the cost of solid-state modules continues to decrease, more
applications will be lled by this high-end antenna technology.
Time Delay for Wide Band Arrays:
TRENDS IN ARRAY ANTENNA DEVELOPMENT
The phenomenon of phased array squint as described in
equation 9 and Figure 7, is the primary factor limiting the
instantaneous bandwidth of large phase steered arrays.
There is no squint if time delay units can be used at every
element, but for very large arrays this option may be far too
costly. The practical, although band limiting, solution is to
group elements into phase steered subarrays and then provide time delay behind each subarray. Figure 13 depicts an
array of linear subarrays with time delay at each subarray
input port, and phase shift across each 4-element subarray.
Cost reduction, coupled with increasing capability, have led
to major new applications for array technology. Array production costs have declined with advances in microwave
solid-state components and circuits and with the development of printed circuit elements. Arrays with modest performance can now satisfy an increasing number of system
needs. Alternatively, array antennas offer immense growth
potential at the high end of performance capability. Though
not inexpensive, a variety of very high performance arrays
provide for multifunction capability and pattern features
like very low sidelobes or adaptive pattern control for clutter and jammer suppression. This capability is available
using digital pattern control, where the phase shift or time
delay functions are performed by digital sampling the RF
or baseband signal and processing the N array outputs to
do all adaptive and deterministic signal processing. Optically controlled arrays with ber optic switched delay lines
and other features provided by optical technology have
Figure 14 shows the behavior of an array of 8-element
linear array subarrays. The array has 16 subarrays with
phase shifters at every element and time delay at each subarray input. The patterns at center frequency are shown
at left. These patterns include the basic subarray pattern
shown in the top gure, the array factor for the widely
spaced subarrays at the center, and at the bottom the array radiation pattern, which is the product of subarray and
array factor. At center frequency the array factor is shown
scanned to u = 0.5 (30 degrees) and because of the 4 wavelength spacing between rectangular subarrays, there are
grating lobes at 0.5 + p(0.25) for p = 1 and 2. Fortunately these locations exactly correspond to the nulls of
the subaray pattern (top), and the envelope of the array
radiation pattern, the periodic sinc function at bottom left,
is a smooth monotonic function.
At a higher frequency f = 1.5fo , the time delay has kept
the main beam of the array factor at the desired scan angle,
12
Antenna Arrays
Figure 14. Subarray patterns, array factors
and radiation patterns of time delay steered
array at u = 0.5 (scanned to 30 degrees).
but now the subarray separation is effectively 4.6 lambda,
and so the grating lobes are closer together. These data
are shown at the right of Figure 14. More importantly, the
subarray pattern, which is phase shifted, squints toward
f0
broadside so that its peak is at
u0 (approximately 0.435
f
instead of 0.5). The nulls are no longer aligned with the
peaks of the grating lobes, and the product of subarray and
array factor produces large quantization lobes.
There are several ways to reduce or eliminate these
quantization lobes. Since they are caused by the periodic
errors that result from quantizing the aperture phase, using an aperiodic array of subarrays instead of a fully periodic one would eliminate the quantization lobes and leave a
spatial pattern with a distribution of higher average sidelobes. This approach results in improved patterns, but a
loss of pattern gain because the array aperture is not completely lled. Recent studies [14] have shown that using
irregular subarrays can result in a signicant reduction
of the quantization lobes without sacricing aperture efciency.
The classic method of eliminating the subarray quantization lobes is to produce a subarray pattern unlike the
sinc function of Figure 14, but instead one with a pulse
shape chosen so that the main beam would be within the
subarray pattern envelope for all frequencies within the
band f, but the quantization lobes outside of the pulse
shaped pattern would be suppressed.
For an array with subarrays spaced Dx apart, and assuming a very large array so that the array beamwidth is
much less than the subarray beamwidth, the idealized subarray pattern is shown in Figure 15 as a pulse (in u-space)
with its center at the scan angle u0 , and pass band from
0.5 min
0.5 min
min
u0
to u0 +
, a width of u =
.
Dx
Dx
Dx
With a subarray pattern of this width, scanned to u0 , the
nearest quantization lobes would be at u = u0 = min /Dx
and would be outside of the subarray pattern for all scan
Figure 15. Ideal subarray pattern for time delay steered arrays.
angles u0 . The accompanying bandwidth is given by equation 9.
There are several ways to synthesize this type of subarray pattern, depending upon the desired bandwidth. Since
the at-topped type subarray pattern is a pulse function,
the feed illumination to produce that radiation must be a
sinc function that spans the whole array. Figure 15 shows
that this required illlunination for each subarray overlaps
all the other subarray illuminations (only two shown for
clarity).
Approximations to this ideal subarray excitation have
been invented by a number of authors including Mailloux
[15], Dufort [16] and others. Most recently Skobelev [17]
Antenna Arrays
13
Figure 16. Chess network of Skobelev
[17] (a) N-cascade (chess) network (b) Pattern of networks with N cascades for N =
1,2,3,4 (curves displaced for legibility)
Figure 17. Overlapped subarray characteristics
for lens-fed time delayed system.
has published a constrained overlapped system that offers
detailed control of sidelobe levels. This network, shown
schematically in Figure 16 is designed with cells of two
elements each, spaced a distance a apart. Adjacent subarrays are therefore two elements apart. The network is
called a chess-network and consists of 2N layers, or N cascades. At each cascade N, the signals are distributed over
a wider aperture, covering 2N+1 cells. As the aperture distribution becomes wider and a better approximation to a
sinc function, the resulting subarray radiation pattern becomes more like a pulse function with sharper skirts and
a atter passband.
The inter-subarray spacing for these types of constrained subarrays is necessarily limited to being rather
small, because large subarrays with many elements would
require extremely complex, and therefore lossy, circuits.
Thus, constrained subarrays are ideal for the wideband
waveforms that accompany small inter-subarray spacings.
At the other extreme, for very large arrays, perhaps tens
to hundreds of thousands of elements, the squint band-
width may only be a percent or so, and then it is important
to use subarrays with spacing as large as possible to reduce
the number of time delays to some affordable number while
increasing the bandwidth to a size appropriate to the radar
or communication system. For such tasks completely overlapped subarrays can be produced by quasi-optical beamformers as shown in Figure 17. This kind of beamformer
was rst introduced as the Hughes Corporation HIPSAF
(High Performance Scanning Array Feed) [18] and comprised of an objective lens fed by a multiple beam feed. The
feed shown in the lower portion of the gure is a small
multiple beam lens or a Butler matrix, or may very likely
be a small digitally beamformed array. Any input to the
multiple beam feed produces a linear progressive phase at
its output, and this feed radiates to the back face of the
objective. Two such ports are shown excited, and result in
radiation patterns shown solid and dashed in the gure.
Each feed produces an approximate sinc function of excitation across the main lens radiating aperture, and so each
radiates like a pulse function in space.
14
Antenna Arrays
Figure 17 also shows some of the geometric constraints
and illustrates that for an equal path lens aperture, the
subarray pattern width is equal to the angle subtended
by the feed array and that the beamwidth f/f0 is given by
/u0 as in equation 9. Thus, for an array with M elements
spaced dx apart, the subarray pattern width is given by
the objective lens size Mdx , divided by the focal length F.
When all of the input ports are excited with a low sidelobe
weight distribution the radiating pattern can have a low
sidelobe radiation over the chosen bandwidth.
These subarray techniques have been developed to reduce system weight and the number of costly time delay
components. Since they reduce the number of time delays,
they are highly compatable with digital beamforming technology.
BIBLIOGRAPHY
1. W. E. Hord Microwave and millimeter wave ferrite phase
shifters, Microwave Journal State of the Art Reference, 32:
(September) 1989.
2. A. K. Sharma Solid state control devices: State of the art, Microwave Journal State of the Art Reference, 32: (September)
1989.
3. C. P. Wu Analysis of nite parallel plate waveguide arrays,
IEEE Trans. Antennas Propag., AP-18 (3): 328334, 1970.
4. P. M. Woodward A method of calculating the eld over a plane
aperture required to produce a given polar diagram, Proc. IEE
(London), Part 3A, 93: 15541555, 1947.
5. J. Butler R. Loe Beamforming matrix simplies design of electronically scanned antennas, Electronic Design, 9: 170173,
(12 April) 1961.
6. S. A. Schelkunov A mathematical theory of linear arrays, Bell
System Tech J., 80107, 1943.
7. C. L. Dolph A current distribution for broadside arrays which
optimizes the relationship between beamwidth and sidelobe
level, Proc. IRE, 34(June): 335345, 1946.
8. T. T. Taylor Design of line source antennas for narrow
beamwidth and low sidelobes, IEEE Trans. Antennas Propag.,
AP-3: 1628, (January) 1955.
9. E. T. Bayliss Design of monopulse antenna difference patterns
with low sidelobes, Bell System Tech. J., 47: 623640.
10. R. S. Elliott On discretizing continuous aperture distributions,
IEEE Trans. Antennas Propag., AP-25: 617621, (September)
1977.
11. H. R. Orchard R. S. Elliott G. J. Stern Optimizing the synthesis
of shaped antenna patterns, IEE Proc. (London) Part H (1):
6366, 1984.
12. F. Ares et al. Application of genetic algorithms and simulated
annealing technique in optimising the aperture distributions
of antenna array patterns, Electronic Lett., 32 (3): 148149,
1996.
13. M. K. Skolnik J. W. Sherman III F. C. Ogg, Jr. Statistically Designed Density-Tapered Arrays, IEEE Trans. Antennas Propag., AP-12: 408417, (July) 1964.
14. R. J. Mailloux, S. G. Santorelli and T. M. Roberts, Wideband
Arrays using Irrregular (Polyomino) shaped Subarrays, electronics Letters, Vol.42, No.18,Aug. 2006, pp. 10191020
15. R. J. Mailloux, An Overlapped Subarray for Limited Scan Applications, IEEE Trans. AP-22, No.3,May 1974, pp. 487489
16. E. C. Dufort, Constrained feeds for limited scan arrays, IEEE
Trans. AP-26,May 1978, pp. 407413
17. S. P. Skobelev, Methods of constructing optimum phasedarray antennas for limited eld of view, IEEE Antennas and
Propagation Magazine, Vol. 40, No.2,April 1998, pp. 3949
18. R. Tang, Survey of Time-Delay Steering Techniques, Phased
Array Antennas, pp. 254260, Artech House, Dedham, MA,
1972
ROBERT J. MAILLOUX
University of Massachusetts,
Dept. of Electrical and
Computer Engineering,
Amherst, MA
J. Webster (ed.), Wiley Encyclopedia of Electrical and Electronics Engineering
c 1999 John Wiley & Sons, Inc.
Copyright
ANTENNA RADIATION PATTERNS
An antenna is used to either transmit or receive electromagnetic waves. It serves as a transducer converting
guided waves into free-space waves in the transmitting mode or vice versa in the receiving mode. Antennas,
including aerials, can take many forms according to the radiation mechanism involved and can be divided
into different categories. Some common types are wire antennas, aperture antennas, reflector antennas, lens
antennas, traveling-wave antennas, frequency-independent antennas, horn antennas, printed and conformal
antennas, etc. (see Antennas). When applications require radiation characteristics that cannot be met by
a single radiating element, multiple elements are employed. Various configurations are utilized by suitably
spacing the elements in one or two dimensions. These configurations, known as array antennas, can produce the
desired radiation characteristics by appropriately feeding each individual element with different amplitudes
and phases, which allows increasing the electrical size of the antenna. Furthermore, antenna arrays combined
with signal processing lead to smart antennas (switched-beam or adaptive antennas), which offer more degrees
of freedom in wireless system design (1). Moreover, active antenna elements or arrays incorporate solid-state
components producing effective integrated antenna transmitters or receivers with many applications (see
Antennas and Ref. 1).
Regardless of the antenna considered, there are some fundamental figures of merit that describe its
performance. The response of an antenna as a function of direction is given by the antenna pattern. This
pattern commonly consists of a number of lobes; the largest one is called the main lobe, and the others are
called sidelobes, minor lobes, or back lobes. If the pattern is measured sufficiently far from the antenna so
there is no change in the pattern with distance, the pattern is the so-called far-field pattern. Measurements at
shorter distances yield near-field patterns, which are a function of both angle and distance. The pattern may be
expressed in terms of the field intensity (field pattern) or in terms of the Poynting vector or radiation intensity
(power pattern). If the pattern is symmetrical, a simple pattern is sufficient to completely specify the variation
of the radiation with angle. Otherwise, a three-dimensional diagram or a contour map is required to show the
pattern in its entirety. However, in practice two patterns, perpendicular to each other and to the main-lobe
axis, may suffice. These are called the principal-plane patterns for the E plane and the H plane, containing the
field vectors E and H, respectively.
Having established the radiation patterns of an antenna, some important parameters can now be considered, such as radiated power, radiation efficiency, directivity, gain, and antenna polarization. All of them will
be considered in detail in this article.
Here scalar quantities are presented in lightface italics, while vector quantities are boldface, e.g., the
electric field E (vector) of magnitude E (=|E|) (scalar). Unit vectors are boldface with a circumflex over the
letter; , , and are the unit vectors in the x, y, z, and r directions, respectively. A dot over a symbol means
that the quantity is harmonically time-varying or a phasor. For example, taking the electric field, represents
a space vector and time phasor, but
El ejt .
is a scalar phasor. The relations between them are
, where
ANTENNA RADIATION PATTERNS
The first section of this article introduces several antenna patterns, giving the necessary definitions
and presenting the common types. The field regions of an antenna are also pointed out. The most common
reference antennas are the ideal isotropic radiator and the very short dipole. Their fields are used to show the
calculation and meaning of the different parameters of antennas covered in this article. The second section
begins with a treatment of the Poynting vector and radiation power density, starting from the general case of an
electromagnetic wave and extending the definitions to a radiating antenna. After this, radiation performance
measures such as the beam solid angle, directivity, and gain of an antenna are defined. In the third section
the concepts of wave and antenna polarization are discussed. Finally, in the fourth section, a general case of
antenna pattern calculation is considered, and numerical solutions are suggested for radiation patterns that
are not available in simple closed-form expressions.
Radiation From Antennas
Radiation Patterns. The radiation pattern of an antenna is generally its most basic requirement, since
it determines the spatial distribution of the radiated energy. This is usually the first property of an antenna that
is specified, once the operating frequency has been stated. An antenna radiation pattern, or antenna pattern,
is defined as a graphical representation of the radiation properties of the antenna as a function of space
coordinates. Since antennas are commonly used as parts of wireless telecommunication systems, the radiation
pattern is determined in the far-field region where no change in pattern with distance occurs. Using a spherical
coordinate system, shown in Fig. 1, with the antenna at the origin, the radiation properties of the antenna
depend only on the angles and along a path or surface of constant radius. A plot of the radiated or received
power at a constant radius is called a power pattern, while the spatial variation of the electric or magnetic field
along a constant radius is called the amplitude field pattern. In practice, the necessary information from the
complete three-dimensional pattern of an antenna can be obtained by taking a few two-dimensional patterns,
according to the complexity of radiation pattern of the specific antenna. For most applications, a number of
plots of the pattern as a function of for some particular values of , plus a few plots as a function of for some
particular values of , give the needed information.
Antennas usually behave as reciprocal devices. This is very important, since it permits the characterization of the antenna either as a transmitting or as a receiving antenna. For example, radiation patterns are
often measured with the test antenna operating in the receive mode. If the antenna is reciprocal, the measured
pattern is identical when the antenna is in the transmit mode. If nonreciprocal materials, such as ferrites and
active devices, are not present in an antenna, its transmitting and receiving properties are identical.
The radiation fields from a transmitting antenna vary inversely with distance, whereas the variation
with observation angles (, ) depends on the antenna type. A very simple but basic configuration antenna is
the ideal, or very short, dipole antenna. Since any linear or curved wire antenna may be regarded as being
composed of a number of short dipoles connected in series, knowledge of this antenna is useful. So we will use
the fields radiated from an ideal antenna to define and understand the properties of radiation patterns. An
ideal dipole positioned symmetrically at the origin of the coordinate system and oriented along the z axis is
shown in Fig. 1. The pattern of electromagnetic fields, with wavelength , around a very short wire antenna
of length L << , carrying a uniform current I0 ejt , is described by functions of distance, frequency, and angle.
Table 1 summarizes the expressions for the fields from a very short dipole antenna as given in Refs. 2 and 3.
We have E = H r = H = 0 for r and L . The variables shown in these relations are I0 = amplitude
(peak value in time) of current (A), supposed to be constant along the dipole; L = length of dipole (m); = 2f
= radian frequency, where f is the frequency in hertz; t = time (s), = 2/ = phase constant (rad/m) =
azimuthal angle, (dimensionless); c = velocity of light 3108 m/s; = wavelength (m); j = complex operator
ANTENNA RADIATION PATTERNS
Fig. 1. Spherical coordinate system for antenna analysis. A very short dipole is shown with the directions of its nonzero
field components.
=
distance from center of dipole to observation point, (m) and 0 = permittivity of free space = 8.85
pF/m.
Note that E and H are in time phase in the far field. Thus, the electric and magnetic fields in the far
field of the spherical wave from the dipole are related in the same manner as in a plane traveling wave. Both
ANTENNA RADIATION PATTERNS
Fig. 2. Radiation field pattern of far field from an ideal (very short) dipole: (a) three-dimensional pattern plot, (b) E-plane
radiation-pattern polar plot, and (c) H-plane radiation-pattern polar plot (HPBW, H-plane beamwidth).
are also proportional to sin . That is, both are at a maximum when = 90 and a minimum when = 0 (in the
direction of the dipole axis). This variation of E or H with angle can be presented by a field pattern, shown
in Fig. 2, the length r of the radius vector being proportional to the value of the far field (E or H ) in that
direction from the dipole. The pattern in Fig. 2(a) is the three-dimensional far-field pattern for the ideal dipole,
while the patterns in Fig. 2(b, c) are two-dimensional and represent cross sections of the three-dimensional
pattern, showing the dependence of the fields on the angles and .
All far-field components of a very short dipole are functions of I0 , the dipole current; L/, the dipole length
in wavelengths; 1/r, the distance factor; jej(tr) , the phase factor; and sin , the pattern factor, which gives
ANTENNA RADIATION PATTERNS
the variation of the field with angle. In general, the expression for the field of any antenna will involve these
factors.
For longer antennas with complicated current distribution the field components generally are functions of
the above factors, which are grouped into the element factor and the space factor. The element factor includes
everything except the current distribution along the source, which is the space factor of the antenna. For
example, consider the case of a finite dipole antenna. The field expressions are produced by dividing the
antenna into a number of very short dipoles and summing all their contributions. The element factor is equal
to the field of the very short dipole located at a reference point, while the space factor is a function of the
current distribution along the source, the latter usually described by an integral. The total field of the antenna
is given by the product of the element and space factors. This procedure is known as pattern multiplication.
A similar procedure is also employed in array antennas, which are used when directive characteristics
are needed. The increased electrical size of an array antenna due to the use of more than one radiating element
gives better directivity and special radiation patterns. The total field of an array is determined by the product
of the field of a single element and the array factor of the array antenna. If we use isotropic radiating elements,
the pattern of the array is simply the pattern of the array factor. The array factor is a function of the geometry
of the array and the excitation phase. Thus, changing the number of elements, their geometrical arrangement,
their relative magnitudes, their relative phases, and their spacing, we obtain different patterns. Figure 3 shows
some of characteristic patterns of an array antenna with two isotropic point sources as radiating elements,
using different values of the above quantities, which produce different array factors.
Common Types of Radiation Patterns. An isotropic source or radiator is an ideal antenna that
radiates uniformly in all directions in space. Although no practical source has this property, the concept of the
isotropic radiator is very useful, and it is often used as a reference for expressing the directive properties of
actual antennas. It is worth recalling that the power flux density S at a distance r from an isotropic radiator
is Pt /4r2 , Pt being the transmitted power, since all the transmitted power is evenly distributed on the surface
of a spherical wavefront with radius r. The electric field intensity is calculated as
(using the relation
from electric circuits, power = E2 /, where is the characteristic impedance of free space, 377
).
On the contrary, a directional antenna is one that radiates or receives electromagnetic waves more
effectively in some directions than in others. An example of an antenna with a directional radiation pattern
is that of an ideal or very short dipole, shown in Fig. 2. It is seen that this pattern, which resembles a
doughnut with no hole, is nondirectional in the azimuth plane, which is the xy plane characterized by the set
of relations [f(), = /2], and directional in the elevation plane, which is any orthogonal plane containing
the z axis characterized by [g(), = constant]. This type of directional pattern is called an omnidirectional
pattern and is defined as one having an essentially nondirectional pattern in a given plane, which for this
case is the azimuth plane, and a directional pattern in any orthogonal plane, in this case the elevation plane.
The omnidirectional patternknown also as broadcast-typeis used for many broadcast or communication
services where all directions are to be covered equally well. The horizontal-plane pattern is generally circular,
while the vertical-plane pattern may have some directivity in order to increase the gain.
Other forms of directional patterns are pencil-beam, fan-beam, and shaped-beam patterns. The pencilbeam pattern is a highly directional pattern, which is used when it is desired to obtain maximum gain and
when the radiation pattern is to be concentrated in as narrow an angular sector as possible. The beamwidths
in the two principal planes are essentially equal. The fan-beam pattern is similar to the pencil-beam pattern
except that the beam cross section is elliptical in shape rather than circular. The beamwidth in one plane
may be considerably broader than in the other plane. As with the pencil-beam pattern, the fan-beam pattern
generally implies a rather substantial amount of gain. The shaped-beam pattern is used when the pattern in
one of the principal planes is desired to have a specified type of coverage. A typical example is the cosecant
pattern, which is used to provide a constant radar return over a range of angles in the vertical plane. The
ANTENNA RADIATION PATTERNS
Fig. 3. Three-dimensional graphs of power radiation patterns for an array of two isotropic elements of the same amplitude
and (a) opposite phase, spaced 0.5 apart, (b) phase quadrature, spaced 0.5 apart, (c) opposite phase, spaced 0.25 apart,
and (d) opposite phase, spaced 1.5 apart.
pattern in the other principal plane is usually a pencil-beam type, but may sometimes be circular, as in certain
types of beacon antennas.
In addition to the above pattern types, there are a number of special shapes used for direction finding
and other purposes. These include the well-known figure-of-eight pattern, the cardioid pattern, split-beam
patterns, and multilobed patterns whose lobes are of substantially equal amplitude. For such patterns, it is
generally necessary to specify the pattern by an actual plot of its shape or by a mathematical relationship.
Antennas are often referred to by the type of pattern they produce. Two terms, which usually characterize
array antennas, are broadside and endfire. A broadside antenna is one for which the main beam maximum is
in a direction normal to the plane containing the antenna. An endfire antenna is one for which the main beam
ANTENNA RADIATION PATTERNS
Fig. 4. Polar plots of a linear uniform-amplitude array of five isotropic sources with 0.5-wavelength spacing between the
sources: (a) broadside radiation pattern (0 phase shift between successive elements), and (b) endfire radiation pattern
(180 phase shift).
is in the plane containing the antenna. For example, the short dipole antenna is a broadside antenna. Figure 4
shows two cases of broadside and endfire radiation patterns, which are produced from a linear uniform array
of isotropic sources of 0.5-wavelength spacing between adjacent elements. The type of radiation pattern is
controlled by the choice of phase shift between the elements. Zero phase shift produces a broadside pattern,
and 180 phase shift (for this case where the spacing between adjacent element is 0.5 ) leads to an endfire
pattern, while intermediate values produce radiation patterns with the main lobes between these two cases.
Characteristics of simple patterns. For a linearly polarized antenna, such as a very short dipole
antenna, performance is often described in terms of two patterns [Fig. 2(b, c)]. Any plane containing the z axis
has the same radiation pattern, since there is no variation in the fields with angle [Fig. 2(b)]. A pattern taken
in one of these planes is called an E-plane pattern, because it is parallel to the electric field vector E and passes
through the antenna in the direction of the beam maximum. A pattern taken in a plane orthogonal to an E
plane and cutting through the short dipole antenna (the xy plane in this case) is called an H-plane pattern,
because it contains the magnetic field H and also passes through the antenna in the direction of the beam
maximum [Fig. 2(c)]. The E- and H-plane patterns, in general, are referred to as the principal-plane patterns.
The pattern plots in Fig. 2(b, c) are called polar patterns or polar diagrams. For most types of antennas it
is a usual practice to orient them so that at least one of the principal-plane patterns coincides with one of
the geometrical principal planes. An illustration is shown in Fig. 5, where the principal planes of a microstrip
antenna are plotted. The xy plane (azimuthal plane, = /2) is the principal E plane, and the xz plane (elevation
plane, = 0) is the principal H plane.
A typical antenna power pattern is shown in Fig. 6. In Fig. 6(a), a polar plot on a linear scale is depicted,
and in Fig. 6(b), the same pattern is shown in rectangular coordinates in decibels. As can be seen, the radiation
pattern of the antenna consists of various parts, which are known as lobes. The main lobe (or main beam or
major lobe) is defined as the lobe containing the direction of maximum radiation. In Fig. 6(a) the main lobe is
pointing in the = 0 direction. In some antennas there may exist more than one major lobe. A minor lobe is
any lobe except the main lobe. Minor lobes comprise sidelobes and back lobes. The term sidelobe is sometimes
reserved for those minor lobes near the main lobe, but is most often taken to be synonymous with minor lobe.
A back lobe is a radiation lobe in, approximately, the opposite direction to the main lobe. Minor lobes usually
ANTENNA RADIATION PATTERNS
Fig. 5. The principal-plane patterns of a microstrip antenna: the xy plane or E plane (azimuth plane, = /2), and the xz
plane or H plane (elevation plane, = 0).
represent radiation in undesired directions, and they should be minimized. Sidelobes are normally the largest
of the minor lobes. The level of side or minor lobes is usually expressed as a ratio of the power density in the
lobe in question to that of the main lobe. This ratio is often termed the sidelobe ratio or sidelobe level and
desired values of it depend on the antenna application.
For antennas with simple patterns, the half-power beamwidth and the sidelobe level in the two principal
planes specify the important characteristics of the patterns. The half-power beamwidth (HPBW) is defined in
a plane containing the major maximum beam, as the angular width within which the radiation intensity is
at least one-half the maximum value for the beam. The beamwidth between first nulls (BWFN) and the beam
widths 10 dB or 20 dB below the pattern maximum are also sometimes used. All of them are shown in Fig. 6.
However the term beamwidth by itself is usually reserved to describe the half-power (3 dB) beamwidth.
The beamwidth of the antenna is a very important figure of merit in the overall design of an antenna
application. As the beamwidth of the radiation pattern increases, the sidelobe level decreases, and vice versa.
So there is a tradeoff between the sidelobe ratio and beamwidth.
In addition, the beamwidth of the antenna is used to describe the resolution of the antenna: its ability to
distinguish between two adjacent radiating sources or radar targets. The most common measure of resolution
is half the first null beamwidth, which is usually used to approximate the half-power beamwidth. This means
that two sources separated by angular distances equal to or greater than the HPBW of an antenna, with a
uniform distribution, can be resolved. If the separation is smaller, then the antenna will tend to smooth the
two signals into one.
Field Regions of an Antenna. For convenience, the space surrounding a transmitting antenna is
divided into several regions, although, obviously, the boundaries of the regions cannot be sharply defined. The
names given to the various regions denote some pertinent prominent property of each region.
In free space there are mainly two regions surrounding a transmitting antenna, the near-field region and
the far-field region. The near-field region can be subdivided into two regions, the reactive near field and the
radiating near field.
The first and innermost region, which is immediately adjacent to the antenna, is called the reactive or
induction near-field region. Of all the regions, it is the smallest. It derives its name from the reactive field, which
lies close to every current-carrying conductor. In this region the reactive field, which decreases with either the
ANTENNA RADIATION PATTERNS
Fig. 6. Antenna power patterns: (a) a typical polar plot on a linear scale, and (b) a plot in rectangular coordinates on a
decibel (logarithmic) scale. The associated lobes and beamwidths are also shown.
square or the cube of distance, dominates over all radiated fields, the components of which decrease with the
from
first power of distance. For most antennas, this region is taken to extend over distances r < 0.62
the antenna as long as D , where D is the largest dimension of the antenna and is the wavelength (2).
For the case of an ideal or very short dipole, for which D = z , this distance is approximately one-sixth
10
ANTENNA RADIATION PATTERNS
Fig. 7. Field regions of an antenna and some commonly used boundaries.
of a wavelength (/2). At this distance from the very short dipole the reactive and radiation field components
are respectively equal in magnitude.
Between the reactive near-field region and far-field regions lies the radiating near-field region, which is
the region where the radiation fields dominate, but the angular field distribution still depends on the distance
from the antenna. For a transmitting antenna focused at infinity, which means that the rays at a large distance
from the antenna are parallel, the radiating near-field region is sometimes referred to as the Fresnel region, a
term taken from the field of optics. This is taken to be that between the end of reactive near-field region (r >
) and the starting distance of the far-field region (r < 2D2 /)(2).
The outer boundary of the near-field region lies where the reactive field intensity becomes negligible with
respect to the radiation field intensity. This occurs at distances of either a few wavelengths or a few times the
major dimension of the antenna, whichever is the larger. The far-field or radiation region begins at the outer
boundary of the near-field region and extends outward indefinitely into free space. In this region the angular
distribution of the field is essentially independent of the distance from the antenna. For example, for the case
of a very short dipole, the sin pattern dependence is valid anywhere in this region. The far-field region is
commonly taken to be at distances r > 2D2 / from the antenna, and for an antenna focused at infinity it is
sometimes referred to as the Fraunhofer region.
All three regions surrounding an antenna and their boundaries are illustrated in Fig. 7.
0.62
ANTENNA RADIATION PATTERNS
11
Antenna Performance Measures
Poynting Vector and Radiation Power Density. In an electromagnetic wave, energy is stored in
equal amounts in the electric and magnetic fields, which together constitute the wave. The power flow is found
by making use of the Poynting vector, S, defined as
where E (V/m) and H (A/m) are the field vectors. Since the Poynting vector represents a surface power density
(W/m2 ), the integral of its normal component over a closed surface always gives the total power through the
surface. That is,
where P is the total power (W) flowing out of closed surface A, and dA = dA, being the unit vector normal
to surface. The Poynting vector S and the power P in the above relations are instantaneous values.
Normally, it is the time-averaged Poynting vector Sav , which represents the average power density, that
is of practical interest. It is given by
where Re stands for the real part of the complex number and the asterisk denotes the complex conjugate.
in Eq. (3) are the electric and magnetic fields written as complex numbers to include the
Note that and
change with time. That is, for a plane wave traveling in the positive z direction with electric and magnetic
field components in the x and y directions, respectively, the electric field is E = Ex0 ejT while in Eq. (1) it is E
= Ex0 . The factor 12 appears because the fields represent peak values; it should be omitted for rms values.
The average power Pav flowing outward through a closed surface can now be obtained by integrating Eq.
(3):
Consider the case that the electromagnetic wave is radiated by an antenna. If the closed surface is taken
around the antenna within the far-field region, then this integration results in the average power radiated by
the antenna. This is called radiation power, Prad , while Eq. (3) represents the radiation power density, Sav , of
the antenna. The imaginary part of Eq. (3) represents the reactive power density stored in the near field of an
antenna. Since the electromagnetic fields of an antenna in its far-field region are predominately real, Eq. (3) is
enough for our purposes.
The average power density radiated by the antenna as a function of direction, taken on a large sphere of
constant radius in the far-field region, results in the power pattern of the antenna.
As an example, for an isotropic radiator, the total radiation power is given by
12
ANTENNA RADIATION PATTERNS
Here, because of symmetry, the Poynting vector Si = Si (r) is taken independent of the spherical coordinate
angles and , having only a radial component.
From Eq. (5) the power density can be found:
The above result can also be reached if we assume that the radiated power expands radially in all
directions with the same velocity and is evenly distributed on the surface of a spherical wavefront of radius r.
As we will see later, an electromagnetic wave may have an electric field consisting of two orthogonal linear
components of different amplitudes, Ex0 and Ey0 , respectively, and a phase angle between them, . Thus, the
total electric field vector, called an elliptically polarized vector, becomes
which at z = 0 becomes
So is a complex vector (phasorvector), which is resolvable into two components
vector associated with , at z = 0, is then
and
y.
The total
where is the phase lag of y with respect to x . From Eq. (9) the complex conjugate magnetic field can be
found changing only the signs of exponents.
Now the average Poynting vector can be calculated using the above fields:
It should be noted that Sav is independent of , the phase angle between the electric field components.
In a lossless medium = 0, because the electric and magnetic fields are in time phase and Ex0 /H x0 = Ey0 /H y0
= , where is the intrinsic impedance of the medium, which is real. If E =
and H =
ANTENNA RADIATION PATTERNS
13
are the amplitudes of the total E and H fields respectively, then
The above expressions are the most general form of radiation power density of an elliptically polarized wave
or of an elliptically polarized antenna, respectively, and hold for all cases, including the linear and circular
polarization cases, that we will introduce later on.
Radiation Intensity. Radiation intensity is a far-field parameter, in terms of which any antenna radiation power pattern can be determined. Thus, the antenna power pattern, as a function of angle, can be
expressed in terms of its radiation intensity as (2,3)
where
U(, ) = radiation intensity (W/unit solid angle)
Sav = radiation density, or radial component of Poynting vector (W/m2 )
E(r,,) = total transverse electric field (V/m)
H(r,,) = total transverse magnetic field (A/m)
r = distance from antenna to point of measurement (m)
= intrinsic impedance of medium (
/square)
In Eq. (12) the electric and magnetic fields are expressed in spherical coordinates.
What makes radiation intensity important is that it is independent of distance. This is because in the far
field the Poynting vector is entirely radial, which means the fields are entirely transverse and E and H vary
as 1/r.
Since the radiation intensity is a function of angle, it can also be defined as the power radiated from an
antenna per unit solid angle. The unit of solid angle is the steradian, defined as the solid angle with its vertex
at the center of a sphere of radius r that is subtended by an area on the sphere equal to r2 . But the area of a
sphere of radius r is given by A = 4r2 , so in the whole sphere there are 4r2 /r2 = 4 sr. For a sphere of radius
r, an infinitesimal area dA on the surface of it can be written as
14
ANTENNA RADIATION PATTERNS
and therefore the element of solid angle d
of a sphere is given by
Thus, the total power can be obtained by integrating the radiation intensity, as given by Eq. (12), over the
entire solid angle of 4 as
As an example, for the isotropic radiator ideal antenna, the radiation intensity U(,) will be independent
of the angles and , and the total radiated power will be
or U i = Prad /4, which is the power density of Eq. (6) multiplied by r2 .
Dividing U(,) by its maximum value U max (,), we obtain the normalized antenna power pattern,
A term associated with the normalized power pattern is the beam solid angle
A defined as the solid
angle through which all the power from a radiating antenna would flow if the power per unit solid angle were
constant over that solid angle and equal to its maximum value (Fig. 8). This means that, for typical patterns,
the beam solid angle is approximately equal to the half-power beamwidth (HPBW), that is,
If the integration is done over the main lobe, the main-lobe solid angle,
M , results, and the difference of
A
M gives the minor-lobe solid angle. These definitions hold for patterns with clearly defined lobes. The
beam efficiency (BE) of an antenna is defined as the ratio
M /
A and is a measure of the amount of power in the
major lobe compared to the total power. A high beam efficiency means that most of the power is concentrated
in the major lobe and that minor lobes are minimized.
Directivity and Gain. A very important antenna parameter, which indicates how well an antenna
concentrates power into a limited solid angle, is its directivity D, defined as the ratio of the maximum radiation
intensity to the radiation intensity averaged over all directions. The average radiation intensity is calculated
ANTENNA RADIATION PATTERNS
15
Fig. 8. Power pattern and beam solid angle of an antenna.
by dividing the total power radiated by 4 sr. Hence,
since from Eq. (16), Prad /4 = U i . So, alternatively, the directivity of an antenna can be defined as the ratio of
its radiation intensity in a given direction (which usually is taken to be the direction of maximum radiation
intensity) to the radiation intensity of an isotropic source with the same total radiation intensity. Equation
16
ANTENNA RADIATION PATTERNS
(19) can also be written
Thus, the directivity of an antenna is equal to the solid angle of a sphere, which is 4 sr, divided by
the antenna beam solid angle
A . We can say that by this relation the value of directivity is derived from
the antenna pattern. It is obvious from this relation that the smaller the beam solid angle, the larger the
directivity, or, stated in a different way, an antenna that concentrates its power in a narrow main lobe has a
large directivity.
Obviously, the directivity of an isotropic antenna is unity. By definition an isotropic source radiates equally
in all directions. If we use Eq. (20),
A = 4, since U n (, ) = 1. This is the smallest directivity value that one
can attain. However, if we consider the directivity in a specified direction, for example D(, ), its value can be
smaller than unity.
As an example let us calculate the directivity of the very short dipole. We can calculate its normalized
radiated power using the electric or the magnetic field components, given in Table 1. Using the electric field E
for the far-field region, from Eq. (12) we have
and
Alternatively, we can work using power densities instead of power intensities. The power flowing in a
particular direction can be calculated using Eq. (3) and the electric and magnetic far-field components given in
Table 1:
ANTENNA RADIATION PATTERNS
17
By integrating over all angles the total power flowing outwards is seen to be
The directivity is the ratio of the maximum power density to the average power density. For the very short
dipole antenna, the maximum power density is in the = 90 direction (Fig. 2), and the average power density
is found by averaging the total power PT from Eq. (24) over a sphere of surface area 4 r2 . So
Thus, the directivity of a very short dipole is 1.5, which means that the maximum radiation intensity is
1.5 times the power of the isotropic radiator. This is often expressed in decibels:
Here, we use a lowercase letter for the absolute value and a capital letter for the logarithmic value of the
directivity, as is a common in the field of antennas and propagation.
The gain of an antenna is another basic property for its characterization. Gain is closely associated with
directivity, which is dependent upon the radiation patterns of an antenna. The gain is commonly defined as the
ratio of the maximum radiation intensity in a given direction to the maximum radiation intensity produced in
the same direction from a reference antenna with the same power input. Any convenient type of antenna may
be taken as the reference. Many times the type of the reference antenna is dictated by the application area, but
the most commonly used one is the isotropic radiator, the hypothetical lossless antenna with uniform radiation
intensity in all directions. So
where the radiation intensity of the reference antenna (isotropic radiator) is equal to the power in the input,
Pin , of the antenna divided by 4.
Real antennas are not lossless, which means that if they accept an input a power Pin , the radiated power
Prad generally will be less than Pin . The antenna efficiency k is defined as the ratio of these two powers:
where Rr is the radiation resistance of the antenna. Rr is defined as an equivalent resistance in which the same
current as that flowing at the antenna terminals would produce power equal to that produced by the antenna.
Rloss is the loss resistance, which allows for any heat loss due to the finite conductivity of the materials used to
construct the antenna or due to the dielectric structure of the antenna. So, for a real antenna with losses, its
radiation intensity at a given direction U(, ) will be
where U 0 (, ) is the radiation intensity of the same antenna with no losses.
18
ANTENNA RADIATION PATTERNS
Using Eq. (29) in Eq. (27) yields the expression for the gain in terms of the antenna directivity:
Thus, the gain of an antenna over a lossless isotropic radiator equals its directivity if the antenna efficiency
is k = 1, and it is less than the directivity if k < 1.
The values of gain range between zero and infinity, while those of directivity range between unity and
infinity. However, while the directivity can be computed from either theoretical considerations or measured
radiation patterns, the gain of an antenna is almost always determined by a direct comparison of measurement
against a reference, usually a standard-gain antenna.
Gain is expressed also in decibels:
where, as in Eq. (26), lowercase and capital letters mean absolute and logarithmic values, respectively. The
reference antenna used is sometimes declared in a subscript; for example, dBi means decibels over isotropic.
Polarization
Wave and antenna polarization. Polarization refers to the vector orientation of the radiated waves
in space. As is known, the direction of oscillation of an electric field is always perpendicular to the direction of
propagation. For an electromagnetic wave, if its electric field oscillation occurs only within a plane containing
the direction of propagation, it is called linearly polarized or plane-polarized. This is because the locus of
oscillation of the electric field vector within a plane perpendicular to the direction of propagation forms a
straight line. On the other hand, when the locus of the tip of an electric field vector forms an ellipse or a circle,
the electromagnetic wave is called an elliptically polarized or circularly polarized wave.
The decision to label polarization orientation according to the electric intensity is not as arbitrary as it
seems; its causes the direction of polarization to be the same as the direction of the antenna. Thus, vertical
antennas radiate vertically polarized waves, and horizontal antennas radiate horizontally polarized waves.
There has been a tendency, over the years, to transfer the label to the antenna itself. Thus people often refer
to antennas as vertically or horizontally polarized, whereas it is only their radiations that are so polarized.
It is a characteristic of antennas that the radiation they emit is polarized. These polarized waves are
deterministic, which means that the field quantities are definite functions of time and position. On the other
hand, other forms of radiation, for example light emitted by incoherent sources, such as the sun or light
bulbs, has a random arrangement of field vectors and is said to be randomly polarized or unpolarized. In
this case the field quantities are completely random and the components of the electric field are uncorrelated.
In many situations the waves may be partially polarized. In fact, this case can be seen as the most general
situation of wave polarization; a wave is partially polarized when it may be considered to be of two parts, one
completely polarized and the other completely unpolarized. Since we are mainly interested in waves radiated
from antennas, we consider only polarized waves.
Linear, Circular, and Elliptical Polarization. Consider a plane wave traveling in the positive z direction, with the electric field at all times in the x direction, as in Fig. 9(a). This wave is said to be linearly
ANTENNA RADIATION PATTERNS
19
Fig. 9. Polarization of a wave: (a) linear, (b) circular, and (c) elliptical.
polarized (in the x direction), and its electric field as a function of time and position can be described by
In general the electric field of a wave traveling in the z direction may have both an x and a y component,
as shown in Fig. 9(b, c). If the two components Ex and Ey are of equal amplitude, the total electric field at a
fixed value of z rotates as a function of time, with the tip of the vector forming a circular trace, and the wave is
said to be circularly polarized [Fig. 9(b)].
Generally, the wave consists of two electric field components, Ex and Ey , of different amplitude ratios
and relative phases. (Obviously, there are also magnetic fields, not shown in Fig. 9 to avoid confusion, with
amplitudes proportional to and in phase with Ex and Ey , but orthogonal to the corresponding electric field
vectors.) In this general situation, at a fixed value of z the resultant electric vector rotates as a function of time,
the tip of the vector describing an ellipse, which is called the polarization ellipse, and the wave is said to be
elliptically polarized [Fig. 9(c)]. The polarization ellipse may have any orientation, which is determined by its
tilt angle, as shown in Fig. 10; the ratio of the major to the minor axis of the polarization ellipse is called the
axial ratio (AR). Since the two cases of linear and circular polarization, can be seen as two particular cases
of elliptical polarization, we will analyze the latter. Thus, for a wave traveling in the positive z direction, the
electric field components in the x and y directions are
20
ANTENNA RADIATION PATTERNS
Fig. 10. Polarization ellipse at z = 0 of an elliptically polarized electromagnetic wave.
where Ex0 and Ey0 are the amplitudes in the x and y directions, respectively, and is the time-phase angle
between them. The total instantaneous vector field E is
At z = 0, we have Ex = Ex0 sin t and Ey = Ey0 sin (t + ). The expansion of Ey gives
Using the relation for Ex , we obtain sin t = Ex /E1 and cos t =
(36) eliminates t, giving after rearranging
If we define
. Introduction of these into Eq.
ANTENNA RADIATION PATTERNS
21
Eq. (37) takes the form
which is the equation of an ellipse, the polarization ellipse shown in Fig. 10. The line segment OA is the
semimajor axis, and the line segment OB is the semiminor axis. The tilt angle of the ellipse is . The axial ratio
is
From this general case, the cases of linear and circular polarization can be found. Thus, if there is only
Ex (Ey0 = 0), the wave is linearly polarized in the x direction, and if there is only Ey (Ex0 = 0), the wave is
linearly polarized in the y direction. When both Ex and Ey exist, for linear polarization they must be in phase
or antiphase with each other. In general, the necessary condition for linear polarization is that the time-phase
difference between the two components must be a multiple of . If = 0, , 2, . . . and Ex0 = Ey0 , the wave is
linearly polarized but in a plane at an angle of /4 with respect to the x axis ( = /4). If the ratio of the
amplitudes Ex0 and Ey0 is different, then the tilt angle will also be different.
If Ex0 = Ey0 and = /2, the wave is circularly polarized. Generally, circular polarization can be achieved
only when the magnitudes of the two components are the same and the time-phase angle between them is an
odd multiple of /2.
Consider the case that = /2. Taking z = 0, from Eq. (33), (34), and (35) at t = 0 one has E = Ey0 ,
and one-quarter cycle later, at t = /2, one has E = Ex0 . Thus, at a fixed position (z = 0) the electric field
vector rotates with time, tracing a circle. The sense of rotation, also referred to as the sense of polarization, can
be defined by the sense of rotation of the wave as it is observed along the direction of propagation. Thus the
above wave rotates clockwise if it is observed looking towards the source (viewing the wave approaching) or
counterclockwise if it is observed looking away from the source (viewing the wave moving away). Thus, unless
the wave direction is specified, there is ambiguity. The most generally accepted notation is that of the IEEE,
by which the sense of rotation is always taken as that with the wave it traveling away from the observer. If the
rotation is clockwise, the wave is right-handed or clockwise circularly polarized (RH or CW). If the rotation is
counterclockwise, the wave is left-handed or counterclockwise circularly polarized (LH or CCW). Yet another
way to define the polarization is with the aid of helical-beam antennas. A right-handed helical-beam antenna
radiates (or receives) right-handed waves regardless of the position from which it is viewed, while a left-handed
one radiates right-handed waves.
Although linear and circular polarizations can be seen as special cases of elliptical, usually, in practice,
elliptical polarization refers to other than linear or circular. A wave is characterized as elliptically polarized
if the tip of its electric vector forms an ellipse. For a wave to be elliptically polarized, its electric field must
have two orthogonal linearly polarized components, Ex0 and Ey0 . If the two components are not of the same
magnitude, the time-phase angle between them must not be 0 or a multiple of , while in the case of equal
magnitude, the angle must not be an odd multiple of /2. Thus, a wave that is not linearly or circular polarized
is elliptically polarized. The sense of its rotation is determined according to the same rule as for circular
polarization. So a wave is right-handed or clockwise elliptically polarized (RH or CW) if the rotation of its
electric field is clockwise, and it is left-handed or counterclockwise elliptically polarized (LH or CCW) if the
electric field vector rotates counterclockwise.
In addition to the sense of rotation, elliptically polarized waves are characterized by their axial ratio
AR and their tilt angle . The tilt angle is used to identify the spatial orientation of the ellipse and can
22
ANTENNA RADIATION PATTERNS
Fig. 11. Polarization states of an electromagnetic wave represented with the aid of the Poincare sphere: (a) one octant of
the Poincare sphere with polarization states, (b) the full range of polarization states in rectangular projection.
be measured counterclockwise or clockwise from the reference direction (Fig. 10). If the electric field of an
elliptically polarized wave has two components of different magnitude with a time-phase angle between them
an odd multiple of /2, the polarization ellipse will not be tilted. Its position will be aligned with the principal
axes of the field components, so that the major axis of the ellipse will be aligned with the axis of the larger field
component and the minor axis with the smaller one.
The Poincare Sphere and Antenna Polarization Characteristics. The polarization of a wave can
be represented and visualized with the aid of a Poincare sphere. The polarization state is described by a
point on this sphere, whose longitude and latitude are related to parameters of the polarization ellipse. Each
point represents a unique polarization state. On the Poincare sphere the north pole represents left circular
polarization, the south pole right circular polarization, and the points along the equator linear polarization of
different tilt angles. All other points on the sphere represent elliptical polarization states. One octant of the
Poincare sphere with polarization states is shown in Figure 11(a), while the full range of polarization states is
shown in Figure 11(b), which presents a rectangular projection of the Poincare sphere.
The polarization state described by a point on Poincare sphere can be expressed in terms of:
(1) The longitude L and latitude of the point, which are related to the parameters of the polarization ellipse
by
where is the tilt angle with values 0 and = cot 1 ( AR) with values /4 +/4. The axial
ratio AR is negative and positive for right- and left-handed polarization respectively.
ANTENNA RADIATION PATTERNS
23
Fig. 12. One octant of the Poincare sphere, showing the relations of the angles , , , and that can be used to describe
a polarization state.
(2) The angle subtended by the great circle drawn from a reference point on the equator, and the angle between
the great circle and the equator:
where = tan 1 (Ey0 /Ex0 ) with 0 /2 and the time-phase difference between the components of the
electric field ( +).
All the above quantities , , , and , are interrelated by trigonometric formulae (4), and knowing , one
can determine , and vice versa. As a result, the polarization state can be described by either of these two
sets of angles. The geometric relation between these angles is shown in Fig. 12.
The polarization state of an antenna is defined as the polarization state of the wave radiated by the
antenna when it is transmitting. It is characterized by the axial ratio AR, the sense of rotation, and the tilt
angle, which identifies the spatial orientation of the ellipse. However, care is needed in the characterization
of the polarization of a receiving antenna. If the receiving antenna has a polarization that is different from
that of the incident wave, a polarization mismatch occurs. In this case the amount of power extracted by the
receiving antenna from the incident wave will be lower than the expected value, because of the polarization
loss. A figure of merit, which can be used as a measure of polarization mismatch, is the polarization loss factor
(PLF). It is defined as the cosine squared of the angle between the polarization states of the antenna in its
transmitting mode and the incoming wave. Another quantity that can be used to describe the relation between
the polarization characteristics of an antenna and an incoming wave is the polarization efficiency, also known
as loss factor or polarization mismatch. It is defined as the ratio of the power received by an antenna from a
given plane wave of arbitrary polarization to the power that would be received by the same antenna from a
plane wave of the same power flux density and direction of propagation, whose state of polarization has been
adjusted for maximum received power.
In general an antenna is designed for a specific polarization. This is the desired polarization and is called
the copolarization or normal polarization, while the undesired polarization, usually taken orthogonal to the
desired one, is known as the cross polarization or opposite polarization. The latter can be due to a change
24
ANTENNA RADIATION PATTERNS
of polarization characteristics during the propagation of waves, which is known as polarization rotation. In
general an actual antenna does not completely discriminate against a cross-polarized wave, due to engineering
and structural restrictions. The directivity pattern obtained over the entire direction on a representative
plane for cross polarization with respect to the maximum directivity for normal polarization is called antenna
cross-polarization discrimination, and it is an important factor in determining the antenna performance.
The polarization pattern gives the polarization characteristics of an antenna and is the spatial distribution
of the polarization of the electric field vector radiated by the antenna over its radiation sphere. The description
of the polarizations is accomplished by specifying reference lines, which are used to measure the tilt angles of
polarization ellipses, or the directions of polarization for the case of linear polarization.
Evaluation Of Antenna Pattern And Directivity (General Case)
Derivation of Electromagnetic Fields. As already pointed out, the radiation pattern of an antenna is
generally its most basic property and it is usually the first requirement to be specified. Of course, the patterns
of an antenna can be measured in the transmitting or receiving mode; in most cases one selects the receiving
mode if the antenna is reciprocal. But to find the radiation patterns analytically, we have to evaluate the fields
radiated from the antenna.
In radiation problems, the case where the sources are known and the fields radiated from these sources
are required is characterized as an analysis problem. It is a very common practice during the analysis process
to introduce auxiliary functions that will aid in the solution of the problem. These functions are known as vector
potentials, and for radiation problems the most widely used ones are the magnetic vector potential A and the
electric vector potential, F. Although it is possible to calculate the electromagnetic fields E and H directly from
the source current densities the electric current J and magnetic current M, it is simpler to calculate A and F
first and then evaluate the fields. The vector potential A is used for the evaluation of the electromagnetic field
generated by a known harmonic electric current density J. The vector potential F can give the fields generated
by a harmonic magnetic current, which, although physically unrealizable, has applications in some cases, as in
volume or surface equivalence theorems. Here, we restrict ourselves to the use of the magnetic vector potential
A, which is the potential that gives the fields for the most common wire antennas.
Using the appropriate equations from electromagnetic theory, one can express the vector potential A as
(2)
where k2 = 2 0 0 , with 0 and 0 the magnetic permeability and electric permittivity of the air, respectively,
and is the radian frequency. r is the distance from any point in the source to the observation point. The fields
are then given by
and
ANTENNA RADIATION PATTERNS
25
where in 41a the scalar function V represents an arbitrary electric scalar potential, which is a function
of position. The fields radiated by antennas with finite dimensions are spherical waves, and in the far-field
region, the electric and magnetic field components are orthogonal to each other and form a TEM (transverse
electromagnetic mode) wave. Thus in the far field region, Eq. (41a) simplify to
and
So the problem becomes that of first evaluating the function A from the specified electric current density
on the antenna; then, using Eq. (42a), the E and H fields are evaluated and the radiation pattern extracted.
For example, for the case of a very short dipole the magnetic vector potential A is given by
Using Eq. (43), the fields shown in Table 1 can be evaluated.
Numerical Calculation of Directivity. Usually, the directivity of a practical antenna is easiest to
evaluate from its radiation pattern using numerical methods. This is especially true when the radiation patterns
are so complex that closed-form mathematical expressions are not available. Even when such expressions
exist, because of their complicated form the necessary integration to find the radiated power is very difficult
to perform. A numerical method of integration, like the Simpson or trapezoidal rule, can greatly simplify the
evaluation of radiated power and yield the directivity, leading in this way to a method of general application
that needs only a function or a matrix supplying the values of radiated field. However, in many cases the
evaluation of the integral that gives the radiated power, using a series approximation, has proven to give the
correct value of the directivity.
Consider the case where the radiation intensity of a given antenna can be written in the following form:
which means that it is separable into two functions, each being a function of one variable only, and A is a
constant. Then Prad from Eq. (15) will be
26
ANTENNA RADIATION PATTERNS
If we take N equal divisions over the interval of the variable , and M equal divisions over the interval
2 of the variable , the two integrals can be calculated by a series approximation, respectively:
and
Introducing Eq. (46a) into Eq. (45), we obtain
A computer program can easily evaluate the above equation. The directivity then is given by Eq. (19),
which is repeated here:
In the case that and variations are not separable, Prad can also be calculated by a computer program
using a slightly different expression,
where we consider that in this case U (,) = BF(,).
For more information about radiation patterns in general and radiation patterns of specific antennas, the
reader should consult Refs. 1 and 5,6,7,8,9,10,11.
BIBLIOGRAPHY
1.
2.
3.
4.
5.
6.
7.
8.
S. Drabowitch et al. Modern Antennas, London: Chapman & Hall, 1998.
C. A. Balanis Antenna Theory, Analysis and Design, New York: Wiley, 1997.
J. D. Kraus Antennas, New York: McGraw-Hill, 1988.
J. D. Kraus K. R. Carver Electromagnetics, New York: McGraw-Hill, 1973.
W. L. Stutzman G. A. Thiele Antenna Theory and Design, New York: Wiley, 1981.
W. L. Weeks Antenna Engineering, New York: McGraw-Hill, 1968.
S. A. Schelknunoff H. T. Friis Antenna Theory and Practice, New York: Wiley, 1952.
E. Jordan K. Balmain Electromagnetic Waves and Radiating Systems, New York: Prentice-Hall, 1968.
ANTENNA RADIATION PATTERNS
27
9. T. A. Milligan Modern Antenna Design, New York: McGraw-Hill, 1985.
10. R. C. Johnson H. Jasik (ed.) Antenna Engineering Handbook, New York: McGraw-Hill, 1993.
11. Y. T. Lo S. W. Lee (ed.) Antenna Handbook: Theory, Applications and Design, New York: Van Nostrand Reinhold, 1988.
MICHAEL CHRYSSOMALLIS
Democritus University of Thrace
CHRISTOS CHRISTODOULOU
The University of New Mexico
ANTENNA THEORY
ANTENNA THEORY
FUNDAMENTALS
Maxwells Equations
Antenna properties are analyzed with basic laws of physics.
These laws have been collected into a set of equations commonly referred to as Maxwells equations. (The presentation
in this section follows the textbook by Stutzman and Thiele
(1) where a more detailed treat may be found.) In the time
domain, these equations are
E =
B
B
M
t
(1)
H =
D
D
+T
t
(2)
D = (t)
(3)
If multiple frequencies are present, the solution to the equations may be found for each frequency separately and the results combined for the total solution. The linearity restriction
was only to ensure that the analysis would be properly restricted to a single frequency. For nonlinear media and some
complex problems, it is advantageous to solve the time-domain equations and obtain the frequency-domain form
through a Fourier (or Laplace) transform process. Computationally, the Fourier transform is usually obtained using a
fast Fourier transform (FFT).
Maxwells equations define relationships between the field
quantities, but do not explicitly provide information about the
media in which these fields exist. The material is characterized by three quantities: permittivity , permeability , and
conductivity . Sometimes the material conductivity is given
in inverse form as the resistivity 1/ . These quantities
relate the density and intensity quantities as well as the portion of the current due to conduction. Thus we have D E,
B H, and J E that lead to
and
B =m
(4)
The cross and dot derivatives are referred to as the curl and
divergence respectively. A supplementary equation that may
be deduced from the second and third equations is
T =
(t)
t
(5)
and is denoted the continuity equation to explicitly describe
the electric current density T in terms of the movement of
volumetric electric charge, . A similar relationship holds for
the magnetic current density M and volumetric magnetic
charge, m. These latter two quantities have not been identified to date as actual physical quantities, but are found to be
extremely useful in analysis. In fact, the concept of magnetic
current is identical to the concept of ideal voltage sources in
electrical networks. The remaining quantities, E , H , D , and
B , describe the physical terms of electric and magnetic field
intensities and the electric and magnetic field densities respectively.
In most antenna applications, we analyze sinusoidally varying sources in a linear environment. For such time-harmonic
fields with a radian frequency of , we use the phasor form of
the fields which may be written in the form
E e jt )
E = (E
(6)
to obtain the phasor form of Maxwells equations and the continuity equation as
E = jB
BM
(7)
D +J
H = jD
(8)
D =
(9)
B =m
(10)
J = j
(11)
and
595
H M)
E = jH
(12)
E + Ji
H = ( + j)E
E = i
j
(13)
H = (m)
(15)
J i = ji
(16)
(14)
and
where the i-subscript denotes the impressed sources in the
system, equivalent to the independent sources of circuit theory. We find the simple media description limited in two
ways in the last equations: (1) the medium is described by
scalar quantities, implying isotropic media, and (2) the material parameters have been extracted from the derivatives, implying a constant, homogeneous media. These simplifications
are valid for a large portion of antenna problems and the generalization is left for specific situations. It should be noted
that Eqs. (14) and (15) can be obtained from Eqs. (13) and
(12), respectively, with the appropriate continuity relations,
such as Eq. (16).
Wave Equations
Along transmission lines and in the far-field of antennas, the
solution of Maxwells equations are solutions to the wave
equation in source-free regions. The wave equation may be
obtained by eliminating either E or H from Eqs. (12) through
(15) with no impressed sources as
E
(k2 + 2 )
=0
(17)
H
where k [ ( /j)]. The quantity k is referred to as
the propagation constant or wave number and may be written
in terms of the phase and amplitude constants as ( j). In
most antenna problems of interest, it is common to use instead of k since the media is generally lossless. Similar steps
may be taken for the transmission line to give a one-dimensional equation in either the voltage or current.
J. Webster (ed.), Wiley Encyclopedia of Electrical and Electronics Engineering. Copyright # 1999 John Wiley & Sons, Inc.
596
ANTENNA THEORY
The solutions to Eq. (19) may be written in terms of either
traveling or standing waves, with traveling waves being more
common for antenna applications. The traveling wave solution to the electric field has a plane-wave solution form of
E (rr ) = E + e jkk r + E e jkkr
(18)
The corresponding magnetic field is given by
H (rr ) =
1
E + e jkkr E e jkkr ], =
k [E
(19)
for the geometry of Fig. 1. This general form can be specialized to the far-field case for an antenna located near the origin by expanding R r r in a binomial series as
R = |rr r | = r2 2rr r + r2
(24)
r2
(rr r )2
r r
+
= r
+
r
2r
2r3
for r sufficiently small. Only the first term in this expansion,
r, needs to be retained for use in the denominator of Eq. (23).
However, more accuracy is needed for R in the exponential to
account for phase changes; so the second term of the expansion is used in the exponential:
R r r r
The form of Eq. (20) is called a generalized plane wave along
k with the restriction that k E 0, since the divergence is
zero. The more general solution requires additional work and
is not presented here. The direct solution of the differential
forms of Maxwells equations may be obtained analytically in
special cases and numerically in most other cases. Numerical
procedures typically use finite differences (FD), the finite difference-time domain (FDTD) method, or finite-element (FE)
techniques. The alternative is to transform the equations into
integral forms for solution, where the solution structure is
written in integal form and the integrals are used to solve for
the field quantities.
The complete far-field approximation is then
A (rr ) =
e jrr
4r
1
A
(20)
The corresponding electric-field intensity in simple media (using the Lorentz gauge for the potential) is given by
E=
1
[ 2A + A]
j
J (rr )e j r rr dv
(26)
1
k A )]
[ 2A k (k
j
(27)
1
k A]
[k
j
(28)
and
H
H=
which is a familiar Fourier transform representation.
In the far-field where Eq. (26) is applicable, we may approximate the corresponding electric and magnetic fields as
Auxiliary Functions
Auxiliary functions are used to extend the solution of the
wave equation beyond the simple traveling plane-wave form.
If the magnetic sources are zero, then we may expand the
magnetic-flux density in terms of the curl of an auxiliary function, the magnetic vector potential A, or
(25)
The second term in Eq. (27) simply removes the radial portion A from the electric field.
Duality
Duality provides an extremely useful way to complete the development of the solution form as well as equating some
forms of antennas. To complete the previous set of equations
for the magnetic current and charge, we simply note that we
may change the variable definitions to obtain an identical
form of equations. Specifically, we replace
E H
(29a)
E
H E
(29b)
J M
(29c)
A F
(29d)
,
(29e)
(21)
The use of a gauge condition completes the specification of the
degrees of freedom for A. The magnetic-vector potential must
satisfy the Helmholtz equation given by
Source volume v
A = J
J
( 2 + 2 )A
(22)
having a solution in free space of
A (rr ) =
V
R = r r
P, field point
J (rr )
e j |rr rr |
dv
4|rr r |
(23)
Figure 1. Coordinates and geometry for solving radiation problems.
ANTENNA THEORY
The vertical electric current in Fig. 2 fed at the ground plane
is called a monopole; it together with its image form a dipole
and
Im
Im
Sources
PEC
Zmonopole =
Images
Figure 2. Images of elemental electric (I) and magnetic (Im) currents
over a perfect electric ground plane.
1
Z
2 dipole
(32)
Since the corresponding field is radiated into only a halfspace, the gain of the antenna defined as the peak power density in the far-field compared to the average power density
over the radiation region of the antenna is double for the
ground plane-fed antenna as
Gmonopole = 2Gdipole
and
, 1/
ANTENNA CHARACTERISTICS
E=
1
F
[ 2A + A ]
j
(30a)
H=
1
A
[ 2F + F ] +
j
(30b)
and
The alternate use of duality is to equate similar dual problems numerically. A classic problem is the relationship between the input impedance of a slot dipole and strip dipole.
The two structures are planar complements, each filling the
void of the other, and have input impedances which satisfy
2
4
(33)
(29f)
where is the intrinsic impedance of the medium. The solution forms for J and M may be combined for the total solution
as
ZslotZstrip =
597
(31)
This relationship incorporates several equivalencies, but most
importantly the electric and magnetic quantities are scaled
appropriately by to preserve the proper units in the dual
relationship. For a 72 strip dipole, we find the complementary slot dipole has an input impedance of Zslot 493.5 .
Self-complementary planar structures such as spirals provide
an input impedance of 188.5 . A self-complementary structure is its own complement.
Images
Many antennas are constructed above a large metallic structure referred to as a ground plane. As long as the structure is
greater than a half-wavelength in radius, the finite plane may
be moldeled as an infinite structure for all but radiation behind the plane. The advantage of the infinite structure which
is a perfect electric conductor (PEC) is that the planar sheet
may be replaced by the images of the antenna elements in the
plane. For the PEC, the images are constructed to provide a
zero, tangential electric field at the plane. Figure 2 shows the
equivalent current structure for the original and the image
problems.
It is common to feed antennas at the ground plane through
a coaxial cable. Then the equivalent voltage for the imaged
problem is twice that of the source above the ground plane.
There are a number of characteristics that describe an antenna as a device. Characteristics such as impedance and
gain are common to any electrical device. On the other hand,
a property such as radiation pattern is unique to the antenna.
In this section we discuss patterns and impedance. Gain is
discussed in the following section. We begin with a discussion
of reciprocity.
Reciprocity
Circuit Form. Reciprocity plays an important role in antenna theory and can be used to great advantage in calculations and measurements. Fortunately, antennas usually behave as reciprocal devices. This permits characterization of
the antenna as either a transmitting or receiving antenna.
For example, radiation patterns are often measured with the
test antenna operating in the receive mode. If the antenna is
reciprocal, the measured pattern is identical when the antenna is in either a transmit or a receive mode. In fact, the
following general statement applies: If nonreciprocal materials are not present in an antenna, its transmitting and receiving
properties are identical. A case where reciprocity may not hold
is when ferrite or active devices are included as a part of the
antenna.
Reciprocity is also helpful when examining the terminal
behavior of antennas. Consider two antennas, a and b shown
in Fig. 3. Although connected through the intervening medium and not by a direct connection path, we can view this
as a two-port network. Two port circuit analysis permits us
to write the following:
Antenna
a
Va = Zaa Ia + Zab Ib
(34a)
Vb = Zba Ia + Zbb Ib
(34b)
Antenna
b
Figure 3. Two-port device representation for coupling between antennas.
598
ANTENNA THEORY
where Va and Vb are the terminal voltages and Ia and Ib are
the currents of antennas a and b, respectively. Zaa and Zbb are
self impedances and Zab and Zba are mutual impedances. To
illustrate the use of these equations, suppose a generator of
current Ia is placed on antenna a. The open circuit (Ib 0)
voltage at antenna b is then,
Vb = Zba Ia
(35)
Therefore, mutual impedance Zba provides the coupling between a transmitting antenna and a receiving antenna. Reversing the situation by using antenna b as the transmitter
and antenna a as the receiver, leads to
Va = Zab Ib
(36)
far field of the antenna a, then we have the radiated electric
field of antenna a along the point dipole as
J a E b M a H b ] dv
(41)
b E a = [J
V
If we go to the extreme of taking the test antenna to the surface of the problem antenna, then Eq. (41) becomes an equation that may be used for the solution of the currents on the
antenna.
The actual form of reaction is to suggest that the field b
reaction with current a is equal to the field a reaction with
current b, or
Ja E b M a H b ) dv = (J
Jb E a M b H a ) dv
(J
(42)
V
It can be seen from Eqs. (35) and (36) that if the applied currents are the same (Ia Ib I), then reciprocity is satisfied
(i.e., Va Vb) if
Zab = Zba
for reciprocal antennas
written symbolically as
b, a
= a, b
(43)
(37)
Antenna Impedance
If one antenna is rotated, the output voltage as a function of
rotation angle becomes the radiation pattern. Since the coupling mechanism is via mutual impedances Zab and Zba, they
must correspond to the radiation patterns. For example, if
antenna b is rotated in the plane of Fig. 3, the pattern in that
plane is proportional to the output of a receiver connected to
antenna b due to a source of constant power attached to antenna a. For reciprocal antennas Eq. (37) implies the transmitting and receiving patterns for the rotated antenna are
the same.
Another interesting result follows from Eq. (34). The input
impedance of antenna a is
Va
Za =
Ia
for reciprocal antennas
Since Va IZ, we may write
Z=
1
I2
J E b ) dv
(J
(45)
(38)
If antennas a and b are far enough apart, such as in the far
field, Zab Za and the input impedance of antenna a becomes
Za = (Zaa Ia + 0)/Ia = Zaa
Reciprocity may be used to obtain the basic formula for the
input impedance of an antenna. If we define the two problems
for Eq. (42) as (a) the antenna current distribution in the
presence of the antenna structure and (b) the same antenna
current in free space, then we can apply Eq. (42) to obtain
J E b ) dv = (J
J E a ) dv = IVa
(J
(44)
(39)
That is, the input impedance equals the self impedance and
antenna a acts as if it is in free space.
Reaction Theorem. Reciprocity may also be stated in integral form by cross multiplying Maxwells equations by the opposite field for two separate problems, integrating and combining to obtain
E a H b E b H a ] dss = [(J
Ja E b M a H b )
[E
(40)
S
V
J b E a M b H a )] dv
(J
For antenna problems, the surface integral on the left of Eq.
(40) is taken to an infinite radius and the integral becomes
zero for finite antennas. This form is the typical field form
of reciprocity. This form also suggests constructing a second
problem that can be used as an auxiliary form to solve the
original problem. For instance, if Mb 0 and Jb is a point
dipole (or test antenna) of vector unit length lb located in the
Thus, if the current distribution on the antenna is known, or
may be estimated, then Eq. (45) provides a means for computing the antenna impedance Z by integrating the near-field
radiated by the antenna current in free space times the current distribution itself. A common approach to this computation results in the induced-EMF method (2).
Radiation Patterns
The radiation pattern is a description of the angular variation
of radiation level around an antenna. This is perhaps the
most important characteristic of an antenna. In this section
we present definitions associated with patterns and develop
the general procedures for calculating radiation patterns.
Radiation Pattern Basics. A radiation pattern (antenna pattern) is a graphical representation of the radiation (far-field)
properties of an antenna. The radiation fields from a transmitting antenna vary inversely with distance, for example,
1/r. The variation with observation angles (, ), however, depends on the antenna and, in fact, forms the bulk of antenna
investigations.
Radiation patterns can be understood by examining the
ideal dipole. The fields radiated from an ideal dipole are
shown in Fig. 4(a) over the surface of a sphere of radius r
which is in the far field. The length and orientation of the
ANTENNA THEORY
599
sin
H-plane
H
HP = 90
E-plane
x
(b)
(a)
y
z
(c)
(d)
Figure 4. Radiation from an ideal dipole. (a) Field components. (b) E-plane radiation pattern
polar plot. (c) H-plane radiation pattern polar plot. (d) Three-dimensional pattern plot.
field vectors follow from Eq. (30a); they are shown for an instant of time for which the fields are peak. The angular variation of E and H over the sphere is sin . An electric-field
probe antenna moved over the sphere surface and oriented
parallel to E will have an output proportional to sin ; see
Fig. 4(b). Any plane containing the z-axis has the same radiation pattern since there is no variation in the fields. A pattern taken in one of these planes is called an E-plane pattern
because it contains the electric vector. A pattern taken in a
plane perpendicular to an E-plane and cutting through the
test antenna (the xy-plane in this dipole case) is called an Hplane pattern because it contains the magnetic field H . The
E- and H-plane patterns, in general, are referred to as principal plane patterns. The E- and H-plane patterns for the ideal
dipole are shown in Fig. 4(b) and (c). These are polar plots in
which the distance from the origin to the curve is proportional
to the field intensity; they are often called polar patterns or
polar diagrams.
The complete pattern for the ideal dipole is shown in isometric view with a slice removed in Fig. 4(d). This solid polar
radiation pattern resembles a doughnut with no hole. It is
referred to as an omni directional pattern since it is uniform
in the xy-plane. Omni directional antennas are very popular
in ground-based applications with the omni directional plane
horizontal. When encountering new antennas the reader
should attempt to visualize the complete pattern in three dimensions.
Radiation patterns in general can be calculated in a manner similar to that used for the ideal dipole if the current
distribution on the antenna is known. This calculation is done
by first finding the vector potential given in Eq. (26). As a
simple example consider a filament of current along the z-axis
and located near the origin. Many antennas can be modeled
by this line source; straight wire antennas are good examples.
In this case the vector potential has only a z-component and
the vector potential integral is one-dimensional
Az =
I(z )
e j R
dz
4R
(46)
600
ANTENNA THEORY
We can do this because in the far field r is very large compared to the antenna size, so r z z cos . In the phase
term R, we must be more accurate when computing the
distance from points along the line source to the observation
point. The integral Eq. (46) sums the contributions from all
the points along the line source. Although the amplitude of
waves due to each source point is essentially the same, the
phase can be different if the path length differences are a
sizable fraction of a wavelength. We, therefore, include the
first two terms of the series in Eq. (52) for the R in the numerator of Eq. (46) giving
z
P(0,y,z)
R
z z
R r z cos
x
Figure 5. Geometry used for field calculations of a line source along
the z-axis.
Using the far-field approximations Eqs. (53) and (54) in Eq.
(46) yields
where has been used for typical radiation media. Due to the
symmetry of the source, we expect that the radiation fields
will not vary with . This lack of variation is because as the
observer moves around the source such that and z are constant, the appearance of the source remains the same; thus,
its radiation fields are also unchanged. Therefore, for simplicity we will confine the observation point to a fixed in the
yz-plane ( 90) as shown in Fig. 5. Then from Fig. 5 we
see that
y2 = y2 + z 2
(47)
z = r cos
(48)
y = r sin
(49)
Applying the general geometry of Fig. 1 to this case, r
yy zz and r zz lead to R r r yy (z z)z and
then
R=
y2 + (z z )2 = y2 + z2 2zz + (z )2
(50)
Substituting Eqs. (47) and (48) into Eq. (49), to put all field
point coordinates into the spherical coordinate system, gives
R = {r2 + [2rz cos + (z )2 ]}1/2
(51)
This result could also be obtained by using Jz(r)
I(z)(x)(y) in Eq. (23) where dv dx dy dz. In order to
develop approximate expressions for R, we expand Eq. (51)
using the binomial theorem:
1
1
[2rz cos + (z )2 ] 3 [2rz cos + (z )2 ]2 +
2r
8r
(z )3 sin2 cos
(z )2 sin2
= r z cos +
+
+
2r
2r2
(52)
R = r+
The terms in this series decrease as the power of z increases
if z is small compared to r. This expression for R is used in
the radiation integral Eq. (46) to different degrees of approximation. In the denominator of Eq. (46) (which affects only the
amplitude) we let
Rr
(53)
(54)
Az =
I(z )
e j (rz cos )
e j r
dz =
4r
4r
I(z )e j z
cos
dz
(55)
where the integral is over the extent of the line source.
The electric field is found from Eq. (27), which is
A
E = jA
j
k A)
k (k
(56)
This far-field result for a z-directed current, as in Eq. (46),
reduces to
E jA = j sin Az
(57)
Note that this result is the portion of the first term of Eq. (56)
which is transverse to r because jA j(Az sin
Az cos r). This form is an important general result for zdirected sources that is not restricted to line sources.
The radiation fields from a z-directed line source (any zdirected current source in general) are H and E, and are
found from Eqs. (27) and (28). The only remaining problem is
to calculate Az, which is given by Eq. (26) in general and by
Eq. (55) for z-directed line sources. Calculation of Az is the
focal point of linear antenna analysis. We shall return to this
topic after pausing to further examine the characteristics of
the far-field region.
The radiation field components given by Eqs. (27) and (28)
yield
E =
H = H
(58)
where / is the intrinsic impedance of the medium.
An interesting conclusion can be made at this point. The radiation fields are perpendicular to each other and to the direction of propagation r and their magnitudes are related by
Eq. (58).
These are the familiar properties of a plane wave. They
also hold for the general form of a transverse electromagnetic
(TEM) wave which has both the electric and magnetic fields
transverse to the direction of propagation. Radiation from a
finite antenna is a special case of a TEM wave, called a spherical wave which propagates radially outward from the antenna and the radiation fields have no radial components.
Spherical wave behavior is also characterized by the ejr /4r
ANTENNA THEORY
factor in the field expressions; see Eq. (55). The ejr phase
factor indicates a traveling-wave propagating radially outward from the origin and the 1/r magnitude dependence leads
to constant power flow just as with the infinitesimal dipole.
In fact, the radiation fields of all antennas of finite extent
display this dependence with distance from the antenna.
Another way to view radiation field behavior is to note that
spherical waves appear to an observer in the far field to be a
plane wave. This local plane wave behavior occurs because
the radius of curvature of the spherical wave is so large that
the phase front is nearly planar over a local region.
If parallel lines (or rays) are drawn from each point on a
line current as shown in Fig. 6, the distance R to the far field
is geometrically related to r by Eq. (54), which was derived by
neglecting high order terms in the expression for R in Eq.
(52). The parallel ray assumption is exact only when the observation point is at infinity, but it is a good approximation
in the far field. Radiation calculations often start by assuming
parallel rays and then determining R for the phase by geometrical techniques. From the general source shown in Fig.
6, we see that
R = r r cos
(59)
Using the definition of dot product, we have
R = r r r
(60)
This form is a general approximation to R for the phase factor
in the radiation integral. Notice that if r zz, as for line
sources along the z-axis, Eq. (60) reduces to Eq. (54).
The definition of the distance from the source where the
far field begins is where errors due to the parallel ray approximation become insignificant. The distance where the far field
begins, rff , is taken to be that value of r for which the path
length deviation due to neglecting the third term of Eq. (52)
is a sixteenth of a wavelength. This corresponds to a phase
error (by neglecting the third term) of 2/ /16 /8
rad 22.5.
If D is the length of the line source, rff is found by equating
the maximum value of the third term of Eq. (52) to a sixteenth of a wavelength; that is, for z D/2 and 90, the
P
R
J
dv
r
r
co
Figure 6. Parallel ray approximation for far-field calculations of radiation from a general source.
601
third term of Eq. (52) is
(D/2)2
=
2rff
16
(61)
Solving for rff gives
rff =
2D2
(62)
The far-field region is r rff and rff is called the far-field distance, or Rayleigh distance. The far-field conditions are summarized as follows:
r>
2D2
(63a)
r D
(63b)
(63c)
The condition r D was mentioned in association with the
approximation R r of Eq. (53) for use in the magnitude
dependence. The condition r follows from r (2r/ )
1 which was used to reduce Eq. (46) to Eq. (55). Usually the
far field is taken to begin at a distance given by Eq. (62)
where D is the maximum dimension of the antenna. This is
usually a sufficient condition for antennas operating in the
ultra high frequency (UHF) region and above. At lower frequencies, where the antenna can be small compared to the
wavelength, the far-field distance may have to be greater
than 2D2 / in order that all conditions in Eq. (63) are satisfied.
The concept of field regions was introduced in an earlier
section and illustrated with the fields of an ideal dipole. We
can now generalize that discussion to any finite antenna of
maximum extent D. The distance to the far field is 2D2 / . This
zone was historically called the Fraunhofer region if the antenna is focused at infinity; that is, if the rays at large distances from the antenna when transmitting are parallel. In
the far-field region the radiation pattern is independent of
distance. For example, the sin pattern of an ideal dipole is
valid anywhere in its far field. The zone interior to this distance from the center of the antenna, called the near field, is
divided into two subregions. The reactive near-field region is
closest to the antenna and is that region for which the reactive field dominates over the radiative fields. This region extends to a distance 0.62D3 / from the antenna, as long as
D . For an ideal dipole, for which D z , this distance
is /2. Between the reactive near-field and far-field regions
is the radiating near-field region in which the radiation fields
dominate and where the angular field distribution depends on
distance from the antenna. For an antenna focused at infinity
the region is sometimes referred to as the Fresnel region. We
can summarize the field region distances for cases where D
as follows:
Region
Distance from antenna (r)
Reactive near field
0 to 0.62 D3 /
Radiating near field 0.62 D3 / to 2D2 /
Far field
2D2 / to
(64a)
(64b)
(64c)
602
ANTENNA THEORY
Steps in the Evaluation of Radiation Fields. The derivation
for the fields radiated by a line source can be generalized for
application to any antenna. The analysis of the line source,
and its generalizations, can be reduced to the following three
step procedure:
1. Find A. Select a coordinate system most compatible
with the geometry of the antenna, using the notation of
Fig. 1. In general, use Eq. (23) with R r in the magnitude factor and the parallel ray approximation of Eq.
(60) for determining phase differences over the antenna.
These yield
j r
e
A=
4r
Je
j r rr
dv
(65)
Jz e j r rr dv
(66)
For z-directed line sources on the z-axis
A = z
e
4r
x = 0, y = 0, |z |
L
2
(73)
elsewhere
where L is the length of the line source; see Fig. 5. We first
find Az from Eq. (67) as follows:
Az =
e j r
4r
L/2
L/2
Io e j z
cos
dz =
e j r
sin[(L/2) cos ]
Io L
4r
(L/2) cos
(74)
The electric field from Eq. (69) is then
e j r
4r
j r
I
I(z ) = o
0
For z-directed sources
A = z
Example: The Uniform Line Source. The uniform line source
is a line source for which the current is constant along its
extent. If we use a z-directed uniform line source centered on
the origin and along the z-axis, the current is
I(z )e j z
cos
dz
(68)
which is transverse to the direction of propagation, r.
This result is expressed formally as
(69)
which arises from the component of A tangent to the
far-field sphere. For z-directed sources this form becomes
E = jAz sin
(70)
which is Eq. (57).
3. Find H. In general, use the plane-wave relation
H=
1
r E
(71)
This equation expresses the fact that in the far field the
directions of E and H are perpendicular to each other
and to the direction of propagation, and also that their
magnitudes are related by . For z-directed sources
H =
(75)
(67)
which is Eq. (55).
2. Find E. In general, use the component of
E = jA
A + j(r A )r = j(A + A )
sin[(L/2) cos ]
e j r
sin
4r
(L/2) cos
The magnetic field is simply found from this form using
H E / .
A
E = jA
E = jAz sin = jIo L
(72)
which is Eq. (58). The most difficult step is the first,
calculating the radiation integral. To develop an appreciation for the process, we present an example. This
uniform line source example will also serve to provide a
specific setting for introducing general radiation pattern concepts and definitions.
Radiation Pattern Definitions. Since the radiation pattern is
the variation over a sphere centered on the antenna, r is constant and we have only and variation of the field. It is
convenient to normalize the field expression such that its
maximum value is unity. This is accomplished as follows for
a z-directed source which has only a -component of E
F (, ) =
E
E (max)
(76)
where F(, ) is the normalized field pattern and E(max) is
the maximum value of E over a sphere of radius r.
In general E can be complex-valued and, therefore, so can
F(, ). In this case the phase is usually set to zero at the
same point the magnitude is normalized to unity. This is appropriate since we are only interested in relative phase behavior. This variation is, of course, independent of r. An element of current on the z-axis has a normalized field pattern
of
F ( ) = sin
(77)
and there is no variation. The normalized field pattern for
the uniform line source is from Eq. (75) in Eq. (76)
F ( ) = sin
sin[(L/2) cos ]
(L/2) cos
(78)
and again there is no variation. The second factor of this
expression is the function sin(u)/u. It has a maximum value
of unity at u 0; this corresponds to 90 where u
(L/2) cos . Substituting 90 in Eq. (78) gives unity and
we see that F() is properly normalized.
In general, a normalized field pattern can be written as
the product
F (, ) = g(, ) f (, )
(79)
ANTENNA THEORY
603
z
z
(b) Intermediate
(c) Endfire
(a) Broadside
where g(, ) is the element factor and f(, ) is the pattern
factor. The pattern factor comes from the integral over the
current and is strictly due to the distribution of current in
space. The element factor is the pattern of an infinitesimal
current element in the current distribution. For example, for
a z-directed current element the total pattern is given by the
element factor:
F ( ) = g( ) = sin
(80)
for a z-directed current element. Actually this factor originates from Eq. (57) and can be interpreted as the projection
of the current element in the -direction. In other words, at
90 we see the maximum length of the current, whereas
at 0 or 180 we see the end view of an infinitesimal current which yields no radiation. The sin factor expresses the
fraction of the size of the current as seen from the observation
angle . On the other hand, the pattern factor f(, ) represents the integrated effect of radiation contributions from the
current distribution, which can be treated as being made up
of many current elements. The pattern value in a specific direction is then found by summing the parallel rays from each
current element to the far field with the magnitude and phase
of each included. The radiation integral of Eq. (65) sums the
far-field contributions from the current elements and when
normalized yields the pattern factor.
Antenna analysis is usually easier to understand by considering the antenna to be transmitting as we have here.
However, most antennas are reciprocal and thus their radiation properties are identical when used for reception; as discussed in the section on reciprocity.
For the z-directed uniform line source pattern Eq. (78) we
can identify the factors as
g( ) = sin
(81)
and
Figure 7. Polar plots of uniform line
source patterns. (a) Broadside. (b) Intermediate. (c) Endfire.
Frequently the directional properties of the radiation from
an antenna are described by another form of radiation pattern, the power pattern. The power pattern gives angular dependence of the power density and is found from the , variation of the r-component of the Poynting vector. For z-directed
sources H E / so the r-component of the Poynting vector
is EH E2 /(2) and the normalized power pattern is simply the square of its field pattern magnitude P() F()2.
The general normalized power pattern is
P(, ) = |F (, )|2
The normalized power pattern for a z-directed current element is
P(, ) = sin2
sin[(L/2) cos ]
(L/2) cos
(82)
For long line sources (L ) the pattern factor of Eq. (82) is
much sharper than the element factor sin , and the total
pattern is approximately that of Eq. (82), that is, F() f().
Hence, in many cases we need only work with f(), which is
obtained from Eq. (67). If we allow the beam to be scanned as
in Fig. 7, the element factor becomes important as the pattern
maximum approaches the z-axis.
(84)
and for a z-directed uniform line source is
2
sin[(L/2) cos ]
P( ) = sin
(L/2) cos
(85)
Frequently patterns are plotted in decibels. It is important
to recognize that the field (magnitude) pattern and power pattern are the same in decibels. This follows directly from the
definitions. For field intensity in decibels
|F (, )|dB = 20 log |F (, )|
(86)
and for power in decibels
P(, )dB = 10 log P(, ) = 10 log |F (, )|2 = 20 log |F (, )|
(87)
and we see that
P(, )dB = |F (, )|dB
f ( ) =
(83)
(88)
Radiation Pattern Parameters. A typical antenna power pattern is shown in Fig. 8 as a polar plot in linear units (rather
than decibels). It consists of several lobes. The main lobe (or
main beam or major lobe) is the lobe containing the direction
of maximum radiation. There is also usually a series of lobes
smaller than the main lobe. Any lobe other than the main
lobe is called a minor lobe. Minor lobes are composed of side
lobes and back lobes. Back lobes are directly opposite the
main lobe, or sometimes they are taken to be the lobes in
604
ANTENNA THEORY
Main lobe maximum direction
half the maximum value:
HP = |HP left HP right|
Main lobe
1.0
Half-power point (left)
Half-power point (right)
0.5
Half-power beamwidth (HP)
Beamwidth between
first nulls (BWFN)
Minor
lobes
Figure 8. A typical power pattern polar plot.
the half-space opposite the main lobe. The term side lobe is
sometimes reserved for those minor lobes near the main lobe,
but is most often taken to be synonymous with minor lobe;
we will use the latter convention.
The radiation from an antenna is represented mathematically through the radiation pattern function, F(, ) for field
and P(, ) for power. This angular distribution of radiation
is visualized through various graphical representations of the
pattern, which we discuss in this section. Graphical representations also are used to introduce definitions of pattern parameters that are commonly used to quantify radiation pattern characteristics.
A three-dimensional plot as in Fig. 4(d) gives a good overall
impression of the entire radiation pattern, but cannot convey
accurate quantitative information. Cuts through this pattern
in various planes are the most popular pattern plots. They
usually include the E- and H-plane patterns; see Figs. 4(b)
and (c). Pattern cuts are often given various fixed values,
leaving the pattern a function of alone; we will assume that
is the case here. Typically the side lobes are alternately positive and negative valued. In fact, a pattern in its most general
form may be complex-valued. Then we use the magnitude of
the field pattern F() or the power pattern P().
A measure of how well the power is concentrated into the
main lobe is the (relative) side lobe level, which is the ratio of
the pattern value of a side lobe peak to the pattern value of
the main lobe. The largest side lobe level for the whole pattern is the maximum (relative) side lobe level, frequently abbreviated as SLL. In decibels it is given by
F (SLL)
(89)
SLL = 20 log
F (max)
where F(max) is the maximum value of the pattern magnitude and F(SLL) is the pattern value of the maximum of the
highest side lobe magnitude. For a normalized pattern
F(max) 1.
The width of the main beam is quantified through halfpower beamwidth, HP, which is the angular separation of the
points where the main beam of the power pattern equals one-
(90)
where HP left and HP right are points to the left and right of the
main beam maximum for which the normalized power pattern has a value of one-half (see Fig. 8). On the field pattern
F() these points correspond to the value 1/2. For example,
the sin pattern of an ideal dipole has a value of 1/2 for
values of H-P left 135 and H-P right 45. Then HP 135
45 90. This is shown in Fig. 4(b). Note that the definition
of HP is the magnitude of the difference of the half-power
points and the assignment of left and right can be interchanged without changing HP. In three dimensions the
radiation pattern major lobe becomes a solid object and the
half-power contour is a continuous curve. If this curve is essentially elliptical, the pattern cuts that contain the major
and minor axes of the ellipse determine what the Institute
of Electrical and Electronics Engineers (IEEE) defines as the
principal half-power beamwidths.
Antennas are often referred to by the type of pattern they
produce. An isotropic antenna, which is hypothetical, radiates
equally in all directions giving a constant radiation pattern.
An omnidirectional antenna produces a pattern which is constant in one plane; the ideal dipole of Fig. 4 is an example.
The pattern shape resembles a doughnut. We often refer to
antennas as being broadside or endfire. A broadside antenna
is one for which the main beam maximum is in a direction
normal to the plane containing the antenna. An endfire antenna is one for which the main beam is in the plane containing the antenna. For a linear current on the z-axis, the
broadside direction is 90 and the endfire directions are
0 and 180. For example, an ideal dipole is a broadside antenna. For z-directed line sources several patterns are possible. Figure 7 illustrates a few f() patterns. The entire pattern (in three dimensions) is imagined by rotating the pattern
about the z-axis. The full pattern can then be generated from
the E-plane patterns shown. The broadside pattern of Fig.
7(a) is called fan beam. The full three dimensional endfire
pattern for Fig. 7(c) has a single lobe in the endfire direction.
This single lobe is referred to as a pencil beam. Note that the
sin element factor, which must multiply these patterns to
obtain the total pattern, will have a significant effect on the
endfire pattern. Intermediate scan angles are also possible,
as shown in Fig. 7(b).
ANTENNA PERFORMANCE MEASURES
Antennas are devices that are used in systems for communications or sensing. There are many parameters used to quantify the performance of the antenna as a device, which in turn
impacts on system performance. In this section we consider
the most important of these parameters when they are employed in their primary application area of communication
links, such as the simple communication link as shown in Fig.
9. We first discuss the basic properties of a receiving antenna.
Receiver
Transmitter
R
Figure 9. A communication link.
ANTENNA THEORY
matched to the wave, and is impedance matched to its load.
The maximum refers to the assumption that there are no
ohmic losses on the antenna.
Maximum effective aperture for the ideal dipole is found
using Eqs. (92) and (93) with Eq. (94) to give
IA
V
ZL
Incident
wave with
power
density, S
VA
ZL
ZA
Aem
(a)
(b)
Figure 10. Equivalent circuit for a receiving antenna. (a) Receive
antenna connected to a receiver with load impedance ZL. (b) Equivalent circuit.
The receiving antenna with impedance ZA and terminated in
load impedance ZL is modeled as shown in Fig. 10. The total
power incident on the receiving antenna is found by summing
up the incident power density over the area of the receive
antenna, called effective aperture. How an antenna converts
this incident power into available power at its terminals depends on the type of antenna used, its pointing direction, and
polarization. In this section we discuss the basic relationships
for power calculations and illustrate their use in communication links.
Directivity and Gain. For system calculations it is usually
easier to work with directivity rather than its equivalent,
maximum effective aperture. The relation can be established
by examining an infinitestimal dipole and generalizing. The
maximum effective aperture of an ideal, lossless dipole of
length z is found by orienting the dipole for maximum response, which is parallel to the incoming linearly polarized
electric field Ei. Then the open circuit voltage is found from
VA = E i z Ideal dipole receiving antenna
1 |VA |2
1 |E i |2
=
(z)2
8 Rr
8 Rr
(92)
where Eq. (91) was used. The available power can also be
calculated by examining the incident wave. The power density
(Poynting vector magnitude) in the incoming wave is
S=
1
1 |E i |2
E H | =
|E
2
2
(93)
The available power is found using the maximum effective aperture Aem, which is the collecting area of the antenna. The
receiving antenna collects power from the incident wave in
proportion to its maximum effective aperture
PAm = SAem
1 |VA |2
PAm
3 2
1
8 Rr
=
=
(z)2 =
=
i 2
S
4
R
8
1 |E |
r
2
(95)
where the ideal dipole radiation resistance value of [2/3
(z/ )2] was used. The maximum effective aperture of an
ideal dipole is independent of its length z (as long as z
). However, it is important to note that Rr is proportional to
(z/ )2 so that even though Aem remains constant as the dipole
is shortened, its radiation resistance decreases rapidly and it
is more difficult to realize this maximum effective aperture
because of the required conjugate impedance match of the receiver to the antenna.
The directivity of the ideal dipole can be written in the
following manner:
D=
4 3 2
3
= 2
2
8
Ideal dipole
(96)
Grouping factors this way permits identification of Aem from
Eq. (95). Thus
D=
4
Aem
2
(97)
Although we derived this for an ideal dipole, this relationship
is true for any antenna. For an isotropic antenna, the directivity by definition is unity; so from Eq. (97) with D 1
Aem =
(91)
The power available from the antenna is realized when the
antenna impedance is matched by a load impedance of ZL
Rr jXA assuming Rohmic 0. Rr is the radiation resistance.
The maximum available power is then
PAm =
605
2
4
Isotropic antenna
(98)
Comparing this to the definition of directivity in (100) below
we see that
2 = Aem A
(99)
which is also a general relationship. We can extract some interesting concepts from this relation. For a fixed wavelength
Aem and A, are inversely proportional; that is, as the maximum effective aperture increases (as a result of increasing its
physical size), the beam solid angle decreases, which means
power is more concentrated in angular space (i.e., directivity
goes up). For a fixed maximum effective aperture (i.e., antenna size), as wavelength decreases (frequency increases)
the beam solid angle also decreases, leading to increased directivity.
Directivity is more directly related to its definition through
this inverse dependence on beam solid angle as
D=
4
A
(100)
(94)
The maximum available power PAm will be realized if the antenna is directed for maximum response, is polarization
where
A =
|F (, )|2 d
(101)
606
ANTENNA THEORY
This directivity definition has a simple interpretation. Directivity is a measure of how much greater the power density at
a fixed distance is in a given direction than if all power were
radiated isotropically. This view is illustrated in Fig. 11. For
an isotropic antenna, as in Fig. 11(a), the beam solid angle is
4, and thus Eq. (100) gives a directivity of unity.
In practice antennas are not completely lossless. Earlier
we saw that power available at the terminals of a transmitting antenna was not all transformed into radiated power.
The power received by a receiving antenna is reduced to the
fraction er (radiation efficiency) from what it would be if the
antenna were lossless. This is represented by defining effective aperture
Ae = er Aem
S=
PA = SAe
S=
4
4
er Aem = 2 Ae
2
Ae = apAp
(105)
It is important to note that although we developed the general relationships of Eqs. (97), (99), and (104) for receiving
antennas, they apply to transmitting antennas as well. The
relationships are essential for communication system computations that we consider next.
Communication Links. We are now ready to completely describe the power transfer in the communication link of Fig. 9.
Um = DUave
Gt Pt Aer
4R2
Pr = Pt
Gt Gr 2
(4R)2
(b)
Figure 11. Illustration of directivity. (a) Radiation intensity distributed isotropically. (b) Radiation intensity from an actual antenna.
(108)
(109)
which gives the available power in terms of the transmitted
power, antenna gains, and wavelength. Or, we could use
Gt 4Aet / 2 in Eq. (108) giving
Pr = Pt
Aet Aer
R2 2
(110)
which is called the Friis transmission formula (2).
The power transmission formula Eq. (109) is very useful
for calculating signal power levels in communication links. It
assumes that the transmitting and receiving antennas are
matched in impedance to their connecting transmission lines,
have identical polarizations, and are aligned for polarization
match. It also assumes the antennas are pointed toward each
other for maximum gain. If any of these conditions are not
met, it is a simple matter to correct for the loss introduced
by polarization mismatch, impedance mismatch, or antenna
misalignment.
The antenna misalignment effect is easily included by using the power gain value in the appropriate direction. The
effect and evaluation of polarization and impedance mismatch
are additional considerations. Figure 10 shows the network
model for a receiving antenna with input antenna impedance
ZA and an attached load impedance ZL, which can be a transmission line connected to a distant receiver. The power delivered to the terminating impedance is
PD = pqPr
Uave
(107)
where Aer is the effective aperture of the receiving antenna
and we assume it to be pointed and polarized for maximum
response. Now from Eq. (104) Aer Gr2 /4, so Eq. (108) becomes
Uave
(a)
GtUave
GP
= t t2
R2
4R
Pr = SAer =
(104)
For electrically large antennas effective aperture is equal to
or less than the physical aperture area of the antenna Ap,
which is expressed using aperture efficiency ap:
(106)
Using this in Eq. (103) gives the available received power as
(103)
This simple equation is very intuitive and indicates that a
receiving antenna acts to convert incident power (flux) density in W/m2 to power delivered to the load in watts. Losses
associated with mismatch between the polarization of the incident wave and receiving antenna as well as impedance mismatch between the antenna and load are not included in Ae.
These losses are not inherent to the antenna, but depend on
how it is used in the system. The concept of gain is introduced
to account for losses on an antenna, that is, G erD. We can
form a gain expression from the directivity expression by multiplying both sides of Eq. (97) by er and using Eq. (102):
Uave
Pt
=
R2
4R2
where Pt is the time-averaging input power (Pin) accepted by
the transmitting antenna. The quantity Uave denotes the time
average radiation intensity given in the units of power per
solid angle see Fig. 11. For a transmitting antenna that is not
isotropic but has gain Gt and is pointed for maximum power
density in the direction of the receiver, we have for the power
density incident on the receiving antenna,
(102)
and the available power with antenna losses included, analogous to Eq. (94), is
G = er D =
If the transmitting antenna were isotropic, it would have
power density at distance R of
(111)
where
PD power delivered from the antenna
Pr power available from the receiving antenna
p polarization efficiency (or polarization mismatch factor), 0 p 1
q impedance mismatch factor, 0 q 1
ANTENNA THEORY
607
An overall efficiency, or total efficiency total, can be defined to
include the effects of polarization and impedance mismatch:
total = pqap
(112)
Temperature distribution
T (, )
Then PD totalPr. It is convenient to express Eq. (111) in dB
form:
PD (dBm) = 10 log p + 10 log q + Pr (dBm)
(113)
where the unit dBm is power in decibels above a milliwatt;
for example, 30 dBm is 1 W. Both powers could also be expressed in units of decibels above a watt, dBW. The power
transmission formula Eq. (109) can also be expressed in dB
form as
Pr (dBm) = Pt (dBm) + Gt (dB) + Gr (dB)
20 log R(km) 20 log f (MHz) 32.44
Effective Isotropically Radiated Power. A frequently used
concept in communication systems is that of effective (or
equivalent) isotropically radiated power, EIRP. It is formally
defined as the power gain of a transmitting antenna in a
given direction multiplied by the net power accepted by the
antenna from the connected transmitter. Sometimes it is denoted as ERP, but this term, effective radiated power, is usually reserved for EIRP with antenna gain relative to that of a
half-wave dipole instead of gain relative to an isotropic antenna. As an example of EIRP, suppose an observer is located
in the direction of maximum radiation from a transmitting
antenna with input power Pt. Then the EIRP may be expressed as
(115)
For a radiation intensity Um, as illustrated in Fig. 11(b), and
Gt 4Um /Pt, we obtain
EIRP = Pt
4Um
= 4Um
Pt
TA
TA
TA
(a)
(b)
Figure 12. Antenna temperature. (a) An antenna receiving noise
from directions (, ) producing antenna temperature TA. (b) Equivalent model.
(114)
where Gt(dB) and Gr(dB) are the transmit and receive antenna gains in decibels, R(km) is the distance between the
transmitter and receiver in kilometers, and f(MHz) is the frequency in megahertz.
EIRP = Pt Gt
Rr
Power pattern
P (, )
(116)
The same radiation intensity could be obtained from a lossless isotropic antenna (with power gain Gi 1) if it had an
input power Pin equal to PtGt. In other words, to obtain the
same radiation intensity produced by the directional antenna
in its pattern maximum direction, an isotropic antenna would
have to have an input power Gt times greater. Effective isotropically radiated power is a frequently used parameter. For
example, FM radio stations often mention their effective radiated power when they sign off at night.
Antenna Noise Temperature and Radiometry
Receiving systems are vulnerable to noise and a major contribution is the receiving antenna, which collects noise from its
surrounding environment. In most situations a receiving antenna is surrounded by a complex environment as shown in
Fig. 12(a). Any object (except a perfect reflector) that is above
absolute zero temperature will radiate electromagnetic
waves. An antenna picks up this radiation through its antenna pattern and produces noise power at its output. The
equivalent terminal behavior is modeled in Fig. 12(b) by considering the radiation resistance of the antenna to be a noisy
resistor at a temperature TA such that the same output noise
power from the antenna in the actual environment is produced. The antenna temperature TA is not the actual physical
temperature of the antenna, but is an equivalent temperature
that produces the same noise power, PNA, as the antenna operating in its surroundings. This equivalence is established
by assuming the model of Fig. 12(b); the noise power available
from the noise resistor in bandwidth f at temperature TA is
PNA = kTA f
(117)
where
PNA available power due to antenna noise [W]
k Boltzmanns constant 1.38 1023 JK1
TA antenna temperature [K]
f receiver bandwidth [Hz]
Such noise is often referred to as Nyquist or Johnson noise
for system calculations. The system noise power PN is calculated using the total system noise temperature Tsys in place
of TA in Eq. (117) with Tsys TA Tr, where Tr is the receiver
noise temperature.
Antenna noise is important in several system applications
including communications and radiometry. Communication
systems are evaluated through carrier-to-noise ratio, which is
determined from the signal power and the system noise power
as
CNR =
PD
PN
(118)
where PN kTsys f is the system noise power. This noise
power equals the sum of PNA and noise power generated in
the receiver connected to the antenna.
Noise power is found by first evaluating antenna temperature. As seen in Fig. 12(a), TA is found from the collection of
608
ANTENNA THEORY
noise through the scene temperature distribution T(, )
weighted by the response function of the antenna, the normalized power pattern P(, ). This is expressed mathematically
by integrating over the temperature distribution:
TA =
1
A
T (, )P(, ) d
(119)
If the scene is of constant temperature To over all angles, To
comes out of the integral and then
TA =
To
A
P(, ) d =
0
To
= To
A A
of technology development, including new antenna designs.
DirecTv (trademark of Hughes Network Systems) transmits
from 12.2 to 12.7 GHz with 120 W of power and an EIRP of
about 55 dBW in each 24 MHz transponder that handles several compressed digital video channels. The receiving system
uses a 0.46 m (18 in) diameter offset fed reflector antenna. In
this example we perform the system calculations using the
following link parameter values:
f = 12.45 GHz (midband)
Pt (dBW) = 20.8 dBW (120 W)
(120)
using Eq. (101) for A. The antenna is completely surrounded
by noise of temperature To and its output antenna temperature equals To independent of the antenna pattern shape.
In general, antenna noise power PNA is found from Eq.
(117) using TA from Eq. (119) once the temperature distribution T(, ) is determined. Of course, this depends on the
scene, but in general T(, ) consists of two components: sky
noise and ground noise. Ground noise temperature in most
situations is well approximated for soils by the value of 290
K, but is much less for surfaces that are highly reflective due
to reflection of low temperature sky noise. Also, smooth surfaces have high reflection for near grazing incidence angles.
Unlike ground noise, sky noise is a strong function of frequency. Sky noise is made up of atmospheric, cosmic, and
manmade noise. Atmospheric noise increases with decreasing
frequency below 1 GHz and is primarily due to lightning,
which propagates over large distances via ionospheric reflection below several MHz. Atmospheric noise increases with
frequency above 10 GHz due to water vapor and hydrometeor
absorption; these depend on time, season, and location. It also
increases with decreasing elevation angle. Atmospheric gases
have strong, broad spectral lines, such as water vapor and
oxygen lines at 22 and 60 GHz, respectively.
Cosmic noise originates from discrete sources such as the
sun, moon, and radio stars as well as our galaxy, which has
strong emissions for directions toward the galactic center. Galactic noise increases with decreasing frequency below 1 GHz.
Manmade noise is produced by power lines, electric motors,
and other sources and usually can be ignored except in urban
areas at low frequencies. Sky noise is very low for frequencies
between 1 and 10 GHz, and can be as low as a few K for high
elevation angles.
Of course, the antenna pattern strongly influences antenna
temperature; see Eq. (119). The ground noise temperature
contribution to antenna noise can be very low for high-gain
antennas having low side lobes in the direction of the earth.
Broad beam antennas, on the other hand, pick up a significant amount of ground noise as well as sky noise. Losses on
the antenna structure also contribute to antenna noise. A
figure of merit used with satellite earth terminals is G/Tsys,
which is the antenna gain divided by system noise temperature usually expressed in dB/K. It is desired to have high
values of G to increase signal and to have low values of Tsys
to decrease noise, giving high values of G/Tsys.
Example: Direct Broadcast Satellite Reception. Reception of
high quality television channels at home in the 1990s, with
inexpensive, small terminals, is the result of three decades
Gt (dB) = EIRP(dBW) Pt (dBW) = 55 20.8 = 34.2 dB
R = 38, 000 km (typical slant path length)
2
4
4
0.46
Gr = 2 apAp =
0.7
= 2538
(0.024)2
2
= 34 dB (70% aperture efficiency)
The received power from Eq. (114) is
Pr (dBm) = 20.8 + 34.2 + 34 20 log(38, 000) 20 log(12450)
32.44 = 116.9 dBW
(121)
This is 2 1012 W! Without the high gains of the antennas
(68 dB combined) this signal would be hopelessly lost in noise.
The receiver uses a 67 K noise temperature low noise block
downconverter. This is the dominant receiver contribution,
and when combined with antenna temperature leads to a system noise temperature of Tsys 125 K. The noise power in
the effective signal bandwidth f 20 MHz is
PN = kTsys f
= 1.38 1023 125 20 106 = 3.45 1014
(122)
= 134.6 dBW
Thus the carrier to noise ratio from Eqs. (118) and (121) is
CNR(dB) = PD (dBW ) PN (dBW )
= 116.9 (134.6) = 17.7 dB
(123)
Antenna Bandwidth
Bandwidth is a measure of the range of operating frequencies
over which antenna performance is acceptable. Bandwidth is
computed in one of two ways. Let f U and f L be the upper and
lower frequencies of operation for which satisfactory performance is obtained. The center (or sometimes the design frequency) is denoted as f C. Then bandwidth as a percent of the
center frequency, Bp, is
Bp =
fU fL
100
fC
(124)
Bandwidth is also defined as a ratio, Br, by
Br =
fU
fL
(125)
The bandwidth of narrow band antennas is usually expressed
as a percent whereas wide band antennas are quoted as a
ratio. Resonant antennas have small bandwidths. For exam-
APERTURE ANTENNAS
ple, half-wave dipoles have bandwidths of up to 16%, ( fU and
f L determined by the VSWR 2.0). On the other hand, antennas that have traveling waves on them rather than standing
waves (as in resonant antennas), operate over wider frequency ranges.
BIBLIOGRAPHY
1. W. L. Stutzman and G. A. Thiele, Antenna Theory and Design, 2nd
ed., New York: Wiley, 1998.
2. E. C. Jordan and K. G. Balmain, Electromagnetic Waves and Radiating Systems, 2nd ed., New York: Prentice-Hall, 1968, p. 555.
WARREN L. STUTZMAN
WILLIAM A. DAVIS
Virginia Polytechnical Institute and
State University
ANTIFERROMAGNETISM. See MAGNETIC MATERIALS.
APD. See AVALANCHE DIODES.
609
ANTENNAS
563
ANTENNAS
HISTORY OF ANTENNAS
Marconis first experiments with transmitting electromagnetic waves in 1901, antennas have found several important
applications over the entire radio frequency range and, numerous designs of antennas now exist. Antennas are an integral part of our everyday lives, used for a multitude of purposes. An antenna is used to either transmit or receive
electromagnetic waves and it serves as a transducer converting guided waves into free space waves in the transmitting mode or vice versa in the receiving mode. All antennas
operate on the same basic principles of electromagnetic theory
formulated by James Clark Maxwell. Maxwell put forth his
unified theory of electricity and magnetism in 1873 (1) in his
famous book, A Treatise on Electricity and Magnetism, incorporating all previously known results on electricity and magnetism and expressing these mathematically through what we
refer to as Maxwells equations which hold over the entire
electromagnetic spectrum. His theory was met with much
skepticism and it wasnt until 1886 that Heinrich Hertz (2),
considered the father of radio, was able to validate this theory
with his experiments. The first radio system, at a wavelength
of 4 m, consisted of a /2 dipole (transmitting antenna) and a
resonant loop (receiving antenna) as shown in Fig. 1 (3). By
turning on the induction coil, sparks were induced across the
A
C
B
c
Figure 1. Heinrich Hertzs radio system.
J. Webster (ed.), Wiley Encyclopedia of Electrical and Electronics Engineering. Copyright # 1999 John Wiley & Sons, Inc.
564
ANTENNAS
gap A which were detected across the gap B of the receiving
antenna.
Almost a decade later, Guglielmo Marconi, in 1901, was
able to receive signals across the Atlantic in St. Johns, Newfoundland, sent from a station he had built in Poldhu, Cornwall, England. Marconis transmitting antenna was a fan antenna with 50 vertical wires supported by two 6 m guyed
wooden poles. The receiving antenna was a 200 m wire pulled
up with a kite (3). For many years since Marconis experiment, antennas operated at low frequencies, up to the ultra
high frequency (UHF) region and were primarily wire type
antennas. The need for radar during World War II launched
antenna design into a new era and opened up the entire radio
frequency spectrum for their use. Since the 1950s many new
antenna types such as reflector, aperture, and horn antennas
came into use, most of them operating in the microwave region. Their applications range from communications to astronomy to various deep space applications. These antennas
have been discussed in several books and some of these have
been included in Ref. 426. A good explanation of how an
antenna radiates is given in Refs. 20 and 23. To understand
how the antenna radiates, consider a pulse of electric charge
moving along a straight conductor. A static electric charge
or a charge moving with a uniform velocity does not radiate.
However, when charges are accelerated along the conductor
and decelerated upon reflection from its end, radiated fields
are produced along the wire and at each end (20,21). The Institute of Electrical and Electronic Engineers (IEEE) standard definitions of terms for antennas (24) and Balanis (25)
provide a good source of definitions and explanations of the
fundamental parameters associated with antennas.
TYPES OF ANTENNAS
Elaborate antennas or antenna systems require careful design and a thorough understanding of the radiation mechanism involved. The selection of the type of antenna to be used
is determined by electrical and mechanical constraints and
operating costs. The electrical parameters of the antenna are
the frequency of operation, gain, polarization, radiation pattern, impedance, and so on. The mechanical parameters of
importance are the size, weight, reliability, manufacturing
process, and so on. In addition, the environment under which
the antenna is to be used also needs to be taken into consideration for example, the effects of temperature, rain, and wind
vibrations. Antennas are shielded from the environment
through the use of radomes whose presence is taken into account while designing the antenna.
Antennas can be classified broadly into the following categories: wire antennas, reflector antennas, lens antennas, traveling wave antennas, frequency independent antennas, horn
antennas, and conformal antennas. In addition, antennas are
very often used in array configurations to improve upon the
characteristics of an individual antenna element.
Wire Antennas
Wire antennas were among the first type of antennas used
and are the most familiar type to the layman. These antennas
can be linear or in the form of closed loops. The thin linear
dipole is used extensively and the half-wavelength dipole has
a radiation resistance of 73 , very close to the 75 charac-
z
z
/2
Figure 2. A half wavelength dipole and its radiation pattern.
teristic impedance of feed lines such as the coaxial cable. It
has an omnidirectional pattern as shown in Fig. 2 with a half
power beamwidth of 78. Detailed discussions on dipole antennas of different lengths can be found in Ref. 25.
Loop antennas can have several different shapes such as
circular, square, and rectangular. Electrically small loops are
those whose overall wire extent is less than one-tenth of a
wavelength. Electrically large loops have circumferences that
are of the order of a wavelength. An electrically small circular
or square loop antenna can be treated as an infinitesimal
magnetic dipole with its axis perpendicular to the plane of the
loop. Various configurations of polygonal loop antennas have
been investigated (27) in the ferrite loop, where a ferrite core
is placed in the loop antenna to increase its efficiency. Loop
antennas are inefficient with high ohmic losses and often are
used as receivers and as probes for field measurements. The
radiation pattern of small loop antennas has a null perpendicular to the plane of the loop and a maximum along the plane
of the loop. An electrically large antenna has the maximum
radiation perpendicular to the plane of the loop and is regarded as the equivalent to the half wavelength dipole.
Dipole and loop antennas find applications in the low to
medium frequency ranges. They have wide beamwidths and
their behavior is greatly affected by nearby obstacles or structures. These antennas are often placed over a ground plane.
The spacing above the ground plane determines the effect the
ground plane has on the radiation pattern and the increase
in the directivity (21).
Thick dipoles are used to improve the narrow bandwidth
of thin dipole antennas. Examples of these are the cylindrical
dipole, the folded dipole, and the biconical antennas. The use
of a sleeve around the input region and the arms of the dipole
also results in broader bandwidths.
Reflector Antennas
Since World War II, when reflector antennas gained prominence due to their use with radar systems, these antennas
have played an important role in the field of communications.
Love (28) has published a collection of papers on reflector antennas. Reflector antennas have a variety of geometrical
shapes and require careful design and a full characterization
of the feed system. Silver (5) presents the technique for analysis based on aperture theory and physical optics. Other methods such as the geometrical theory of diffraction (GTD) and
the fast Fourier transform (FFT) along with various optimiza-
ANTENNAS
tion techniques (29) are now used for a more accurate design
of these antennas.
The plane reflector is the simplest type of a reflector and
can be used to control the overall system radiation characteristics (21). The corner reflector has been investigated by
Kraus (30) and the 90 corner reflector is found to be the most
effective. The feeds for corner reflectors are generally dipoles
placed parallel to the vertex. These antennas can be analyzed
in a rather straightforward manner using the method of images. Among curved reflectors, the paraboloid is the most
commonly used. The paraboloid reflector shown in Fig. 3 is
formed by rotating a parabolic reflector about its axis. The
reflector transforms a spherical wave radiated from a feed at
its focus into a plane wave.
To avoid blockage caused by the feed placed at the focal
point in a front fed system, the feed is often off-set from the
axis (31). The Cassegrain reflector is a dual reflector system
using a paraboloid as the primary and a hyperboloid as the
secondary reflector with a feed along the axis of the paraboloid.
The Gregorian dual reflector antenna uses an ellipse as the
subreflector. The aperture efficiency in a Cassegrain antenna
can be improved by modifying the reflector surfaces (28). Most
paraboloidal reflectors use horn antennas (conical or pyramidal) for their feeds. With a paraboloidal reflector, beam scanning by feed displacement is limited. A spherical reflector provides greater scanning but requires more elaborate feed
design since it fails to focus an incident plane to a point.
Spherical reflectors can suffer from a loss in aperture and increased minor lobes due to blockage by the feed.
565
/(n-1)
Figure 4. Zoned lens.
Lens Antennas
At larger wavelengths, reflectors become impractical due to
the necessity of having large feed structures and tolerance
requirements. At low frequencies, the lens antenna is prohibitively heavy. Both lens antennas and parabolic reflectors use
free space as a feed network to excite a large aperture. The
feed of a lens remains out of the aperture and thus eliminates
aperture blockage and high side lobe levels. Dielectric lens
antennas are similar to optical lenses and the aperture of the
antenna is equal to the projection of the rim shape. Lenses
are divided into two categories: single-surface and dual-surface. In the single-surface lens refraction occurs only at one
surface. The other surface is an equiphase surface of the incident or emergent wave and the waves pass through normal to
the surface without refraction. Single-surface lenses convert
either cylindrical or spherical waves to plane waves. In a
dual-surface lens, refraction occurs at both lens surfaces. The
far field is determined by diffraction from the aperture. Dualsurface lenses allow more control of the pattern characteristics. Both surfaces are used for focusing and the second surface can be used to control the amplitude distribution in the
aperture plane. These simple lenses are many wavelengths
thick if their focal length and aperture are large compared to
a wavelength. The surface of the lens can be zoned by removing multiples of wavelengths from the thickness. The zoning
can be done either in the refracting or non-refracting surface
as shown in Fig. 4. The zoned lens is frequency sensitive and
can give rise to shadowing losses at the transition regions (5).
Artificial dielectric lenses in which particles such as metal
spheres, strips, disks, or rods are introduced in the dielectric
have been investigated by Kock (32). The size of the particles
has to be small compared to the wavelength. Metal plate
lenses using spaced conducting plates are used at microwave
frequencies. Since the index of refraction of a metal-plate medium depends on the ratio of the wavelength to the spacing
between the plates, these lenses are frequency sensitive. The
Luneberg lens is a spherically symmetric lens with an index
of refraction that varies as function of the radius. A plane
wave incident on this lens will be brought to a focus on the
opposite side. These lens antennas can be made using a series
of concentric spherical shells each with a constant dielectric.
Traveling Wave Antennas
Figure 3. A parabolic reflector antenna with its feed. (Courtesy,
NASA Lewis Center)
Traveling wave antennas (33) are distinguished from other
antennas by the presence of a traveling wave along the struc-
566
ANTENNAS
Dummy coax.
cable
Driven
element
Reflector
Directors
Coax.
cable
Figure 5. A YagiUda antenna.
Figure 6. A two-arm balanced conical spiral antenna.
ture and by the propagation of power in a single direction.
Linear wire antennas are the dominant type of traveling
wave antennas. Linear wave antennas with standing wave
patterns of current distributions are referred to as standing
wave or resonant antennas, the amplitude of the current distribution is uniform along the source but the phase changes
linearly with distance. There are in general two types of traveling wave antennas. The surface wave antenna is a slow
wave structure, where the phase velocity of the wave is
smaller than the velocity of light in free space. The radiation
occurs from discontinuities in the structure. A leaky wave antenna is a fast wave structure, the phase velocity of the wave
being greater than the velocity of light in free space. The
structure radiates all its power with the fields decaying in the
direction of wave travel.
A long wire antenna, many wavelengths in length, is an
example of a traveling wave antenna. The Beverage antenna
is a thin wire placed horizontally above a ground plane. The
antenna has poor efficiency but can have good directivity and
is used as a receiving antenna in the low to mid-frequency
range. The V antenna is formed by using two Beverage antennas separated by an angle and fed from a balanced line. By
adjusting the angle, the directivity can be increased and the
side lobes can be made smaller. Terminating the legs of the
V antenna in their characteristic impedances makes the wires
nonresonant and greatly reduces back radiation. The rhombic
antenna consists of two V antennas. The second V antenna
brings the two sides together and a single terminating resistor can be used to connect the balanced lines. An inverted V
over a ground plane is another configuration for a rhombic antenna.
The pattern characteristics can be controlled by varying
the angle between the elements, the lengths of the elements,
and the height above the ground. The helical antenna (21)
is a high gain broadband end-fire antenna. It consists of a
conducting wire wound in a helix. It has found applications
as feeds for parabolic reflectors and for various space communications systems. A popular and practical antenna is the
YagiUda antenna (34,35) shown in Fig. 5. It uses an arrangement of parasitic elements around the feed element to
act as reflectors and directors to produce an end-fire beam.
The elements are linear dipoles with a folded dipole used as
the feed. The mutual coupling between the standing wave
current elements in the antenna is used to produce a traveling wave unidirectional pattern.
Frequency Independent Antennas
Frequency independent antennas or self scaling antennas
were introduced in the early 1950s extending antenna bandwidths by greater than 40% (36). Ideally an antenna will be
frequency independent if its shape is specified only in terms
of angles. These antennas have to be truncated for practical
use and the current should attenuate along the structure to a
negligible value at the termination. Examples of these antennas are the bidirectional planar spiral, and the unidirectional
conical spiral antenna shown in Fig. 6.
Horn Antennas
The electromagnetic horn antenna is characterized by attractive qualities such as a unidirectional pattern, high gain, and
purity of polarization. Horn antennas are used as feeds for
reflector and lens antennas and as a laboratory standard for
other antennas. A good collection of papers on horn antennas
can be found in Ref. 37. Horns can be of a rectangular or
circular shape as shown in Fig. 7.
Rectangular horns, derived from a rectangular waveguide,
can be pyramidal or sectoral E plane and H plane horns. The
E plane sectoral horn has a flare in the direction of the E field
of the dominant TE10 mode in the rectangular waveguide and
the H plane sectoral horn has a flare in the direction of the H
field. The pyramidal horn has a flare in both directions. The
radiation pattern of the horn antenna can be determined from
a knowledge of the aperture dimensions and the aperture
field distribution. The flare angle of the horn and its dimen-
Sectoral H-plane
Sectoral E-plane
Pyramidal
Conical
Figure 7. Examples of horn antennas.
ANTENNAS
b2
a2
h2
b1
a1
h1
Figure 8. A coaxial fed (a) microstrip antenna and (b) stacked microstrip antenna.
sions affect the radiation pattern and its directivity. Circular
horns derived from circular waveguides can be either conical,
biconical, or exponentially tapered.
The need for feed systems that provide low cross polarization and edge diffraction and more symmetrical patterns led
to the design of the corrugated horn (38). These horns have
corrugations or grooves along the walls which present equal
boundary conditions to the electric and magnetic fields when
the grooves are /4 to /2 deep. The conical corrugated horn,
referred to as the scalar horn, has a larger bandwidth than
the small flare angle corrugated horns.
Conformal Antennas
Microstrip antennas have become a very important class of
antennas since they received attention in the early 1970s.
These antennas are light weight, easy to manufacture using
printed circuit techniques, and are compatible with monolithic microwave integrated circuits (MMICs). In addition, an
attractive property of these antennas is that they are low profile and can be mounted on surfaces, that is, they can be made
to conform to a surface, hence they are referred to as conformal antennas. The microstrip antenna consists of a conducting patch or radiating element which can be square, rectangular, circular, or triangular etched on a grounded dielectric
substrate as shown in Fig. 8.
These antennas are an excellent choice for use on aircraft
and spacecraft. Microstrip antennas have been investigated
extensively over the past twenty years and the two volumes
published by Hall and Wood (39) provide an excellent description of various microstrip antennas, their design, and usage.
Microstrip antennas are fed either using a coaxial probe, a
microstrip line, proximity coupling, or through aperture coupling. A major disadvantage of these antennas is that they
are poor radiators and have a very narrow frequency bandwidth. They are often used in an array environment to
achieve the desired radiation characteristics. Larger frequency bandwidths are obtained by using stacked microstrip
antennas.
Antenna Arrays
Antenna arrays are formed by suitably spacing radiating elements in a one or two dimensional lattice. By suitably feeding
these elements with relative amplitudes and phases, these
arrays produce desired directive radiation characteristics.
The arrays allow a means of increasing the electric size
of the antenna without increasing the size of the individual
567
elements. Most arrays consist of identical elements which can
be dipoles, helices, large reflectors, or microstrip elements.
The array has to be designed such that the radiated fields
from the individual elements add constructively in the desired directions and destructively in the other directions.
Arrays are generally classified as end-fire arrays that produce
a beam directed along the axis of the array, or broadside
arrays with the beam directed in a direction normal to the
array. The beam direction can be controlled or steered using
a phased array antenna in which the phase of the individual
elements is varied. Frequency scanning arrays are an example where beam scanning is done by changing the frequency.
Adaptive array antennas produce beams in predetermined directions. By suitably processing the received signals, the antenna can steer its beam toward the direction of the desired
signal and simultaneously produce a null in the direction of
an undesired signal.
APPLICATIONS AND IMPACT ON SYSTEMS
Antennas enjoy a very large range of applications, both in the
military and commercial world. The most well-known applications of antennas to the average person are those associated
with radio, TV, and communication systems. Today, antennas
find extensive use in biomedicine, radar, remote sensing, astronomy, navigation, radio frequency identification, controlling space vehicles, collision avoidance, air traffic control,
GPS, pagers, wireless telephone, wireless local area networks
(LANs) etc. These applications cover a very wide range of frequencies as shown in Table 1 (2,3,40):
Antennas in Communication Systems
Antennas are one of the most critical components in a communication system since they are responsible for the proper
transmission and reception of electromagnetic waves. The antenna is the first part of the system that will receive or transmit a signal. A good design can relax some of the complex
system requirements involved in a communication link and
increase the overall system performance.
The choice of an antenna for a specific application (cellular,
satellite based, ground based, etc.), depends on the platform
to be used (car, ship, building, spacecraft, etc.), the environment (sea, space, land), the frequency of operation, and the
nature of the application (video, audio data, etc.). Communication systems can be broken into several different categories:
Direct (Line-of-Site) Links. A transmission link established
between two highly directional antennas. The link can be between two land-based antennas (radio relays); between a
tower and a mobile antenna (cellular communication); between a land-based antenna and a satellite antenna (earthspace communication); between two satellite antennas (space
communication). Usually these links operate at frequencies
between 1 to 25 GHz. A typical distance between two points
in a high capacity, digital microwave radio relay system is
about 30 miles.
Satellites and Wireless Communications. Antennas on orbiting satellites are used to provide communications between
various locations around the earth. In general, most telecommunication satellites are placed in a geostationary orbit
568
ANTENNAS
Table 1. Frequency Bands and General Usage
Band
Designation
Frequency
Range
Very low frequencies
(VLF)
330 kHz
Low frequency
(LF)
30300 kHz
Medium frequency (MF)
3003000 kHz
High frequency
(HF)
330 MHz
Very high frequency (VHF)
30300 MHz
Ultra high frequency (UHF)
L
3001000 MHz
24 GHz
48 GHz
812 GHz
Ku
1218 GHz
K
Ka
Submillimeter
waves
1827 GHz
2740 GHz
12 GHz
Usage
Long distance telegraphy, navigation. Antennas are physically large but electrically
small. Propagation is accomplished using earths surface
and the ionosphere. Vertically polarized wave
Aeronautical navigation services, long distance communications, radio broadcasting. Vertical polarization
Regional broadcasting and
communication links, AM
radio
Communications, broadcasting, surveillance, CB radio
(26.96527.225 MHz). Ionospheric propagation. Vertical
and horizontal propagation
Surveillance, TV broadcasting
(5472 MHz), (7688 MHz),
and (174216 MHz), FM radio (88108 MHz). Wind profilers
Cellular communications, surveillance TV (470890 MHz)
Long range surveillance, remote sensing
Weather, traffic control,
tracking, hyperthermia
Weather detection, long range
tracking
Satellite communications, missile guidance, mapping
Satellite communications, altimetry, high resolution
mapping
Very high resolution mapping
Airport surveillance
Experimental stage
satellites operate at the L, S, or Ku band, but increasing demand for mobile telephony and high speed interactive data
exchange is pushing the antenna and satellite technology into
higher operational frequencies (47). Future satellites will be
equipped with antennas at both the Ku and the Ka bands.
This will lead to greater bandwidth availability. For example,
the ETS-VI (A Japanese satellite comparable to NASAs
TDRS), carries five antennas: an S-band phased array, a 0.4
m reflector for 43/38 GHz, for up and down links, an 0.8 m
reflector for 26/33 GHz, a 3.5 m reflector for 20 GHz, and a
2.5 m reflector for 30 GHz and 6/4 GHz. Figure 10 shows a
few typical antennas used on satellites. It is expected that
millions of households, worldwide, will have access to dual
Ku/Ka band dishes in the twenty-first century. These households will be able to enjoy hundreds of TV channels from
around the world. Moreover, low cost access to high speed,
voice, data, and video communications will be available to
more customers (48).
Personal/Mobile Communication Systems. The vehicular antennas used with mobile satellite communications constitute
the weak link of the system. If the antenna has high gain,
then tracking of the satellite becomes necessary. If the vehicle
antenna has low gain, the capacity of the communication system link is diminished. Moreover, hand-held telephone units
require ingenious design due to lack of real estate on the
portable device.
There is more emphasis now in enhancing antenna technologies for wireless communications, especially in cellular
communications, which will enhance the link performance
and reduce the undesirable visual impact of antenna towers.
Techniques that utilize smart antennas, fixed multiple
beams, and neural networks are now being utilized to increase the capacity of mobile communication systems,
whether it is land-based or satellite-based (49). It is anticipated that in the twenty-first century the wire will no
longer dictate where we must go to use the telephone, fax, email, or run a computer. This will lead to the design of more
compact and more sophisticated antennas.
Satellite
(GEO), about 22,235 miles above the earth as shown in Fig.
9. There are also some satellites at lower earth orbits (LEOs)
that are used for wireless communications. Modern satellites
have several receiving and transmitting antennas which can
offer services such as video, audio, data transmission, and
telephone in areas that are not hard-wired. Moreover, directTV is now possible through the use of a small 18-inch reflector
antenna with 30 million users in the U.S. today (41,42).
Satellite antennas for telecommunications are used either
to form a large area-of-coverage beam for broadcasting or spot
beams (small area-of-coverage) for point-to-point communications. Also, multibeam antennas are used to link mobile and
fixed users that cannot be linked economically via radio, landbased relays (4346).
The impact of antennas on satellite technology continues
to grow. For example, very small aperture terminal dishes
(VSAT) at Ku band that can transmit any combination of
voice, data, and video using satellite networking, have become
valuable tools for several small and large companies. Most
22500 miles
Satellite dish
Satellite dish
Figure 9. A satellite communication system.
ANTENNAS
569
15.2 ft
C-band omni antenna
Dual subreflectors
47.1 ft
29.9 ft
Solar array
10.8-ft,
20-GHz
transmitting
antenna
Ka-band
command,
ranging, and
telemetry
antennas
Beam-forming
networks
7.2-ft, 30-GHz
receiving antenna
3.3-ft steerable antenna
Figure 10. Typical antennas on a satellite. (Courtesy, NASA Lewis Center)
Solar array
Antennas for Biomedical Applications
In many biological applications the antenna operates under
very different conditions than the more traditional free-space,
far-field counterparts. Near fields and mutual interaction
with the body dominate. Also, the antenna radiates in a lossy
environment rather than free space. Several antennas, from
microstrip antenna to phased arrays, operating at various frequencies, have been developed to couple electromagnetic energy in or out of the body. Most medical applications can be
classified into two groups (50): (1) therapeutic and (2) informational. Examples of therapeutic applications are hyperthermia for cancer therapy, enhancement of bone and wound
healing, nerve simulation, neural prosthesis, microwave angioplasty, treatment of prostatic hyperlastia, and cardiac ablation. Examples of informational applications are tumor detection using microwave radiometry, imaging using microwave
tomography, measurement of lung water content, and dosimetry.
Therapeutic applications are further classified as invasive
and noninvasive. Both applications require different types of
antennas and different restrictions on their design. In the
noninvasive applications (not penetrating the body), antennas
are used to generate an electromagnetic field to heat some
tissue. Antennas such as helical-coils, ring capacitors, dielectrically loaded waveguides, and microstrip radiators are attractive because of their compactness. Phased arrays are also
used to provide focusing and increase the depth of penetration. The designer has to choose the right frequency, size of
the antenna, and the spot size that the beam has to cover
in the body. The depth of penetration, since the medium of
propagation is lossy, is determined by the total power applied
or available to the antenna. Invasive applications require
some kind of implantation in the tissue. Many single antennas and phased or nonphased arrays have been extensively
used for treating certain tumors. A coaxial cable with an
extended center conductor is a typical implanted antenna.
This type of antenna has also been used in arteries to
soften arterial plaque and enlarge the lumen of narrowed
arteries.
Antennas have also been used to stimulate certain nerves
in the human body. As the technology advances in the areas
of materials and in the design of more compact antennas,
more antenna applications will be found in the areas of biology and medicine.
Radio Astronomy Applications
Another field where antennas have made a significant impact
is the field of astronomy. A radio telescope is an antenna system that astronomers use to detect radio frequency (RF) radiation emitted from extraterrestrial sources. Since radio wavelengths are much longer than those in the visible region,
radio telescopes make use of very large antennas to obtain
the resolution of optical telescopes. Today, the most powerful
radio telescope is located in the Plains of San Augustin, near
Sorocco, N.M. It is made of an array of 27 parabolic antennas,
each about 25 m in diameter. Its collecting area is equivalent
to a 130-m antenna. This antenna is used by over 500 astronomers to study the solar system, the Milky Way Galaxy, and
extraterrestrial systems. Puerto Rico is the site of the worlds
largest single-antenna radio telescope. It uses a 300-m spherical reflector consisting of perforated aluminum panels. These
panels are used to focus the received radio waves on movable
antennas placed about 168 meters above the reflector surface.
The movable antennas allow the astronomer to track a celestial object in various directions in the sky.
Antennas have also been used in constructing a different
type of a radio telescope, called radio interferometer. It consists of two or more separate antennas that are capable of
receiving radio waves simultaneously but are connected to
one receiver. The radio waves reach the spaced antennas at
different times. The idea is to use information from the two
antennas (interference) to measure the distance or angular
position of an object with a very high degree of accuracy.
Radar Applications
Modern airplanes, both civilian and military, have several antennas on board used for altimetry, speed measurement, collision avoidance, communications, weather detection, naviga-
570
ANTENNAS
Transmitter
Duplexer
Low noise
amplifier
Display
Pulse
modulator
LO
Mixer
IF
amp.
Detector
Video
amp.
Figure 11. A basic radar system.
tion, and a variety of other functions (40,5153). Each
function requires a certain type of antenna. It is the antenna
that makes the operation of a radar system feasible. Figure
11 shows a block diagram of a basic radar system.
Scientists in 1930 observed that electromagnetic waves
emitted by a radio source were reflected back by aircrafts
(echoes). These echoes could be detected by electronic equipment. In 1937, the first radar system, used in Britain for direction finding of enemy guns, operated around 20 MHz to
30 MHz. Since then, several technological developments have
emerged in the area of radar antennas. The desire to operate
at various frequencies led to the development of several, very
versatile and sophisticated antennas. Radar antennas can be
ground-based, mobile, satellite-based, or placed on any aircraft or spacecraft. The space shuttle orbiter, for example, has
23 antennas. Among these, four C-band antennas are used for
altimetry, two to receive and two to transmit. There are also
six L-band antennas and 3 C-band antennas used for navigation purposes.
Today, radar antennas are used for coastal surveillance,
air traffic control, weather prediction, surface detection
(ground penetrating radar), mine detection, tracking, air-defense, speed-detection (traffic radar), burglar alarms, missile
guidance, mapping of the surface of the earth, reconnaissance, and other uses.
In general, radar antennas are designed as part of a very
complex system that includes high power klystrons, traveling
wave tubes, solid state devices, integrated circuits, computers, signal processing, and a myriad of mechanical parts. The
requirements on the radar antennas vary depending on the
application (continuous wave, pulses radar, Doppler, etc.) and
the platform of operation. For example, the 23 antennas on
the space shuttle orbiter must have a useful life of 100,000
operational hours over a ten-year period or about 100 orbital
missions. These antennas are required to operate at temperatures from 150 F to 350 F during re-entry. The antennas
also have to withstand a lot of pressure and a direct lightning
strike. The antenna designer will have to meet all of these
constraints along with the standard antenna problems of polarization, scan rates, frequency agility, etc.
Impact of Antennas in Remote Sensing
Remote sensing is a radar application where antennas such
as horns, reflectors, phased arrays, and synthetic apertures
are used from an airplane or a satellite to infer the physical
properties of planetary atmosphere and surface or take images of objects.
There are two types of remote sensing: active and passive
(radiometry) and both are in wide use. In the active case a
signal is transmitted and the reflected energy, intercepted by
the radar as shown in Figure 12, is used to determine several
characteristics of the illuminated object such as temperature,
wind, shape, etc. In the passive case the antenna detects the
amount of microwave energy radiated by thermal radiation
from the objects on the earth. Radiometers are used to measure the thermal radiation of the ground surface and/or atmospheric condition (13,5456).
Most antennas associated with remote sensing are downward-looking, whose radiation patterns possess small, closein sidelobes. Remote sensing antennas require a very careful
design to achieve high beam efficiency, low antenna losses,
low sidelobes, and good polarization properties. Ohmic losses
in the antenna is perhaps the most critical parameter since it
can modify the apparent temperature observed by the radiometer system.
The degree of resolution of a remote map depends on the
ability of the antenna system to separate closely space objects
in range and azimuth. To increase the azimuth resolution a
technique called synthetic aperture is employed. Basically,
as an aircraft flies over a target the antenna transmits pulses
assuming the value of a single radiating element in a long
array. Each time a pulse is transmitted, the antenna, due to
the aircrafts motion, is further along the flight path. By storing and adding up the returned signals from many pulses, the
single antenna element acts as the equivalent of a very large
antenna, hundreds of feet long. Using this approach, an antenna system can produce maps approaching the quality of
good aerial photographs. This synthetic aperture antenna becomes a radio camera that can yield excellent remote imagery. Figure 13 depicts a reflectivity map of the earth taken by
NASAs scatterometer.
Today, antennas are used in remote sensing applications
for both the military and civilian sectors. For example, in the
1960s the US used remote sensing imaging from satellite and
airplanes to track missile activities over Cuba. In the 1970s,
remote sensing provided NASA with needed maps of the lunar surface before the Apollo landing. Also in July 1972,
NASA launched the first earth resource technology satellite
(ERTS-1). This satellite provided data about crops, minerals,
soils, urban growth, and other earth features. This program
Receiver
Pr
Transmitter
Pt
max
min
Figure 12. Active remote sensing (microwave scatterometer).
ANTENNAS
571
Figure 13. A reflectivity map of the earth taken by NASAs Scatterometer. (Courtesy, NASA/
JPL)
6. S. A. Schelkunoff and H. T. Friis, Antenna Theory and Practice,
New York: Wiley, 1952.
still continues its original success using the new series of satellites called the Landsats. In 1985, British scientists noted
the ozone depletion over Antarctica. In 1986, US and French
satellites sensed the Chernobyl nuclear reactor explosion that
occurred in Ukraine. Landsat images from 1975 to 1986
proved to be very instrumental in determining the deforestation of the earth, especially in Brazil. In 1992, hurricane Andrew, the most costly natural disaster in the history of the
United States, with winds of 160 miles per hour, was detected
on time by very high resolution radar on satellites. Because
of the ability to detect the hurricane from a distance, on time,
through sophisticated antennas and imagery, the casualties
from this hurricane were low. In 1993, during the flooding of
the Mississippi River, antenna images were used to assist in
emergency planning, and locating threatened areas (56). In
1997, NASA, using antennas, managed to receive signals
from Mars and have the entire world observe the pathfinder
maneuver itself through the rocky Martian terrain.
10. R. S. Elliot, Antenna Theory and Design, New York: PrenticeHall, 1981.
BIBLIOGRAPHY
16. R. E. Collin, Antennas and Radiowave Propagation, New York:
McGraw-Hill, 1985.
1. J. C. Maxwell, A Treatise on Electricity and Magnetism, London:
Oxford Univ. Press, 1873, 1904.
2. H. R. Hertz, Electric Waves, London: McMillian, 1893, New York:
Dover, 1962.
3. J. D. Kraus, Antennas since Hertz and Marconi, IEEE Trans.
Antennas Propag., 33: 131137, 1985.
4. J. Aharoni, Antennae, London: Oxford Univ. Press, 1946.
5. S. Silver, Microwave Antenna Theory and Design, MIT Radiation
Lab. Series, vol. 12. New York: McGraw-Hill, 1949.
7. S. A. Schelkunoff, Advanced Antenna Theory, New York: Wiley,
1952.
8. E. A. Laport, Radio Antenna Engineering, New York: McGrawHill, 1952.
9. R. E. Collin and F. J. Zucker (eds.), Antenna Theory Pts. 1 and 2,
New York: McGraw-Hill, 1969.
11. W. L. Stutzman and G. A. Thiele, Antenna Theory and Design,
New York: Wiley, 1981.
12. A. W. Rudge et al. (eds.), The Handbook of Antenna Design, vols.
1 and 2, London: Peter Peregrinus, 1982.
13. R. C. Johnson and H. Jasik, Antenna Engineering Handbook, New
York: McGraw-Hill, 1961, 1984.
14. K. F. Lee, Principles of Antenna Theory, New York: Wiley, 1984.
15. W. L. Weeks, Antenna Engineering, New York: McGraw-Hill,
1984.
17. J. R. Wait, Introduction to Antennas and Propagation, London: P.
Perigrinus, 1986.
18. L. V. Blake, Antennas. New York: Wiley, 1966, Norwood, MA:
Artech House, 1987.
19. E. Wolff, Antenna Analysis, New York: Wiley, 1966, Norwood,
MA: Artech House, 1988.
20. Y. T. Lo and S. W. Lee (eds.), Antenna Handbook: Theory Applications and Design, New York: Van Nostrand Reinhold, 1988.
21. J. D. Kraus, Antennas, New York: McGraw-Hill, 1950, 1988.
572
ANTENNAS FOR HIGH-FREQUENCY BROADCASTING
22. F. R. Connor, Antennas. London: Edward Arnold, 1989.
23. C. A. Balanis, Antenna Theory: Analysis and Design, New York:
Wiley, 1982, 1996.
24. IEEE standard definitions of terms for antennas, IEEE standard,
1451993, IEEE Trans. Antennas Propag., 27: 329, 1993.
25. C. A. Balanis, Antenna theory: a review, Proc. IEEE, 80: 723,
1992.
26. T. A. Mulligan, Modern Antenna Design, New York: McGrawHill, 1985.
27. T. Tsukiji and S. Tou, On polygonal loop antennas, IEEE Trans.
Antennas Propag., 28: 571575, 1980.
28. A. E. Love (ed.), Reflector Antennas, New York: IEEE Press, 1978.
29. P. J. Wood, Reflector Analysis and Design, London: Peter Peregrinus, 1980.
30. J. D. Kraus, The corner reflector antenna, Proc. IRE, 28: 513
519, Nov. 1940.
31. A. W. Rudge, Off-set parabolic reflector antennas: a review, Proc.
IEEE, 66: 15921618, 1978.
32. W. E. Kock, Metal lens antennas, Proc. IRE, 34: 828836, 1946.
33. C. H. Walter, Traveling Wave Antennas, New York: McGrawHill, 1965.
34. S. Uda, Wireless beam of short electric waves, J. IEEE (Japan),
12091219, Nov. 1927.
35. H. Yagi, Beam transmission of ultra short waves, Proc. IEEE, 26:
715741, 1928.
36. V. H. Rumsey, Frequency Independent Antennas, Academic
Press, 1966.
37. A. W. Love (ed.), Electromagnetic Horn Antennas, New York:
IEEE Press, 1976.
38. P. J. B. Clarricoats and A. D. Olver, Corrugated Horns for Microwave Antennas, London: Peter Peregrinus, 1984.
39. J. R. James and P. S. Hall, Handbook of microstrip antennas,
Vols. 1 and 2, London: Peter Peregrinus, 1989.
40. J. L. Eaves and E. K. Reedy (eds.), Principles of Modern Radar,
New York: Van Nostrand Reinhold, 1987.
41. J. Griffiths, Radio Wave Propagation and Antennas, London:
Prentice-Hall International, Chaps. 8 and 9, 1987.
42. F. J. Ricci, Personal Communications Systems Applications, Englewood, New Jersey: Prentice-Hall, 1997.
43. T. T. Ha, Digital Satellite Communications, New York: MacMillan, 1986.
44. M. Rubelj, P. F. Wahid, and C. G. Christodoulou, A microstrip
array for direct broadcast satellite receivers, Microw. Opt. Technol. Lett., 15 (2): 6872, June 1997.
45. W. L. Pritchard and J. A. Sciulli, Satellite Communications Systems Engineering, Englewood Cliffs, NJ: Prentice-Hall, 1986.
46. L. H. Van Tress (ed.), Satellite Communication Systems, New
York: IEEE Press, 1979.
47. S. D. Dorfman, Satellite communications in the 21st century,
Strategies Summit, Telecom 95 (IUT), Geneva, Switzerland, Oct.
10, 1995.
48. A. Jagoda and M. de Villepin, Mobile Communications, New
York: Wiley, 1993.
49. A. H. El Zooghby, C. G. Christodoulou, and M. Georgiopoulos,
Performance of radial basis functions for direction of arrival estimation with antenna arrays, IEEE Trans. Antennas Propag., 45:
16111617, 1997.
50. C. H. Durney, Antennas and other electromagnetic applicators in
biology and medicine, Proc. IEEE, 80: 194199, 1992.
51. M. I. Skolnik, Introduction to Radar Systems, 2nd ed., New York:
McGraw-Hill, 1980.
52. D. K. Barton, Radar Systems Analysis, Dedham, MA: Artech
House, 1976.
53. G. W. Stimson, Introduction to Airborne Radar, Hughes Aircraft
Company, Radar Systems Group, El Segundo, CA, 1983.
54. C. T. Swift, Passive microwave remote sensing of the oceana
review, Boundary Layer Meteorology, 18: 2554, 1980.
55. R. H. Dicke, The measurement of thermal radiation at microwave
frequencies, Rev. Sci. Instrum., 17: 268275, 1946.
56. Remote sensing in history, in NASAs Observatorium (Internet
pages).
C. CHRISTODOULOU
P. F. WAHID
University of Central Florida
ANTENNAS,
ANTENNAS,
ANTENNAS,
ANTENNAS,
ANTENNAS,
ACTIVE. See ACTIVE ANTENNAS.
CONFORMAL. See CONFORMAL ANTENNAS.
CONICAL. See CONICAL ANTENNAS.
DIPOLE. See DIPOLE ANTENNAS.
DIRECTIVE. See DIRECTIVE ANTENNAS.
APERTURE ANTENNAS
APERTURE ANTENNAS
Aperture antennas are most commonly used in the microwave- and millimeter-wave frequencies. There are a large
number of antenna types for which the radiated electromagnetic fields can be considered to emanate from a physical aperture. Antennas that fall into this category include several
types of reflectors: planar (flat plate) arrays, lenses, and
horns. The geometry of the aperture geometry may be square,
rectangular, circular, elliptical, or virtually any other shape.
Aperture antennas are very popular for aerospace applications because they can be flush mounted onto the spacecraft
or aircraft surface. Their opening can be covered with an electromagnetic (dielectric) window material or radome to protect
the antenna from environmental conditions (1). This installation will not disturb the aerodynamic profile of the vehicle,
which is of critical importance in high-speed applications.
In order to evaluate the distant (far-field) radiation patterns, it is necessary to know the internal currents that flow
on the radiating surfaces. However, these current distributions may not be exactly known and only approximate or experimental measurements can provide estimates for these
data. To expedite the process, it is necessary to have alternative methods to compute the radiation patterns of the aperture antennas. A technique based on the equivalence principle allows one to make a reasonable approximation to the
fields on, or in the vicinity of, the physical antenna structure
and subsequently to compute the radiation patterns.
Field equivalence, first introduced by Schelkunoff (2), is a
principle by which the actual sources on an antenna are replaced by equivalent sources on an external closed surface
that is physically outside of the antenna. The fictitious
sources are said to be equivalent within a region because they
produce the same fields within that region. Another key concept is Hugenss principle (3), which states that the equivalent source at each point on the external surface is a source
of a spherical wave. The secondary wave front can be con-
609
structed as the envelope of these secondary spherical waves
(4).
Using these principles, the electrical and/or magnetic
fields in the equivalent aperture region can be determined
with straightforward approximate methods. The fields elsewhere are assumed to be zero. In most applications, the closed
surface is selected so that most of it coincides with the conducting parts of the physical structure. This is preferred because the vanishing of the tangential electrical components
over the conducting parts of the surface reduces the physical
limits of integration. The formula to compute the fields radiated by the equivalent sources is exact, but it requires integration over the closed surface. The degree of accuracy depends on the knowledge of the tangential components of the
fields over the closed surface.
Aperture techniques are especially useful for parabolic reflector antennas, where the aperture plane can be defined immediately in front of the reflector. Parabolic reflectors are
usually large, electrically. More surprisingly, aperture techniques can be successfully applied to small aperture waveguide horns. However, for very small horns with an aperture
dimension of less than approximately one wavelength, the assumption of zero fields outside the aperture fails unless the
horn is surrounded by a planar conducting flange (5). In this
section, the mathematical formulas will be developed to analyze the radiation characteristics of aperture antennas. Emphasis will be given to the rectangular and circular configurations because they are the most commonly used geometries.
Due to mathematical complexities, the results will be restricted to the far-field region.
One of the most useful concepts to be discussed is the farfield radiation pattern that can be obtained as a Fourier
transform of the field distribution over the equivalent aperture, and vice versa. The existing relationship of the Fourier
transforms theory is extremely important since it makes all
of the operational properties of the Fourier transform theory
available for the analysis and synthesis of aperture antennas.
Obtaining analytical solutions for many simple aperture distributions in order to design aperture antennas is useful.
More complex aperture distributions, which do not lend themselves to analytical solutions, can be solved numerically. The
increased capabilities of the personal computer (PC) have resulted in its acceptance as a conventional tool of the antenna
designer. The Fourier-transform integral is generally well behaved and does not present any fundamental computational
problems.
Considering the use of the Fourier transform, first consider
rectangular apertures in which one aperture dimension is
large in wavelength and the other is small in terms of wavelength. This type of aperture is approximated as a line source
and is easily treated with a one-dimensional Fourier transform (6). For many kinds of rectangular aperture antennas
such as horns, the aperture distributions in the two principal
plane dimensions are independent. These types of distributions are said to be separable. The total radiation pattern is
obtained for separable distributions as the product of the pattern functions obtained from a one-dimensional Fourier
transform, which corresponds to the two principal plane distributions.
If the rectangular aperture distribution is not able to be
separated, the directivity pattern is found in a similar manner to the line-source distribution except that the aperture
J. Webster (ed.), Wiley Encyclopedia of Electrical and Electronics Engineering. Copyright # 1999 John Wiley & Sons, Inc.
610
APERTURE ANTENNAS
field is integrated over two dimensions rather than one dimension (7). This double Fourier transform can also be applied to circular apertures and can be easily evaluated on a
PC.
For all aperture distributions, the following observations
are made (8):
1. A uniform amplitude distribution yields the maximum
directivity (nonuniform edge-enhanced distributions for
supergain being considered impractical), but at high
side-lobe levels.
2. Tapering the amplitude at the center, from a maximum
to a smaller value at the edges, will reduce the side-lobe
level compared with the uniform illumination, but it results in a larger (main-lobe) beam width and less directivity.
3. An inverse-taper distribution (amplitude depression at
the center) results in a smaller (main-lobe) beam width
but increases the side-lobe level and reduces the directivity when compared with the uniform illumination
case.
4. Depending on the aperture size in wavelengths and
phase error, there is a frequency (or wavelength) for
which the gain peaks, falling to smaller values as the
frequency is either raised or lowered.
Lastly, we consider aperture efficiencies. The aperture efficiency is defined as the ratio of the effective aperture area to
the physical aperture area. The beam efficiency is defined as
the ratio of the power in the main lobe to the total radiated
power. The maximum aperture efficiency occurs for a uniform
aperture distribution, but maximum beam efficiency occurs
for a highly tapered distribution. The aperture phase errors
are the primary limitation of the efficiency of the antenna.
HUYGENSS PRINCIPLE
The principle proposed by Christian Huygens (16291695) is
of fundamental importance in the development of the wave
theory. Huygenss principle states that, Each point on a primary wavefront serves as the source of spherical secondary
wavelets that advance with a speed and frequency equal to
those of the primary wave. The primary wavefront at some
later time is the envelope of these wavelets (9,10). This is
illustrated in Fig. 1 for spherical and plane waves modeled
Incoming
wave
Outgoing
wave
Figure 2. Diffraction of waves through a slit based on the Huygens
principle.
as a construction of Huygens secondary waves. Actually, the
intensities of the secondary spherical wavelets are not uniform in all directions, but vary continuously from a maximum
in the direction of wave propagation to a minimum of zero
in the backward direction. As a result, there is no backward
propagating wavefront. The Huygens source approximation is
based on the assumption that the magnetic and electrical
fields are related as a plane wave in the aperture.
Let us consider the situation shown in Fig. 2, in which an
infinite electromagnetic plane wave is incident on an infinite
flat sheet that is opaque to the waves. This sheet has an opening that is very small in terms of wavelengths. Accordingly,
the outgoing wave corresponds to a spherical wavefront propagating from a point source. That is, when an incoming wave
comes against a barrier with a small opening, all but one of
the effective Huygens point sources are blocked, and the energy coming through the opening behaves as a single point
source. In addition, the outgoing wave emerges in all directions, instead of just passing straight through the slit.
On the other hand, consider an infinite plane electromagnetic wave incident on an infinite opaque sheet shown in Fig.
3 that has an opening a. The field everywhere to the right of
the sheet is the result of the section of the wave that passes
through the slot. If a is large in terms of wavelengths, the
field distribution across the slot is assumed, to a first approximation, to be uniform. The total electromagnetic field at a
point to the right of the opening is obtained by integrating
the contributions from an array of Huygens sources distributed over the length a. We calculate the electrical field at
point P on a reference plane located at a distance R0 behind
Spherical
wave
front
R0
ri
Incoming
wave
Plane-wave front
(a) Spherical
(b) Plane
Figure 1. Spherical and plane-wave fronts constructed with Huygens secondary waves.
Figure 3. Plane wave incident on an opaque sheet with a slot of
width a.
the plane by Huygenss principle (11):
e jkr
dy
r
(1)
For points near to the array, the integral does not simplify
but can be reduced to the form of Fresnel integrals.
The actual evaluation of this integral is best achieved on a
PC computer, which reduces the integral to the summation of
N Huygens sources:
E=
N
E0
i=1
e jkr i
ri
R0
2a2
5
10
15
(2)
where ri is the distance from the ith source to point P. The
field variation near the slot that is obtained in this way is
commonly called a Fresnel diffraction pattern (4).
For example, let us consider the case in which the slot
length a is 5 cm and the wavelength is 1.5 cm (20 GHz). We
can use Eq. (2) to compute the field along a straight line parallel to the slot and distance R0 from it. The field variation
for R0 2.5 cm shown in Fig. 4(a) is well within the near field
(Fresnel region). As we continue to increase R0 the shape of
the field variation along this line continues to vary with R0
until we reach the far field or Fraunhofer region. [See the
trends in Figs. 4(b), 4(c), and 4(d)]. Once we have entered the
Fraunhofer region, the pattern is invariant to range. For the
point to be in the far field, the following relationship must
exist:
(3)
where a is the width of the slot and is the wavelength. Thus,
the larger the aperture or the shorter the wavelength, the
greater the distance at which the pattern must be measured
if we wish to avoid the effects of Fresnel diffraction.
Huygenss principle is not without limitations as it neglects the vector nature of the electromagnetic field space. It
also neglects the effect of the currents that flow at the slot
edges. However, if the aperture is sufficiently large and we
confine our attention to directions roughly normal to the aperture, the scalar theory of Huygenss principle gives very satisfactory results.
Geometric optic techniques are commonly applied in reflector antennas to establish the fields in the reflector aperture plane. This procedure is referred to as the aperture field
method and it is employed as an alternative to the so-called
induced current method, which is based upon an approximation for the electric current distribution on the reflector surface. The fields in the aperture plane can be thought of as an
ensemble of Huygens sources. The radiation pattern can be
computed via a numerical summation of the sources.
EQUIVALENCE PRINCIPLE
The ability to determine electromagnetic waves radiated
fields via field-equivalence principles is a useful concept and
the development can be traced back to Schelkunoff (2). The
equivalence principle often makes an exact solution easier to
obtain or suggests approximate methods that are of value in
611
20
20
Relative amplitude (dB)
E0
0
Distance along the y axis (cm)
(a) R0 = 2.5 cm
20
0
Distance along the y axis (cm)
(b) R0 = 5 cm
20
0
Distance along the y axis (cm)
(c) R0 = 15 cm
20
0
Distance along the y axis (cm)
(d) R0 = 20 cm
20
0
5
10
15
20
20
Relative amplitude (dB)
E=
0
5
10
15
20
20
Relative amplitude (dB)
Relative amplitude (dB)
APERTURE ANTENNAS
0
5
10
15
20
20
Figure 4. Electromagnetic field versus distance along the Y axis.
simplifying antenna problems. Field-equivalence principles
are treated at length in the literature and we will not consider the many variants here. The book by Collin and Zucker
(12) is a useful source of references in this respect. The basic
concept is illustrated in Fig. 5. The electromagnetic source
region is enclosed by a surface S that is sometimes referred
to as Huygenss surface.
612
APERTURE ANTENNAS
In essence, Huygenss principle and the equivalence theorem shows how to replace actual sources by a set of equivalent sources spread over the surface S (13). The equivalence
principle is developed by considering a radiating source, electrically represented by current densities J1 and M1. Assume
that the source radiates fields E1 and H1 everywhere. We
would like to develop a method that will yield the fields outside the closed surface. To accomplish this, a closed surface S
is shown by the dashed lines that enclose the current densities J1 and M1. The volume inside S is denoted by V. The
primary task is to replace the original problem [Fig. 5(a)] by
an equivalent one that will yield the same fields E1 and H1
[Fig. 5(b)]. The formulation of the problem can be greatly
aided if the closed surface is judiciously chosen so that the
fields over most of the surface, if not the entire surface, are
known a priori.
The original sources J1 and M1 are removed, and we assume that there exists a field E and H inside V. For this field
to exist within V, it must satisfy the boundary conditions on
the tangential electrical and magnetic field components on
surface S. Thus on the imaginary surface S, there must exist
equivalent sources (14):
H1 H )
J S = n (H
(4)
n (E
E1 E )
M S = n
(5)
These equivalent sources radiate into an unbounded space.
The current densities are said to be equivalent only outside
region V, because they produce the original field (E1, H1).
A field E or H, different from the original, may result
within V.
The sources for electromagnetic fields are always, apparently, electrical currents. However, the electrical current distribution is often unknown. In certain structures, it may be a
complicated function, particularly for slots, horns, reflectors,
and lenses. With these types of radiators, the theoretical work
is usually not based on the primary current distributions.
Rather, the results are obtained with the aid of what is
known as aperture theory (15). This simple, sound theory is
based upon the fact that an electromagnetic field in a sourcefree, closed region is completely determined by the values of
tangential E or tangential H on the surface of the closed re-
E 1, H 1
E 1, H 1
V1
S
E, H
J1
M1
JS = n (H1 H)
MS = n (E1 E)
(a) Original problem
(b) Equivalent problem
Figure 5. Equivalence principle with a closed Huygens surface S enclosing sources.
(a) Horn
(b) Parabola
(c) Lens
Figure 6. Some apertures yielding the same electromagnetic fields
to the right side of the Huygens surface S.
gion. For exterior regions, the boundary condition at infinity
may be employed, in effect, to close the region. This is exemplified by the following case.
Without changing the E and H fields external to S, the
electromagnetic source region can be replaced by a zero-field
region with appropriate distributions of electrical and magnetic currents (Js and Ms) on the Huygens surface. This example is overly restrictive and we could specify any field within
S with a suitable adjustment. However, the zero internal field
approach is particularly useful when the tangential electrical
fields over a surface enclosing the antenna are known or can
be approximated. In this case, the surface currents can be
obtained directly from the tangential fields, and the external
field can be determined.
Assuming zero internal field, we can consider the electromagnetic sources inside S to be removed, and the radiated
fields outside S are then determined from the electrical and
magnetic surface current distributions alone. This offers significant advantages when the closed surface is defined as a
two-hemisphere region, with all sources contained on only one
side of the plane. If either the electrical or magnetic fields
arising from these sources can be determined over the planar
Huygens surface S, then the radiated fields on the far side of
the plane can be calculated. The introduction of an infinite
conducting sheet just inside the Huygens surface here will
not complicate the calculations of the radiated fields in the
other half-space (16). This infinite-plane model is useful for
antennas the radiation of which is directed into the right
hemisphere (Fig. 6), and has found wide application in dealing with aperture antennas. For instance, if the antenna is a
rectangular horn, it is assumed the horn transitions into a
infinite flange. All tangential fields outside the rectangular
boundary along the infinite Huygens surface are taken to be
zero.
When the limitations of the half-space model are acceptable, it offers the important advantage that either the electrical or magnetic currents need to be specified. However,
knowledge of both is not required. It must be emphasized that
any of the methods described before will produce exact results
over the Huygens surface.
In the analysis of electromagnetic problems, often it is easier to form equivalent problems that will yield the same solution only within a region of interest. This is the case for aperture antenna problems.
APERTURE ANTENNAS
The steps that must be used to form an equivalent problem
and solve an aperture antenna problem are as follows:
1. Select an imaginary surface that encloses the actual
sources (the aperture). The surface must be judiciously
chosen so that the tangential components of the electrical field and/or the magnetic field are known, exactly or
approximately, over its entire span. Ideally, this surface
is a flat plane extending to infinity.
2. Over the imaginary surface, form equivalent current
densities JS and MS over S, assuming that the E and
H fields within S are not zero.
3. Lastly, solve the equivalent-aperture problem.
source is positioned in a ground plane of infinite extent. This
model is simple and yet the analysis gives results that illustrate the main features of the most practical of the two-dimensional apertures. The line-source distribution does have a
practical realization, namely, in a long one-dimensional array
that has sufficient elements to enable it to be approximated
to a continuous distribution. The applicable transform pair is
(7,17)
E(sin ) =
There are many kinds of antennas for which the radiated
electromagnetic fields emanate from a physical aperture. This
general class of antennas provides a very convenient basis for
analysis and permits a number of well-established mathematical techniques to be applied that provides expressions for the
distant radiation fields.
Horns or parabolic reflectors, in particular, can be analyzed as aperture antennas. Incident fields are replaced by
equivalent electrical and magnetic currents. With use of vector potentials, the far fields are found as a superposition of
each source. Generally one can assume that the incident field
is a propagating free-space wave, the electrical and magnetic
fields of which are proportional to each other. This will give
the Huygens source approximation and allow us to use integrals of the electric field in the aperture. Each point in the
aperture is considered a source of radiation.
The first step involved in the analysis of aperture antennas
is to calculate the electromagnetic fields over the aperture
due to the sources on the rearward side of the infinite plane
and to use these field distributions as the basis for the prediction of the distant fields in the forward half-space. The electromagnetic fields in the aperture plane cannot be determined
exactly but approximation distributions can be found by many
different methods, which are dependent upon the antenna.
One can find the far-field radiation pattern for various distributions by a Fourier-transform relation.
For instance, consider a line source of length Lw using the
coordinate system as illustrated in Fig. 7. Assume that the
(6)
sin
(7)
E(sin )e jkx
E( ) =
L w /2
E(x)e jkx sin dx
(8)
+L w /2
Note that Eq. (8) is a relative relation.
For example, consider a uniform distribution for which
E(x) =
1
Lw
(9)
The field distribution pattern can be found by incorporating
this into Eq. (8):
E( ) =
1
Lw
L w /2
e( j2 x/) sin dx
(10)
L w /2
We complete the straightforward integration to get the final
result:
E( ) =
L
sin
Lw sin
(11)
This sin(x)/x distribution is very important in antenna theory
and is the basis for many antenna designs. It has a first sidelobe level of 13.2 dB.
Another popular continuous aperture distribution is the cosine raised to power n,
E(x) = cosn
0
d(sin )
where k 2/ . For real values of , 1 sin 1, the field
distribution represents radiated power, while outside this region it represents reactive or stored power (18). The field distribution E(sin ), or an angular spectrum, refers to an angular distribution of plane waves. The angular spectrum for a
finite aperture is the same as the far-field pattern, E(). Thus,
for a finite aperture the Fourier integral representation of Eq.
(6) may be written (8):
sin
y
E(x)e jkx sin dx
and
E(x) =
RECTANGULAR APERTURES
613
x
Lw
(12)
Lw
Figure 7. Coordinate system used to analyze a linear aperture of
length Lw.
where Lw /2 x Lw /2. This is shown in Fig. 8 for n 0, 1,
2, and 3. To make a relative comparison of the various distributions, we must first normalize to the transmitted power of
the uniform case. To do this, we multiply the pattern function
614
APERTURE ANTENNAS
Relative illumination (dimensionless)
0.8
0.6
0.4
Uniform
cos
cos2
cos3
0.2
Figure 8. Some popular aperture distributions.
LW
2
0
Position along the x axis (cm)
by the normalization constant:
Cp =
1
L w /2
L w /2
(13)
E (x) dx
E(x) = 1 +
To demonstrate the principles, we computed the antenna radiation pattern of a 1 meter long line-source antenna for
cosine0 (uniform), cosine1, and cosine2 distributions. The operating wavelength is 3 cm. The resulting patterns are shown
in Fig. 9. These data indicate that the more heavily tapered
illuminations result in decreased side-lobe levels, but at a
penalty of main beam peak gain.
Many distributions actually obtained in practice can be approximated by one of the simpler forms or by a combination
of simple forms. For example, a common linear aperture distribution is the cosine on a pedestal p:
x
(14)
Lw
x
Lw /2
(15)
x
Lw /2
(16)
for Lw /2 x 0, and
E(x) = 1
for 0 x Lw /2.
In practice, the rectangular aperture is probably the most
common microwave antenna. Because of its configuration, the
rectangular coordinate system is the most convenient system
to express the fields at the aperture. The most common and
convenient coordinate system used to analyze a rectangular
aperture is shown in Fig. 10. The aperture lies in the xy
plane and has a defined tangential aperture distribution E(x,
y). In keeping with the equivalence principle we shall assume
Relative decibels (dB)
10
Uniform
cos
cos2
20
30
40
Figure 9. Radiation patterns of line
sources for three different aperture distributions (Lw 1 m, 3 cm).
50
LW
2
where 0 p 1. This is a combination of a uniform plus a
cosine type distribution. The triangular distribution is popular:
E(x) = p + (1 p) cos
(deg)
APERTURE ANTENNAS
Bw
Aw
y
Figure 10. Coordinate system used to analyze rectangular aperture
of dimensions Aw, Bw.
the xy plane is a closed surface that extends from to
in the xy plane. Outside the rectangular aperture boundaries we shall assume that the field distribution is zero for all
points on this infinite surface. The task is to find the fields
radiated by it, the pattern beam widths, the side-lobe levels
of the pattern, and the directivity.
Note that a horn of aperture size Aw by Bw, with Aw / 1
and Bw / 1, can be analyzed as a continuous line source. If
these conditions are not met, the pattern must be obtained by
the integral (19):
E(, ) =
B w /2
B w /2
A w /2
E(x, y)e j(k x x+k y y) dx dy
(17)
A w /2
where
kx k sin cos
ky k sin sin
These are the x and y components of the propagation vector
k (20).
For many types of antennas, such as the rectangular horn,
the x and y functions are separable and may be expressed by
the form
E(x, y) = E(x)E(y)
sin kx
E(kx , ky ) =
Aw
2
Aw
kx
2
For nonseparable distributions, the integration of Eq. (17)
is best carried out on a PC computer using numerical methods. Figure 11 is a listing of a simple program written in Basic that can be run on any PC computer.
In running the program, 0 corresponds to the principal
plane pattern in the xz plane while 90 is the principal
plane pattern in the yz plane. For example, consider an aperture with Aw 75 cm, Bw 125 cm, and 3 cm. Assume
cosine distribution in each plane. The principal plane patterns in the x plane and y plane and the pattern in the intercardinal plane ( 45) that result are shown in Fig. 12.
We applied the computer code to compute the secondary
pattern characteristic produced by uniform, cosine raised to
power n, cosine on a pedestal p, and triangular aperture distributions. The results shown in Table 1 compare the gain,
beam width, and the first side-lobe levels. All gain levels are
compared with the uniform illumination case.
A uniform line-source or rectangular aperture distribution
produces the highest directivity. However, the first side lobe
is only about 13.2 dB down. Thus, aperture distributions
used in practice must be a trade-off or a compromise between
the desired directivity (or gain) and side-lobe level.
CIRCULAR APERTURES
Circular antennas form the largest single class of aperture
antennas. For instance, the circular parabolic reflector is used
extensively in telecommunications and radar applications.
The most common and most convenient coordinate system
used to analyze the radiation from a circular aperture of diameter Dw is shown in Fig. 13. The radiation field pattern for
a circular aperture can be calculated by applying Huygenss
principle in a similar way to that for a rectangular aperture
(a). The simplest form of a circular aperture distribution is
one in which the field does not vary with , that is, one that
is rotationally symmetric. This is not always true in practice;
however, we will assume that case here in order to demonstrate the methodology of analyzing circular apertures.
As was the case with rectangular apertures, a Fouriertransform relationship exists between the antenna distribution and the far-field radiation pattern. For a circular symmetric aperture distribution, the radiation pattern can be
written in normalized form (6):
(18)
For this particular distribution, the pattern in the principal
xz plane can be determined from a line-source distribution
E(x) while the pattern in the yz plane can be determined
from a line-source distribution E(y).
We use the Fourier transform and ignore the polarization
of the electric field in the aperture to get the result
sin ky
Bw
2
Bw
ky
2
615
E(u) =
1
3
2
0
E(P)e jPu cos( ) P dP d
(20)
where
u=
Dw sin
and the normalized radius is
(19)
This pattern in both planes is given by a k-space function. We
conclude that the result quoted for a line-source distribution
applies for the principal planes of separable rectangular antenna apertures.
P=
2r
Dw
For a uniformly illuminated circular aperture, the normalized
field pattern as a function of and Dw is (8)
E(u) =
2J1 (u)
u
(21)
CLS
Program to compute rectangular aperture radiation patterns
REDIM A(101, 101), patternx(200), patterny(200), intercardinal(200)
PI 3.14159265358#
Results stored in file Rect.dat
OPEN Rect.dat FOR OUTPUT AS #1
------ input data -----frequency 10
Aw 75: Bw 125
--------------------wavelength 30 / frequency
Awave Aw / wavelength: Bwave Bw / wavelength
Load aperture distribution: all dimensions in cm
FOR I 1 TO 101
FOR J 1 TO 101
X Aw * (51 - I) / 100: Y Bw * (51 - J) / 100
A(I, J) COS(PI * X / Aw) * COS(PI * Y / Bw)
Note: a(I, J) is the aperture distribution in the X and Y plane
NEXT J
NEXT I
Normalize power over aperture to that in a uniform distribution
power 0
FOR I 1 TO 101
FOR J 1 TO 101
power power A(I, J) 2
NEXT J
NEXT I
power power / 101 2 : power SQR(power)
Cp 1 / power
Cp is the desired normalization constant
FOR phi 0 TO 90 STEP 45
Note: Phi 0 results in principal X-plane pattern
Phi 45 results in intercardinal plane pattern
Phi 90 results in principal Y-plane pattern
K 0: G0DB 0
FOR theta 0 TO 10 STEP .1
PRINT Computing for theta; theta; Phi; phi
RE 0: IM 0
K K 1
Integration over aperture
FOR I 1 TO 101
FOR J 1 TO 101
X Aw * (51 - I) / 100: Y Bw * (51 - J) / 100
psi (2 * PI * X / wavelength) * SIN(theta * PI / 180) * COS(phi * PI / 180)
psi psi (2 * PI * Y / wavelength) * SIN(theta * PI / 180) * SIN(phi * PI / 180)
RE RE A(I, J) * COS(psi)
IM IM A(I, J) * SIN(psi)
NEXT J
NEXT I
TMM Cp * SQR(RE 2 IM 2) / 101 2
IF theta 0 THEN
G0DB 20 * LOG(TMM) / LOG(10): GV TMM
END IF
TOT 20 * LOG(TMM) / LOG(10)
IF phi 0 THEN patternx(K) TOT
IF phi 45 THEN intercardinal(K) TOT
IF phi 90 THEN patterny(K) TOT
NEXT theta
NEXT phi
K 0
FOR theta 0 TO 10 STEP .1
K K 1
PRINT #1, K, patternx(K), intercardinal(K), patterny(K)
PRINT Angle; theta; Ex; patternx(K); Exy; intercardinal(K): Ey; patterny(K)
NEXT theta
Print
PRINT Peak Gain G0 (dB); G0DB; G0(voltage); GV; G0(power); GV 2
CLOSE #1
END
Figure 11. Computer program to compute a radiation pattern from a rectangular aperture.
616
APERTURE ANTENNAS
617
x plane
Intercardinal
10
Relative decibels (dB)
y plane
20
30
40
50
Angle from boresight (deg)
in which J1 is a first-order Bessel function. This can also be
expressed as
D
sin
2J1
E( ) =
Dw sin
Figure 12. Radiation patterns for a rectangular aperture (Aw 75 cm, Bw cm,
3 cm).
Other types of circular aperture distributions include the
cosine raised to power n:
E(r) = cosn
(22)
r
(23)
Dw
where 0 r Dw /2.
The cosine on a pedestal p distribution is here defined:
The uniformly illuminated circular aperture radiation pattern
has a first side-lobe level of 17.6 dB compared to 13.2 dB
for that of the uniformly illuminated rectangular aperture.
E(r) = p + (1 p) cos
r
(24)
D2
Table 1. Radiation Pattern Characteristics Produced by Various Linear Aperture Distributions
Distribution
Normalized
Half-Power
Beam Width
(deg)
HPbw/K
Normalized
Null-to-Null
Beam Width
(deg)
NULLbw*/K
Side-Lobe
Level
(dB):
SLL dB
Normalized
Side-Lobe
Angle (deg)
SLpos/K
50.67
114.67
13.26
82.00
1
2
3
4
5
68.67
82.67
95.33
106.00
116.67
172.00
229.33
286.67
344.00
402.00
23.00
31.46
39.29
46.74
53.93
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0
68.67
64.67
62.00
59.33
58.00
56.00
54.67
54.00
52.67
52.00
50.67
172.00
162.00
152.67
144.67
138.00
132.67
127.67
123.33
120.00
117.33
114.67
73.34
114.59
Comments
Uniform
Cosine raised
to power n
Cosine on a
pedestal p
Triangular
n
n
n
n
n
p
p
p
p
p
p
p
p
p
p
p
Gain Relative
to Uniform
(dB)
G0 dB
Power Gain
Factor Relative
to Uniform
G0 power
Voltage Gain
Factor Relative
to Uniform
G0 volts
0.00
1.000
1.000
108.33
135.50
163.00
191.00
219.00
0.91
1.76
2.40
2.89
3.30
0.810
0.666
0.576
0.514
0.468
0.900
0.816
0.759
0.717
0.684
23.01
23.00
21.66
20.29
18.92
17.65
16.53
15.55
14.69
13.93
13.26
108.33
98.00
97.00
93.67
90.67
88.33
86.67
85.00
83.83
82.67
82.00
0.91
0.68
0.50
0.35
0.24
0.15
0.09
0.05
0.02
0.00
0.00
0.810
0.855
0.892
0.923
0.947
0.966
0.979
0.989
0.995
0.998
1.000
0.900
0.925
0.944
0.961
0.973
0.983
0.989
0.995
0.998
0.999
1.000
26.52
164.00
1.25
0.749
0.866
618
APERTURE ANTENNAS
efficiency M.
and the parabolic raised to a power n distribution (16) is
E(r) = 1
r
Dw /2
2
n
M =
(25)
M
A
(27)
Using these symbols, the directivity of an antenna is given by
To analyze the various circular aperture distributions we
can utilize a PC using numerical methods to perform the aperture integration. Actually a variant of the two-dimensional
Fourier-transform relation used in the rectangular aperture
analysis can be used. It works for any aperture rim shape
including a circle. A listing of the program written in Basic
appears in Fig. 14.
To demonstrate the behavior of various distributions, the
computer code was applied to compare the secondary pattern
characteristic produced by uniform, cosine raised to power n,
cosine on a pedestal p, and parabolic raised to power n distributions. The results shown in Table 2 compare the gain, beam
width, and the first side-lobe levels. All gain levels are compared with the uniform illumination case.
BEAM EFFICIENCY
This discussion considers the effect of the aperture field distribution on the beam and aperture efficiencies.
For many applications, the fraction of the total radiated
energy that is in the main (null-to-null) antenna beam is important. This quantity is called the beam efficiency (21). The
beam efficiency can be used to judge the ability of the antenna
to discriminate between signals received through its main
lobe and those through the minor lobes.
Before we go into this subject, it is helpful to review some
fundamentals. The main beam is comprised of the solid angle
M HP HP
(26)
where HP and HP are the half-power beam widths (HPBW) in
the two principal planes, minor lobes being neglected.
The (total) beam area A (or beam solid angle A) consists
of the main-beam area (or solid angle) plus the minor-lobe
area (or solid angle). Furthermore, the ratio of the main-beam
area to the (total) beam area defines what is called the beam
D=
4
4
= 2 Ap
A
(28)
where Ap is the physical area of the aperture. The aperture
efficiency is defined as the ratio of the effective aperture area,
Ae, to the physical aperture, or
ap =
Ae
Ap
(29)
so that the ratio of the aperture and beam efficiencies is (8)
A e A
k 2
ap
=
= 0
M
A p M
A p M
(30)
where M is the main beam solid angle (sr), A the total beam
solid angle (sr), and k0 the free space wavenumbers (k0
2/ ). It is important to recognize, then, that beam efficiency
and aperture efficiency are related to each other.
In terms of the radiated intensity E(, ) of a pencil beam
with boresight at ( 0, 0), the beam efficiency can be
defined by (14)
nB =
n n
2
0
E(, )E(, ) sin d d
(31)
E(, )E(, ) sin d d
where n is the angle from boresight to first null in and n
the angle from boresight to first null in . Also, E(, )* denotes the conjugate of E(, ).
In general, the aperture and beam efficiencies must both
be multiplied by a gain-degradation factor due to phase errors
given by (22)
kg = e(2 /)
(32)
where is the rms phase error over the aperture. It is assumed that the correlation intervals of the deviations are
greater than the wavelength. The controlling effect of the taper on the efficiencies (beam and aperture) tends to decrease
as the phase error increases. The efficiencies are also reduced
by the presence of the phase error.
The curves of Fig. 15 show that the beam efficiency tends
to increase with an increase in taper but the aperture efficiency decreases. Maximum aperture efficiency occurs for a
uniform aperture distribution, but maximum beam efficiency
occurs for a highly tapered distribution. In most cases a taper
is used that is intermediate between the two extremes.
r
x
y
Figure 13. Coordinate system used to analyze a circular aperture of
diameter Dw.
APERTURE SYNTHESIS
To simplify the discussion, we shall discuss a one-dimensional
line source or length Lw. Earlier in Eqs. (6) and (7) a Fourier-
APERTURE ANTENNAS
CLS
Program to compute rectangular aperture radiation patterns
REDIM A(101, 101), pattern(200)
PI 3.14159265358#
Results stored in file Circ.dat
OPEN Circ.data FOR OUTPUT AS #1
------ input data -----frequency 10: Dw 100
--------------------wavelength 30 / frequency: Dwave Dw / wavelength
Get reference power for uniformly illuminated circular array:
reference 0
FOR I 1 TO 101
FOR J 1 TO 101
X Dw * (51 - I) / 100: Y Dw * (51 - J) / 100
radius SQR(X 2 Y 2)
IF radius (Dw / 2) THEN
A(I, J) 0
ELSE
A(I, J) 1
END IF
reference reference A(I, J) 2
NEXT J
NEXT I
Load aperture distribution: all dimensions in cm
FOR I 1 TO 101
FOR J 1 TO 101
X Dw * (51 - I) / 100: Y Dw * (51 - J) / 100
radius SQR(X 2 Y 2)
IF radius (Dw / 2) THEN
A(I, J) 0
ELSE
A(I, J) COS(PI * radius / Dw)
Note: A(I, J) is aperture distribution in the X and Y plane
END IF
NEXT J
NEXT I
Normalize power over aperture to that in a uniform distribution
power 0
FOR I 1 TO 101
FOR J 1 TO 101
power power A(I, J) 2
NEXT J
NEXT I
power power / reference: power SQR(power)
Cp 1 / power
Cp is the desired normalization constant
phi 0
Note: Phi0 results in principal X-plane pattern
Phi45 results in intercardinal plane pattern
Phi90 results in principal Y-plane pattern
For a symmetrical circular distribution, all these are the same
K 0: G0DB 0
FOR theta 0 TO 10 STEP .1
PRINT Computing for theta; theta
RE 0: IM 0
K K 1
Integration over aperture
FOR I 1 TO 101
FOR J 1 TO 101
X Cw * (51 - I) / 100: Y Dw * (51 - J) / 100
psi (2 * PI * X / wavelength) * SIN(theta * PI / 180) * COS(phi * PI / 180)
psi psi (2 * PI * Y / wavelength) * SIN(theta * PI / 180) * SIN(phi * PI / 180)
RE RE A(I, J) * COS(psi)
IM IM A(I, J) * SIN(psi)
NEXT J
Figure 14. Computer program to compute the radiation pattern from a circular aperture.
619
620
APERTURE ANTENNAS
NEXT I
TMM Cp * SQR(RE 2 IM 2) / reference
IF TMM 0 THEN TMM 10 -6
IF theta 0 THEN
G0DB 20 * LOG(TMM) / LOG(10): GV TMM
END IF
pattern(K) 20 * LOG(TMM) / LOG(10)
NEXT theta
K 0
FOR theta 0 TO 10 STEP .1
K K 1
PRINT #1, theta, pattern(K)
PRINT Angl; theta; E ; pattern(K)
NEXT theta
Print
PRINT Peak Gain G0 (dB); G0DB; G0(voltage); GV; G0(power); GV
CLOSE #1
END
Figure 14 (Continued)
transform pair was defined for a line source relating the aperture distribution and the far-field radiation pattern, and vice
versa. In the synthesis process, we wish to determine the aperture distribution for a desired radiation pattern. To do this
we first express the illumination function as a sum of N uniform distributions:
E(x) =
N
cm e j m x
E( ) =
Lw
N
L w m=1
sin + m )x
dx
yielding for a line source of length Lw
E( ) =
(33)
N
sin (k sin + m )
Cm
m=1
m=1
The Fourier transform can then be written as
cm e j(k
(k sin + m )
Lw
2
(34)
(35)
Lw
2
Thus each coefficient Cm is responsible for a (sin x)/x type of
Table 2. Radiation Pattern Characteristics Produced by Various Circular Aperture Distributions
Distribution
Normalized
Half-Power
Beam Width
(deg)
HPbw/K
Normalized
Null-to-Null
Beam Width
(deg)
NULLbw*/K
Side-Lobe
Level
(dB):
SLL dB
Normalized
Side-Lobe
Angle (deg)
SLpos/K
59.33
140.00
17.66
93.67
1
2
3
4
5
74.67
88.00
99.33
110.00
120.00
194.67
250.00
306.67
362.67
420.00
26.07
33.90
41.34
48.51
55.50
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0
74.67
70.67
68.67
66.00
64.67
63.33
62.00
61.33
60.67
60.00
59.33
194.67
183.33
174.00
166.00
159.60
154.67
150.67
147.33
144.00
142.00
140.00
59.33
72.67
84.67
94.67
104
140
187.33
232.67
277.2
320.33
Comments
Uniform
Cosine raised
to power n
Cosine on a
pedestal p
Parabolic
raised to
power n
n
n
n
n
n
p
p
p
p
p
p
p
p
p
p
p
n
n
n
n
n
0
1
2
3
4
Gain Relative
to Uniform
(dB)
G0 dB
Power Gain
Factor Relative
to Uniform
G0 power
Voltage Gain
Factor Relative
to Uniform
G0 volts
0.00
1.000
1.000
119.33
145.50
173.00
200.30
228.17
1.42
2.89
4.04
4.96
5.73
0.721
0.514
0.394
0.319
0.267
0.849
0.717
0.628
0.564
0.517
26.07
25.61
24.44
23.12
21.91
20.85
19.95
19.18
18.52
17.96
17.66
119.33
112.67
107.83
104.17
101.33
99.47
97.83
96.47
95.27
94.57
93.67
1.42
0.98
0.66
0.43
0.27
0.17
0.09
0.05
0.02
0.00
0.00
1.000
0.799
0.859
0.905
0.9388
0.963
0.9789
0.9895
0.9958
0.9991
1.000
1.000
0.894
0.927
0.951
0.9689
0.981
0.9894
0.9947
0.9979
0.9995
1.000
17.66
24.64
30.61
35.96
40.91
93.67
116.33
138.67
160.17
181.33
0.00
1.244
2.547
3.585
4.432
1.00
0.701
0.556
0.438
0.36
1.00
0.866
0.746
0.662
0.6
APERTURE ANTENNAS
621
2. S. A. Schelkunoff, Some equivalence theorems of electromagnetics and their application to radiation problems, Bell System Tech.
J., 15: 92112, 1936.
9. C. Huygens, Traite de la Lumiere, Leyden, 1690; Max Born, Optik,
Berlin: Springer-Verlag, 1933.
10. A. Sommerfeld, Theorie der Beugung, in P. Frank and R. von
Mises (eds.), Die Differential und Integralgleichungen der Mechanik und Physik, Braunschweig: Vieweg, 1935.
11. J. D. Kraus, Radio Astronomy, 2nd ed., New York: CygnusQuasar, 1986.
12. R. E. Collin and Z. J. Zucker, Antenna Theory, New York:
McGraw-Hill, 1969.
13. W. L. Weeks, Antenna Engineering, New York: McGraw-Hill,
1968.
14. T. A. Milligan, Modern Antenna Design, New York: McGrawHill, 1985.
15. A. W. Rudge et al., The Handbook of Antenna Design, Stevenage,
UK: Peregrinus, 1986.
16. S. Silver, Microwave Antenna Theory and Design, New York:
McGraw-Hill, 1949.
17. H. G. Booker and P. C. Clemmow, The concept of an angular spectrum of plane waves and its relation to that of polar diagram and
aperture distribution, Proc. IEE, London, Ser. 3, 97: 1117, 1950.
18. D. R. Rhodes, The optimum line source for the best mean square
approximation to a given radiation pattern, IEEE Trans. Antennas Propag., AP-11: 440446, 1963.
19. R. S. Elliot, Antenna Theory and Design, Englewood Cliffs, NJ:
Prentice-Hall, 1987.
20. I. S. Sokolnikoff and R. M. Redhefer, Mathematics of Physics and
Modern Engineering, New York: McGraw-Hill, 1958.
21. R. C. Hansen, Linear Arrays, in A. W. Rudge et al. (eds.), The
Handbook of Antenna Design, Stevenage, UK: Peregrinus, 1986.
22. R. T. Nash, Beam efficiency limitations for large antennas, IEEE
Trans. Antennas Propag., AP-12: 918923, 1964.
23. L. I. Vaskelainen, Virtual array synthesis method for planar
array antennas, IEEE Trans. Antennas Propag., 46: 922928,
1998.
24. R. J. Mailoux, Phased Array Antenna Handbook, Norwood, MA:
Artech House, 1994.
25. E. Botha and D. A. McNamara, A contoured beam synthesis technique for planar antenna arrays with quadrantal and centrosymmetry, IEEE Trans. Antennas Propag., 41: 12221231, 1993.
26. B. P. Ng, M. H. Er, and C. Kot, A flexible array synthesis method
using quadratic programming, IEEE Trans. Antennas Propag.,
41: 15411550, 1993.
27. H. J. Orchard, R. S. Elliott, and G. J. Stern, Optimizing the synthesis of shaped beam antenna patterns, IEE Proc. H, 132 (1):
6368, 1985.
28. R. F. E. Guy, General radiation-pattern synthesis technique for
array antennas of arbitrary configuration and element type, Proc.
Inst. Electron. Eng., 135, Pt. H, No. 4: 241248, 1988.
29. P. M. Woodward, A method of calculating the field over a plane
aperture required to produce a given polar diagram, IEE J., UK,
Part III A, 93: 15541558, London, 1947.
3. C. Huygens, Traite de La Lumiere, Leyden, 1690. Translated into
English by S. P. Thompson, Chicago: Univ. Chicago Press, 1912.
Reading List
100
Efficiency (%)
80
60
40
20
0
(Full taper)
Beam efficiency
Aperture efficiency
Taper (dB)
0
(No taper)
Figure 15. Form of beam and aperture efficiencies for an aperture
as a function of taper.
beam, and there are N beams. The coefficients may be obtained manually by estimating the number of independent
beams and their relative magnitudes and positions to approximate the desired radiation pattern. Alternatively they may
be obtained mathematically via a Fourier-series representation. Also, the results may be extended by the reader to a twodimensional aperture.
The preceding equations form the basis for Woodwards aperture synthesis technique (6,29) that enables the aperture
illumination required to produce a given beam shape to be
approximated. A new array antenna synthesis method, called
the virtual array synthesis method, was recently published by
Vaskelainen (23). In this method, the excitation values of a
virtual array are synthesized using some known synthesis
method. The geometry of the virtual array can be chosen so
that there will be a suitable synthesis method for that geometry, and the synthesis of the virtual array can be done accurately enough. In the synthesis method, the excitation values
of the virtual array are transformed into the excitation values
of the actual array geometry. Matrix operations are simple
and large arrays can be easily synthesized. Further references on recent synthesis techniques appear in Refs. 24 to 28.
BIBLIOGRAPHY
1. D. J. Kozakoff, Analysis of Radome Enclosed Antennas, Norwood,
MA: Artech House, 1997.
4. J. D. Kraus and K. R. Carver, Electromagnetics, 2nd ed., New
York: McGraw-Hill, pp. 464467.
5. C. A. Balanis, Antenna Theory Analysis and Design, New York:
Harper and Row, 1982.
6. A. D. Oliver, Basic Properties of Antennas, in A. W. Rudge et al.
(eds.), The Handbook of Antenna Design, IEE Electromagnetic
Wave Series UK, London: UK, Peter Peregrinum, 1986, chap. 1.
7. H. Jasik, Fundamentals of Antennas, in R. C. Johnson (ed.), Antenna Engineer Handbook, 3rd ed., New York: McGraw-Hill, 1993.
8. J. D. Kraus, Antennas, 2nd ed., New York: McGraw-Hill, 1988.
Baker and Copson, The Mathematical Theory of Huygens Principle,
New York: Oxford University Press, 1939.
C. A. Balanis, Antenna Theory Analysis and Design, New York:
Harper & Row, 1982.
E. T. Bayliss, Design of monopulse difference patterns with low sidelobes, Bell Syst., Tech. J., 47: 623650, 1968.
R. N. Bracewell, Tolerance theory of large antennas, IRE Trans. Antennas Propag., AP-9: 4958, 1961.
W. N. Christiansen and J. A. Hogbom, Radiotelescopes, Cambridge,
UK: Cambridge Univ. Press, 1985.
622
APPLICATION PROGRAM INTERFACES
C. L. Dolph, A current distribution for broadside arrays which optimizes the relationship between the beamwidth and the sidelobe
level, Proc. IRE, 34: 335348, 1946.
R. C. Hansen, A one parameter circular aperture with narrow beamwidth and low sidelobe levels, IEEE Trans. Antennas Propag., AP24: 477480, 1976.
R. C. Hansen, Microwave Scanning Antennas, New York: Academic
Press, 1966.
R. C. Hansen, Tables of Taylor distributions for circular aperture antennas, IRE Trans. Antennas Propag. AP-8: 2326, 1960.
R. F. Hyneman, A technique for the synthesis of line source antenna
patterns having specified sidelobe behavior, IEEE Trans. Antennas
Propag., AP-16: 430435, 1968.
H. Jasik, Fundamentals of Antennas, in R. C. Johnson (ed.), Antenna
Engineering Handbook, 3rd ed., New York: McGraw-Hill, 1993.
J. D. Kraus, Radio Astronomy, 2nd ed., New York: Cygnus-Quasar,
1986.
R. T. Nash, Beam efficiency limitations of large antennas, IEEE
Trans. Antennas Propag., AP-12: 918923, 1964.
J. F. Ramsay, Fourier transforms in aerial theory, Marconi Rev., 10:
17, 41, 81, 157, 1947.
J. F. Ramsay, Fourier transforms in aerial theory, Marconi Rev., 9:
139, 1946.
D. R. Rhodes, Synthesis of Planar Antenna Sources, London: Oxford
Univ. Press, 1974.
J. Ruze, Physical limitations of antennas, MIT Research Laboratory
Electronic Technical Report No. 248, 1952.
S. A. Schelkunoff, Some equivalence theorences of electromagnetics
and their application to radiation problems, Bell Syst. Tech. J., 15:
92112, 1936.
H. E. Shanks, A geometrical optics method of pattern synthesis for
linear arrays, IRE Trans. Antennas Propag., AP-8: 485490, 1960.
M. I. Skolnik, Introduction to Radar Systems, 2nd ed., New York:
McGraw-Hill, 1980.
J. C. Slater and N. H. Frank, Introduction to Theoretical Physics, New
York: McGraw-Hill, 1933.
T. T. Taylor, Design of circular apertures for narrow beamwidth and
low sidelobe levels, IEEE Trans. Antennas Propag., AP-8: 1722,
1960.
T. T. Taylor, Design of line-source antennas for narrow beamwidth
and low sidelobes, IRE Trans. Antennas Propag., AP-3: 1628,
1955.
I. Wolf, Determination of the radiating system which will produce a
specified directional characteristic, Trans. IRE, 25: 630643, 1937.
P. M. Woodward, A method of calculating the field over a plane aperture required to produce a given polar diagram, IEE J., Part III
A, 93: 15541558, 1947.
P. M. Woodward and J. D. Lawson, The theoretical precision with
which an arbitrary radiation pattern may be obtained from a
source of a finite size, IEE J., Part III, 95: 363370, 1948.
DENNIS KOZAKOFF
Millimeter Wave Technology
APPARATUS, RADIOLOGICAL IMAGING. See XRAY APPARATUS.
APPARATUS, X-RAY IMAGING. See X-RAY APPARATUS.
APPLIANCES, DOMESTIC. See DOMESTIC APPLIANCES.
APPLICATION FRAMEWORKS. See USER INTERFACE
MANAGEMENT SYSTEMS.
RADIOWAVE PROPAGATION CONCEPTS
181
RADIOWAVE PROPAGATION CONCEPTS
An advantage of electronic communication is that an artificial
link (e.g., transmission line, cable, or waveguide) is not necessarily required between the signal source and the receiver.
Indeed, the term wireless is used in many parts of the world
in preference to radio, and even in the United States wireless is often used to describe commercial personal communiJ. Webster (ed.), Wiley Encyclopedia of Electrical and Electronics Engineering. Copyright # 1999 John Wiley & Sons, Inc.
182
RADIOWAVE PROPAGATION CONCEPTS
Table 1. Frequency Band Designations
Band Designation
Ultralow frequency
Extremely low frequency
Very low frequency
Low-frequency
Medium-frequency
High-frequency
Very high frequency
Ultrahigh frequency
Super high frequency
Extremely high frequency
Submillimeter
Abbreviation
ULF
ELF
VLF
LF
MF
HF
VHF
UHF
SHF
EHF
Frequency Range
3
3
3
30
300
3
30
300
3
30
300
Free-Space Wavelength Range
100
100
100
10
1
100
10
1
10
1
1
Hz
Hz to 3 kHz
kHz to 30 kHz
kHz to 300 kHz
kHz to 3 MHz
MHz to 30 MHz
MHz to 300 MHz
MHz to 3 GHz
GHz to 30 GHz
GHz to 300 GHz
GHz to 1 THz
Mm
Mm to 100 km
km to 10 km
km to 1 km
km to 100 m
m to 10 m
m to 1 m
m to 10 cm
cm to 1 cm
cm to 1 mm
mm to 300 m
Source: The New IEEE Standard Dictionary of Electrical and Electronics Terms, 5th ed. The Institute of Electrical and Electronics
Engineers, 1993, Ref. 1, with modifications by the author. By permission of the IEEE.
cation by radio. Although radio waves also propagate in waveguides, cables, and transmission lines, the term radiowave
propagation is used generally, and specifically in this article,
for signal travel between a source and a receiver without
guidance by such artificial devices. Examples of such systems
are radio and television broadcasting, radio point-to-point
communication, radar, radio navigation, and remote sensing.
The concept of radiowave propagation allows dividing the
total system problem into three separate parts: the transmitter or signal source, the receiver, and propagation. Thus propagation is defined as what happens between the source and
the receiver. For this concept to be valid, the distances must
be such that far-field criteria for the antennas are satisfied;
this allows separating antenna effects from propagation effects. In radar two propagation paths are involved: from the
transmitter to the target and from the target to the receiver.
Much has been learned about radiowave propagation since
Hertz first demonstrated wireless transmission from one
room to another. Now radiowave propagation is considered a
moderately mature science, yet new understanding continues
to emerge. This article presents an overview of radiowave
propagation concepts. More detailed information will be found
in articles on specific topics in radiowave propagation.
THE INFLUENCE OF FREQUENCY AND ENVIRONMENT
Frequency
The spectrum of useful electromagnetic waves covers an extraordinarily wide range of frequencies from as low as 70 Hz
to X rays at approximately 10 EHz (1019 Hz). For convenience,
the lower part of the spectrum has been divided into frequency bands as shown in Table 1. The VLF to SHF band
designations have been adopted as an international standard.
The others are commonly used, but not all authors adhere to
those definitions. Because the propagation behavior of radio
waves does not change abruptly with frequency, the band designations are often used somewhat loosely with respect to radiowave propagation. For example, a system operating in the
range 200 MHz to 500 MHz might be described as a UHF
system. In the microwave region of the spectrum, frequency
bands are often denoted by letter codes originally applied to
radar systems. These are given in Table 2.
Environment
The most important environmental influences on terrestrial
propagation are those of the troposphere, the ionosphere, and
the ground. The troposphere is the lower region of the atmosphere where ionization is too small to affect radio waves appreciably. Tropospheric effects are caused by refractive index
variations, absorption of energy by atmospheric gases, and
absorption and scattering by precipitation. The ionosphere is
the region of the atmosphere where free electrons produced by
ionization strongly affect radio waves in the frequency ranges
below about 30 MHz to 50 MHz. The lower altitudinal limit
of the ionosphere is in the range of 60 km to 70 km. The term
ground designates the surface of the earth, including bodies
of water. Local structures, such as buildings, may also have
important effects.
Table 2. Standard Radar Frequency-Band Nomenclature
Band
Designation
HF
VHF
UHF
L
S
C
X
Ku
K
Ka
V
W
millimeter (mm)a
submillimeter
Frequency Range
3
30
300
1
2
4
8
12
18
27
40
75
110
300
MHz to 30 MHz
MHz to 300 MHz
MHz to 1 GHz
GHz to 2 GHz
GHz to 4 GHz
GHz to 8 GHz
GHz to 12 GHz
GHz to 18 GHz
GHz to 27 GHz
GHz to 40 GHz
GHz to 75 GHz
GHz to 110 GHz
GHz to 300 GHz
GHz to 3 THz
Approx. Free-Space
Wavelength Range
100 m to 10 m
10 m to 1 m
1 m to 30 cm
30 cm to 15 cm
15 cm to 7.5 cm
7.5 cm to 3.75 cm
3.75 cm to 2.5 cm
2.5 cm to 1.67 cm
1.67 cm to 1.11 cm
1.11 cm to 7.5 mm
7.5 mm to 4 mm
4 mm to 2.73 mm
2.73 mm to 1 mm
1 mm to 100 m
Source: The New IEEE Standard Dictionary of Electrical and Electronics Terms,
5th ed. The Institute of Electrical and Electronics Engineers, 1993, Ref. 2,
with modifications by the author. By permission of the IEEE.
a
The mm designation is sometimes used for the entire frequency range 40 GHz
to 300 GHz when general information for this range is to be conveyed.
RADIOWAVE PROPAGATION CONCEPTS
Interaction of Frequency and Environment
The effect of the environment depends greatly on the system
frequency. For example, in frequency ranges below about 30
MHz, the ionosphere is likely to be of prime importance. At
VHF and UHF, under certain meteorological conditions, tropospheric layering may form ducts which guide waves very
efficiently in one frequency band, but not in another. The system frequency is also important with respect to the ground.
At microwave frequencies, the wavelength and depth of penetration into the ground are on the order of centimeters; thus
the dielectric constant and conductivity at the very surface
are important, and even intuitively smooth ground may need
to be considered rough if its vertical variations are not small
compared to wavelength. In contrast, at VLF the penetration
depth is many hundreds of meters; therefore the dielectric
constant and conductivity far below the visible surface are
important, and the ground may appear smooth even in mountainous areas. As these examples show, it is important to
characterize the environment by parametric values appropriate to the system frequency.
The system design also influences the extent to which the
environment affects signal propagation. For example, in
many satellite communication system calculations the effects
of the ground are neglected because the ground station antennas are highly directive and pointed upward. Thus little energy hits the ground on earth-to-space transmission, and little of the energy impinging on the ground is received by the
antenna on the space-to-earth path.
The effects of the troposphere, ionosphere, and ground
vary with frequency and also with geographic location. In addition, the troposphere and ionosphere vary in time. This
variability adds complexity to predicting signal propagation.
In many cases, such predictions are limited to statistical estimates.
Noise Considerations
System performance depends on signal strength and also on
noise that corrupts the signal. Depending on the system frequency, significant external noise sources may be impulses
due to lightning propagated via the ionosphere, emissions
from astronomical sources (e.g., the sun, radio stars), or the
blackbody noise emanating from the transmission medium.
PROPAGATION MECHANISMS AND MODELS
An exact solution of the radiowave propagation problem entails solving Maxwells equations in the presence of the source
and the environment. Such calculations are beyond the state
of the art. Instead, approximations based on physical processes are used. For example, the reception of two rays from
the transmitter might be considered, one directly and the
other via reflection by the ground. Such physically based,
mathematical or empirical approximations for calculating signal strength are called propagation models (3). The term
propagation mechanism is used here for the physical processes which lead to relatively simple propagation models.
In principle, all of the propagation mechanisms discussed
later might be considered in every application, but generally
one or a few account sufficiently well for the system performance, so that the others are neglected. However, the specific
183
application must be kept in mind. For example, a very small
interfering signal may be objectionable, and thus an interference calculation may require including a propagation mechanism that would otherwise be unimportant.
Direct Propagation
The simplest propagation mechanism is direct propagation in
which the signal travels from the emitter to the receiver unaffected by any propagation medium, except possibly for attenuation and mild refraction. When attenuation and refraction
are absent, propagation occurs as in free space. Direct propagation takes the form of a spherical wave, which can be approximated locally at the receiver as a plane wave.
Although free-space propagation is a highly idealized approximation, it has important applications. Many radars and
many satellite communication systems operating in the UHF
to C-band frequency range have antennas sufficiently directive to exclude ground effects, and at these frequencies atmospheric effects may generally be neglected. In these cases, the
free-space model is applicable. For satellite communications
at higher frequencies, atmospheric effects become important,
but the direct propagation model is still applicable when attenuation and refractive effects are included. Predictions are
very much more complicated if the refractive index is inhomogeneous and time-varying.
In general, the direct propagation model applies when the
emitter and receiver are in plain view with respect to one
another, provided the effects of other propagation mechanisms are negligible. This situation is most frequently encountered in the atmosphere at UHF to SHF with systems
utilizing highly directive antennas.
Terrain Reflections
When antennas are not highly directional, signals may travel
directly from the emitter to the receiver and also by reflection
from the ground. In this case, both paths must be considered
in evaluating the system performance. Typical examples are
ground-to-air, air-to-ground, and air-to-air communication at
UHF. In air-to-air communication, the size limitation of aircraft antennas makes it impossible to use highly directive antennas in this frequency range, so that it is not possible to
keep the signal from reaching the ground. In practical
ground-to-air and air-to-ground communication, the ground
station antenna is situated somewhat above the ground and
is not sufficiently directive to discriminate against the
ground-reflected ray. Thus the propagation model for these
systems must include both the direct and the ground-reflection mechanisms.
Ducting
The effects of gravity and meteorological conditions make atmospheric density and humidity functions of altitude. Though
the resulting variations of the refractive index are small, significant bending of the signal path in the vertical plane can
occur. This refraction may be sufficiently strong to guide the
wave along the surface of the earth, a behavior called ducting.
Tropospheric ducts are most commonly observed at VHF and
UHF. They may also exist at higher frequencies, but the more
directional antennas employed at these frequencies are less
likely to couple efficiently into a duct. Ducting is much more
184
RADIOWAVE PROPAGATION CONCEPTS
common at some geographical locations than others because
it is closely related to meteorological phenomena. Although
ducting is a fairly reliable means of communication in some
locations, it is generally more likely to be a source of potential interference.
the radio frequency spectrum have become exceedingly
crowded.
Waveguide Modes
When the emitter and receiver are not within plain view of
each other at VHF and higher frequencies, diffraction is an
appropriate propagation mechanism. When the intervening
terrain is hilly or mountainous, the diffracting obstacles are
often modeled as cylinders or even as vertical half planes
(knife-edge diffraction). Diffraction by a conducting sphere
is the most common model for relatively flat terrain. More
detailed computational models are becoming increasingly
available as a result of increased computer power, more theoretical insights, and better algorithms.
Ray theory requires that the medium vary slowly with respect
to wavelength. Thus it fails for radiowave propagation at frequencies lower than about 100 kHz because the corresponding
wavelengths no longer meet the requirement that they be
much smaller than the earth-ionosphere distance. Instead, at
VLF and ELF the earth-ionospheric space is modeled as a
concentrically spherical waveguide with the ground as the
lower boundary and the ionosphere as the upper. In the ELF
and lower VLF bands only one, or at most a few, modes need
to be considered. The number of modes and therefore the computational difficulty increases with frequency. Calculations in
the LF frequency band are especially laborious, and a combination of ray and mode theory is often used.
Ground Wave
Ionospheric Scatter
When antennas operate near (in terms of wavelength) and on
the ground, it is found that the direct and ground-reflected
waves cancel almost completely. In this case, however, it is
found that a wave can be excited that travels along the surface and decreases rapidly with altitude. Since efficient transmitting antennas at MF and lower frequencies are necessarily
large structures because the wavelength is long, they are generally on or near the ground, and ground-wave propagation is
important at these frequencies. It is the dominant mechanism
for local standard AM broadcast reception in the 540 kHz to
1.7 MHz band. It is also dominant at LF and VLF for relatively short distances.
Signals with a frequency too high to be reflected coherently
from the ionosphere may, nevertheless, still be affected by it.
One possible effect is the scattering of small amounts of energy out of the beam by local irregularities. This is analogous
to the scattering by local irregularities of the troposphere, as
discussed previously. Ionospheric scattering is most pronounced in the frequency range immediately above that
which supports coherent ionospheric reflections. Ionospheric
D-region scatter systems have been operated successfully in
the VHF band over 1000 km to 2000 km distances with a
bandwidth on the order of 100 Hz.
Terrain Diffraction
Meteor Scatter
Tropospheric Scatter
The troposphere is never truly homogeneous, as common experience with weather indicates. In fact, variations over a
wide range of dimensional scales are ever present. These variations can be used to advantage when very reliable wireless
point-to-point communications are needed over a path of hundreds of kilometers. When very strong signals are beamed at
an atmospheric region within the line of sight of both stations, the relatively small signal scattered out of the beam
toward the receiving station may, nevertheless, be sufficient
to enable communication at rates on the order of 5 megabits
per second over 150 km to 400 km distances at UHF. This
technique is useful for communication with communities in
arctic regions, where other forms of communication (e.g., satellite communications, ionospheric reflection) are less reliable.
Ionospheric Reflection
In the upper LF, MF, and HF bands, signals give the appearance of traveling in rays that are reflected by the ionosphere
above and by the ground below, resulting in a series of hops.
Actually, the rays are bent rather than sharply reflected in
the ionosphere, but the effect is essentially the same. Signal
transmission by this means is very efficient, and great distances are spanned with modest power and equipment. For
these reasons the short-wave bands, as they are often called
popularly, are utilized for broadcasting, point-to-point communications, and amateur use, and portions of this band of
Meteors are popularly considered rare phenomena, perhaps
because our daytime habits and increasingly urbanized existence allow many of us to see them only rarely. Actually, visually observable meteors are not rare by any means, but those
too small to observe visually are even more abundant. The
number of meteors of a given mass entering the atmosphere
approximates an inverse exponential function of the mass. As
these meteors enter the atmosphere, they heat and ionize the
gas around them, forming trails of ionization. Where this ionization is substantially more intense than that of the ambient
ionosphere, the trails reflect signals of a higher frequency
than normally reflected from the ionosphere. Meteor-scatter
communication systems are operated successfully in the VHF
band. Because suitably oriented trails are not always present,
such systems must store the information to be transmitted
between the occurrence of suitable trails and transmit the information in bursts when such trails are present. This limits
the average data rate. Also, a means must be provided for
sensing the occurrence of suitable trails. Usually each station
continually transmits a pilot signal, and the other station is
assured that a trail exists when it receives that signal.
Whistlers
Electromagnetic signals in the ELF/VLF frequency range can
propagate through the ionosphere in a peculiar mode in which
they closely follow the lines of the earths magnetic field. It is
not easy to launch artificial waves with such very long wavelengths, but lightning generates electromagnetic energy in
RADIOWAVE PROPAGATION CONCEPTS
this frequency range quite effectively. Lightning-generated
signals travel along the earths magnetic field lines, often going out to a distance of several earth radii, and so are guided
to the antipode, i.e., the point on the opposite hemisphere
where the field line terminates. Part of the signal may be reflected from the antipode along the same field line to the point
of origin and may be reflected there again, and so forth. The
frequency components of the signal are dispersed in the
travel, so that the frequency at the receiver changes with
time. When amplified, the resulting audio has a sound reminiscent of whistling; hence the name of the propagation mechanism. Although not useful for information transmission, this
propagation mechanism is utilized to obtain information
about the upper ionosphere. The same propagation mechanism, but using much stronger artificial magnetic fields, has
been investigated for communicating at much higher frequencies through the much denser ion sheath which forms around
space vehicles as they reenter the atmosphere.
185
System Loss
System loss (Ls or As) is the ratio of the RF power input to the
transmitting antenna terminals and the resultant RF power
available at the receiving antenna terminals. It is expressed
in decibels by
Ls = 10 log(pt /pa ) = Pt Pa
(1)
where pt denotes the RF power input to the transmitter antenna terminals, pa the available power at the receiving antenna terminals, both measured in the same units, and Pt,
Pa are the corresponding power relative to a common reference level (e.g., 1 W) in decibels. Because the system loss is
referred to the antenna terminals, it excludes all losses in
feeder lines but includes all losses associated with the antenna, for example, antenna-grounding losses, dielectric
losses in the antenna, antenna loading-coil losses, and antenna-terminating resistor losses, if applicable.
Nonatmospheric Propagation
Transmission Loss
The propagation mechanisms discussed so far are those commonly encountered with electromagnetic signal propagation
through space or the earths atmosphere. A different environment prevails when signals propagate through the ocean,
through the earths crust, or through some planetary atmospheres. In the ocean, only signals at ELF or in the visiblelight frequency band penetrate to substantial depths. Propagation to substantial depths through the ground also presents
difficulties. Nevertheless, radar systems for locating objects,
such as water and gas lines at typical shallow depths, are
quite successful.
Transmission loss (L or A) is defined as the ratio between the
power radiated by the transmitting antenna and the power
that would be available at the receiving antenna terminals if
this antenna were lossless but its radiation diagram, that is,
its directional and polarization characteristics, were unchanged. In Fig. 1 the antenna losses Ltc and Lrc for transmitting and receiving, respectively, are defined as 10 log (R/R),
where R is the resistive component of the impedance of the
designated antenna and R is its radiation resistance.
TRANSMISSION LOSS
The most significant question relating to propagation for a
radar or telecommunication system is the adequacy of the received signal level. The ratio, usually expressed in decibels
(dB), between the power at some point in the transmitting
system and a corresponding point in the receiving system is
called a loss. Although time delay or phase shift may also be
important parameters in some systems, loss is a universal
and primary consideration because it relates directly to signal
strength. In many cases, the loss varies randomly in time or
with location. Then statistical parameters relating to the loss
are often the quantities of interest, for example, the median
loss.
Several types of losses are defined; see Fig. 1. The usual
symbol for loss is L, although A (for attenuation) is also used,
and subscripts denote the specific loss under consideration.
The loss definitions below follow, with some emendations,
those in The concept of transmission loss for radio links,
Recommendation ITU-R P.341-4, of the International Telecommunication Union (4).
Total Loss
The total loss (Ll or Al) of a link is the ratio between the power
supplied by the transmitter and the power supplied to the
corresponding receiver under real installation, propagation,
and operational conditions. Note: it is necessary to specify the
points at which these powers are determined.
Basic Transmission Loss
Basic transmission loss, denoted by Lb or Ai, is the transmission loss that would occur if the antennas were replaced by
hypothetical isotropic antennas with the same polarizations
in the direction of propagation as the real antennas. The
propagation path is retained, but the effects of obstacles close
to the antennas are disregarded. The symbols Gt and Gr in
Fig. 1 denote the directivity gains of the antennas for the polarization and direction of propagation being considered. The
basic transmission loss is the ratio of the equivalent isotropic
radiated power (EIRP) of the transmitter system to the available power from a hypothetical isotropic, lossless receiving
antenna with the same polarization as that of the real antenna in the propagation direction.
Free-Space Basic Transmission Loss
The free-space basic transmission loss (Lbf or A0) is the transmission loss that would occur if the antennas were replaced
with hypothetical isotropic antennas, matched in polarization
for maximum power transfer and located in an unbounded,
lossless, homogeneous, isotropic environment, retaining the
distance between antennas. Free-space basic transmission
loss is a function only of the ratio of the distance d between
transmitter and receiver to the wavelength . It is given in
decibels by
Lbf = 20 log
4d
where d and are measured in the same units.
(2)
186
RADIOWAVE PROPAGATION CONCEPTS
Free-space basic transmission
loss, Lbf
Isotropic
antenna
Isotropic
antenna
Free space
Propagation
medium,
etc.
Basic transmission loss, Lb
Lb = Lbf + Lm
Transmitting
antenna
Gr
Gt
Receiving
antenna
Transmission loss, L
L = Lb Gt Gr
Transmitting
antenna
losses
Ltc
Lrc
Receiving
antenna
losses
Transmitter
Figure 1. Definitions of losses. Adapted
from The concept of transmission loss for
radio links (Recommendation ITU-R
P.341-4), International Telecommunication Union, 1996, Ref. 4. By permission.
Receiver
Filters,
feeder,etc.
Filters,
feeder,etc.
Total loss, Lb
(reference points should be specified )
Loss Relative to Free Space
Loss relative to free space (Lm or Am) is defined as the difference between the basic transmission loss and the free-space
basic transmission loss in decibels:
Lm = Lb Lbf
System loss, Ll
Ls = L + L tc + + L tc = Pt Pa
(3)
Loss relative to free space may be divided into losses of different types, for example, extinction loss, diffraction loss, waveinterference loss, reflection loss, polarization-coupling loss,
aperture-to-medium coupling loss. Because these losses are
directly connected with propagation mechanisms, the end result of propagation studies is frequently presented in the form
of loss relative to free space.
The resulting decrease in useful power density is considered
a diffraction loss.
Wave-Interference Loss. When several rays combine to produce the effective received wave, interference may cause a
signal decrease relative to that of the ray directly received.
This effect can be characterized as a wave-interference loss,
although this terminology is not encountered frequently in
practice. Of course, signals may also be enhanced by interference, resulting in a negative wave-interference loss.
Extinction Loss. Extinction is the total attenuation due to
absorption of signal power in the propagation medium and
scattering of power out of the beam by the medium. In the
troposphere at X-band and higher frequencies, oxygen, water
vapor, and precipitation absorb power from a propagating
wave. Precipitation also scatters power out of the beam in the
frequency range where particle or droplet size is not small
compared to the wavelength. In the ionosphere, absorption is
caused primarily by collisions between electrons and neutral
molecules. In ground-wave propagation, currents in the
ground cause absorption losses.
Reflection Loss. Several phenomena result in reflection
loss. If reflection from the ground occurs at a non-grazing incidence, the amplitude of the effective plane-wave reflection coefficient of the interface is less than unity, resulting in a reflection loss in the reflected ray. If the surface is rough on a
scale comparable to the wavelength, energy is reflected in a
range of angles about the specular, resulting in an effective
specular reflection coefficient less than unity, and hence a reflection loss in the specular direction. If the reflecting surface
is curved, the defocusing (or focusing) effect of the curvature
decreases (or enhances) the reflected power density in the
propagation direction of interest, resulting in a positive (or
negative) reflection loss. The concave curvature of the ionosphere, as seen from the ground, causes focusing. The convex
curvature of the earth surface, as seen from above, causes defocusing.
Diffraction Loss. On a propagation path using diffraction
by terrain features or by the spherical earth, only a fraction
of the power is scattered in a direction useful to the receiver.
Polarization-Coupling Loss. In an isotropic medium, polarization-coupling loss occurs when the receiving and transmitting antenna polarizations are not matched for optimum
RADIOWAVE PROPAGATION GROUND EFFECTS
power transfer. This situation arises most frequently when
interfering signals are being considered.
In an anisotropic medium, such as the ionosphere, plane
waves of arbitrary polarization are split into two rays of characteristic polarizations. These rays propagate with generally
different attenuations and phase shifts, and they recombine
into a single plane wave when emerging from the medium
or even within the medium. Polarization-coupling loss occurs
when the transmitted wave polarization does not match that
of a desired characteristic mode and when the receiving antenna does not match the polarization of the wave received.
Aperture-to-Medium Coupling Loss (Antenna-Gain Degradation). For some propagation mechanisms under certain conditions, the free-space directivity gain of the antennas is not
realized. Then the directivity gain is said to be degraded, and
the resulting signal loss is termed an aperture-to-medium
coupling loss. As an example, in tropospheric scatter propagation a highly directive antenna is used to illuminate a region
of the turbulent troposphere that is visible from both transmitter and receiver. The receiver utilizes a highly directive
antenna to receive the signal scattered from the illuminated
region. The part of the region within both the transmitting
antenna beam and the receiving antenna beam is called the
common volume. Increasing the directivity of each antenna
increases the signal level, but the increase is not proportional
to the directivity increase because the narrower beams reduce
the common volume. Thus the full free-space directivity of the
antennas is not realized. The loss in directivity, when considered an antenna effect, is called antenna-gain degradation;
when considered as a propagation effect, it is called antennato-medium coupling loss. It is most pronounced for antennas
with very high directivity.
ACKNOWLEDGMENT
By permission. Responsibility for selection and modification
of this material rests solely with the author and not the ITU.
The ITU-R Recommendations may be obtained from: International Telecommunication Union, General Secretariat of Sales
and Marketing Service, Place des Nations CH 1211 GENEVA
20 (Switzerland), Telephone: 41 22 730 61 41 (English)/41
22 730 61 42 (French), Telex: 421 000 uit ch/Fax: 41 22 730
5194, X.400: SSales; Pitu; Cch, Internet: Sales@itu.int.
BIBLIOGRAPHY
1. The New IEEE Standard Dictionary of Electrical and Electronics
Terms, 5th ed. (IEEE Standard 100-1992), New York: IEEE, 1993,
pp. 10621063.
2. Ref. 1, p. 1053.
3. Ref. 1, p. 1018.
4. The concept of transmission loss for radio links (Recommendation
ITU-R P.341-4), in ITU-R Recommendations 1995 P, Geneva: International Telecommunication Union, 1996, pp. 611.
187
BACKSCATTER
The scattering of waves from an object is a diffraction process
that transforms the incident wave into the waves propagating
from the object in all possible directions. The backscatter, or
backcattering, is the scattering of waves back toward the
source of the incident wave. This process substantially depends on material properties of the scattering object, its
shape, size, and spatial orientation relative to the incident
wave, as well as on the frequency and polarization of the incident wave. One distinguishes three frequency regions with
different physical properties of scattered waves. They are
quasi-static, resonance, and quasi-optical regions.
QUASI-STATIC SCATTERING
In the quasi-static region (sometimes called the Rayleigh region), the wavelength of the incident wave is much greater
than the maximum linear dimension l of the scattering object
( l). At a certain time t, the scattered field at small distances (r ) from the object is approximately the static field
created by dipoles and multipoles induced by the incident
wave in the scattering object at the same time t. Far from the
object, the scattered field is an outgoing spherical wave. Its
average power flux density (over one period of oscillations)
relative to that of the incident wave, or radar cross-section
(RCS), is determined by the objects volume rather than the
shape of the object, and it is proportional to 4. Specifically,
this dependence explains the blue color of the cloudless sky
during the day. This color is due to the predominance of blue
light scattered by small inhomogeneities of the air caused by
fluctuations of its density. The violet light, which is shorter
in wavelength, undergoes stronger relative scattering and as
a result suffers higher attenuation while propagating through
the atmosphere. Analytical expressions for quasi-static RCS
of some scattering objects can be found in (13). For example,
the axial RCS of perfectly conducting bodies of revolution is
determined by the following approximation (p. 145 of Ref. 1):
4
e
k 4V 2 1 +
64 4 6
k a
9
RESONANCE SCATTERING
In the resonance frequency region, linear dimensions of scattering objects are comparable to the wavelength of the incident wave. Eigen-oscillations excited by the incident wave in
the scattering object can substantially influence the scattering property. Frequencies of these oscillations are complex
quantities. Their imaginary parts determine both the internal
(thermal) losses inside the object and the external losses that
are due to radiation into the surrounding medium. A major
contribution to RCS is given by the radiation of those eigenoscillations, whose eigenfrequency real part and polarization
are close to the incident wave frequency and polarization. If
the quality factor of these oscillations is quite large, the amplitude and the intensity of the scattered wave sharply increase as the frequency of the incident wave approaches the
real part of the frequency of eigen-oscillations. This phenomenon is referred to as resonance scattering. It reveals itself,
for example, in scatterings from thin metallic half-wavelength
long wires and narrow strips (pp. 293303 of Ref. 2). These
scatterers are used, in particular, to create a chaff clutter for
radars. Analytical and numerical data for scattering by thin
wires are collected in chapter 12 of Ref. 3. Recent analytical
results for resistive wires are presented in Ref. 8.
The resonance scattering from wires can also be explained
as being due to the constructive interference of multiple current waves arising from the wire ends. This process is investigated in detail in Ref. 9. The total current generated by the
incident plane wave in the perfectly conducting wire (Fig. 1)
equals
2
J(z) = J0 (z) +
(1)
[Jn+ (z) + Jn (z)]
(3)
n=1
where k 2/ is the wave number, V is the objects volume,
and is the characteristic length-to-width ratio of the object.
This quantity is found for each objects shape by allowing
the axial dimension of the object to go to zero so as to obtain
the correct result
=
power of the wave number k 2/ . The expansion coefficients are found from the solution of the recursive system of
boundary value problems in potential theory (pp. 848856 of
Ref. 4). In practice it is possible to find only a few first coefficients. Direct numerical methods are efficient tools for the solution of quasi-static scattering problems (57).
(2)
for the circular disk with radius a. Table 8.2 in Ref. 2 (Vol. 2,
pp. 558561) contains explicit expressions for RCS found in
this manner for a variety of bodies of revolution. The first
term in this table (Eq. 8.1-87a on p. 558) contains a misprint.
The letter b should be replaced by h.
In this frequency region, the scattered field can be expressed in terms of a convergent series in positive integer
with the time dependence exp(it) assumed and suppressed
here and later. The term J0(z) is the current induced by the
incident wave in the infinitely long wire ( z ). The
terms Jn(z) and Jn(z) are multiple current waves. Waves
Jn(z) run in the positive z-direction from the left wire end
z l to the right end z l. Waves Jn(z) run in the negative z-direction from the right end z l to the left end
z l. The total length of the wire is L 2l. When the wave
Jn(z) reaches the opposite end it undergoes diffraction and
transforms into the wave Jn1
(z). At the end points of the wire
the total current and its components satisfy the conditions
J(l) = 0,
+
Jn+1
(l)
J1+ (l) = J0 (l),
Jn (l),
Jn+1
(l)
J1 (l) = J0 (l)
Jn+ (l),
(4)
n = 1, 2, 3, . . .
(5)
175
J. Webster (ed.), Wiley Encyclopedia of Electrical and Electronics Engineering. Copyright # 1999 John Wiley & Sons, Inc.
176
BACKSCATTER
z
z=l
z=+l
Figure 1. A thin wire excited by an incident wave. The radius of the
wire is small compared with the wavelength. Such a wire can support
traveling waves due to the multiple edge diffractions. A constructive
interference of these waves results in the resonance behavior of the
surface current and scattered field.
In thin wires (ka 0.2, a is the wire radius), the multiple
current waves are described by the following approximate expressions
+
J2n
(z) = J1 (l)[ (kL)eikL]2n2[k(l + z)]eik(l+z)
+
J2n+1
(z) = J1+ (l)[ (kL)eikL]2n1[k(l + z)]eik(l+z)
J2n
(z) = J1+ (l)[ (kL)eikL]2n2[k(l z)]eik(lz)
(6)
J2n+1
(z) = J1 (l)[ (kL)eikL]2n1[k(l z)]eik(lz)
with n 1, 2, 3, . . .. Function (kz) is defined in (10) as
i
2 ln
ka
(kz) =
2ikz
E(2kz)e2ikz
ln
q
real frequencies ( 0), the denominator D(ka, kL) does not
vanish. But it acquires minimum values when the frequency
of the incident wave is close to the real part of the resonant
frequency ( res). This occurs when kL n or L n /2
with n 1, 2, 3, . . . and results in the current resonance.
Under the normal incidence (the direction of the incident
wave is perpendicular to the wire axis), only the odd resonances (n 1, 3, 5, . . .) are realized due to the symmetry of
inc
the incident field [Einc
z (z) Ez (z)]. Figure 2, taken from the
classic paper (11), illustrates the resonance behavior of scattering from thin wires. The incident wave direction is perpendicular to the wire axis. The quantity A in Fig. 2 is the total
power of the field scattered in all directions. The maximum
scattering occurs in the directions perpendicular to the wire
and therefore in the backscattering direction as well. In Fig.
2, the quantity 1/[2ln(k)] depends on the wire radius,
2l L is the wire length, and kl.
A similar interference of a specular reflection with surface
diffracted rays (Fig. 3) explains the backscattering from perfectly conducting spheres and prolate spheroids at the upper
end of the resonance region [p. 149 of (2) and pp. 822848
of (4)]. However, an important difference exists between the
resonances in scattering from wires and spheres. The resonance backscattering from wires is caused by the current resonance in the wires and it is accompanied by a simultaneous
A
(2l)2
(7)
1.0
= 0.1
where
0.1
and
E(x) =
x
eit
dt = Ci(x) + iSi(x)
t
(8)
= 0.05
0.01
Functions Ci(x) and Si(x) are the well-tabulated cosine and
sine integrals, respectively. For small arguments (x 1),
function E(x) reduces to E(x) ln(x) ln(i) O(x) and ensures the equality (0) 1. Equation (6) shows that all multiple edge waves starting with secondary waves (n 2, 3, 4,
. . .) are expressed approximately by the same function (x).
As a result, the substitution of expressions (6) into Eq. (3)
leads to the geometric series
[Jn+ (z) + Jn (z)] = f (k, z, l, a)
{[ (kL)eikL ]2 }m
n=2
m=0
f (k, z, l, a)
=
D
(9)
which contains the resonance denominator D(ka, kL) 1
[(kL)eikL]2. The equation D(ka, kL) 0 defines the complex
resonant frequencies res ckres res ires, where c is the
light velocity in vacuum. Due to the radiation loss, the quantity res Im() is always negative (res 0). Therefore, for
0
0.001
0.5
1.0
1.5
2.0
2.5
3.0
3
2
5
2
2l
0.0001
Figure 2. Integral cross section of thin wires (from Ref. 11). This
quantity has the maximum (resonance) values for wires with the total
length L 2l (2n 1) /2, n 1, 2, 3, . . .. Along such wires from
one edge to another, each multiple edge wave acquires the phase shift
of (2n 1). Due to reflection at the edge, it acquires an additional
phase shift of . As a result, this wave becomes equi-phased with all
other multiple edge waves. This leads to the resonance behavior for
the current and scattered field.
BACKSCATTER
1
2
Figure 3. Backscattering from a body of revolution. The total scattered field consists of two components. One of them is a specular reflected ray (1) and the other is a beam of diffracted rays (2) radiated
by creeping waves traveling along the shadow side of the scattering
body. The equi-phase interference of these two components results in
the resonance-like increase of the scattered field.
increase of the scattered field in other directions. This is a
true resonance effect. The resonance scattering from spheres
and spheroids is a simple equiphase interference in the single
(backward) direction without the field increase in other directions. Additional data for the resonance backscattering can
be found in (4) (pp. 822848). As in the case of quasi-static
scattering, direct numerical methods are also efficient for
RCS calculations in the resonance frequency region (57).
QUASI-OPTICAL SCATTERING
In the quasi-optical frequency region, which is often referred
to as the high-frequency region, linear dimensions of scatterers are much greater than the wavelength of the incident
wave. For example, this occurs in the scattering of decimeter
and centimeter radar waves by such objects as ships, airplanes, and missiles. In contrast to the quasi-static and resonance frequency regions, the scatterings by objects in the
quasi-optical region are determined mainly by the objects local properties rather than by their whole volume.
Large dimensions and complex shapes of scattering objects
allow the existence of various types of scattered fields. Some
of them are illustrated in Fig. 4. Geometrical optics rays and
beams (A) reflected from the object provide the main contributions to backscattering. Diffraction of the incident wave at
edges and at lines of curvature discontinuity or material dis-
177
continuity creates edge waves (B and C), which can be interpreted as diffracted rays. They represent the second-order
contributions (1215). Diffracted waves arising from corners
provide the third-order contributions (1214). At the shadow
boundary on a smooth scattering surface, the incident wave
excites creeping waves (D), which propagate along the shadow
side of the object and radiate surface diffracted rays (E). Due
to continuous radiation of these rays, the creeping waves attenuate exponentially and for this reason the contribution of
surface diffracted rays (E) to backscattering is small (12,13).
However, for the objects with dimensions comparable to the
wavelength, surface diffracted rays can give appreciable contributions, as it is mentioned already in the previous section
(see also Fig. 2). Diffracting waves propagating along the
scattering object can undergo multiple diffractions at geometrical and material discontinuities and can transform into
other types of waves. This process creates high-order contributions to backscattering (1215). A visual description of
scattering from large objects, simple quantitative estimations
of some contributions to backscattering, and relevant references are presented in (16).
High-frequency asymptotic methods are widely used to
predict scatterings in this frequency region. They include geometrical optics (GO) and its extension, geometrical theory of
diffraction (GTD); physical optics (PO) and its extension,
physical theory of diffraction (PTD); and various modifications and extensions of GTD and PTD. These asymptotic techniques are discussed in ELECTROMAGNETIC WAVE SCATTERING
and RADAR CROSS-SECTIONS. The present article supplements
these and concentrates mainly on the physical optics. This
method is not so precise as GTD, PTD, and their extensions,
but it allows useful estimations for the scattered fields in
which many practical problems cannot be treated with other
techniques. Direct numerical methods, in their classical
forms, are not efficient in the high-frequency region. Various
combinations of these methods with the asymptotic techniques (so-called hybrid methods) represent a promising direction in the prediction of high-frequency scattering (17). Additional information about numerical, hybrid, and asymptotic
techniques used for the solution of scattering problems can be
found in Refs. 6, 13, and 18 and in the reading list at the end
of this article.
Geometrical Optics Approximation
E
B
A
D
B
C
E
E
Figure 4. Backscattering from a convex opaque object. The main
contributor to the scattered field is beam A reflected from the front
planar facet of the object. Edge waves B are created at the edges.
Edge waves C are created at the curvature discontinuities. At the
shadow boundary, the incident wave excites creeping waves D which
propagate along the objects surface and radiate surface diffracted
rays E. Additional creeping waves are excited at the curvature discontinuities.
GO is used for approximate estimations of backscattering in
many practical problems. The basic notion of GO involves the
concept of rays. A ray is an infinitely narrow stream of the
wave field moving with the light velocity along the lines perpendicular to the phase fronts. These lines are called ray trajectories. In free space they are straight lines. Electric and
magnetic vectors of the ray field are perpendicular to each
other and to the direction of propagation. GO reflected rays
obey simple rules (19,20): the reflected ray lies in the incidence plane which contains the incident ray and the normal
to the scattering surface at the reflection point (Fig. 5); the
angle of reflection equals the angle of incidence (Snells reflection law); the power inside an elementary hypothetical
tube confined by neighboring rays is constant.
GO is a good approximation for the field reflected from
large smooth scattering objects. It provides the leading term
in the exact high-frequency asymptotic expansion of the re-
178
yy
;;
BACKSCATTER
vature is infinite. In this case, the rays reflected by the object
form the so-called reflected beams, which undergo the transverse diffusion while propagating from the object and for this
reason lose their geometrical optics structure in the far zone.
Physical Optics Approximation
Figure 5. Reflection from a planar surface. The reflected ray lies in
the plane which contains the incident ray and the unit normal vector
n to the reflecting surface.
flected field. The reflection coefficient is found from the solution of an appropriate canonical problem. For opaque homogeneous objects, this is the Fresnel reflection coefficient, which
determines the amplitude and phase of plane waves reflected
from a planar boundary of a semi-infinite homogeneous medium [pp. 474479 of (2)]. For opaque objects coated with thin
layers, the canonical problem is the reflection of plane waves
from an infinite planar layer. This canonical layer is tangential to the scattering object (Fig. 6). It is homogeneous in the
directions parallel to its surface and has the same material
structure in depth as a real layer at the reflection point T.
The canonical layer is placed on the planar boundary of a
homogeneous medium with the same material properties as a
real object at the tangency point. This implies that the field
on a real coated object is determined exclusively by its local
properties in the vicinity of the reflection point. Nonlocal contributions from various waves propagating along the object
are not treated with this approach. Creeping and traveling
waves [pp. 120 and 130 of (2)] are examples of such waves.
According to this GO approach, the backscattering RCS of
smooth coated objects equals
= |r(0)| R1 R2
2
[(m) ]
(10)
where r(0) is the reflection coefficient for the normal incidence
( 0) and R1, R2 are principal radii of the curvature of the
scattering surface at the reflection point T. In the case of isotropic objects and coatings, the reflection coefficient r(0) does
not depend on the incident wave polarization. Equation (10)
is not valid for objects with planar faces when R1 R2 .
It also fails for objects that contain ruled elements (cylindrical, conical) with a rectilinear generatrix whose radius of cur-
This method goes back to MacDonald (21) and is based on
three concepts which are GO, canonical planar layer, and
equivalency principle.
The first step in the physical optics (PO) approximation is
to use GO for the description of fields right on the scattering
surface where GO approximation is still valid. The second
step is to calculate the scattered field outside the object using
the equivalency principle described in the following. Tangential components (n E, n H) of electric and magnetic vectors of the total field on the scattering surface (with the external unit normal n) can be interpreted as equivalent magnetic
and electric currents
n E
j m = n
(V/m)
je = n H
(11)
(A/m)
In the PO approach, the equivalent currents are defined in
the GO approximation. The total electromagnetic field on
the scattering object is considered approximately as the sum
of the GO incident and reflected waves (EGO Einc Eref ,
HGO Hinc Href ). Thus, the PO surface currents are defined
as
GO
j PO
e = n H
(12)
n E GO
j PO
m = n
This equation defines equivalent currents only on the illuminated side of the opaque scattering object. On the shadow
side, these currents are assumed to be zero. In the particular
case of perfectly conducting objects, the magnetic current does
not exist (jPO
m 0) due to the boundary condition n E 0,
and the electric current equals jPO
2n Hinc according to
e
the GO approximation.
The scattered field is found by the integration of equivalent
currents over the scattering surface S. Geometry of a sample
scattering problem is shown in Fig. 7, where the quantity r is
the distance between the integration (, , ) and observation
n
z
Canonical
Layer
Opaque object
with a layer
Figure 6. A scattered field at the reflection point on a coated scattering object is equal asymptotically (with kR1,2 ) to the field which
would be reflected from a tangential layer with the same material
properties. Due to losses, the contributions of rays and waves propagating along the object (inside the coating) become small and can be
neglected.
y
x
Figure 7. Schematics of a scattering problem: S is the surface of the
scattering object; the dashed part of this surface (with the boundary
) is located in the shadow region which is hidden from the incident
rays.
BACKSCATTER
179
2
(R, , ) points. In the far zone (R kmax
), the scattered field
is determined as
E = Z0 H = ik(Z0 Ae + Am
)
E = Z0 H = ik(Z0 Ae Am
)
ikR
1 e
A e,m =
j e,m eik cos dS
4 R S
cos = cos cos + sin sin cos( )
(13)
(15)
Accuracy of PO. Approximate estimations for the PO scattered field [Eq. (13)] can be found by the application of asymptotic techniques to the integrals [Eq. (14)] with the PO currents [Eq. (12)]. The first term of the asymptotic expansion
found in this way is correct for the fields scattered by smooth
convex objects and planar plates in the specular directions
predicted by GO. All higher-order terms in the PO asymptotic
expansion are incorrect. Only two exceptions exist when PO
provides the exact solution. The first is the scattering from
the infinite perfectly conducting plane. The second is the scattering from the semi-infinite perfectly conducting paraboloid
of revolution illuminated by the plane wave incident along
the symmetry axis (22). In the latter case, the scattered field
consists of only the GO reflected rays. Reference 22 also
shows that PO provides the correct second term in the highfrequency asymptotic expansion for the specular backscattering from any convex perfectly conducting bodies of revolution
when the incident wave propagates in the direction parallel
to the symmetry axis.
The first term of the PO asymptotic expansion for the field
scattered by smooth convex objects in specular directions represents the GO reflected rays [pp. 5062 of (2)]. Therefore, for
such objects the PO value of RCS in specular directions is
asymptotically (with k ) equivalent to the GO estimation.
However, it is well known that GO is valid only away from
the forward direction, i.e., from the shadow boundary of the
incident rays. But PO is more general than GO and is applicable in the vicinity of this direction. All known results show
that the first term of the PO asymptotic expansion for the
field scattered in the forward direction is correct and leads to
the following RCS for large opaque objects:
A2
2
[(m)2 ]
(14)
Here, E, is the electric field intensity (V/m); H, is the
magnetic field intensity (A/m); Ae is the electric potential vector (A); Am is the magnetic potential vector (V); and Z0
0 / 0 377 () is the impedance of vacuum.
The PO approximation for the scattered field follows from
Eqs. (13) and (14) when the PO approximation given by Eq.
(12) is used for equivalent surface currents and the integration region is restricted to the illuminated part of the scattering surface. The line shown in Fig. 7 is the boundary between the illuminated and shadow sides of the scattering
surface S. The PO approach is usually applied to large convex
objects. However, it is also applicable to concave objects when
the multiple GO reflections are taken into account.
= 4
(16)
Here, the quantity A is the area of the scattering object projection on the plane perpendicular to the direction of the incident wave propagation.
Figure 8. Directions of the forward ( ) and specular (
) scattering from the plate S. The dashed line A denotes the projection of the plate S.
Reference 23 contains similar PO estimations for thin
semi-transparent plates. The field on the plate surface is defined by complex reflection and transmission coefficients
which depend on the incidence angle () and polarization of
the incident wave. The incident wave with an arbitrary linear
polarization can be decomposed into two independent waves
with orthogonal polarizations. A decomposition is chosen such
that either the electric or magnetic vector of the incident
wave is parallel to the plate. In the first case, denote the reflection and transmission coefficients for the electric vector as
re() and te(), respectively. Similar coefficients, rh() and
th(), for the magnetic vector describe the plate when the magnetic vector of the incident wave is parallel to the plate. Reference 2 (pp. 479499) contains instructions for the calculation
of these coefficients. Two directions of scattering are of greatest interest. The first is the direction of specular reflection,
, and the second is the forward direction,
(Fig. 8). According to (23), the PO approximations of RCS in
the specular direction are given by
A2
|re ( )|2
2
A2
h ( ) = 4 2 |rh ( )|2
e ( ) = 4
(17)
and in the forward direction by
A2
|1 te ( )|2
2
A2
h ( ) = 4 2 |1 th ( )|2
e ( ) = 4
(18)
where the quantity A is the same as in Eq. (16). This is the
area of the plate projected on the plane perpendicular to the
direction of the incident wave. Equations (17) and (18) are
applicable for planar plates of an arbitrary shape under the
condition A 2. This means that the grazing angles (
/2) cannot be treated with these equations.
Known results for perfectly conducting plates (re,h() 1,
te,h() 0) show that PO estimations given in Eqs. (17) and
(18) are correct. These equations also give the correct result,
h(/2) 0, for perfectly conducting plates under the grazing
180
BACKSCATTER
1
2
Figure 9. The PO current given by Eq. (12) is discontinuous at the
shadow boundary of the scattering surface. This leads to spurious
waves (1 and 2) in the case of a smooth scattering surface.
incidence. In this case, the incident wave does not undergo
diffraction because its electric vector is perpendicular to the
plate surface. PO describes satisfactorily the field scattered
from large conducting plates not only in the specular and forward directions corresponding to main lobes in the directivity
pattern, but also in the directions of neighboring side lobes.
However, PO fails to predict a field level in minimums of the
directivity pattern [Figs. 7-19 and 7-20 on p. 509 of (2)] and
does not satisfy the reciprocity principle.
The PO currents given by Eq. (12) are discontinuous on
the shadow boundary of a scattering surface. The PO field
contains spurious waves from such a boundary in the case of
smooth scattering surfaces (Fig. 9). A similar current discontinuity on scattering objects with edges results in edge waves.
If the scattering edge is visible from the observation point,
such an edge wave does exist. The PO edge waves coming
from invisible edges are spurious shooting-through waves
(Fig. 10). Such shooting-through waves do not occur in the
backscattering direction. All PO spurious waves can be removed by neglecting the corresponding terms in the asymptotic expansion of the integral in Eq. (14). For real edge
waves, even the first-order term of their PO asymptotic
expansion is incorrect. This defect is remedied in PTD by the
inclusion of the field radiated by the so-called nonuniform
currents arising from the diffraction of the incident wave at
edges (14,15).
One should emphasize a special role of PO in PTD. PO is
a constitutive part of PTD. Therefore, the POs first- and
higher-order asymptotic terms are integral parts of the PTD
asymptotic expansions for the total scattered field. For example, the terms with coefficients (3d/16a) in the PTD equations
(99) and (100) of (15) are exactly the POs second-order terms
in the asymptotic expansion of the field scattered by a perfectly conducting cylinder of finite length.
Polarization of the PO Scattered Field. In general cases, components E and E in Eq. (13) have different amplitudes and
phases. This results in the elliptic polarization of the scattered field even in the case when the incident wave is linearly
polarized. This means that the electric vector of the scattered
field rotates with the angular frequency k c and its endpoint traces an ellipse. The direction of rotation (clockwise or
counterclockwise) is determined by the phase shift between
components E and E. The lengths of the elliptical axes are
determined by the amplitudes of these components. Due to
diffraction, the scattered field can contain an electric field
component perpendicular to the incident wave polarization.
This phenomenon is known as depolarization, or crosspolarization.
The PO field scattered by arbitrary perfectly conducting
objects in the backscattering direction does not contain the
crosspolarized component [p. 56 of (2)]. It is assumed only
that no multiple GO reflections occur on the objects surface.
This PO result is correct for scattering objects with certain
symmetry. These are objects with a symmetry plane parallel
both to the electric (or magnetic) vector of the incident wave
and to the direction of its propagation. Each element of such
a scattering object may create the crosspolarized component.
But due to the symmetry, the crosspolarized components from
symmetrical elements cancel each other in the backscattering
direction (Fig. 11). A convex smooth body of revolution whose
symmetry axis is parallel to the incident wave direction is a
simple example of such an object. A symmetrical plate, illuminated by the plane wave whose electric (or magnetic) vector
and direction of propagation are parallel to the symmetry
plane, is another example where the backscattered field does
not contain a crosspolarized component (Fig. 12).
As previously stated, the first term of the PO high-frequency asymptotic expansion represents the GO reflected ray.
This ray contains the crosspolarized component when the
electric vector of the incident ray makes any angle different
from 0 and 90 with the incidence plane at the reflection
point. This is the case when PO correctly describes depolarization of the scattered field in the high-frequency asymptotic
limit (k ). However, PO fails to predict all depolarization
effects caused by the diffraction part of equivalent surface
currents.
Ecr
1
Ecr
Einc
S
y
3
x
Figure 10. A scattered field is generated by the induced surface currents. Any approximations for these currents can result in the appearance of nonphysical components in the scattered field. In particular, the PO currents create spurious shooting-through edge waves (1,
2, and 3) passing through an opaque object.
Figure 11. Backscattering without depolarization from a symmetrical perfectly conducting surface S. The incident wave direction is parallel to the symmetry plane y-z. Vectors Ecr are the cross-polarized
components of the reflected field. Due to the symmetry, they cancel
each other.
BACKSCATTER
Einc
x
Figure 12. Backscattering without depolarization from a perfectly
conducting plate S. The incident wave direction is parallel to the symmetry plane y-z. Cross-polarized components scattered by the left and
right parts of the plate are symmetrical and completely cancel each
other.
Bistatic RCS. Bistatic RCS determines the power flux density of electromagnetic waves scattered by the object in an
arbitrary direction. The angle between the directions to the
transmitter and receiver is called the bistatic angle. Monostatic, or backscattering, RCS is a particular case of bistatic
RCS when the bistatic angle equals zero. Some PO results for
bistatic RCS have already been discussed. This section addresses the interrelationships between bistatic and monostatic RCS.
Using PO, one can prove the following statement:
For perfectly conducting bodies which are sufficiently smooth, in
the limit of vanishing wavelength, the bistatic cross section is
equal to the monostatic cross section at the bisector of the bistatic
angle between the direction to the transmitter and receiver.
[pp. 157160 of (1) and p. 11 of (2)]
There is a simple physical explanation for this result. As
already stated, the first term of the PO asymptotic expansion
for the field scattered by smooth objects exactly equals the
GO expression for the reflected rays. The monostatic RCS
caused by these rays is given by Eq. (10). In the case of perfectly conducting objects, this equation reduces to
= R1 R2
(19)
It should be noted that this equation is valid also for the bistatic RCS, which therefore does not depend on the bistatic
angle 2 (Figs. 5 and 6). This follows directly from Eqs.
(5.32), (6.19), and (6.20), given in Chapter 8 of (19):
1
E (0) R1 R2
2
1
H (r) = H (0) R1 R2
2
E (r) =
eiks
s
eiks
s
181
GO reflected field really does not depend on the incidence
angle. As a result, the bistatic RCS does not depend on the
bistatic angle and is the same as the monostatic RCS at the
bisector direction that is perpendicular to the scattering surface at the reflection point. Thus, the cited equivalence between the bistatic and monostatic RCS is a pure GO effect
and is fulfilled asymptotically (with k ) only in the ray
region, away from the shadow boundary behind the scattering
object. It is also clear that this equivalence is not applicable
when the scattered field contains multiple reflected rays arising from concave parts of the scattering surface. Reference 1
(pp. 160183) presents additional results for bistatic RCS of
some typical objects found using PO and other approximations.
In the case of coated smooth objects, Eq. (20) leads to the
bistatic RCS
( ) = |re,h ( )|2 R1 R2
(21)
where the reflection coefficients re,h() depend on the polarization and direction of the incident wave. Therefore, the asymptotic equivalence between bistatic and monostatic RCS can be
valid only for those bistatic angles where re,h() re,h(0).
This discussion relates to the situation where the bistatic
scattered field is produced by a single scattering source (the
reflection point). The field scattered by complex objects can
have many sources (scattering centers) on the objects surface.
Figure 13 illustrates the bistatic scattering from the center
located at the point (xn, yn, zn). The origin of Cartesian coordinates is somewhere inside the scattering object. The z-axis is
directed along the bisector of the bistatic angle . In the far
zone, the bistatic scattered field can be represented as the
sum of partial contributions from all scattering centers [pp.
983988 of (4)]:
u ( ) =
eikR
v n e2ikz n cos 2
R n
(22)
Here, u is either the electric or magnetic vector of the total
scattered field; R is the distance from the origin to the observation point. Vector vn determines the amplitude and polarization of the wave generated by the nth scattering center.
Suppose that vectors vn and the number of scattering centers
are constant inside the angular sector 0 max. Assume
also that coordinates zn of scattering centers do not depend on
the bistatic angle , while coordinates xn and yn can be func-
(xn , yn , zn) y
(20)
These expressions describe the field reflected by smooth convex objects at a far distance (s R1,2) from the reflection point
for any incidence angle (0 /2). In the case of reflection
from concave surfaces, the reflected field acquires the additional phase shift of (/2) in passing through a focus of reflected rays. Vectors E(0) and H(0) denote the reflected field
at the reflection point. Expressions (20) clearly show that the
/2
z
/2
R
Figure 13. Geometry of the bistatic scattering problem. A solid winding line represents a scattering object with many scattering centers
(xn, yn, zn). The scattering direction forms the angle with the direction of the incident wave.
182
BACKSCATTER
tions of this angle. Under these conditions, the monostatic
field scattered in the bisector direction equals
u (0) =
eikR
v n e2ikz n
R n
(23)
Comparisons of Eqs. (22) and (23) show that the bistatic
RCS, (, k), at the frequency c k will be equal to the
monostatic RCS, [0, kcos(/2)], at the frequency c k
cos(/2). This equality requires the additional assumption
that each vector vn is constant in the frequency band
[c kcos(/2) c k]. The derivation, some applications,
and restrictions of this equivalence relation are presented in
(4) (pp. 983988). In particular, this reference notes that this
equivalence is not true for the bistatic scattering from spheres
when the bistatic angle exceeds one degree and the sphere
radius is less than 6. Before applying this equivalence in
practice, we must first check carefully that all assumptions
made in this scattering model are really fulfilled. One can
expect that this approximate model can be reasonable only
for small bistatic angles.
PTD as an Extension of PO. PTD is a natural extension of
PO (14,15,24). In PTD, the PO current given by Eq. (12) is
considered as the uniform component ( j0) of the total surface
current and is supplemented by the additional, nonuniform
component ( j1). In contrast to the PO current that has the GO
origin, the nonuniform current is caused by diffraction at
smooth bendings, sharp edges, corners, and any other geometrical discontinuity and material inhomogeneity on the scattering surfaces. Creeping and edge current waves are examples
of such a current. The field generated by the nonuniform current represents the PTD contribution to the scattered field.
Exact analytical expressions for nonuniform currents are not
available. Therefore, one has to find their high-frequency approximations by the solution of appropriate canonical problems. In this manner, Fock developed special functions which
describe the nonuniform current on smooth convex objects in
the vicinity of the shadow boundary (19). The Sommerfeld solution of the wedge canonical problem is used for the asymptotic description of the nonuniform current near perfectly conducting edges (14,15,24). The concept of uniform and
nonuniform currents plays a key role in PTD and those hybrid
techniques that combine direct numerical methods with highfrequency asymptotic approximations (6,17,18). Reference 15
shows that PTD properly defines the leading term in the highfrequency asymptotic expansions for primary and multiple
edge waves. A close connection exists between PTD and GTD.
The latter automatically follows from the PTD integrals when
they are evaluated by the stationary phase technique [pp.
136138 of (15)]. Some PTD results are presented in the
next section.
Exact, numerical solutions of scattering problems for bodies
of revolution can be found, for example, in (6), (18), and (25).
Semitransparent Disk
The geometry of this scattering problem is shown in Fig. 8.
The backscattering direction is determined by the spherical
coordinates , /2. The disk radius is denoted
by the letter a. The incident wave can have either E- or Hpolarization. In the first case, the electric vector is perpendicular to the incidence plane and parallel to the disk face. The
disk properties are described by the reflection and transmission coefficients, re(), te(), with respect to the electric vector.
In the case of H-polarization, the magnetic vector of the incident wave is perpendicular to the incident plane and parallel
to the disk face. The reflection and transmission coefficients,
rh(), th(), determine the magnetic vector on the front (z
0) and rear (z 0) faces of the disk, respectively. According to Eq. (67) in (23), the backscattering RCS is given by
ePO ( ) = |re ( )|2 a2 [J1 (2ka sin )]2 cot2
(24)
hPO ( ) = |rh ( )|2 a2 [J1 (2ka sin )]2 cot2
where J1(x) is the Bessel function and the incidence angle is
restricted by the values 0 /2. For perfectly conducting
disks, one should put re() 1 and rh() 1. Then, in the
case of the normal incidence ( 0), Eq. (24) reduces to
ePO = hPO = a2 (ka)2
(25)
Figure 7-24 on p. 514 of (2) shows that this equation is in
good agreement with the exact results when ka 5. Note also
that Eq. (18), with A a2 cos , determines the PO bistatic
RCS of this disk for the forward direction ( ). PTD estimations for RCS of a perfectly conducting disk are presented
in Chapters 2 and 5 of (14). See also pp. 514521 of (2). Some
important corrections in the PTD expressions for bistatic scattering from a disk are given in (24). Contributions of multiple
edge waves to forward scattering are presented in (15) (pp.
149151).
Circular Cone
Geometrical parameters of a perfectly conducting cone are
shown in Fig. 14. The incident wave direction is parallel to
the symmetry axis of the cone. The PO backscattering RCS is
2a
BACKSCATTERING RCS OF SIMPLE SHAPES
This section contains examples of PO estimations for RCS of
simple objects. Whenever possible, these estimations are accompanied by more precise PTD counterparts that include the
contributions of primary edge waves generated by the nonuniform edge currents. Only objects with symmetry of revolution
are considered. All given data are taken from (15) and (16).
l
Figure 14. Backscattering from a truncated cone. The base diameter
of the cone (2a) is large compared to the wavelength. The length of
the cone (l) can be arbitrary. In the limiting case l 0, the cone
transforms into a disk.
BACKSCATTER
perfectly conducting paraboloid equals
PO = 4a2 tan2 sin2 kl
(30)
This equation can be written in another form as
2a
PO = a2 tan2 |eikl eikl |2
l
Figure 15. Backscattering from a truncated paraboloid. The base diameter of the paraboloid (2a) is large compared to the wavelength.
The length of the paraboloid (l) can be arbitrary. In the limiting case
l 0, the paraboloid transforms into a disk.
given by Eqs. (17.06) and (17.09) in (15),
2
1
tan2 sin kl tan eikl
PO = a2
ka
(26)
where the cone length equals l a cot . To clarify the physics in this equation, we rewrite it as
2
i
i
tan eikl tan +
tan eikl
PO = a2
2ka
2ka
(27)
The first term (with exponential eikl) is related to the wave
scattered by the cone tip. Comparison with the exact solution
[Fig. 18.15 on p. 691 of (3)] shows that this PO approximation
is quite satisfactory for all cone angles (0 /2). The
second term (with the exponential eikl) describes the edge
wave contribution. This PO approximation is incorrect. PTD
takes into account the additional contribution from the nonuniform (diffraction) currents located near the cone edge and
provides a more accurate result, given by Eqs. (17.06) and
(17.08) in (14),
2
2
1
sin
2
PTD
2
n
n
ikl
tan sin kl +
e
= a
2
ka
cos cos
n
n
(28)
(31)
which is more convenient for the physical analysis. The term
with the exponential eikl gives the correct contribution of the
specular reflection from the paraboloid tip. The term with the
exponential eikl represents the edge wave contribution and is
wrong. PTD includes the additional contribution from the
nonuniform edge currents and provides the correct result,
given by Eq. (18.04) in (14):
2
2
sin
PTD
2
n
n
2ikl
e
= a tan +
2
cos cos
n
n
(32)
where n 3/2 /. When the paraboloid transforms into
the disk ( /2 and l 0), these expressions reduce to
Eq. (29).
Truncated Sphere
The geometry of this scattering problem is shown in Fig. 16.
The angle is formed by the tangent to the sphere generatrix and the symmetry axis. The sphere radius equals
a/cos , where a is the base radius. The length of the truncated sphere equals l (1 sin ). It is assumed that l
. The PO backscattering RCS of a perfectly conducting
sphere equals [Eq. (19.05) in (14)]
2
1
i
i
tan
PO = a2
e2ikl
cos
2ka
2ka
(33)
In this equation, the first two terms represent the specular
reflection from the sphere, and both are correct. The third
term (with the exponential e2ikl) gives the contribution from
the edge and it is wrong. With ka 1, Eq. (33) simplifies to
PO
2
1
2ikl
tan e
= a
cos
2
where n 3/2 /. When the cone transforms into the disk
( /2, l 0) the previous expressions reduce to
PO = PTD = a2 (ka)2
183
(34)
(29)
2a
which coincides with Eq. (25).
Paraboloid
The directrix of a paraboloid is given by the equation r
2pz where p a tan (Fig. 15). The length of the paraboloid
equals l a2 /(2p) (a/2)cot . The angle is formed by the
symmetry axis z and the tangent to the directrix at the point
z l. The radius of the paraboloid base equals a. The incident
wave propagates in the positive direction of the z-axis. According to Eq. (18.02) in (14), the PO backscattering RCS of a
l
Figure 16. Backscattering from a truncated sphere. The base diameter of the sphere (2a) is large compared to the wavelength. The length
of the sphere (l) can be arbitrary. In the limiting case l 0, the
sphere transforms into a disk.
184
BACKSCATTER
When 0, the latter gives the RCS of a hemisphere,
a2. The PTD backscattering RCS is determined by Eq.
(19.12) in (14),
2
2
1
sin
n
n
+
e2ikl
PTD = a2
2
cos
cos cos
n
n
(35)
where n 3/2 /. When the sphere transforms into the
disk ( /2, , l 0), Eqs. (34) and (35) reduce exactly
to Eq. (29).
Circular Cylinder with Flat Ends
The diameter and length of a perfectly conducting cylinder
are assumed to be large as compared with the wavelength of
the incident wave. PO and PTD estimations for backscattering RCS are developed in Chapter 3 of (14). They are also
presented in (2) (pp. 308312). PTD asymptotic expressions
for bistatic RCS are given in (15) (pp. 152154).
BACKSCATTERING FROM COMPLEX
OBJECTS AND STEALTH PROBLEMS
Computer codes based on GTD, PTD, and on their hybridizations have been developed for prediction of high-frequency
scattering from complex perfectly conducting objects. Relevant references can be found in (16), (18), and in special issues of Proc. IEEE (1989), IEEE Trans. Antennas Propag.
(1989), and Annales des Telecommunications (1995), which are
mentioned in the reading list. Note also the XPATCH code
(based on the shooting-and-bouncing ray technique and PTD),
which allows the calculation of backscattering from complex
geometries. Information about this code is published in IEEE
Trans. Anntennas Propagat. Magazine, 36 (1), pp. 6569,
1994. Computer codes interfaced with graphical utilities of
workstations can display three-dimensional chromatic views
of scattering centers and magnitudes of their contributions
to RCS. This is the end result of complicated computations.
However, a part of this can be obtained without any computations. Nature can show us the location of all scattering centers if we bring a small metallized model of the scattering
object into an anechoic optical chamber and illuminate the
model by the light. Bright shining points (scattering centers)
seen on a scattering object are exactly those from which the
radar waves will be reflected toward the radar, if we look at
the object from the light source direction. (The following text
is taken from Ref. 16 and slightly modified.)
The locations of these points do not depend on the frequency of incident electromagnetic waves, and they are determined completely by the location of the light source (the radar), the observer, and the scattering object. These shining
points obey the Fermat principle. This means that the path
along the ray between the source, the reflecting point, and
the observer is extremal (minimal or maximal) in comparison with similar paths corresponding to neighboring points
on the objects surface. A more detailed description of the
Fermat principle is presented for example in Section 3.3.2
in Ref. 20.
Waves reflected from discrete shining points located on
the smooth parts of the scattering object represent the
usual geometrical optics reflected rays. Waves reflected
from discrete shining points located on edges, tips, and
corners are diffracted rays. The farthest shining points on
a smooth object, i.e., those located on the boundary between
visible and invisible sides of the object, create surface diffracted rays.
As the orientation of the object is changed, the shining
points move along the object. Some of them can merge with
each other and create a brighter point. In this case our eyes
(i.e., the radar) are located on a caustic is the envelope of
merged rays.
We can also observe bright shining lines and bright shining spots on the object, which contain an infinite number of
continuously distributed shining points. The important property is that the optical path through a shining point from the
source to the observer is constant for all of these points. It is
assumed here that the source and observer are far from the
scattering object. All reflected waves from these points reach
the observer with the same phase. From the mathematical
point of view, each such point is a stationary point of the infinite order: the derivatives (of any higher order) of the wave
phase along the shining line (or along the shining spots) are
zero at these points.
Shining spots and lines located on smooth parts of the scattering surface generate powerful reflected beams (such as
those radiated by reflector antennas) which represent the
strongest contributors to RCS. Shining edge lines create edgediffracted beams whose contributions can be comparable with
those from ordinary reflected rays.
It is difficult to model in optics the electromagnetic properties of realistic scattering surfaces for the radar frequency
band. But the optical modeling can be used to identify the
scattering centers and to control them by an appropriate
shaping of the scattering surface. As it is well known, one of
the basic ideas of the current stealth technology is to use an
appropriate body shaping and to shift all reflected beams and
rays away from the directions to the radar. See, for example,
Refs. 2, 16, and the radar cross-section handbooks mentioned
in the reading list. Some interesting details about the development of stealth technology in the United States are presented in Refs. 2628.
The second idea of stealth technology is traditional: to use
radar absorbing materials (RAMs) and composite structures
in order to reduce the intensity of reflected beams and rays.
References 2, 16, 29, and radar handbooks (mentioned in the
reading list) describe fundamental concepts used in the design and application of RAMs. We present here some details
taken from Ref. 16. In order to use RAMs efficiently, it is necessary to place an electric (magnetic) RAM in the region
where the average electric (magnetic) field is maximal. Location of these regions in the vicinity of real objects depends on
many factors, such as the radar frequency, geometry, size,
and electrical properties of the object, as well as properties of
materials intended for absorption. Identification of such regions and optimization of the RAM parameters to minimize
RCS is a very complex problem. Its solution is attainable only
in some simple cases. Most of these relate to absorbing layers
on an infinite metallic plane. From the physical point of view
such absorbing layers can be considered as open resonators
that can support eigen-oscillations. Frequencies of eigen-oscillations are complex quantities. Their imaginary part is responsible for the loss inside the resonator and radiation
BACKSCATTER
from the resonator. It turns out that the minimal reflection
from such resonators happens when the frequency of an
incident wave is close to the real part of the resonator
eigenfrequency.
Note that thin electric RAMs are not efficient when applied
on metallic objects. This is due to the boundary condition: the
tangential component of the electric field is very small on the
metal surface. On the contrary, magnetic absorbing materials
can be applied directly to the surface of a metallic object. This
is an important advantage of magnetic materials over electric ones.
However, any RAMs (electric, magnetic, and hybrid) homogeneous in the direction parallel to the reflecting plate
are not efficient for grazing incidence ( 90, in Fig.
5). In this case, the reflection coefficient tends to unity
independently of the incident wave polarization when
90. This is a fundamental limitation of ordinary RAMs.
They do not work against grazing incident waves. That is
why ordinary RAMs do not reduce forward scattering. Actually, the RAM terminology is justified only for incidence
angles that are not too far from 0 and when the
reflection coefficient is small enough.
Various geometrical and material inhomogeneities on the
scattering surface can partially transform the incident wave
into surface waves propagating along absorbing layers. This
may be used to further reduce the RCS. However, this idea
has two essential defects. First, any inhomogeneity creates an
additional undesirable scattered field. Second, it is not a simple problem to design an absorbing layer that would allow the
propagation of surface waves. To support surface waves with
the electric vector parallel to the incidence plane, the surface
impedance must be inductive. But the surface impedance
must be capacitive to support surface waves with the electric
vector perpendicular to the incidence plane. This means that
the surface impedance, and therefore the absorbing layer,
must depend on the radar polarization with respect to the
incidence plane. But this plane is different at different points
of the scattering surface and different at the same point when
the scattering object changes its orientation with respect to
radar. It is very difficult and probably impossible to design
such an absorber, especially against radars with circular polarization. However, for some chosen orientations of the object
and for an appropriate polarization of the incident wave, this
might not be a hopeless problem.
Development of efficient hybrid techniques and computer
codes to predict RCS of large complex objects with realistic
materials and research efforts to overcome the above physical
limitations in RCS reduction represent challenging problems
for future stealth technology. One can expect that future advanced computer codes will contain as necessary constitutive
components the known high-frequency techniques (such as
GTD, PTD, and the Uniform Theory of Diffraction) extended
for coated and composite objects. Diffraction coefficients used
in these techniques can be determined by the numerical solution of appropriate canonical problems. Direct numerical
methods should be used for calculation of scattering from
those elements of the scattering object that cannot be treated
by high-frequency methods. Diffraction interaction between
the objects elements handled by high-frequency techniques
and by direct numerical methods can be described by the surface integral equations.
185
BIBLIOGRAPHY
1. J. W. Crispin Jr. and K. M. Siegel (eds.), Methods of Radar CrossSection Analysis. New York: Academic Press, 1968.
2. G. T. Ruck, D. E. Barrick, W. D. Stuart, and C. K. Kirchbaum,
Radar Cross-Section Handbook, Vol. 1 and 2. New York: Plenum
Press, 1970.
3. J. J. Bowman, T. B. A. Senior, and P. L. E. Uslenghi (eds.), Electromagnetic and Acoustic Scattering by Simple Shapes. New York:
Hemisphere Publishing Corp., 1987
4. P. C. Fritch (ed.), Special Issue on Radar Reflectivity, Proc. IEEE,
53 (8): August 1965.
5. M. N. O. Sadiku, Numerical Techniques in Electromagnetics. Boca
Raton, FL: CRC Press, 1992.
6. E. K. Miller, L. Medgyesi-Mitschang, and E. H. Newman, Computational Electromagnetics, New York: IEEE Press, 1991.
7. P. P. Silvester and G. Pelosi (eds.), Finite Elements for Wave Electromagnetics, New York: IEEE Press, 1994.
8. P. Ya. Ufimtsev and A. P. Krasnozhen, Scattering from a straight
thin wire resonator, Electromagnetics, 12 (2): 133146, 1992.
9. L. A. Vainshtein, Waves of current in a thin cylindrical conductor, II. The current in a passive oscillator, and the radiation of a
transmitting antenna, Sov. Phys., Tech. Phys., 4 (6): 617626,
December 1959.
10. L. A. Vainshtein, Waves of current in a thin cylindrical conductor, III. Variational method and its application to the theory of
ideal and impedance conductors, Sov. Phys., Tech. Phys., 6 (1):
1929, July 1961.
11. M. A. Leontovich and M. L. Levin, Towards a theory on the simulation of oscillations in dipole antennas, Zhurnal Technicheskoi
Fiziki, 14 (9): 481506, 1944 (in Russian). [The English translation is published in the report: K. C. Chen (ed.), SAND91-0720,
UC-705, Sandia National Laboratories, Albuquerque, NM and
Livermore, CA, Contract DE-AC04-76DP00789, January 1992].
12. J. B. Keller, Geometrical theory of diffraction, J. Opt. Soc. Am.,
52: 116130, 1962.
13. R. C. Hansen (ed.), Geometrical Theory of Diffraction, New York:
IEEE Press, 1981.
14. P. Ya. Ufimtsev, Method of Edge Waves in the Physical Theory of
Diffraction. Moscow: Soviet Radio Publishing House, pp. 1243,
1962. Translated by U.S. Air Force, Foreign Technology Division,
Wright-Patterson AFB, Ohio, 1971; Technical Report AD No.
733203, DTIC, Cameron Station, Alexandria, VA.
15. P. Ya. Ufimtsev, Elementary edge waves and the physical theory
of diffraction, Electromagnetics, 11 (2): 125160, 1991.
16. P. Ya. Ufimtsev, Comments on diffraction principles and limitations of RCS reduction techniques, Proc. IEEE, 84 (12): 1830
1851, 1996.
17. D. P. Bouche, F. A. Molinet, and R. Mittra, Asymptotic and hybrid techniques for electromagnetic scattering, Proc. IEEE, 81:
16581684, 1993.
18. W. R. Stone (ed.), Radar Cross Section of Complex Objects, New
York: IEEE Press, 1990.
19. V. A. Fock, Electromagnetic Diffraction and Propagation Problems. London: Pergamon Press, 1965.
20. M. Born and E. Wolf, Principles of Optics, New York: Pergamon,
1975.
21. H. M. Macdonald, The effect produced by an obstacle on a train
of electric waves, Phil. Trans. Roy. Soc. Lond., Series A, Math.
Phys. Sci., 212: 299337, 1912.
22. C. E. Schensted, Electromagnetic and acoustic scattering by a
semi-infinite body of revolution, J. Appl. Phys., 26: 306308,
1955.
186
BACKSCATTER
23. P. Ya. Ufimtsev, Diffraction of electromagnetic waves at blackbodies and semi-transparent plates, Radiophys. Quantum Electr.,
11: 527538, 1968.
24. P. Ya. Ufimtsev, Comments on Comparison of three high frequency diffraction techniques, Proc. IEEE, 63: 17341737, 1975.
25. R. D. Graglia et al., Electromagnetic scattering for oblique incidence on impedance bodies of revolution, IEEE Trans. Antennas
Propag., 43 (1): 1126, 1995.
26. M. W. Browne, Two Rival Designers Led the Way to Stealthy
Warplanes, in The New York Times, Science Times Section,
US, May 14, 1991.
27. S. F. Brown, The Secret Ship, in magazine Popular Science,
US, October 1993.
28. B. Rich and L. Janos, Skunk Works, Boston-New York-London:
Little, Brown & Company, 1994.
29. K. J. Vinoy and R. M. Jha, Radar Absorbing Materials, Boston:
Kluwer Academic Publishers, 1996.
Reading List
This section contains short comments on some related references.
P. C. Fritch (ed.), Radar Reflectivity, Special issue of the Proc. IEEE,
53 (8), August 1965.
The first attempt to sum up basic results in the field of RCS.
Includes a comprehensive subject index, about 1500 titles (pp.
10251064).
J. W. Crispin Jr. and K. M. Siegel (eds.), Methods of Radar Cross
Section Analysis. New York: Academic Press, 1968.
Includes a short historical survey of high-frequency approximations. Contains results of such approximations for monostatic and
bistatic RCS of simple objects. States a strategy of RCS calculations for complex objects. Some results are conveniently summarized in tables. Table 5 on p. 147 contains expressions for RCS in
the Rayleigh region (for objects small in comparison with wavelength). High-frequency monostatic and bistatic RCS are given in
Tables 7 and 8 on p. 168, 169, 171.
G. T. Ruck, D. E. Barrick, W. D. Stuart, and C. K. Kirchbaum, Radar
Cross-Section Handbook. New York: Plenum Press, 1970.
This is a real encyclopedia of RCS, which includes most results
obtained before 1970. It contains numerous theoretical and experimental results for both perfectly conducting and absorbing objects.
The physical theory of diffraction (PTD) is mentioned here as the
SommerfeldMacdonald technique. Equation (5.1-54), presented
on p. 351, for the bistatic RCS of an ellipsoid is incorrect. In the
particular case when an ellipsoid transforms into a sphere, this
equation does not provide the bistatic RCS for the sphere,
a2. Instead it leads to the wrong quantity a2 /(1 cos )2,
where is the bistatic angle.
J. J. Bowman, T. B. A. Senior, and P. L. E. Uslenghi (eds.), Electromagnetic and Acoustic Scattering by Simple Shapes. New York:
Hemisphere Publishing Corp., 1987.
Contains a comprehensive collection of theoretical results for
RCS of simple objects which allow the exact solutions of diffraction
problems. Both low-frequency and high-frequency approximations
are presented as well.
M. Skolnik (ed.), Radar Handbook. New York: McGraw-Hill, 1970.
Contains many results of measurements and calculations for
RCS. Calculations were carried out mostly by the physical optics
approach and GTD. Analytical expressions for RCS are not given.
W. R. Stone (ed.), Radar Cross Sections of Complex Objects, Special
issue of the Proc. IEEE, 77 (5), May 1989.
W. R. Stone (ed.), Radar Cross Sections of Complex Objects, Special
issue of the IEEE Trans. Antennas Propag. 37 (5), May 1989.
These two references contain many theoretical results concerning RCS for complex objects (perfectly conducting objects with
complex shapes and simple objects with complex boundary conditions).
W. R. Stone (ed.), Radar Cross Sections of Complex Objects, New York:
IEEE Press, 1989.
This book consists of a collection of articles. It includes expanded versions of about half of the papers published in two previously mentioned special issues. It also contains papers written
especially for this book and reprints of some earlier key papers.
J. M. Bernard, G. Pelosi, and P. Ya. Ufimtsev (eds.),, Radar Cross
Sections of Complex Objects, Special issue of the French journal
Annales des Telecommunications, 50 (56), MayJune 1995. It is
published in English with abstracts translated into French.
Contains the asymptotic analysis of RCS for higher-order
curved surfaces, physical theory of slope diffraction, PO and PTD
analysis of trihedral corner reflectors, a selective review of some
numerical methods for electromagnetic scattering, and some
other results.
The following three books contain additional information on RCS.
They include concise descriptions of basic exact and approximate
techniques for prediction of RCS, they introduce methods of RCS
enhancement and reduction, and they contain a large number of
calculated and measured data for RCS of many typical simple and
complex objects. The books complement each other, with emphasis
on different aspects in the field of RCS.
A. L. Maffett, Topics for a Statistical Description of Radar Cross Section. New York: John Wiley & Sons, 1989.
This book treats the subject of RCS with special emphasis on
statistical aspects and applications. It reflects broad interests of
the author: from historical background and perspective through
analytical and numerical methods of RCS calculation and RCS
measurements to elements of detection theory, investigation of anisotropic layers, and the inverse problem for anisotropic materials
with diagonal permittivity and permeability tensors.
A. K. Bhattacharyya and D. L. Sengupta, Radar Cross Section Analysis and Control. Boston-London: Artech House, 1991.
The book concentrates its attention on deliberate changes of
RCS (enhancement and reduction). It contains a useful table (p.
108) which represents in concise form the comparison of different
methods available for RCS analysis, with discussion of their advantages, disadvantages, and possible applications. It also complements other books by inclusion of the Maluzhinets function, which
plays a fundamental role in the theory of scattering by absorbing
objects with sharp edges or with impedance discontinuities. The
computer code for the calculation of this important function is provided in the appendix of the book.
E. F. Knott, J. F. Schaffer, and M. T. Tuley, Radar Cross Section, 2nd
Ed. Boston-London: Artech House, 1993.
This book presents updated material which covers most aspects of RCS: radar fundamentals, radar detection, RCS prediction, RCS reduction, radar absorbing materials, and RCS measurements. Chapter 14 can be especially useful for a brief review.
It contains a table (p. 562) with RCS estimations, as well as the
RCS data presentation formats and data reduction recipes.
P. Ya. Ufimtsev, Comments on diffraction principles and limitations
of RCS reduction techniques, Proc. IEEE, 84: 18301851, 1996.
RCS reduction techniques are discussed briefly from the physical point of view. Attention is concentrated on the physical structure of radar waves scattered from large objects. Possible passive
and active techniques to control and reduce reflected beams, rays,
and shadow radiation as well as potential limitations of these
techniques are considered. In particular, it is emphasized that
grazing reflected rays and shadow radiation cannot be eliminated
by absorbing materials.
R. C. Hansen (ed.), Geometrical Theory of Diffraction. New York: IEEE
Press, 1981.
BACKTRACKING
This book consists of a collection of articles. It contains key
papers on GTD, asymptotic solutions of some canonical problems,
and applications-oriented papers.
Many scattering objects contain nonmetallic materials, composites, and various layered structures. To simplify the solution of
scattering problems for such objects, it is often practical to apply
approximate boundary conditions. These conditions are enforced
on the external surface of the object and contain important information about the internal structure of the scattering object. As a
result, this approximation allows one to substantially reduce the
spatial region under investigation. The two following books present the development and applications of this approximation technique.
T. B. A. Senior and J. L. Volakis, Approximate Boundary Conditions
in Electromagnetics. London: The Institution of Electrical Engineering, 1995.
D. J. Hoppe and Y. Rahmat-Samii, Impedance Boundary Conditions
in Electromagnetics, Washington, D.C.: Taylor & Francis, 1995.
PYOTR YA. UFIMTSEV
University of California
at Los Angeles
BACKSCATTERING. See BACKSCATTER.
187
J. Webster (ed.), Wiley Encyclopedia of Electrical and Electronics Engineering
c 1999 John Wiley & Sons, Inc.
Copyright
CABLE TELEVISION
Cable TV (CATV) as a successful business can be considered an American invention. Even though first commercial cable television installations date back as early as 1936 in Europe (United Kingdom and Germany),
they did not create a market at that time.
Right after World War II, the first CATV systems in the United States appeared almost simultaneously
in Pennsylvania and Oregon. Community antenna television was conceived as a system for delivery by cable
of signals broadcast by wireless television stations. The system was primarily targeted to deliver television
in rural areas with poor wireless reception. A community antenna provided broadcast television to these
neighborhoods via coaxial cables. The abbreviation CATV meant community antenna television at this early
stage of the system.
For distant places the operator used several microwave links to bring the three major network signals
across hundreds of kilometers and then distributed them around the local community by a cable network. If
mountains blocked the radio path, a single antenna on the mountaintop fed the cable trunk down into the town.
This kind of service was known as basic cable, and the customer paid few dollars per month for a high-quality
TV signal of three existing channels. Advertisers on those channels were the real promoters of CATV at that
time.
The need for signal amplification approximately every kilometer limited the area that could be served.
Also, a separate amplifier was needed for each channel, which restricted the number of channels to three or
four. In the mid-1960s, improvements in transistors boosted channel numbers progressively. By 1970 there
were 2750 operators serving six million homes.
In 1974 satellites (so far used only for government purposes) became commercial. Cable operators started
to use satellite transponders to rebroadcast to their facilities programs from multiple producers, and their large
capacity made it possible to offer many channels using up to a total of 300 MHz of bandwidth. Service providers
started to produce programming and sell their services to cable as well as to direct broadcast satellite (DBS)
operators, thus motivating rapid deployment in large metropolitan areas, where cable distribution systems
competed with DBS.
Networks specifically designed to be distributed by the cable system appeared in the United States by
1975 with Time Inc.s Home Box Office (HBO); soon others emerged, such as Ted Turners Superstation (soon
renamed WTBS) in 1976, and C-SPAN (live broadcasts of the House of Representatives), ESPN (sports), and
Nickelodeon (childrens programming), all in 1979. The 24-hour news Cable News Network (CNN) followed in
1980.
Though CATV offered a wider range of programming than ordinary broadcasting, the capacity of pure
coaxial systems became insufficient, being still small compared to the number of DBS channels. To overcome
this lack, with the invention of the linear light source in the 1980s, cable operators started to replace coaxial
cable trunks with fiber systems. The benefit was not only in reducing the number of amplifiers (thus increasing
available bandwidth) but also in improving the reliability while minimizing maintenance costs. These networks
are referred to as CATV HFC networks, where CATV no longer means community antenna television but cable
television, and HFC stands for hybrid fiber coaxial. New channels were pay-per-view, bringing in new revenues
1
CABLE TELEVISION
to broadcasters while broadening opportunities to the creative community. This development reached a peak
by the mid-1980s and congestion by the early 1990s.
It is worth noting that cable operators must negotiate franchises with municipalities that authorize them
to lay cable in the ground, the use of poles, and so on, in exchange for fees to the local government. In this
respect CATV is similar to public utilities, since it uses public rights and places to deploy a capital-intensive
network. It distributes services from content providers, and subscribers are billed on a monthly basis.
CATV networks have their main market in United States. Their development followed different patterns
in other countries. CATV penetration in Southern European countries is still small, but current deployments
are full digital and with a large optical trunk plant to target directly not only broadcasting but the broadband
market. CATV is full-grown in Central European countries such as in Germany.
Worldwide, CATV networks serve over 200 million residences, while telephony serves more than 500 million. CATV is currently extending its capabilities by migrating from analog narrowband to digital broadband.
Indeed, it has recently emerged as a promising access network infrastructure for the delivery of voice, video,
and new broadband applications. Current research efforts are focused on the design of protocols for CATV to
deliver different levels of quality of service (QoS) for diverse user applications.
Catv Network Architecture
The architecture of a CATV network comprises essentially the following three basic operations in traditional
cable networks:
(1) Signal Reception Cable programming is obtained from satellite or terrestrial broadcasting at places called
headends, where signals are also appropriately processed and combined into an analog stream to be broadcast by cable.
(2) Signal Distribution The stream obtained at the headend is distributed to the households via coaxial cable
and optical fiber. In order to ensure the quality of signal to households, the signal is amplified (roughly
every kilometer) and/or attenuated, depending on the users proximity to the headend and network nodes.
Signals must also be equalized, since higher frequencies undergo higher attenuations.
(3) Signal Delivery Equipment in the home converts cable signals into tunable TV channels. Descramblers
decode encrypted programming, and additional equipment allows delivery on demand.
CATV network architecture has evolved drastically since its first implementations. Early CATV networks picked up analog TV signals from satellites or landlines, and each signal was downstreamed through
a tree-and-branch topology and one-way delivery services (in contrast to the telephone star topology, which
maximizes interconnection). Technical upgrading of CATV networks is infrequent and very capital-intensive,
and consequently some geographical areas will overtake others. Nevertheless, their chronological evolution
from a technical point of view can be outlined as follows:
From analog to digital
From all-coaxial to HFC
From one-way to two-way
Migration to digital improved the system capacity substantially, since every 6 MHz analog video channel
yields 27 Mbit/s of raw throughput with 64-QAM modulation and RS FEC. Such a data rate can convey up to six
digital video channels through MPEG-2 compression with the same quality as analog channels. Migration to
CABLE TELEVISION
Fig. 1. CATV architectures: (a) tree-and-branch all-coaxial network; (b) hybrid fiber coaxial network with an optical trunk
ring.
hybrid coaxial has increased robustness by reducing the number of households in serving areas and increasing
the bandwidth (>750 MHz) by decreasing the coaxial cascade depth (1).
Figure 1(a) shows an early tree-and-branch all-coaxial CATV network architecture. Figure 1(b) shows
the upgrading to an optical trunk ring (HFC). The dowstream traffic flows from the headend and is directly
injected into the trunk fiber (originally coaxial) ring. The optical section is terminated with a Hub, or node,
which converts optical signals to electrical. Feeder coaxial cable distribution is arranged from the node in a
tree-and-branch topology where traffic is split at branching points to up to 2000 houses. Propagation through
the cable attenuates the signal by an amount proportional to the square root of frequency, resulting in higher
losses for higher frequencies, which is mitigated with amplifiers and equalizers, especially in the UHF bands.
The HFC access network represents the natural evolution of the existing mature tree-and-branch CATV
networks for introducing the new digital broadcast or interactive services. An HFC network provides transparent communication channels between the subscribers and the CATV hubs or headends, since no signal
processing occurs between the subscriber and the hub, besides optical-to-electrical conversion and eventual
frequency translation and amplification. The hub provides the interface between the core network and the
HFC distribution network. At this level, communication equipment is used to transmit an analog frequency
multiplex comprising both the TV broadcast signals (analog or digital) and the modulated digital interactive
ones. Each node generally supplies up to 1000 subscribers (2,3).
CABLE TELEVISION
Fig. 2. HFC CATV headend elements and functionalities.
Figure 2 shows a scheme of the headend, comprising a number of possible elements and functionalities
of a two-way digital CATV HFC network. Analog or digital broadcast TV is not only the content source of the
cable system, which can also supply video and data with local or remote (Internet) servers. Telephone services
are also contemplated through connection to the PSTN.
Figure 3 shows the elements at user premises. Different modems are necessary for video and data
applications, and also an MPEG-2-compliant set-top decoder is needed to transform the digital transport
stream for the analog TV set to be able to present the image.
Catv Bandwidth And Capacity
Figure 4 shows an example of CATV bandwidth allocation. This spectrum is an expansion of the standard
analog video broadcast spectrum with space reserved for digital video services and an upstream spectrum for
interactive services. Current systems allocate a band of 30 MHz to 65 MHz for upstream communication; since
the total bandwidth may be up to 860 MHz, the allocation is clearly asymmetric. This arrangement is due to
the allocation of analog TV legacy channels from 55 MHz up to 300 MHz to 400 MHz. Bandwidth available for
interactive communications lies below 50 MHz and above 550 MHz.
The downstream frequency band is divided in two parts. The wider one is devoted to the broadcast services,
among which are analog and digital TV, pay TV, Near Video on Demand (NVOD), FM radio, digital radio, and
broadcast data. A small number of these downstream channels will be reserved for the interactive services.
CABLE TELEVISION
Fig. 3. CATV elements for data and video applications at user premises.
Fig. 4. Example of CATV frequency allocation.
Each channel carries a potential payload of 30 to 40 Mbit/s using a complex multilevel modulation scheme
(64-QAM typical).
Assuming downstream channels of 6 MHz with an effective bandwidth of 5.4 MHz and channels from
550 MHz to 750 MHz, the total capacity is approximately of 1 Gbit/s, which should be sufficient at least until
high-definition TV (HDTV) arrives. For the upstream capacity let us assume also 6 MHz channels with QPSK
modulation, as is being proposed by standardization groups, and a total bandwidth of 30 MHz (more or less
as shown in Figure 4). With these assumptions the upstream capacity becomes 10 Mbit/s per channel and 60
Mbit/s in total. Taking into account up to 20% of overhead for upper-layer protocols, the upstream capacity
may be reduced to 5% of downstream capacity. This worst-case capacity may be enough, though, since peak
CABLE TELEVISION
hours can be assumed to be below 25% of the day. However, voice applications, for example, need real-time
delivery, taking priority over available TCP/IP connections. Bandwidth assignment and bit-rate management
by differentiating best-effort traffic from traffic with guaranteed bandwidth are QoS problems to be solved by
cable networks. Interested readers may refer to Refs. 4 and 5.
Upstream Transmission
CATV networks were designed for video broadcasting, but cable operators soon realized the usefulness of
a return path for movie selection, pay-per-view, billing information, monitoring, and so on. One-way cable
networks had an inherent capability for upstream transmission, since raw wire can transport signals in both
directions simultaneously. However, amplifiers boost signals only in one direction, blocking the return path.
Thus, upgrading to the two-way network architecture must include two-way amplifiers, laser transmitters to
the headend, and additional equipment at the user premises. It should be noted that downstream video may
be broadcast or on demand, but the reverse path is a shared medium, which means access must be controlled
via some medium access (MAC) protocol (6).
An important issue for the reverse path is ingress noise degrading the quality of service. Home electrical
devices (hair dryers, vacuum cleaners, etc.) create noise bursts within the 5 MHz to 40 MHz frequency range
(within the upstream bandwidth; see Fig. 4), which are propagated back to the network through the amplifiers.
Amateur radio and AM radio are also potential sources of interference in the return path. A number of
techniques are used to mitigate the interference and noise present in the upstream spectrum:
Low-pass filtering at the cable drop in the data terminal in the home
Modulation techniques such as frequency hopping and spread-spectrum cable modems
Standards
The rapid evolution of broadband service has resulted in cable television operators, providers of telephone
service, and other service providers seeking to provide voice, data, and video services to their residential
and commercial subscribers over existing and new infrastructures. In this context, standardization is needed,
mainly to make possible retail distribution of cable network elements in the home for both video and data
applications. The standardization of technological features of these elements provides a means to reduce costs
by creating a competitive market. Standardization activities cover not only the physical layer (PHY), but also
the MAC layer, security, and service management. Organizations involved in cable TV standardization are the
following (1):
Digital Audio Visual Council (DAVIC)
Digital Video Broadcasters Project (DVB)
IEEE 802.14
International Telecommunications Union (ITU)
Multimedia Cable Network System (MCNS) Partners Ltd.
Society of Cable Television Engineers (SCTE)
Society of Motion Picture and Television Engineers (SMPTE)
Video Electronics Standards Association (VESA)
Standards from some of these organizations are summarized in the following.
CABLE TELEVISION
Davic And Dvb. DAVIC was started by Leonardo Charliglione, the Motion Picture Experts Group
(MPEG) committee head. It was established in 1994 as a nonprofit association based in Switzerland, and it was
constituted with companies involved in all sectors of the audiovisual industry (manufacturing and service),
as well as a number of government agencies and research organizations from more than 25 countries. The
association was disbanded after 5 years of activity, remaining active only through its Web site (www.davic.org).
DAVIC specifications were primarily aimed at promoting the success of interactive digital audiovisual applications and services, and therefore they are based on DVB specifications. Since DAVIC aims at interoperability
across applications, it does not specify systems but components (tools) that are non-system-specific in that they
have to be usable by different industries in different systems and still guarantee interoperability.
The DVB Project is a consortium of companies from more than 30 countries worldwide in the fields of
broadcasting, manufacturing, network operation, and regulatory matters. The DVB Project Office is based in
Geneva, Switzerland. It aims at establishing common international standards for the migration from analog to
digital broadcasting through the design of a global standard for the delivery of digital television (www.dvb.org).
Their standards are based on the common MPEG-2 coding. Through the use of MPEG-2 stream packets, DVB
is potentially able to deliver TV service [from HDTV to multiple-channel standard-definition TV (PAL/NTSC
or SECAM)], broadband multimedia data, and interactive services effortlessly from one medium to another. In
particular, DVB signals can move from satellite to cable and from cable to terrestrial. DVB standards are published by the European Telecommunications Standards Institute (ETSI). ETSI, the Centre for Electrotechnical
Standards (CENELEC), and the European Broadcasting Union (EBU) have formed a joint technical committee
(JTC) to handle the DVB family of standards, which can be grouped as follows with regard to transmission
standards:
DVB-C (cable systems)
DVB-DSNG (digital satellite news-gathering systems)
DVB-MC [multipoint video distribution systems (MVDSs) below 10 GHz]
DVB-MS (MVDSs at 10 GHz and above)
DVB-S (framing structure, channel coding, and modulation for 12 GHz satellite services)
DVB-SFN (megaframe for single-frequency network)
DVB-SMATV (DVB interaction channel for satellite master antenna TV distribution systems)
DVB-T (terrestrial systems)
The original specification is DAVIC 1.1, ratified by 1996, covering different generic elements as follows:
High- and mid-layer Protocols
DAVIC subsystems (service provider, delivery system, and service consumer)
System-wide issues (DAVIC functionalities, system reference models and scenarios, reference points, interfaces, and dynamics)
Specifications 1.2 to 1.4 were issued subsequently. The last specification, 1.5, dealt with specific technical
issues:
Jitter concealment tools
Applicability of DAVIC 1.5 intranet architecture to TV-anywhere and TV-anytime scenarios
DAVIC cable modem
DAVIC intranet technical platform specification
TV anytime and TV anywhere
CABLE TELEVISION
DVB has submitted to ETSI a complete set of standards defining issues related to the provision of
interactive services on cable and LMDS networks:
ETS 300 802 defines network-independent layers for both media.
ETS 300 800 defines the network-dependent layers for cable networks.
ETS 300 429 defines the downstream physical layer for cable networks.
The DVB RCCL [return channels for cable and local multipoint distribution systems (LMDSs)] includes the
specifications related to network-dependent layers and has been developed in close consultation with DAVIC.
The result is the adoption by DAVIC of ETS 300 800 as the only standard for cable modem applications. This
is a major step towards common specifications between DVB and DAVIC.
Working in this direction, the DVB/DAVIC Interoperability Consortium provides multivendor interoperability based on the open international standards DVB RCCL (ETSI ETS 300 800) and DAVIC 1.5 for broadband
communication (www.dvb-davic.org). The Consortium intends to build a multiservice platform enabling a single system to deliver video, audio, data, and voice to residential devices such as cable modems, set-top boxes,
multimedia home platforms, and residential gateways. The consortium of European multisystem operators
(MSOs) has recognized it as the preferred technology, so that the DVB-RCCL/DAVIC specification represents
a technology that may become a standard for international deployment.
The following vendors have announced their intention to comply with the specification: Alcatel, Cisco,
DiviCom, Hughes Network Systems, Nokia Multimedia Network Terminals, Sagem, The Industree, Thomson
Broadcast Systems (a subsidiary of Thomson Multimedia), and Thomson Multimedia. It is apparent that while
European MSOs have committed to DVB-RCCL/DAVIC, other big providers have already chosen to deploy
DOCSIS (data-over-cable service interface specification) technology. DVB-RCC is available as ETSI standard
ETS 300 800, Digital Video Broadcasting (DVB); Interaction Channel for Cable TV Distribution Systems
(CATV). DVB-C is available as ETSI standard ETS 300 429, Digital Video Broadcasting (DVB); Framing
Structure, Channel Coding and Modulation for Cable Systems.
Key physical-layer features of specifications for HFC networks are the following:
Simple Network Management Protocol (SNMP), network management
Downstream:
Passband unidirectional PHY on coax tool (DAVIC 1.3)
QAM modulation [called in-band signaling (IB)].
MPEG and Asynchronous Transfer Mode (ATM), framing
Bit rates up to 56 Mbit/s (with 8 MHz carriers and 256-QAM modulation)
Upstream:
Passband bidirectional PHY on coax tool (DAVIC 1.3)
QPSK modulation [called out-of-band signaling (OOB)]
ATM framing
Bit rate up to 3 Mbit/s
Support for a variety of MAC techniques: time-division multiple access (TDMA), contention, and reservation
ATM signaling either with UNI (ITU-T Q.2931) or by proxy using DSMCC (digital storage media
configuration and control, ISO/IEC 13818-6)
CABLE TELEVISION
Fig. 5. DOCSIS architecture model. CM: cable modem; CMTS: cable modem termination system.
Docsis And Mcns. MCNS is a consortium of North American entities that came together to develop
the specification DOCSIS for transmitting data over a cable network (www.cablemodem.com). DOCSIS 1.0 was
proposed in March 1997. In March 1998 the ITU accepted DOCSIS as a cable modem standard (J.112). To
deliver DOCSIS over a cable television (CAT) network, one 6 MHz RF channel in the 50 MHz to 750 MHz
range is typically allocated for downstream traffic to homes, and another channel in the 5 MHz to 42 MHz
band is used to carry upstream signals. Cable modems were described as external devices that connect to a
personal computer through a standard 10Base-T (10 Mbit/s Ethernet over CAT3-5 cabling) Ethernet card or
USB interface. CableLabs manages a certification process to ensure DOCSIS cable modems manufactured by
different vendors comply with the standard and are interoperable.
CableLabs is an organization dedicated to the certification of DOCSIS-based equipment. In April 1999 it
issued a second-generation specification called DOCSIS 1.1, which adds key enhancements to the original standard, such as improved QoS and hardware-based packet-fragmentation capabilities, to support IP telephony
and other constant-bit-rate services. The next-generation standard is designed to be backward compatible, enabling DOCSIS 1.0 and 1.1 modems to operate in the same spectrum on the same network. A third-generation
DOCSIS standard will add an advanced PHY to the core specification to increase the upstream transmission
capacity and reliability by using FA-TDMA (frequency-agile TDMA) and S-CDMA (synchronous code division
multiple access).
Vendors supporting the DOCSIS standard are 3Com, Bay Networks, Cisco Systems, General Instruments,
Hewlett-Packard, Hybrid Networks, Intel, Motorola, NEC, Panasonic, Scientific Atlanta, Sharp Electronics,
Toshiba, US Robotics, and Zenith Electronics.
The DOCSIS architecture model is shown in Figure 5. The PC generates Internet protocol (IP) over
Ethernet packets. The cable modem (CM) acts as a bridge and forwards the Ethernet frames to the network.
DOCSIS specifies a new MAC layer from the CM to the cable modem termination system (CMTS) in the
upstream direction. The Ethernet frame is encapsulated by the CM in a DOCSIS MAC frame and sent to the
CMTS. The CMTS (a router or a bridge) de-encapsulates the Ethernet frame and forwards it upstream.
10
CABLE TELEVISION
The relationship between a CM and its CMTS is a master-slave relationship. The CMTS controls the
bandwidth allocation on the upstream channel. The CMTS sends on the downstream channel bandwidth
allocation messages called upstream bandwidth allocation maps (referred to as MAPs), which define how the
time units (mini-slots) on the upstream channel must be used. The cable modem boot process used by the CM
influences architecture design, and DOCSIS concludes with an explanation of how the CM boots up.
Key PHY features of specifications for HFC networks are the following:
Upstream:
QPSK or 16-QAM modulation with no interleaving
Symbol rates 160, 320, 640, 1280, and 2560 ksymbol/s.
Downstream:
256- and 64-QAM modulation with variable interleaving
MPEG-2 Transport Stream (TS), framing
Variable packet length
Grant-based bandwith assignment
DOCSIS standards have not yet been accepted in Europe and Asia. The alternative standard DAVIC/DVB
for cable modem systems, also called the EuroModem standard, has been formalized, and, as was stated before,
a significant number of European cable system operators have embraced it.
IEEE 802.14. The IEEE 802.14 Working Group is a committee of engineers representing the vendor
community that has developed a specification for data-over-cable networking. The group was formed in 1994
and intended to develop a specification that would be recognized as an international standard. However, MCNS
defined an international specification first. Despite ITU recognition of DOCSIS as an international standard in
1999, the IEEE 802.14 Working Group continued to work on its specification, and MCNS indicated that it would
implement IEEE 802.14s advanced PHY specification. Overall the future of the IEEE 802.14 specification
is unknown. The group had good intentions and its specification was undoubtedly a better technological
alternative than that developed by MCNS. However, timing is critical in developing standards.
At the current time, performance characteristics of the standardized DOCSIS and the DVB/DAVIC CM
specifications have not been well assessed. For both specifications an OPNET model has been created by
CableLabs and MIL3 for the former, and by EuroCableLabs Centre of Competence for the latter. However, both
models support only the features of the upstream channel. Current research aims at the development of a new
simulation model for the downstream channels for both the DOCSIS and the DVB/DAVIC protocols using the
OPNET simulation package. Its performance should be based on the system throughput versus offered load
and the end-to-end delay versus system throughput with respect to:
(1) Number of CMs in the CATV network
(2) Variable offered load
(3) Different scheduling algorithms.
Technical Differences Between Standards
As we described before, DVB/DAVIC and DOCSIS/MCNS are two groups of standards defined by different
organizations currently focused on delivering data to set-top boxes and PC cable modems, respectively. Due to
market dynamics, while both standards look to deliver data to broadband services, the technical evolution has
CABLE TELEVISION
11
been different, though it is expected that these two courses will probably converge in some way. Table 1 shows
the technical differences between the standards. Recently OpenCable has started to design a retail-model
set-top box using the DOCSIS standard for data, but also using the DAVIC standard. This box has DAVIC as
well as DOCSIS components: DAVIC is chosen as a core requirement, and DOCSIS as an extension for the
specification.
Applications
Two-way CATV HFC applications should be grouped into two major markets, in part because such applications are driving different (not necessarily divergent in principle) paths in the technology upgrading and
standardization:
Digital video over cable TV networks (although in the transition from analog to digital broadcasting, cable
systems will also deliver analog channels)
Data services over cable TV networks.
In fact the two markets will converge, since multimedia applications relate to both. The road to such
a confluence is being paved now through the standardization process, and the success achieved by each of
the parts will depend on both cost and technological accomplishments. In either case digital video or data
application services are to be delivered both in broadcast and in interactive mode.
12
CABLE TELEVISION
Digital Video over Cable TV Networks. First it is interesting to point out differences and key
technological and performance aspects of analog versus digital video. Unlike analog video, which degrades in
quality, digital video is virtually identical to the parent. There is a major difference between the way computers
and television handle video. When a television set displays its analog video signal, it displays the odd lines (the
odd field) first and the even lines (the even field) next. Each pair forms a frame, and there are 60 of these fields
displayed every second (30 frames every second). This is referred to as interlaced video. Instead, the computer
displays each line in sequence, from top to bottom. This entire frame is displayed 30 times every second. This
is often referred to as noninterlaced video.
Analog formats include NTSC (National Technical Standards Committee), used in the United States,
Mexico, Japan and Canada. Phase alternation line (PAL) is used in western Europe, and Sequential Couleur a`
Memoire (SECAM) is used in France, Russia, and Eastern Europe (Hewlett Packard, p. 4).
Digital formats include Joint Photographic Experts Group (JPEG), Motion JPEG (MJPEG), MPEG, Advanced Video Interleave (AVI) from Microsoft, Indeo from Intel, and CellB for use on Sun SPARCstations. Of
these digital formats, only MPEG is an internationally recognized compression standard.
Benefits of digital video are in random access to stored information, the ability to compress the video,
and the low cost of reproduction without loss of quality. From a transmission point of view, a number of digital
channels can be sent where only one analog channel was allocated. VCR video is in analog format running
at 30 frames/s. To achieve the same quality on the desktop, without compromising on frame rate, resolution,
color resolution, or image quality, one needs hundreds of gigabytes of disk space. Twenty-five centimeters (10
inches) of videotape is required to record one second of video. Compression is the solution, and there are two
types:
Intraframe compression compresses each individual video frame. JPEG and MJPEG use intraframe for
compression ratios of 20 : 1 to 40 : 1. The larger the ratio, the poorer the quality.
Interframe compression only looks at the video data that have changed. This also reduces the bandwidth
necessary to carry the video stream. MPEG uses interframe compression and can achieve up to 200 : 1
compression rates. Roughly, 9 Mbyte of hard-disk space is required for every minute of MPEG compressed
video. CATV and DBS use MPEG-2.
TV broadcasting and video conferencing, which allows a group of users to communicate in real time
through the use of live streaming audio and video, are the main applications of digital video.
Data Services over Cable TV Networks. Data services can be divided into narrowband and broadband applications according to the amount of bandwidth required. Telephony is narrowband, and multimedia
contents are broadband.
Telephony. We may include voice services within data interactive applications. Telephony can be offered
by cable by using IP to deliver voice through the use of a gateway located at the cable headend. Telephony
networks can either be private or allow access across the Internet or to and from the PSTN. Videotelephony is
a natural successor to telephony, although more complex. There is a tradeoff between quality, bandwidth, and
real-time.
PUSH Applications. PUSH technology is the delivery of information from server to client over IP.
Webcasting and multicasting (transmission of files or streaming audio and video to preselected multiple users)
are PUSH applications.
Virtual Private Networks. Since a cable system acts as a broadband network, operators are capable of
establishing virtual private networks (VPNs) as completely closed environments (intranets) allowing users to
access the Internet.
CABLE TELEVISION
13
Home Monitoring and Security. Cable networks can be used to monitor homes for fires, break-ins, and
medical emergencies. The available bandwidth of a cable network, combined with the systems being always
on, enables emergency information to be quickly transmitted.
Telemedicine. Medical doctors, patients, hospitals, clinics, and mobile units can use the CATV broadband
technology for quick transfer of huge files to provide diagnosis, treatment, consulting, and education.
Immersive Environments. An immersive environment is a virtual social environment, managed by a
computer program, where the user is represented within the program as an animated character (avatar)
visible to all of the users. This is clearly a broadband multimedia application, where sound, data, and video
come together, requiring a great amount of bandwidth along with critical constraints on latency.
One goal for a media immersion environment (MIE) is for people to interact, communicate, collaborate,
and entertain themselves naturally in a shared virtual space while they reside in distant physical locations.
The MIE has applications in many domains where physical presence is expensive (e.g., distance learning);
impossible (e.g., space exploration), unsafe (e.g., nuclear studies), or inconvenient (e.g., entertainment), or
where more than one person must be involved (e.g., remote medicine). The principal function of MIEs is to
synthesize multimodal perceptions that do not exist in the current physical environment, thus immersing users
in a seamless blend of visual, aural, and (potentially) haptic information.
Alternative Technologies
A CATV HFC is called an access network; it may also be referred to as a last-mile solution or (for solutions
proposed by telephone companies) a local loop solution. An access network connects customers premises to
the network termination and performs operations interfacing with the transport network, content provider,
and home network elements. Figure 6 shows graphically the technological frame for the access network. All
technologies share a common element, which is the network termination (NT), or network interface, in the
home. The main functions of these networks are the following:
Connection to the core transport network by switching, routing, and multiplexing
Classification of user traffic by QoS
Security procedures and handling of packet encapsulation
Registration of hardware and updating of software in the household equipment
Measurements for billing
A number of alternative cable and wireless competitive technologies are becoming feasible for the access
network, so that major CATV providers have had to accelerate standardization of their products.
Competitive wire technologies are the following:
xDSL (asymmetric, symmetric, high- and very high-bit-rate digital subscriber line)
FTTx (fiber to the neighborhood, curb, building, etc.)
ISDN (Integrated Service Digital Network).
Competitive wireless technologies are the following:
MMDS (multichannel multipoint distribution system)
LMDS (local multipoint distribution system)
Satellite [deosynchronous earth orbit (GEO) and low earth orbit (LEO)]
In the following a brief description is given of each of these networks.
14
CABLE TELEVISION
Fig. 6. Alternative technologies for access networks: wired and wireless.
xDSL (Digital Subscriber Line). xDSL refers to a series of networking technologies, comprising ADSL,
HSDL, VDSL, and SDSL, that are capable of supporting high data rates over the existing telephone network.
The technology bases its competitiveness on the existing extensive infrastructure and heavy capitalization
associated with telcos.
Standard telecom modems establish a data stream between two arbitrary points using the entire telecom
systemthat is, from the senders local loop, through the telephone switching system (mostly digital switches
now), and then to the receivers local loop. Standard modem connections can span continents, with one end
thousands of kilometers from the other end. DSL modems, on the other hand, establish a connection from
one end of a copper wire to the other end of that copper wire: the signal does not pass into the telephone
switching system. Consequently, DSL modems are not limited to using the voice frequencies passed by the
standard telephone system (typically 0 to 4 kHz); DSL modems typically use more than 100 kHz. To reiterate,
one end of the DSL link will be at the consumer site, the other end must be at the other end of the copper cable,
usually at the local telephone exchange, where data and voice are split. The voice frequencies are wired into a
traditional plain ordinary telephone service (POTS) switch and enter the usual telephone switching network.
The data frequencies are wired into a corresponding DSL modem, and the resulting high-speed digital data
stream coming from (or going to) the consumer is then handled as ordinary data (not analog voice) and may be
hooked into any number of networking technologies for further connection to the datas destination. Thus, the
data never enter the standard telephone switching system. Typically the data will be routed over a local-area
network (LAN) or wide-area network (WAN) connection (10Base-T Ethernet, T1, T3, ATM, frame relay) to a
business office.
Over the next five years, xDSL (and particularly ADSL) is considered to be the greatest threat to the cable
modem industry.
FTTx (Fiber to the Neighborhood, Curb, Building, Etc.). The growing demand for interactivity and
more bandwidth per subscriber is being satisfied by pushing fiber closer to the home and by the availability
of the required electro-optic components. FTTx refers to a series of networking technologies that run optical
CABLE TELEVISION
15
fiber from the central office to a users neighborhood (FTTN), curb (FTTC), building (FTTB), home (FTTH), etc.
Though the optical fiber is relatively inexpensive, optical transmitters and receivers are very costly. However,
continued growth of broadband services is fueled by the clear advantages that optical fiber systems offer in
cost, reliability, and performance for broadcast networks.
FTTH is not a practical solution for the moment to deliver residential data services. The present idea is
to replace long copper lines with fiber optic lines (not only in the telephone network, but also in CATV). The
major drawback is the cost of the replacement of the copper-based infrastructure. A critical point is what part
of the network will be replaced with fiber optic lines: that is what distinguishes FTTH, FTTB, fiber to the office,
FTTN, fiber-to-the-street, FTTC, etc. On the other hand, users of a CATV network again use copper lines, but
all the rest of the infrastructure is fiber-optics-based. With the exception of FTTH, all the above approaches
use some form of high-speed metallic access technology for service delivery to the customers premises.
ISDN (Integrated Service Digital Network). ISDN is a service provided by local telephone companies
that modifies regular telephone lines so that they can transmit data almost five times as fast as the fastest
analog modems currently available. In addition to the significant increase in transmission speed, ISDN also
allows the transmission of not only data, but a combination of data, voice, and video simultaneously on one
line. ISDN provides higher speeds than POTS by allowing data to be transferred digitally from end to end.
In contrast, POTS converts the digital data to analog within the local loop that extends into the home or
office, significantly reducing transmission speed. An ISDN line can carry up to 128 kbit/s of data. Converting
to an ISDN-compatible configuration for the consumer only requires an additional piece or two of relatively
inexpensive hardware. Telecommuting to an office or base requires additional equipment at the base. While
ISDN usage costs are slightly higher than those for analog telephone lines, users enjoy more than commensurate
benefits.
Though ISDN has a share of the high-speed connectivity market, the technology is limited to 128 kbit/s
and thus is in a different class than technologies that can support megabytes per second. In particular, it is
very limited for video applications; it can only be used for videoconferencing with 6 to 8 frames per second.
Satellites. There are two basic types of satellite systems being proposed: GEO and LEO.
GEOs orbit in the Clarke belt, approximately 35,000 km (22,000 miles) above the equator. With this
orbit, the satellite can stay over the same area of the earth for an indefinite period of time. Each GEO serves
one geographic area, and can theoretically cover about 41% of the earths surface. Companies proposing GEO
systems are planning on using between three and fifteen satellites to deliver worldwide service. The primary
advantage of GEO systems is that they are a proven technology. Most current communications satellites are
GEOs. A GEO system is also far less expensive than an LEO system, and also GEO ground stations can be
relatively simple because they need only target a fixed point relative to the earth. The main drawback for
GEO systems is called the latency factor. In order to obtain information from an Internet server, a signal has
to travel 35,000 km to the satellite, then 35,000 km back to the earth. This round trip takes approximately
one-quarter second.
LEOs orbit 20 times closer to the earth, between 700 km (450 miles) and 1350 km (700 miles) above the
earths surface. Each LEO is moving relative to the earth, covering a particular area for only a few seconds.
Because of this, a network of many satellites is required to cover the world.
Teledesic plans to launch a large number of LEOs that will be capable of offering high-speed Internet
access anywhere in the world. Service providers will include Teledesic, Globalstar, OrbComm, and SkyBridge
and M-Star (backed by Motorola). M-Star, while a broadband LEO system, is not aimed at the consumer market
like Teledesic; it is planned to offer high-bandwidth intercontinental links between network providers rather
than end users.
Satellites require a dedicated piece of the spectrum. Currently, the ITU has allocated 2.5 GHz of spectrum
for fixed satellite services in the 28 GHz Ka band. There are fourteen satellite applicants vying for pieces of
that 2.5 GHz. Requests for single applicants range from 750 MHz to the full 2.5 GHz, with most applicants
requesting 1 GHz of spectrum.
16
CABLE TELEVISION
The reason the Ka band has not been used in the past is that such high-frequency transmissions are
easily blocked. Buildings, trees, and other solid objects can cause a loss of signal. This makes these frequencies
unsuitable for use by ground-based systems, because they would require a large number of transmitters to
be able to avoid all obstacles. Satellites avoid many of the problems associated with blocking because their
signals come from directly overhead. Buildings and trees do not present an impediment to signals coming from
overhead satellites.
Power-Line Area Networks (Access via Utility Power Grids). There have been proposals lately
for traditional power companies to provide high-speed access via their existing grids. The biggest obstacle
to this technology is that data are scrambled when they pass through transformers. This obstacle is slowly
being overcome as several companies continue to work towards a solution. Nortel Networks has successfully
tested networks in Europe and Asia, where the transformer-to-customer ratio is 1/300. They have successfully
bypassed the transformers, but their speed has been limited, being comparable with cable modems and xDSL.
Data are transferred through the actual power wiring.
NIU (Network Interface Unit). NIU refers to high-speed connectivity through hybrid fiber-coax or
FTTH or FTTC networks using a network interface unit at the customer premises rather than an external or
PC-installed cable modem. Systems using NIUs usually support both data and telephony. Since the technology
can be utilized by cable operators, it is not actually a competitor to the cable industry but to cable modems.
Multipoint Multichannel Distribution Service, and Local Multipoint Distribution Systems.
Multipoint multichannel distribution service (MMDS), also referred to as wireless cable, delivers broadband
services to subscribers through microwave transmitting and receiving antennas. The channels allocated to
MMDS are generally used to provide a multichannel video programming service that is similar to cable television, but, rather than being hard-wired, MMDS uses microwave frequencies.
MMDS is a wireless technology for access networks that operates at frequencies 200 MHz to 2700 MHz.
Its range may be 50 km to 60 km, and its main advantage is invulnerability to weather conditions. Operators
that offer wireless video service can also offer data services. MMDS operators can offer telco return service
and, with the emergence of LMDS (see below), will be capable of supporting bidirectional services.
The local multipoint distribution system (LMDS) is a broadband wireless technology used to deliver
voice, data, Internet, and video services in the at frequencies of 25 GHz and higher (depending on the license).
Spectrum in the millimeter band has been allocated for LMDSs to deliver broadband services in a point-to-point
or point-to-multipoint configuration. Due to the propagation characteristics of signals in this frequency range,
LMDSs use a cellularlike network architecture (normally the cells are large and a big city can be fully covered
with four or five cells). The services provided are fixed (not mobile) and are seriously impaired by adverse
weather.
Cable Modem versus Set-Top Box. Numerous companies are working towards high-quality, fullscreen, real-time delivery of video programming that can be delivered via a cable modem or broadband Internet
connection. As the cable television networks transition from a broadcast-only network to a high-bandwidth
two-way network, the importance and functionality of the set-top box increases. Currently, mixed cable modem
and set-top box network architectures and services exist. Developments in the set-top box industry have been
restrained by the control the cable companies have over their equipment. Currently, a set-top box includes
closed, proprietary technology, which prohibits its use on other cable systems networks. Technology is evolving
in both directions, from the cable modem to the set-top box and vice versa. Current technology development
can be summarized as follows:
Cable Modem Technology.
Cable modem from DOCSIS, IEEE 802.14, IETF, DVB, and ATMF
EuroModem from DVB-RCC and ETS 300800
CABLE TELEVISION
17
EuroDOCSIS from DOCSIS (DOCSIS cable modem with some DVB technical compliance)
Set-Top Box Technology.
EuroBox
A number of cable operators from across Europe have developed the Eurobox Platform. This concept is
based on a common set-top box and a common application program interface API and conditional access
method. The box specification is used as a reference model by the cable industry and manufacturers.
Viaccess has been selected as the conditional access system, and Open TV as the API. The Eurobox
Platform has been successfully implemented, for example, in France, Sweden, and Denmark. However,
some cable operators, notably in the UK, do not appear to be following this platform in its entirety.
OpenCable
This initiative from CableLabs includes guidelines for building advanced set-top boxes including
feature enrichments to support broadband applications.
Multimedia Home Platform
MHP includes set-top boxes, integrated TV receivers, in-home digital networks, personal computers,
network computers, and so on. The first specification for MHP, covering home access networks (HANs)
with an active NT and based on an ATM interface operating at 25 Mbit/s or 51 Mbit/s, was approved by
DVBEBU JTC and published by ETSI as TS 101 224. The MHP API consists of a software specification
that will be implemented in set-top boxes, integrated digital TV receivers, and multimedia PCs. The
MHP will connect the worlds of broadcast television, Internet computing, and telecommunications
through these devices and their associated peripherals.
BIBLIOGRAPHY
1. G. Abe Residential Broadband, 2nd ed., Indianapolis, IN: Cisco Press, 2000.
2. K. Maxwell K. Maxwell Residential Broadband: An Insiders Guide to the Battle for the Last Mile, New York, Wiley:
1998.
3. B. C. Lindberg Digital Broadband Networks and Services, New York, McGraw-Hill: 1995.
4. C. A. Eldering N. Himayat F. M. Gardner CATV return path characterization for reliable communications, IEEE Commun. Mag., 33 (8): 6269 August 1995.
5. R. Rabbat K. Y. Siu QoS support for integrated services over CATV, IEEE Commun. Mag., 37 (1): 6468, January 1999.
Available www:http://www.comsoc.org/pubs/free/private/1999/jan/Rabbat.html
6. J. Karaoguz J. Yu V. Eyuboglu Comparison of single-carrier, multi-carrier and spread spectrum modulations for upstream
PHY layer in HFC CATV networks, IEEE 802.14a/98-018, San Diego, July 1998.
18
CABLE TELEVISION
INTERNET RESOURCES
Broadband Bob: http://catv.org
CableLAbs, research organization for US cable operators: http://www.cablelabs.com
CATV CyberLab: http://www.catv.org
DAVIC: http://www.davic.org
DVB/DAVIC Interoperability Consortium: http://www.dvb.org
DVB Web site: http://www.dvb.org
DOCSIS specifications: http://www.cablemodem.com
ECCA Web site: http://www.ecca.be
ETSI: http://www.etsi.org
FCC cable: www.fcc.gov
ITU: http://www.itu.org
MARYAN VAZQUEZ
-CASTRO
Carlos III University of Madrid
CHIRALITY
INTRODUCTION
Chiral media have the ability to discriminate between
left-handed and right-handed electromagnetic (EM) elds.
These media can be classied into two types: (1) isotropic
chiral media and (2) structurally chiral media. The
molecules of a naturally occurring isotropic chiral medium
are handed, while an articial isotropic chiral medium can
be made by randomly dispersing electrically small, handed
inclusions (such as springs) in an isotropic achiral host
medium. The molecules of a structurally chiral medium,
such as a chiral nematic liquid crystal, are randomly positioned but have helicoidal orientational order. Structurally
chiral media can also be articially fabricated either as
stacks of uniaxial laminae or using thin-lm technology.
Whereas considerable theoretical and experimental work
on isotropic chiral media has been reported at microwave
frequencies during the 1980s and the 1990s, microwave
research on structurally chiral media remains in an embryonic stage at the time of this writing (1). Therefore, the
major part of this article is devoted to isotropic chiral media.
NATURAL OPTICAL ACTIVITY
Ordinary sunlight is split into its spectral components by
a prism. A spectral component is monochromatic (i.e., it
has one and only one wavelength 0 in vacuum). The wavelength 0 of one of the visible spectral components lies anywhere between 400 nm (violet) and 700 nm (red). A spectral
component can be almost isolated from other spectral components by carefully passing sunlight through a series of
lters. Although ltering yields quasimonochromatic light,
many experiments have been and continue to be performed
and their results analyzed, assuming that the ltered light
is monochromatic.
Light is an EM wave with spectral components to which
our retinal pigments happen to be sensitive, and the consequent images, in turn, happen to be decipherable in our
brains. All optical phenomena can be generalized to other
electromagnetic spectral regimes.
Suppose that a monochromatic EM wave is propagating in a straight line in air, which is synonymous with vacuum (or free space) for our present purpose. Its electric
eld vector vibrates in some direction to which the propagation direction is perpendicular; the frequency of vibration is f=c/0 , where c=3 108 m/s is the speed of light in
vacuum. Its magnetic eld vector also vibrates with the
same frequency, but is always aligned perpendicular to the
electric eld vector as well as to the propagation direction.
Suppose that we x our attention on a certain plane that is
transverse to the propagation direction. On this plane, the
locus of the tip of the electric eld vector is the so-called
vibration ellipse, which is of the same shape as the locus
of the tip of the magnetic eld vector. A vibration ellipse is
shown in Fig. 1. Its shape is characterized by a tilt angle
as well as an axial ratio; in addition, it can be left-handed
Figure 1. The tip of the electric eld vector of a plane-polarized
monochromatic electromagnetic wave traces the socalled vibration ellipse in a plane transverse to the propagation direction.
if the tip of the electric eld vector rotates counterclockwise, or right-handed if otherwise. Similarly, an EM wave
is said to be elliptically polarized, in general; however, the
vibration ellipse can occasionally degenerate into a circle
(circular polarization) or even a straight line (linear polarization).
The shape of the vibration ellipse of monochromatic
light is altered after traversal through a certain thickness
of a so-called optically active medium. This phenomenon,
known as optical activity, was discovered around 1811
by F. Arago while experimenting with quartz. Crystals
are generally anisotropic, but J.-B. Biot observed around
1817 the optical activity of turpentine vapor, denitely an
isotropic medium. Isotropic organic substances were believed to have exclusively biological provenances, and in
1860 L. Pasteur argued that turpentine vapor exhibited
natural optical activity, but the optical activity of crystals
could not be similarly qualied. Pasteur was unduly restrictive. Isotropic optically active media, of biological or
other origin, are nowadays called isotropic chiral media,
because EM elds excited in them necessarily possess a
property called handedness (Greek cheir=hand). Facsimile
reproductions of several early papers are available (2).
CHIRAL MEDIA: NATURAL AND ARTIFICIAL
The molecules of an isotropic chiral medium are mirror
asymmetric (i.e., they are noncongruent with their mirror images). A chiral molecule and its mirror image are
called enantiomers (3). As examples, the two enantiomers
of 2-butanol are shown in Fig. 2. Enantiomers can have
different properties, although they contain identical atoms
in identical numbers. One enantiomer of the chiral compound thalidomide may be used to cure morning sickness,
during pregnancy, but its mirror image induces fetal malformation. Aspartame, a common articial sweetener, is
one of the four enantiomers of a dipeptide derivative. Of
these four, one (i.e., aspartame) is sweet, another is bitter, while the remaining two are tasteless. Of the approximately 1850 natural, semisynthetic, and synthetic drugs
marketed these days, no less than 1045 can exist as two or
more enantiomers; but only 570 were being marketed in the
J. Webster (ed.), Wiley Encyclopedia of Electrical and Electronics Engineering. Copyright 2007 John Wiley & Sons, Inc.
Chirality
2. P. Drude showed that chiral molecules can be modeled as spiral oscillators and theoretically veried a
rule Biot had given regarding OR spectra (6).
Figure 2. The two enantiomers of 2-butanol are mirror images
of each other, as shown by the directed circular arrangements of
the OH, CH2 CH3 , and CH3 groups.
late 1980s as single enantiomers, of which 61 were totally
synthetic. But since 1992, the U.S. Food and Drug Administration (FDA) has insisted that only one enantiomer of
a chiral drug be brought into the market. Biological chirospecicity, once the subject of speculations by Pasteur on
the nature of the life force (vis viva), is now the topic of
conferences on the origin of life (4).
An isotropic chiral medium is circularly birefringent
(i.e., both left-handed and right-handed circularly polarized light can propagate in a region lled with a homogeneous isotropic chiral medium, with different phase velocities and attenuation rates). Therefore, when monochromatic, elliptically polarized light irradiates an isotropic
chiral slab, the tilt angle and the axial ratio of the transmitted light are different from those of the incident light. The
change in the tilt angle is quantied as optical rotation
(OR) and alteration of the axial ratio as circular dichroism (CD). Both OR and CD depend on the wavelength 0 ,
and the dependences are reasonably material-specic that
spectroscopies based on their measurements have long had
industrial importance. Biot himself had pioneered these attempts by cataloging the OR spectra of a large number of
syrups and oils, and went on to found the science of saccharimetry for which he was awarded the Rumford Medal
in 1840 by the Royal Society of London. The rst edition of
Landolts tables on optical activity appeared in the German
language in 1879; the English translation of the second edition of 1898 appeared in 1902.
Although Maxwells unication of light with electromagnetism during the third quarter of the nineteenth century
came to mean that natural optical activity is an EM phenomenon, the term optical rotation persisted. By the end of
the nineteenth century, several empirical rules had evolved
on OR spectrums of isotropic chiral mediums. Then, in the
late 1890s, two accomplishments of note were reported:
1. J. C. Bose constructed several articial chiral materials by twisting jute bers and laying them end to end,
and experimentally veried OR at millimeter wavelengths. These materials were anisotropic, but Bose
went on to infer from his experiments that isotropic
chiral materials could also be constructed in the same
way (5). Thus, he conclusively demonstrated the geometric microstructural basis for optical activity, and
he also constructed possibly the worlds rst articial anisotropic chiral medium to alter the vibration
ellipses of microwaves.
Experimental verication of Drudes spiral oscillator
hypothesis had to wait for another two decades. As electromagnetic propositions can be tested at lower frequencies if the lengths are correspondingly increased and other
properties proportionally adjusted, K. F. Lindman made
2.5-turn, 10-mm-diameter springs from 9-cm-long copper
wire pieces of 1.2 mm cross-sectional diameter. Springs are
handed, as illustrated in Fig. 3. Each spring was wrapped
in a cotton ball, and about 700 springs of the same handedness were randomly positioned in a 262626-cm cardboard box with an eye to achieving tolerable isotropy. Then
the box was irradiated with 13-GHz (30 cm0 10 cm)
microwave radiation and the OR was measured. Lindman
veried Drudes hypothesis remarkably well. He also determined that (1) the OR was proportional to the number
of (identically handed) springs in the box, given that the
distribution of springs was rather sparse; and (2) equal
amounts of left-handed or right-handed springs brought
about the same OR, but in opposite senses (7). Lindmans
experiments were extensively repeated during the 1990s
by many research groups in several countries (8, 9), and
several patents have even been awarded on making articial isotropic chiral mediums with miniature springs.
CONSTITUTIVE RELATIONS OF AN ISOTROPIC CHIRAL
MEDIUM
Electromagnetic elds are governed by the Maxwell postulates, in vacuum as well as in any material medium. These
four postulates have a microscopic basis and are given in
vacuum as follows:
t) and B(r,
t) are the primitive or the fundamental
Thus, E(r,
EM elds, both functions of the three-dimensional position
vector r and time t; 0 =8.8541012 F/m and 0 =4107
H/m are, respectively, the permittivity and the permeability of vacuum; tot (r, t) is the electric charge density and
J tot (r, t) is the electric current density.
Equations (1a), (1b), (1c) and (1d) apply at any length
scale, whereas the charge and the current densities must
be specied not continuously but over a set of isolated
points. Electromagnetically speaking, matter is nothing
but a collection of discrete charged particles in vacuum.
As per the HeavisideLorentz procedure to get a macroscopic description of continuous matter, spatial averages
of all elds and sources are taken, while both tot (r, t) and
J tot (r, t) are partitioned into matter-derived and externally
impressed components. Then the Maxwell postulates at the
Chirality
invariant response characteristics may be stated as
Four constitutive property kernels appear in these equations; the dyadic -e(t) is the dielectric susceptibility kernel,
m(t)
susceptibility kernel, while the dyadics
- em(t)is the magnetic
em(t)
and
are
called
the magnetoelectric kernels. Although a dyadic may be understood as a 33 matrix for
the purpose of this article, Chens textbook (10) is recommended for a simple introduction to the use of dyadics in
EM theory.
All four dyadic kernels in Eqs. (4a) and (4b) are causal
[i.e., -e(t) -0-0 for t0, etc.], because all materials must
exhibit delayed response. In addition, when we substitute
Eqs. (4a) and (4b) in Eqs. (2c) and (2d), respectively, a redundancy emerges with respect to Eqs. (2a) and (2b). Elimination of this redundancy leads to the constraint (11)
Figure 3. An enantiomeric pair of springs. An articial isotropic
chiral medium can be made by randomly dispersing springs in an
isotropic achiral host medium, with more springs of one handedness than the springs of the other handedness.
macroscopic level can be stated as
which has never been known to be violated by a physical material. Finally, crystallographic symmetries may also
impose additional constraints on the constitutive kernels.
A medium described by Eqs. (4a) and (4b) is said to be bianisotropic, since the constitutive kernels indicate anisotropy,
and both D(r,
t) and H(r,
t) depend on both E(r,
t) and B(r,
t)
Suppose next that the linear mediums constitutive
properties are direction-independent. Equations (4a) and
(4b) then simplify to
t) are the externally impressed source
Here, (r,
t) and J(r,
densities, while the new elds
contain two matter-derived quantities: the polarization
P(r,
t) and the magnetization M(r,
t)
Constitutive relations must be prescribed to relate the
matter-derived elds D(r,
t) and H(r,
t) to the basic elds
E(r, t) and B(r, t) in any material medium. The construction
of these relations is primarily phenomenological, although
certain epistemologically mandated proprieties must be
adhered to. The constitutive relations appropriate for a
general, linear, homogeneous, material medium with time-
in consequence of Eq. (5), where the scalar chi(t) is the chirality kernel. Equations (6a), (6b) describe the isotropic
chiral mediumthe most general, isotropic, linear electromagnetic material known to exist (12, 13).
Most commonly, EM analysis is carried out in the
frequency domain, not the time domain. Let all timedependent quantities be Fourier-transformed; thus
and so on, where w = 2 p f is the angular frequency. In
the remainder of this article, phasors such as D(r, ) are
Chirality
called elds, following normal practice. The four Maxwell
postulates Eqs. (2a), (2b), (2c) and (2d) assume the form
The time-averaged Poynting vector
denotes the direction of power ow. In any linear medium,
the monochromatic Poynting theorem reads as
while the constitutive equations [Eqs. 6] for an isotropic
chiral medium simultaneously transform into
Using Eqs. (8b) and (8d) with J(r, ) = 0 in Eqs. (9a)
and (9b), respectively, we obtain the DrudeBornFedorov
(DBF) constitutive relations of an isotropic chiral medium:
Their great merit is that the necessary mirror asymmetry
is transparently reected in them, because E(r, ) and
H(r, ) are not true vectors but only pseudovectors. A
chiral medium is thus described by three constitutive properties; the permittivity and permeability in Eqs. (10a) and
(10b), respectively, may be formally dened as the ratios
but the chirality parameter () can be regarded as either
or
where the asterisk denotes the complex conjugate. Equations (11a), (11b), (11c) and (11d) make it clear that while
() and () are true scalars, () has to be a pseudoscalar
since the numerator in either of its two denitions contains
a pseudovector. Other constitutive relationsequivalent
to Eqs. (9a) and (9b) and Eqs. (10a) and (10b)are also
used in the frequency-domain EM literature, but this article is restricted to the DBF constitutive relations Eqs. (10a)
and (10b), as they bring out the essence of chirality at the
very rst glance. An isotropic chiral medium and its mirror image share the same () and (), and their chirality
parameters differ only in sign.
For specialization to an isotropic chiral medium, we have to
substitute Eqs. (10a) and (10b) in Eq. (12b). The resulting
expression is not particularly illuminating.
An isotropic chiral medium is Lorentz-reciprocal. Suppose that all space is occupied by a homogeneous isotropic
chiral medium and all sources are conned to regions of
bounded extent. Let sources labeled a radiate elds Ea (r, )
and Ha (r, ), while sources labeled b radiate elds Eb (r, )
and Hb (r, ), all at the same frequency. Then the relations
(12)
arise in a source-free region, in consequence of the Lorentz
reciprocity of the medium.
ARTIFICIAL ISOTROPIC CHIRAL MEDIA
That matter is discrete has long been established. Furthermore, when we probe matter at length scales at which it
appears continuous, whether the microstructure is molecular or merely comprises electrically small inclusions is of no
consequence. The linear dimensions of an electrically small
inclusion are less than about a tenth of the maximum wavelength, in the media outside as well as inside the inclusion,
at a particular frequency. Articial isotropic chiral media
active at microwave frequenciescan be constructed with
this thought in mind. Consider a random suspension of
identical, electrically small, inclusions in a host medium,
which we take here to be vacuum for simplicity. The number of inclusions per unit volume is denoted by N, and the
volumetric proportion of the inclusions in the composite
medium is assumed to be very small. Our objective is to
homogenize this dilute particulate composite medium and
estimate its effective constitutive properties (13). Homogenization is much like blending apples into apple sauce or
tomatoes into ketchup.
Any inclusion scatters the EM wave incident on it. Far
away from the inclusion, the scattered EM eld phasors can
be conceptualized, equivalently, as being radiated by an ensemble of multipoles. Multipoles are necessarily frequencydomain entities; and adequate descriptions of electrically
larger inclusions require higher-order multipoles, but homogenizing composite media with electrically large inclusions is fraught with conceptual perils.
The lowest-order multipoles are the electric dipole p and
the magnetic dipole m. In formalisms for isotropic chiral
media, both are accorded the same status. As all inclu-
Chirality
sions in our composite medium are electrically small, we
can think that an inclusion located at position r is equivalent to the colocated dipoles characterized by the following
relations:
Here, Eexc (r , ) and Hexc (r , ) are the elds exciting the
particular inclusion; while and -hh () are the four linear
polarizability dyadics that depend on the frequency, the
constitution, and the dimensions of the inclusion. As the
inclusions are randomly oriented and any homogenizable
chunk of a composite medium contains a large number of
inclusions, -ee () and other terms in Eqs. (14) can be replaced by their orientationally averaged values. If the homogenized composite medium is isotropic chiral, this orientational averaging process must yield
The polarizability dyadics of electrically small, handed inclusions (e.g., springs) may be computed either with standard scattering methods such as the method of moments
(14) or using lumped-parameter circuit models (15). Provided that dissipation in the composite medium can be ignored, at a certain angular frequency, ee (), hh (), and
chi () are purely real-valued.
On applying the Maxwell Garnett homogenization approach, the constitutive relations of the homogenized composite medium (HCM) are estimated as follows (12):
are the DBF constitutive relations of the HCM, with
as the constitutive parameters. Clearly, if chi () = 0 the
composite medium has been homogenized into an isotropic
chiral medium. In passing, other homogenization approaches are also possible for chiral composites (1, 13).
BELTRAMI FIELDS IN AN ISOTROPIC CHIRAL MEDIUM
In a source-free region occupied by a homogeneous
isotropic chiral medium, (r, ) = 0 and J(r, ) = 0. Equations (8a), (8c), and (10a) and (10b) then show that
E(r, ) = 0 and H(r, ) = 0. Thus all four elds
E(r, ), H(r, ), D(r, ) and B(r, )are purely solenoidal.
Next, Eqs. (8a), (8b), (8c) and (8d) and (10a) and (10b) together yield the following vector Helmholtz-like equations:
In the limit () 0 the medium becomes achiral and
these equations reduce to the familiar vector Helmholtz
equation, and so on.
In lieu of the second-order differential equations [Eqs.
(23)], rst-order differential equations can be formulated.
Thus, after dening the auxiliary elds
where
and using the wavenumbers
Equivalently
we get the two rst-order differential equations
Chirality
and vice versa (17). Thus, the decomposition
is possible, as the rst parts on the right sides of Eqs. (22)
are toroidal and the second parts are poloidal. The scalar
functions v (r, ) satisfy the scalar Helmholtz equation as
follows:
Figure 4. Optical rotation (OR) and circular dichroism (CD) spectra of a simple isotropic chiral medium. When the OR changes
sign, the CD records either a maximum or a minimum, which phenomenon is called the Cotton effect.
which are easier to analyze than Eqs. (18). The denominators on the left sides of Eqs. (20a) and (20b) suggest that
2 ()()2 () = 1 is not permissible for an isotropic chiral medium, as both wavenumbers must have nite magnitudes.
According to Eqs. (21a) and (21b), Q1 (r, ) and Q2 (r, )
are Beltrami elds (12). A Beltrami eld is parallel to its
own circulation. The concept arose early in the nineteenth
century, and has often been rediscovered. The easiest way
to think of a Beltrami eld is as a spiral staircase or a
tornado.
While Q1 (r, ) is a left-handed Beltrami eld, the negative sign on the right side of Eq. (15b) means that Q2 (r, )
is a right-handed Beltrami eld, because the two complexvalued wavenumbers 1 () and 2 () must have positive
real parts. Both wavenumbers also must have positive
imaginary parts in a causal material medium, since causal
materials must exhibit delayed response in the time domain and therefore must demonstrate EM loss (or attenuation) in the frequency domain.
As an isotropic chiral medium displays two distinct
wavenumbers at a specic frequency, it is birefringent.
More specically, because Q1 (r, ) and Q2 (r, ) have planewave representations possible only in terms of circularly
polarized plane waves, an isotropic chiral medium is often said to be circularly birefringent. The difference between 1 () and 2 () gives rise to natural optical activity.
While OR is proportional to the real part of [1 () 2 ()],
CD is proportional to the imaginary part of [1 () 2 ()].
The OR and CD spectra must be consistent with the
KramersKronig relations (16). The CD spectrum has a local maximum or minimum at the frequency where the sign
of the OR changes; this feature is labeled as the Cotton effect after H. Cotton, who reported it in 1895 (2). The OR
and CD spectra of a simple chiral medium are illustrated
in Fig. 4.
REPRESENTATION OF BELTRAMI FIELDS
A Beltrami eld is represented in terms of toroidal and
poloidal elds because the curl of a toroidal eld is poloidal
Solutions of Eqs. (23) in the Cartesian, the circular cylindrical, and the spherical coordinate systems are commonplace
(18).
Beltrami plane waves propagating in the +z direction
may be represented as
with A as the amplitudes, while x , y , and z are the Cartesian unit vectors.
In the circular cylindrical coordinate system (, , z) Beltrami elds with an exp(iz) dependence may be expressed
as the sums
for regular behavior as while the expansions
are well behaved at = 0 with An and B n as the coefcients of expansion. The vector cylindrical wavefunctions
are given as
1/2
where = +( 2 2 ) ; ,
and z are the unit vectors in
the cylindrical coordinate system; J n () are the cylindrical Bessel functions of order n, and J n () are the respective rst derivatives with respect to the argument; while
Hn(1) () are the cylindrical Hankel functions of the rst
kind and order n, and Hn(1) () are the rst derivatives with
respect to the argument. For quasi-two-dimensional problems, = 0 because /z = 0 Parenthetically, in this paragraph denotes the radial distance in the xy plane and
should not be confused with the use of for charge density
elsewhere in this article.
Finally, with A smn and B smn as the coefcients of expansion, in the spherical coordinate system (r, , ), we have
Chirality
and (21b), we get
for elds regular as r , and
for elds regular at r = 0. The well-known vector spherical
( j)
( j)
wavefunctions, Msmn
(r) and Nsmn
(r) are stated for j = 1,
3 as
where-I is the identity dyadic. The axial components satisfy
the reduced scalar Helmholtz equations
appropriate solutions of which are commonly worked out
in many different ways for waveguides of different crosssectional geometries (18).
SOURCES IN AN ISOTROPIC CHIRAL MEDIUM
where the angular functions
and are the
have been used. In these expressions, r , ,
unit vectors in the spherical coordinate system; Pnm (cos )
are the associated Legendre functions of order n and degree
m; j n (r) are the spherical Bessel functions of order n; and
h(1)
n (r) are the spherical Hankel functions of the rst kind
and order n.
Boundary-value problems involving scattering by
isotropic chiral half-spaces, cylinders, and spheres can be
analytically solved using Eqs. (24,25a) and (25b), (26a),
(26b) and (27b), (27a) and (27b), (28a) and (28b), (29a) and
(29b). Boundary-value problems involving more complicated geometries generally require numerical treatment,
which necessitates the use of Green functions.
Isotropic chiral waveguides for use at microwave frequencies have been theoretically studied extensively, although no practical realization thereof has yet come to
light. Theoretical investigations on propagation in the socalled chirowaveguides generally consist of decomposing
the Beltrami elds into axial and transverse components
as
where the z coordinate is measured on the waveguide axis
while two other mutually orthogonal coordinates are specied in the transverse plane. Assuming that all elds have
an exp(iz) dependence on z, and making use of Eqs. (21a)
Let us now assume the existence of a magnetic charge
density m (r, ) and a magnetic current density J m (r,
), because they assist in the solution of dual problems
(19). Inaddition, let us dene the intrinsic impedance
() = ()/() as well as the auxiliary wavenumber
k() = ()/() and drop the explicit indication of dependences on for notational simplicity. Now Eqs. (8), (8b),
(8c) and (8d) may be written as
which yield the relations
for a chiral medium, where
are the Beltrami source current densities (12).
Since Eqs. (34) are linear, they can be solved using standard techniques. Their complete solution can be compactly
stated for all r as
where
are the particular solutions due to the source densities
W (r) which are wholly conned to the region V s , and Qcf
(r)
are the complementary functions satisfying the relations
Chirality
identically. Substituting Eqs. 3638 in Eqs. (34), we obtain
the dyadic differential equations
where () is the Dirac delta function.
The solutions of Eqs. (39) are the BeltramiGreen
dyadic functions
wherein
is the familiar dyadic Green function for free space. As
the properties of -Gfs (, r, r0 ) can be found in almost any
graduate-level EM textbook (20, 21), those of -G (r, r0 ) can
be easily determined, as illustrated in Ref. 12.
As an example of the use of Eqs. (37), let us consider
an electric dipole moment Plocated at the origin: J(r) =
ip(r) and Jm (r) = 0 The radiated Beltrami elds turn
out be
which show clearly that the radiation eld of a point electric dipole in an isotropic chiral medium consists of lefthanded as well as right-handed components. If we have
instead a point magnetic dipole m located at the origin,
the source current densities are specied as J(r) = 0 and
Jm (r) = im(r) so that
are the corresponding radiated Beltrami elds. A major
difference between isotropic chiral and achiral media is
shown by the two sets of radiated elds, Eqs. (42a), (42b)
and (43a), (43b). Without loss of generality, let the source
dipole moments be aligned parallel to the z axis. Then, if
the dipole moments are radiating in an achiral medium
(i.e., =0), there is no magnetic eld due to p and there is
no electric eld due to m at any point on the z axis. On the
other hand, the wavenumber difference between the lefthanded and the right-handed Beltrami elds guarantees
that, in an isotropic chiral medium, both Erad (r) and Hrad (r)
are not generally null-valued on the z axis, regardless of
which one of the two dipole moments is radiating.
Canonical sources of Beltrami elds are possible. If
there is a source distribution such that J(r) (1/i)J m (r)
for all r, then Q rad 2 (r) 0 from Eqs. (35a) and (35b) and
(37). Likewise, a source distribution containing electric and
magnetic current densities in the simple proportion J(r) =
(1/i)J m (r) for all r radiates only a right-handed eld, because Q rad 1 (r) 0 emerges from the same equations.
Radiation by complex sources has to be generally
treated using integral equations. Both the Maue and the
Pocklington integral equations for radiation in a homogeneous isotropic chiral medium are available (12). Cerenkov
radiation in an isotropic chiral medium has also been described using Beltrami elds (12).
The foregoing developments make it clear that a description involving differentials of only the rst order sufces for monochromatic radiation and propagation in an
isotropic chiral medium. True, there are terms in
-G1 (r, r0 ) and -G2 (r, r0 ) but dyadic Green functions are not
elds, being instead solutions of dyadic differential equations.
Finally, although the left-handed and the right-handed
Beltrami elds are capable of being independently radiated
and propagated as per Eqs. (34), they do indeed couple in
an isotropic chiral medium. This coupling takes place only
at bimedium boundaries where conditions on the tangential components of E(r) and H(r) must be satised; that is,
the boundary conditions are specied not on Q1 (r) or Q2 (r)
singly, but on the tangential components of the combinations E(r) = Q 1 (r) iQ 2 (r) and H(r) = Q 2 (r) + (1/i)Q 1 (r).
THEOREMS FOR SCATTERING IN AN ISOTROPIC
CHIRAL MEDIUM
Equations 36-39 sufce to set up certain often-used principles for monochromatic scattering and radiation problems,
when all space is lled with a homogeneous isotropic chiral
medium.
The sourceregion Beltrami elds can be obtained from
Eqs. (37) using the Fikioris approach (22). Let S be the
surface of the convex-shaped source region V s , where n
0 is
the unit outward normal at r 0 S (see Fig. 5). Then, Eqs.
(37) and (40) yield the following relations:
The depolarization dyadic
in Eqs. (44) is dependent on the shape of the region V s ,
while
If the maximum linear extent of the region V s times
the magnitude of the greater of the two wavenumbers, 1
Chirality
Figure 5. For the evaluation of elds in the region V s , when the
sources are also conned to the same region and all space is occupied by a homogeneous chiral medium.
and 2 , is much smaller than unity, we may make the quasistatic approximation: W1 (r0 )
=W1 (r) and W2 (r0 )
=W2 (r)
for all r 0 V S . Then, Eqs. (44) simplify to
Figure 6. Relevant to the Huygens principle, the exterior surface
equivalence principle, and the EwaldOseen extinction theorem,
when all space is occupied by a homogeneous chiral medium.
The Huygens principle allows the enunciation of the exterior surface equivalence principle. Consider a problem
in which surface Beltrami current densities W S 1 (r) and
W S 2 (r) exist on the exterior side of the surface S (see Fig.
6). As per Eqs. (37), these surface current densities act as
sources of the radiated elds
where the dyadics
On comparing Eqs. (50a) and (51) to ensure the equivalence
Q rad (r) Q (r) for all r V ext , we obtain the relationships
(12)
depend on the shape as well as on the size of V s . Finally,
the Rayleigh approximation requires that we ignore the
dyadics -M(|r) and -N(|r) completely to obtain the estimates
when V s is an extremely small region. The right sides of
Eqs. (47) and (49) are useful in homogenizing isotropic
chiral composites as well as for devising the method of
moments and the coupled dipole method for scattering by
bianisotropic objects in isotropic chiral environments (12,
23).
Turning now to the mathematical realizations of the
Huygens principle and its progeny, we suppose that all
space is divided into two regions, as shown in Fig. 6. The
external region V ext extends to innity in all directions but
is separated from an internal region V int by the convex and
once-differentiable surface S. Then the Huygens principle
in a homogeneous isotropic chiral medium reads as follows
(12):
as the exterior surface equivalence principle for Beltrami
elds and sources, r0 in Eqs. (52) lying on the exterior side
of S.
The EwaldOseen extinction theorem is a cornerstone
of the extended-boundary-condition method (12, 24). For
scattering in an isotropic chiral medium, this theorem may
be stated as
where Q cf (r) play the role of the incident Beltrami elds.
Once Q (r 0 ), r 0 S have been determined from Eqs. (53),
the total elds in the exterior region may be determined as
From Eqs. (53) and (54), the plane-wave scattering dyadics
for an object in an isotropic chiral environment can be derived, as can the forward plane-wave scattering amplitude
theorems (12).
Thus, the Cauchy data for the elds in a chiral medium
comprise the components of the Beltrami elds that are
tangential to a boundary. When these data are prescribed
on the surface S, we can nd the Beltrami elds everywhere in the region V ext .
STRUCTURALLY CHIRAL MEDIA
The molecules of a naturally occurring isotropic chiral medium are mirror-asymmetric, and so are the inclusions in an articial isotropic chiral medium. As a
10
Chirality
gle. At a low enough frequency, this laminate appears as a
continuously nonhomogeneous medium whose constitutive
properties vary helicoidally. Thus
are the frequency-domain constitutive relations of a CNLC,
where
is the relative permittivity dyadic in a reference plane designated as z=0. The rotation dyadic
Figure 7. Schematic depiction of the arrangement of needle-like
molecules in a chiral nematic liquid crystal. The gaps between
the consecutive sheets as well as the sheets are ctitious, as they
are merely aids to visualization. Only half of the electromagnetic
period is shown.
randomly dispersed and randomly oriented collection
of mirror-asymmetric molecules or inclusions is also
mirror-asymmetric, isotropic chiral media emerge with
direction-independent constitutive properties. In contrast,
the molecules or inclusions of a structurally chiral medium
are not mirror-asymmetric, but their orientation is.
In chiral nematic liquid crystals (CNLCs)also called
cholesteric liquid crystalsneedle-like molecules are randomly positioned on parallel sheets, with all molecules on
any one sheet oriented parallel to one another and with the
orientation rotating helicoidally as one moves across consecutive sheets. The situation is schematically depicted in
Fig. 7. From 1850 to 1888, several scientists came across
CNLCs but were unable to capitalize on their observations (25). Then in 1888 the biochemist F. Reinitzer observed that a CNLC named cholesteryl benzoate has two
distinct melting pointsit is a solid at temperatures below 145.5 C, a clear liquid at temperatures above 178.5 C,
and a cloudy liquid in between. Reinitzers observation of
the mesophasewhen positional order is absent as in a
liquid, but orientational order is still strong as in a solid
opened up the area of liquid crystal research in continuum
mechanics as well as in optics (2628).
Earlier, however, (in 1869), E. Reusch had anticipated
the CNLC structure as a laminate of uniaxial dielectric
sheets, with the crystallographic axes of any two adjacent
sheets offset in the transverse plane by a xed small an-
denotes that the CNLC structure varies helicoidally in the
axial (i.e., z) direction with a period 2 ; however, the electromagnetic period is . The upper sign in Eq. (57) applies
for structural right-handedness; the lower, for structural
left-handedness.
Reuschs model of a CNLC has been often implemented
with either uniaxial crystals or brous laminae, and appears promising for microwave and RF applications as well
(29). More recently, thin-lm technology has been pressed
into service to realize the CNLC structure by releasing a
directed evaporant ux toward a rotating substrate (30,
31). The reference permittivity dyadic of these chiral sculptured thin lms (STFs) differs from Eq. (56), being
instead, and the electromagnetic period is 2 .
The reference permittivity dyadics in Eqs. (56) and (58)
are uniaxial and biaxial, respectively; that is, they have either one or two crystallographic axes. Biaxial -ref () is displayed by chiral smectic liquid crystals also (26, 27). Thus
in general -ref () displays orthorhombic symmetry (32).
Moreover, particularly with advances in thin-lm technology, there is no reason for a chiral STF to be necessarily dielectric only. These considerations led to the proposal of the
helicoidal bianisotropic medium (HBM), whose frequencydomain constitutive relations may be stated as (33)
subject to the constraint
The launching and propagation of EM waves in HBMs is
best studied using a 44 matrix differential equation formalism (31, 34).
Chirality
9.
10.
11.
12.
13.
Figure 8. Scanning electron micrograph of a 10-period chiral
sculptured thin lm made of silicon oxide. (From Professor Russell
Messier, Pennsylvania State University, with permission.)
14.
15.
16.
Although chiral STFs made of uorites, and singlefrequency OR measurements on them, were reported in
1959 (35), systematic experimental studiesalong with
scanning electron microscopic verication of the microstructural geometryappear to have begun only in
1995 (30). Figure 8 shows the scanning electron micrograph of a chiral STF made of silicon oxide. As typical
values of realized today range from 30 nm to 10 m,
microwave applications of these lms are yet not feasible,
but are likely to become an active area of research once
lms with 100 m become available. Many possible applications have been anticipated as the concept of STFs
for biological, optical, electronic, chemical, and other applications is beginning to take root, while many optical and
related applications have already been implemented (30,
31). Large-scale production appears feasible as well, with
adaptation of ion-thruster technology (36).
BIBLIOGRAPHY
1. O. N. Singh andA. Lakhtakia, eds., Electromagnetic Fields in
Unconventional Materials and Structures, Wiley, New York,
2000.
2. A. Lakhtakia, ed., Selected Papers on Natural Optical Activity,
SPIE Optical Engineering Press, Bellingham, WA, 1990.
3. J. Jacques, The Molecule and Its Double, McGraw-Hill, New
York, 1993.
4. B. Holmstedt, F. Hartmut, andB. Testa, eds., Chirality and
Biological Activity, Alan R. Liss, New York, 1990.
5. J. C. Bose, On the rotation of plane of polarisation of electric
waves by a twisted structure, Proc. Roy. Soc. Lond. 63: 146152
(1898) .
6. P. Drude, Lehrbuch der Optik, S. Hirzel, Leipzig, 1900.
7. K. F. Lindman, Uber
eine durch ein isotropes system von spiralformigen resonatoren erzeugte rotationspolarisation der
elektromagnetischen wellen, Ann. Phys. Leipzig. 63: 621644
(1920) .
8. R. Ro, Determination of the Electromagnetic Properties of
Chiral Composites, Using Normal Incidence Measurements,
17.
18.
19.
20.
21.
11
Ph.D. thesis, Pennsylvania State Univ., University Park, PA,
1991.
F. Guerin, Contribution a` Letude Theorique et Experimentale
des Materiaux Composites Chiraux et Bianisotropes dans
le Domain Microonde, Ph.D. thesis, Univ. Limoges, Limoges,
France, 1995.
H. C. Chen, Theory of Electromagnetic Waves, TechBooks, Fairfax, VA, 1993.
A. Lakhtakia and W. S. Weiglhofer, Constraint on linear, spatiotemporally nonlocal, spatiotemporally nonhomogeneous
constitutive relations, Int. J. Infrared Millim. Waves. 17:
18671878 (1996) .
A. Lakhtakia, Beltrami Fields in Chiral Media, World Scientic, Singapore, 1994.
A. Lakhtakia, ed., Selected Papers on Linear Optical Composite
Materials, SPIE Optical Engineering Press, Bellingham, WA,
1996.
J. J. H. Wang, Generalized Moment Methods in Electromagnetics, Wiley, New York, 1991.
C. H. Durney and C. C. Johnson, Introduction to Modern Electromagnetics, McGraw-Hill, New York, 1969.
A. Moscowitz, Theoretical aspects of optical activity: small
molecules, Adv. Chem. Phys. 4: 67112 (1962) .
S. Chandrasekhar, Hydrodynamic and Hydromagnetic Stability, Oxford Univ. Press, Oxford, UK, 1961.
P. Moon and D. E. Spencer, Field Theory Handbook, Springer,
Berlin, 1988.
R. F. Harrington, Time-Harmonic Electromagnetic Fields,
McGraw-Hill, New York, 1961, Chapter 3.
J. Van Bladel, Electromagnetic Fields, Hemisphere Publishing,
New York, 1985.
W. C. Chew, Waves and Fields in Inhomogeneous Media, IEEE
Press, New York, 1995.
22. J. G. Fikioris, Electromagnetic eld inside a current-carrying
region, J. Math. Phys. 6: 16171620 (1965) .
23. B. Shanker and A. Lakhtakia, Extended Maxwell Garnett
model for chiral-in-chiral composites, J. Phys. D: Appl. Phys.
26: 17461758 (1993) .
24. P. C. Waterman, Scattering by dielectric obstacles, Alta Frequenza (Speciale) 38:348352 (1969) .
25. P. J. Collings, Liquid Crystals, Princeton Univ. Press, Princeton, NJ, 1990, Chapter 2.
26. S. Chandrasekhar, Liquid Crystals, Cambridge Univ. Press,
Cambridge, UK, 1992.
27. P. G. de Gennes and J. Prost, The Physics of Liquid Crystals,
Clarendon Press, Oxford, 1993.
28. S. D. Jacobs, ed., Selected Papers on Liquid Crystals for Optics,
SPIE Optical Engineering Press, Bellingham, WA, 1992.
29. A. Lakhtakia, G. Ya. Slepyan, and S. A. Maksimenko, Towards
cholesteric absorbers for microwave frequencies, Int. J. Infrared Millim. Waves. 22: 9991007 (2001) .
30. A. Lakhtakia, R. Messier, M. J. Brett, and K. Robbie, Sculptured thin lms (STFs) for optical, chemical and biological applications, Innov. Mater. Res. 1: 165176 (1996) .
31. A. Lakhtakia and R. Messier, Sculptured Thin Films: Nanoengineered Morphology and Optics, SPIE Press, Bellingham,
WA, (2005) .
32. J. F. Nye, Physical Properties of Crystals, Clarendon Press,
Oxford, 1985.
33. A. Lakhtakia and W. S. Weiglhofer, Axial propagation in general helicoidal bianisotropic media, Microwave Opt. Technol.
Lett. 6: 804806 (1993) .
12
Chirality
34. A. Lakhtakia, Director-based theory for the optics of sculptured thin lms, Optik. 107: 5761 (1997) .
35. N. O. Young and J. Kowal, Optically active uorite lms, Nature. 183: 104105 (1959) .
36. M. W. Horn, M. D. Pickett, R. Messier, and A. Lakhtakia,
Blending of nanoscale and microscale in uniform largearea sculptured thin-lm architectures, Nanotechnology 15:
303310 (2004) .
AKHLESH LAKHTAKIA
Pennsylvania State University,
University Park, PA
202
CONFORMAL ANTENNAS
example, in Ref. 2). However, there are significant differences
between planar and conformal arrays that must be taken into
account during the design of the latter. The individual elements on curved bodies point in different directions that
make it necessary to turn off those elements that radiate primarily away from the desired beam direction. For this reason
also, one cannot factor out the element pattern out of the total
radiation patternthis makes the conformal array analysis
and synthesis more difficult. The element orientation may
also cause severe crosspolarization. In addition, the mutual
coupling effects between the elements can be severe in some
cases.
Within the limitations of space allowed, it is not possible
to describe here every aspect of conformal antennas and antenna arrays. Instead, we shall at first describe briefly certain
aspects of a few basic antennas that are commonly used either singly or as array elements for conformal applications.
Then we give brief descriptions of a selected number of conformal antennas and antenna arrays. Specifically, this article
describes the following:
1. The specific considerations that must be given to the
performance of basic slot and microstrip or patch antenna elements when mounted on nonplanar conducting
surfaces
CONFORMAL ANTENNAS
A conformal antenna may be defined as an antenna whose
radiating aperture conforms to the surface of the body on
which it is mounted. Ideally, such antennas are flush
mounted or low profile (i.e., they do not protrude appreciably
out of the mounting surface). Basic slot and microstrip (patch)
antennas are typical examples of conformal antenna elements. The term conformal array has no unique definition.
Kummer (1) defines it as an array that is nonplanar. We shall
assume here that a conformal array consists of conformal (or
low-profile) antenna elements placed on a nonplanar surface.
The array surface is not generally at the disposal of the antenna designer and is often dictated by the specific application. For ground-based application, a conformal phased array
requiring coverage over 360 in azimuth (omnidirectional coverage) or coverage over a hemisphere the array surface may
be cylindrical or spherical, respectively. For conformal arrays
on aircraft, missiles, satellites, and surface ships, the array
shape may assume another form dictated by the contour of
the vehicle. Basic slot and microstrip antennas are extensively discussed in the literaturefor example, the textbook
by Balanis (2) is a typical reference. These antennas provide
ideal performance only when they are mounted on planar surfaces. During conformal application the curvature of the
mounting surface can affect their impedance and radiation
properties; such effects must be taken into account during the
design of such antennas.
The need for conformal phased arrays for aircraft and missile applications, and for ground-based arrays with omnidirectional coverage in azimuth or complete hemispherical coverage in space, has grown continually with requirements that
emphasize maximum utilization of available space and minimum cost. Many of the developments in conformal arrays
have been extensions of the concepts for planar phased
arrays, which are extensively discussed in the literature (for
2. A selected number of conformal antennas: for example,
microstrip conformal antennas and dielectric filled edge
slot (DFES) antennas
3. A class of wraparound antennas and antenna arrays
4. Cylindrical and spherical phased arrays used for omnidirectional and hemisphereical coverage, respectively.
All of the aforementioned antennas have found practical applications. Detailed descriptions of their development, design
procedures, and analysis of their performance are described
in the references cited at appropriate places.
Literature on conformal antennas is vast and ranges from
technical journal articles to numerous textbooks and specialized books, of which Refs. 28 are typical.
BASIC ANTENNA ELEMENTS
Elementary slot and microstrip patch antennas are commonly
used singly or as array elements for conformal application.
However, these radiators provide ideal performance only
when they use plane conducting surfaces. Ideal theory can be
used when the radii of curvature of the surfaces are large
compared to the operating wavelength. In other cases both
the impedance and radiation characteristics may be affected
significantly.
Slots on Curved Surfaces
The radiation patterns of slot antennas can be significantly
altered by the curved mounting surface. Pathak and Kouyoumjian (9) give a convenient extension of the geometrical
theory of diffraction (GTD) for apertures in curved surfaces.
J. Webster (ed.), Wiley Encyclopedia of Electrical and Electronics Engineering. Copyright # 1999 John Wiley & Sons, Inc.
CONFORMAL ANTENNAS
1.6
Slot
k =1.5
1.4
k = 2.0
Relative amplitude
1.2
r1 r2
1.0
k = 6.0
0.6
0.4
Exact solution
Creeping wave
Surface ray modes
0.2
0
Circumferential slots
b
a
or
Radial slots
a
k = 3.0
0.8
20
203
b
r3
40 60
80 100 120 140 160 180
Azimuth angle () degrees
Figure 1. Patterns of a thin axial slot on a perfectly conducting cylinder. (After Ref. 9.)
Figure 3. Slotted cone geometry.
Figure 1, taken from Ref. 9, shows the patterns of an axial
slot element on perfectly conducting circular cylinders of various radii; the results indicate the accuracy of the approximate
theory. The effects of the cylinder radius on the patterns
shown in Fig. 1 should be noticed. A similar slot on a flat
ground plane would have a constant pattern from 0 to
180. Radiation patterns of slots on a variety of other generalized surfaces are discussed in Refs. 1012. Mailloux (13) summarizes some of the results of Pathak and Kouyoumjian (9)
shown in Fig. 2, which gives the radiated power pattern in
the upper half plane ( 90) for an infinitesimal slot in a
cylinder of radius a. The angular extent of the transition zone
is on the order of (k0a)1/3 on each side of the shadow boundary, k0 2/ 0 being the propagation constant in free space.
The results indicate that above the transition zone (i.e., the
illuminated zone) the circumferentially polarized radiation is
nearly constant but the axially polarized radiation has a cos
pattern. Compared with the field strength in the 0
direction, the field strengths in the 90 area are found to
be about 0.7 and 0.4(2/k0a)1/3 for circumferential and axial polarizations, respectively. It should be noted that in the case of
flat surface the field reduces to zero in the 90 area.
Axial
polarization
f( ) = cos
Circumferential
polarization
f( ) = 1
Transition
zone
2
0.38 ( )1/3
k0a
g(0)/2~0.7
Figure 2. Approximate pattern of a thin slot on a conducting cylinder of radius a; k0 is the free-space propagation constant. (After Ref.
6.)
Slots on Metallic Cones
Slots and slot arrays on metallic cones are found advantageous to use for missile or missilelike bodies. For efficient design of such arrays, the self- and mutual admittances must
be taken into account. Theoretical and experimental investigation of slot antennas on metallic cones are discussed in Ref.
14, where the effects of scattering from a sharp tip on the
mutual admittances have been investigated for pairs of circumferential and radial slots on a semi-infinite metallic cone.
The base of the conical model used in the experimental study
was terminated in a spherical cap to minimize scattering from
the finite length of the apparatus. The two slot antennas configurations considered are shown in Fig. 3. Self- and mutual
admittance expressions for pairs of slots shown in Fig. 3 have
been derived by Golden, Stewart, and Pridmore-Brown (14),
and the results have been confirmed by measurements. These
admittance results can be immediately applied to determine
the aperture voltages required for the analysis of N-element
slots on cones.
In Ref. 14 the circumferential slot results illustrate interference effects between the direct coupling from slot-to-slot
via the geodesic path over the conical surface and the tip back
scattering. For the radial slot configuration, the results indicate negligible tip scattering effects. Golden and Stewart (15)
have found that the current distribution near a slot for a
sharp cone can be approximated by the distribution on an
equivalent cylinder if scattering from the apex (on tip) is
small. Thus, the mutual admittance between two slots can be
approximately calculated by using a cylindrical model with
the same local radii of curvature as the cone, provided the
wave scattering from either the tip or the base region of the
vehicle is negligible. The slotted cone and equivalent cylinder
are shown in Fig. 4, which reveals that the cylinder has a
radius equal to the radius of the circular cross section of the
cone midway between the two slots antennas. For small-angle
cones (0 180), the radial separation of the slots on the
cone can be equated to the axial separation of the slots on the
equivalent cylinder.
204
CONFORMAL ANTENNAS
80
z0
Metal cone
70
20 log10 S12 , dB
Tap-scattered ray
Circumferential slots
or
radial slots
z0
60
50
40
Frequency = 9.0 GHz
Theory
Measurements, z0 = 3.81 cm
30
20
Direct ray
10
30
60
90
120
150
180
210
0 , deg
Figure 6. Mutual coupling for axial slots on cylinder, 0 5.057 cm.
(After Ref. 14.)
2 0
Equivalent
cylinder
z0
cient; in the case of radial slots (azimuthal electric fields)
there is no radial component of the magnetic field in the far
field of tip and therefore no contribution to the mutual admittance. More detailed results and discussions are given in Refs.
14 and 15.
Direct ray
Figure 4. Slotted cone and equivalent cylinder.
Mutual coupling (S12 parameter) results versus azimuthal
separation for two circumferential and axial slots on a cylinder are shown in Figs. 5 and 6, respectively. The mutual coupling between two radial slots on a 12 2 half-angle cone is
shown in Fig. 7 as a function of frequencies. Figure 8 shows
the mutual coupling versus frequencies for circumferential
slots on a 12 2 (half-angle) cone. The results illustrate the
interference effects between the direct and tip scattered components. The mutual coupling between circumferential slots
on an 11 (half-angle) cone is shown in Fig. 9. Using the results given in Ref. 14, it may be concluded that for the case
of circumferential slots (radial electric fields) the tip scattered
portion of the azimuthal magnetic field at the slot aperture
can be expressed in terms of an appropriate diffraction coeffi-
70
z0
50
Microstrip or patch is a popular low-profile, flush-mounted
antenna developed in the 1970s. Detailed descriptions of the
research and development of microstrip antennas can be
found in Refs. 16 and 17. Such antennas generally use a metallic patch on a dielectric substrate backed by a planar
ground plane, and they are excited either by a strip line or a
coaxial line. The shape of the patch can be rectangular, circular, or some other shape, in general, of which the first two are
the most popular. We shall mostly describe the basic rectangular patch antenna whose one dimension is /2 at the operating wavelength in the substance and the other dimension
is slightly less than the former. Ideally, such antennas produce similar E- and H-plane patterns that have maxima in
the broadside direction; generally, the polarization is linear
and parallel to the patch plane but they can be designed to
produce circular polarization also. For conformal applications,
it is necessary to take into account the effects of nonplanar
surfaces on the performance of such antennas.
Cylindrical-Rectangular Patch Antenna
The geometry of a rectangular microstrip patch antenna
mounted on a conducting cylinder is shown in Fig. 10. Reso-
40
45
30
Frequency = 9.0 GHz
Theory
Measurements, z0=10.16 cm
Measurements, z0=7.62 cm
Measurements, z0=5.08 cm
20
10
0
20 log10 S12 , dB
20 log10 S12 , dB
60
Microstrips on Curved Surfaces
20
40
60
80
100
40
35
30
25
Cylinder calculations
Cone measurements
r1 = r2 = 45.53 cm
0 = 60.8 deg
c = 12.2 deg
120
0 , deg
Figure 5. Mutual coupling for circumferential slots on cylinder, 0
5.057 cm. (After Ref. 14.)
20
8.5
9.0
9.5
Figure 7. Mutual coupling for radial slots versus frequency, 0
9.622 cm. (After Ref. 14.)
CONFORMAL ANTENNAS
20 log10 S12 , dB
90
r1 =
r3 =
0 =
c =
r 2 = 45.53 cm
92.39 cm
60.8 deg
12.2 deg
Cylinder calculations
Cone measurements
Cylinder calculations
QQ
;;
205
2b
z
a
80
1 1
70
60
8.5
9.0
9.5
Figure 8. Mutual coupling for circumferential slots versus frequency, 0 9.622 cm. (After Ref. 14.)
Figure 10. Geometry of a cylindrical-rectangular microstrip patch
antenna.
nant frequencies and radiation characteristics of this antenna
are discussed in Refs. 18 and 19. For thin substrate satisfying
h a, Luk, Lee, and Dahele (19) give the following expression
for the (transverse magnetic mode with respect to ) TM resonant frequencies for the antenna
the TM10 mode, there is less radiation in the lower hemisphere for the TM01 mode. Wong and Ke (21) describe the design of this antenna for circular polarization by using the
TM01 and TM10 modes excited by a single coaxial feed located
on a diagonal line and the operating frequency chosen between the two lowest frequencies f 01 and f 10 given by Eq. (1).
Kashiwa, Onishi, and Fukai (22) describe the application
of a strip-line-fed cylindrically curved rectangular patch antenna as a small, portable antenna for mobile communication.
It has been found that near the resonant frequency the real
part of the input impedance approaches 50 . The radiation
patterns near the broadside direction are found to be similar
to those of the equivalent planar antenna; however, significant differences have been found in large off-broadside directions.
Radiation patterns of a cavity-backed microstrip patch antenna on a cylindrical body of arbitrary cross section have
been investigated theoretically and experimentally by Jin,
Berrie, Kipp, and Lee (23). The finite-element method has
been used to characterize the microstrip patch antennas, and
then the reciprocity theorem is applied in conjunction with a
two-dimensional method of moments to calculate the radiated
field. The method can be extended to characterize the radiation patterns of conformal microstrip patch antennas on general three-dimensional bodies.
f mn
c
=
2 r
m
2(a + h)1
2 n 2 1/2
+
(1)
2b
where c is the velocity of light in free space, r is the dielectric
constant of the substrate, and m, n 0, 1, 2, . . ., but m
n 0. Equation (1) indicates that if the dimensions of the
patchthat is, 2(a h) and 2bare fixed, the resonant frequencies of the TM modes are not affected by the curvature
of the thin substrate. However, to account for fringing fields,
effective values of the dimensions are to be used in Eq. (11),
as mentioned by Carver and Mink (20). Luk, Lee, and Dahele
(19) discuss the E- and H-plane radiation patterns produced
by the antenna using r 1.06, r 2.32, and different values
a. It is found that the patterns are not sensitive to the thickness. For a curved patch, there is significant radiation in the
lower hemisphere for the TM01 mode; the deviation from the
flat patch results increases for larger value of r. Compared to
20 log10 S12 , dB
80
r 1 = 27.03 cm
r 2 = 25.88 cm
r 3 = 77.47 cm
0 = 80 deg
c = 11 deg
Cone calculations
Cylinder calculations
Cone measurements
Microstrip Patch Antennas on Conical Surfaces
70
60
50
8.2
8.4
8.6
8.8
9.0
9.2
9.4
Signal frequency, GHz
Figure 9. Mutual coupling for circumferential slots versus frequency; 0 5.041 cm. (After Ref. 14.)
The use of microstrip antennas on conical surfaces is of interest for aerospace vehicles with portions of their bodies conically shaped. Performance of a basic rectangular patch antenna on a metallic cone has been investigated theoretically
by Descardeci and Giarola (24). In the analysis the substrate
thickness is assumed to be very small compared with the distance of the patch to the cone apex, and the curvature radius
of the cone surface large compared with the operating wavelength. The capacitive effects and losses associated with surface wave have been neglected. Except for these assumptions,
the cavity model analysis used is general and applies to any
conical surface. Within the approximations made, the resonant frequency is not significantly affected by the conical surface. However, the radiation pattern is affected, with a conse-
206
CONFORMAL ANTENNAS
quent influence on the input impedance and the total quality
factor. Details can be found in Ref. 24.
;;
;
;
;
;
yy;
;;
;;
yy
Input connector
Copperclad
CONFORMAL ANTENNAS
The Omni Microstrip Antenna
The omni microstrip or spiral slot antenna discussed in Refs.
2527 is essentially a short-circuited quarter-wavelength microstrip patch wrapped around a cylindrical surface to form a
spiral, as shown in Fig. 11. The cylinder is an epoxy fiberglass
dielectric, and the copper conduction are added using an electroless plating, masking, and electroplating technique. The
lower end and the inside of the patch are similarly plated to
form a short circuit and ground plane. The spiral slot antenna
has a height and diameter of 0.060 but, unlike conventional
small antennas, has well-matched input voltage standing
wave ratio (VSWR) of less than 2 : 1 over a 2% bandwidth at
238 MHz. The radiation patterns are similar to those of a
dipole oriented parallel to the cylinder axis, and the 1 dB
gain indicates an efficiency of better than 50%. The spiral slot
has also been developed for 42 MHz application in which the
antenna has to be contained in a 0.040 0.150 cylindrical
volume.
Dielectric-Filled Edge Slot Antenna
A class of circumferential slot antennas, called the DFES antennas, that are ideally suited for conformal mounting on conducting bodies of revolution has been described by Schaubert,
Jones, and Reggia (28). As shown in Fig. 12, the simplest form
of the antenna consists of a disk of dielectric substrate that is
copper coated on both sides and mounted between the two
halves of a conducting body so that the radiating aperture
coincides with the surface. The antenna is excited at the center by a coaxial line whose outer conductor is connected to the
lower conducting surface, and the inner conductor is extended
through the dielectric and finally connected to the conducting
surface at the upper end of the substrate. The input reflection
coefficients of the antenna are found to assume minimum values at some discrete frequencies, called the operating frequencies, where the antenna also radiates most efficiently.
The DFES antenna can be tuned for a desired operating frequency by using a number of axially oriented passive metallic
h
Inductive posts
plated through
Dielectric
(Teflon fiberglass)
Soldered
Figure 12. Two-element edge-slot antenna.
posts. The antenna without the tuning posts is referred to as
the basic DFES antenna, which generally profides the highest
operating frequency. By varying the number and location of
the inductive posts, the operating frequency of the antenna
can be tuned over a 6:1 range; instantaneous bandwidths of
3% are typical. Theory and design of basic and tuned DFES
antennas mounted on a conducting cylinder have been developed and discussed by Sengupta and Martins-Camelo (29).
The radiation patterns of edge-slot excited conducting bodies
of revolution display a high degree of azimuthal symmetry.
The radiation pattern of DFES antennas is strongly influenced by the body on which it is mounted. The patterns in
Fig. 13 are typical of the performance of the antenna when
mounted on a conducting cylinder.
Sometimes it is not possible to place a flat disk across the
body, and at times the antenna must be mounted near the
top of a conical body where the diameter is not sufficient to
build an antenna operating at the desired frequency. In such
cases the planar disk can be deformed (symmetrically) to fit
in the available space and to operate at the required frequency. A conical edge-slot antenna is described in Ref. 28,
and its radiation patterns are shown in Fig. 14. The DFES
antenna is a versatile and useful radiator. Because the azimuthal symmetric radiation pattern can be obtained at any
desired frequency within a very wide range, system designers
are not restricted in their choice of operating frequency. Also,
DFES antennas can be integrated into a variety of structures
because their shape can be varied to conform to the body and
the available space.
Microstrip Wraparound Antennas
g/4
(a)
(b)
Figure 11. (a) Linear shorted g /4 microstrip resonator. (b) Omni
microstrip antenna: a cylindrical /4 microstrip resonator. (After
Ref. 27.)
Microstrip wraparound antennas consisting of continuous
metal strips that wrap around missiles, rockets, and satellites
can provide omnidirectional coverage. Various forms of such
antennas are described in Refs. 3034. Munson (30) proposes
a continuous radiator for linear polarization, as shown in Fig.
15, which shows that the microstrip feed network is a parallel
(or corporate) feed network where two-way power splits are
equal phase to all of the feed points. The number of power
divisions can be 2, 4, 8, 16, etc. The specific number of feeds
and power divisions required is dictated by the microstrip radiator. The number of feed points NF must exceed the number
of wavelengths in the dielectric in the L direction (i.e., NF).
CONFORMAL ANTENNAS
207
Wrapped
on missle
15.2 cm
7.6 cm
0
0
+2
10 20 30
270
+2
dBi
30 20 10 90
270
Relative power
10 20 30
30 20 10 90
L = D
180
Feed
points
0
+2
;;;
Wrapped
to shape
dBi
Relative power
2 Elements
F = 660 MHz
180
+2
dBi
dBi
Microstrip
radiator
Feed
network
Input
Figure 15. Microstrip wraparound antenna. (After Ref. 30.)
270
30 20 10 90
10 20 30
Relative power
180
270
Relative power
180
2 Elements
F = 1210 MHz
0
+2
270
10 20 30
30 20 10 90
10 20 30
+2
dBi
30 20 10 90
Relative power
180
270
10 20 30
dBi
30 20 10 90
Relative power
180
4 Elements
F = 2000 MHz
Figure 13. Radiation patterns of a 7.6 cm DFES antenna mounted
on cylinder. (After Ref. 28.)
The following design relations can be used for the antenna
shown in Fig. 15:
= 0 / r
(2)
w = 0 = /2
2 r
(3)
L = D
(4)
L(r )1/2
LD =
0
(5)
NF > LD
(6)
with 0 = wavelength in free space
0
7 dBi
90
30 2
The pattern coverage of the omnidirectional antenna
shown in Fig. 15 depends on the diameter of the missile. A
typical measured E-plane pattern of a wraparound antenna
mounted on an 8 in. (203 mm) cylinder given in Ref. 31 is
reproduced in Fig. 16. The limiting factor in the omnidirectional coverage is a hole at the tip and tail of the missile that
gets narrower as the diameter of the missile increases.
Reference 33 studies radiation patterns of wraparound microstrip antenna on a spherical body for different radii of the
conducting sphere, frequencies, dielectric constant, and thickness of the dielectric. Specifically, the antenna studied consists of a metal strip of width d wrapped around a conducting
sphere of radius a covered with a dielectric substrate of chosen thickness d. A -symmetric transverse electromagnetic
mode of excitation is used. The parameter d is kept equal to
half a wavelength () inside the dielectric for constructive interference to occur in the broadside direction. The following
(7)
320
40
310
50
300
60
350
340 10
330 20
30
10
20
30
40
50
60
290
70
70
280
80
270
90
80
90
100
0 10
270
90
30 2
100
110
0 10
270
110
120
120
130
130
140
180
Elevation
pattern
E ( ) +0
180
F = 6330 MHz
Azimuthal
pattern
E ( ) 90
Figure 14. Radiation patterns of four-element edge-slot antenna
mounted on conical base (After Ref. 28.)
140
150
160
170 180 170
160
150
Figure 16. Measured E-plane pattern of the 8 in. (203 mm) wraparound microstrip antenna. The antenna pattern is a figure of revolution about the missile axis. (After Ref. 31.)
208
CONFORMAL ANTENNAS
comments summarize the findings of the investigation reported in Ref. 33:
the formation of grating lobes is given by
D
1. Radiation patterns are almost independent of the pressure of the patch when the radius of the sphere is much
larger than the strip width.
2. The larger the radius of the sphere (a 0), the better
the omnidirectional pattern.
3. The dielectric constant r does not have significant influence on the pattern shape. The radiation intensity
tends to increase with increase of the dielectric constant.
4. The shape of the radiation patterns remains almost unchanged for different substrate thickness (h) for h 0.
However, sidelobe levels increase with increase of r.
Radiation patterns of rectangular microstrip patches arrayed circumferentially on a circular cylinder (wraparound
array) have been computed in Ref. 34. Both axial and circumferential patches, using axial and circumferential modes of
excitation, respectively, have been used. In general, it has
been found that the number of circumferential patches required for a given ripple in the gain pattern is considerably
less than that required in the axial case, thus simplifying the
feed network for the former case. Results given in Ref. 34
compared favorably with reported measurements.
A Patch Array for Aircraft
A patch array designed for an aircraft to satellite communication link is described by Sanford (35) and is shown in Fig. 17.
Eight patches are mounted together with the phase shifting
and feeding circuitry to scan the beam in the elevation direction. Designed for operation at 1.5 GHz, the array, including
radome, is 3.6 mm thick. Element phasing was optimized for
maximum multipath rejection at low scan angles and to account for the curvature of the mounting surface. Each element was pointed in a different direction and has an inherent
phase error relative to the center elements. A digital computer was used to determine how the design parameters actually affect the performance of the array. The spacing of the
array elements must be greater than 0.32 (in free space) because the physical size of the radiating element on teflon fiberglass requires thin space. The spacing required to prevent
Phase shifting
circurity
Patch
radiators
Aircraft
body
Figure 17. Conformal array for aircraft application. (After Ref. 35.)
1 + sin
(8)
where D is the separation distance between the patch element, is the wavelength in free space, and is the maximum beam steering angle. For a 50 maximum steering
angle, D must be less than 0.57.
Concentric Microstrip Ring Arrays
Bhattacharryya and Garg (36) describe the design of a concentric annular ring microstrip antenna array that can be excited by means of a single feed by interconnecting two consecutive rings with an impedance transformer. The feasibility of
such an antenna is based on the observation that annular
rings with different mean radii can be designed to resonate
at the same frequency for the TM12 mode. An impedance
bandwidth of about 5% for VSWR 2 has been reported in
Ref. 36. It has been found possible to control the principal
plane patterns for concentric arrays independently of each
other by appropriately designing the feed system.
Saha-Misra and Chowdhury (37) describe electromagnetically fed concentric microstrip ring arrays using the log periodic principle that have been reported to have increased impedance and radiation pattern bandwidths. Specifically,
circular, square, and triangular concentric rings have been
investigated. Generally, these antennas work at multiple
bands of frequencies with some bands having larger bandwidths than standard microstrip antennas. With a nonuniformly spaced concentric annular ring array, almost 20%
bandwidth for VSWR 2 has been reported. A planar, wideband feed for a slot spiral antenna has been described by
Nurnberger and Volakis (38). The antenna has been developed for operation at very high frequency (VHF) frequencies.
In contrast to most traditional printed spiral antenna designs, the one reported in Ref. 38 incorporates a completely
planar spiral microstrip balun feed, thereby making it attractive for a variety of conformal applications.
CONFORMAL ARRAYS
Antenna arrays conforming to a nonplanar surface are suitable and may even be a requirement for a number of applications. For example, phased arrays of flush-mounted elements
conformally mounted on the surface of an aircraft or missile
reduce the aerodynamic drag and hence are preferable. Also,
in some cases a nonplanar array surface may provide some
natural advantage for broad-beam coverage in space. Spherical, cylindrical, and conical arrays have been developed for
ground, airborne, and missile applications. We shall consider
here the class of conformal arrays where the radiating surface
is nonplanar with a radius of curvature large compared to
the operating wavelength. Conformal arrays that are highly
curved are generally difficult to design because of the following reason (1,3,8,4):
1. Array elements point in different directions and so it is
often necessary to switch off those elements that radiate
primarily away from the desired direction of radiation.
CONFORMAL ANTENNAS
This, in turn, requires more sophisticated switching
mechanisms for activation of elements.
2. The fact that element patterns cannot be factored out
of the total radiation pattern makes the analysis and
synthesis of such antennas more complicated.
3. Mutual coupling effects can be very severe and difficult
to ascertain.
4. Nonplanar arrangement of elements may give rise to
severe cross-polarization effects.
Spherical Arrays
Certain applications require phased arrays capable of steering the beam over a complete hemisphere. For this requirement a spherical array surface seems to provide some natural
advantage for beam steering. Schrank (39) discusses the manner in which an array of radiating elements placed on a
sphere provides a natural configuration for obtaining hemispherical coverage with nearly identical highly directive
beams. A spherical phased array consisting of circularly polarized flat spiral antenna elements has been developed by
Sengupta, Smith, and Larson (40) and Sengupta, Ferris, and
Smith (41). Theoretical design and other considerations are
given in Ref. 42, and experimental fabrication and results are
given in Ref. 41. As described in Ref. 40, a special element
distribution was obtained from the consideration of icosahedron geometry resulting in a best possible uniformity of element spacing. It was found that the array could operate with
widely spaced elements. The special element distribution developed for this purpose considerably suppressed the grating
lobes in the pattern and thereby made the array significantly
broadband. Figure 18 shows the icosahedron geometry of element locations on a spherical surface. The choice of circularly
polarized elements made the antenna beam retain its circular
polarization fairly well over the entire range of beam steering
directions (39,40). Experimental results given in Ref. 41 demonstrate the capability of a spherical array of 16 flat spiral
antennas over a frequency range 0.6 to 3 GHz. The work reported in Ref. 41 used manual control of phase and illuminated aperture area; consequently, the results obtained were
limited in scope. However, with the availability of modern sophisticated computer control mechanisms, it seems that such
spherical arrays could provide almost complete hemispherical
phased coverage over a broad band.
Figure 18. Icosahedron geometry of element locations. (After Ref.
39.)
209
Cylindrical Arrays
Conformal elements like microstrips and slots arrayed around
the circumference of large metal cylinders have been used to
obtain omnidirectional pattern coverage. Such coverage may
also be obtained with the help of wraparound antenna, as discussed earlier. References 4246 show that an array of slots
equally spaced around the circumference of a cylinder can
produce a pattern with very low ripple. Croswell and Knop
(43) have obtained extensive numerical data using realistic
patterns for slots on perfectly conducting planes. In such
arrays the design parameters are the numbers of elements,
radiating elements, and feed network. The number of elements is chosen to provide a nearly omnidirectional pattern;
the minimum number of elements is decided by the allowed
amplitude ripple. The evaluation of the amplitude ripple can
be given in terms of the fluctuation, which is defined as the
ratio of maximum F to minimum F, where F is the total farfield pattern of an S-element circular array and is given by
(42)
F =S
N
An ( j)n
n=0
dn
[J (z) + 2( j)SJS (z) cos S]
dz n 0
(9)
where JS(z) is the Bessel function of the first kind of order S,
z k0a sin; k0 is the free-space wave number; a is the radius
of the circular array; , are the usual coordinates, and the
z-axis is the axis of the cylinder; and N S.
The preceding expression assumes that the single element
pattern f(1) can be expressed by a Fourier cosine series
f ( ) =
An cosn
(10)
n=0
A practical single element pattern can be approximated by
f ( ) =
(1 + cos /2)/2
(2 + 3 cos + cos 2 )/6
(11)
Pattern fluctuations as a function of size of cylinder and number of elements for the preceding two single-element patterns
are given in Ref. 43. Cylindrical phased arrays, where selected sections are illuminated to provide a beam in a certain
direction, are sometimes found advantageous to use for some
requirements. Sophisticated types of electronic switches for
such circular arrays are based on a concept originally proposed by Shelton (45) and developed by Sheleg (46). The antenna uses a Butler matrix-fed circular array with fixed phase
shifters to execute current modes around the array and variable phase shifters to provide continuous scanning of the radiated beam over 360. The operation was experimentally demonstrated with a 32-dipole circular array.
The principles involved in scanning a multimode array are
readily seen by considering a continuous distribution of current, as described by Sheleg (47). Figure 19 shows the configuration of a continuous cylindrical sheet of vertical current
elements around a vertical conducting cylinder of radius a.
Referring to Fig. 19, consider a current distribution I() to be
the sum of a finite number of continuous current modes Ine jn
with N n N. The radiation pattern for I() Ine jn is
210
CONFORMAL ANTENNAS
then given by
E() =
N
Cn ein
(12)
n=N
Equal line
lengths
where Cn are complex constants given by
Cn = 2K j n In Jn
2a
with K a constant, being the wavelength of operation, and
Jn being the Bessel function defined earlier. If, in the antenna
being considered, it is desired that the pattern mode be equal
in magnitude and be in phase at 0, the excitation of the
current modes must be
In =
2K j n Jn
Butler matrix
(13)
2a
Variable phase
shifters
Fixed phase
N shifters
N1
Power divider
(14)
Figure 20. Schematic diagram of scanning multimode array. (After
Ref. 46.)
Under this condition, the radiation pattern is given by
E() =
N
2N + 1
e jn = sin
n=N
2
sin
(15)
If the phase difference between the adjacent modes is 0 (i.e.,
multiply In by ejn], the resultant radiation pattern is
E() =
sin[(2N + 1)( 0 )/2]
sin[( 0 )/2]
(16)
which indicates a beam in the 0 direction. As described by
Sheleg (46), it was possible to excite simultaneously and independently all the modes both positive and negative n, from 0
to N/2, by connecting a single ring on N elements to the outputs of a Bulter matrix. A schematic diagram of a scanning
J = N1 x
J=N x
= 0
J=1
x
J=2
multimode array is shown in Fig. 20. The desired phase and
amplitude distribution is established over the inputs of the
Bulter matrix by fixed phase shifts and a corporate structure.
Once the pencil beam pattern is formed at some azimuth
angle, it is scanned just as in a linear array; the mode amplitudes are held fixed and a linear phase progression is set up
on the mode inputs by operating the variable phase shifters.
Conical Array
Kummer (1) discusses a number of difficulties associated with
antenna pattern synthesis utilizing conical surfaces. An array
on a conical surface generally looks different at different aspect angles; also, the geometry is such that all elements do
not contribute equally to the main beam direction, thereby
causing cross-polarization problems. In spite of this, for their
obvious applications to missile and other similar vehicles,
conical arrays have been considered for conformal array development. Theoretical and experimental investigations of
various aspects of conical arrays are discussed in Refs. 15, 47,
and 48. The experimental studies of Munger 48) provide some
data on the characteristics of several conical arrays. Balzano
and Dowling (47) developed an effective method to evaluate
the pattern of elements in a conical array. The method takes
into account the mutual coupling between array elements and
aperture matching conditions. By properly matching the
array aperture, the radiation in a certain direction can be
substantially increased, thus allowing the designer to meet
specific design goals in the application of conical arrays to airborne or missile-borne systems. Moreover, it has been shown
that in some cases, the element pattern can be approximated
by much simpler planar and cylindrical models.
Figure 19. Coordinates for continuous cylindrical sheet of vertical
current elements.
BIBLIOGRAPHY
1. W. H. Kummer, Preface, IEEE Trans. Antennas Propag., AP-22:
13, 1974.
2. C. A. Balanis, Antenna Theory, 2nd ed., New York: Wiley, 1997.
CONFORMAL ANTENNAS
3. R. C. Johnson and H. Jasik, (eds.), Antenna Engineering Handbook, 2nd ed., New York: McGraw-Hill, 1984.
4. R. C. Hansen, Significant Phased Array Papers, Dedham, MA:
Artech House, 1973.
5. Y. T. Lo and S. W. Lee, (eds.), Antenna Handbook: Theory, Applications, and Design, New York: Van Nostrand Reinhold, 1988.
6. R. J. Mailloux, Phased Array Antenna Handbook, Boston: Artech
House, 1994.
7. W. H. Kummer, guest editor, Special Issue on Conformal Arrays,
IEEE Trans. Antennas Propag., AP-22: 1150, 1974.
8. R. C. Hansen, Microwave Scanning Antennas, Vols. 1-III, New
York: Academic Press, 1964.
9. P. H. Pathak and R. G. Kouyoumjian, An analysis of the radiation from aperture in curved surfaces by the geometrical theory
of diffraction, Proc. IEEE, 62: 14381447, 1974.
10. P. H. Pathak et al., A uniform GTD solution for the radiation
from sources on a curvex surface, IEEE Trans. Antennas Propag.
AP-29: 602622, 1981.
11. W. D. Burnside, R. J. Marhefka, and C. L. Yu, Roll plane analysis
of on-aircraft antennas, IEEE Trans. Antennas Propag., AP-21:
780786, 1973.
12. W. D. Burnside et al., A Study of KC-135 aircraft antenna patterns, IEEE Trans. Antennas Propag., AP-23: 309316, 1975.
13. R. J. Mailloux, Conformal and low-profile arrays, in Ref. 3.
14. K. E. Golden, G. E. Stewart, and D. C. Pridmore-Brown, Approximation techniques for the mutual admittance of slot antennas on
metallic cones, IEEE Trans. Antennas Propag., AP-22: 4348,
1974.
15. K. E. Golden and G. F. Stewart, Self and mutual admittance of
rectangular slot antennas in the presence of inhomogeneous
plasma layer, IEEE Trans. Antennas Propag., AP-17: 763771,
1969.
16. D. C. Chang, (ed.), IEEE Trans. Antennas Propag., AP-29: 1
182, 1981.
17. I. J. Bahl and P. Bhartia, Microstrip Antennas, Norwood, MA:
Artech House, 1980.
18. C. M. Krowne, Cylindrical-rectangular microstrip antenna, IEEE
Trans. Antennas Propag., AP-31: 194199, 1983.
19. K.-M. Luk, K.-F. Lee, and J. S. Dahele, Analysis of the cylindrical-rectangular patch antenna, IEEE Trans. Antennas Propag.,
AP-37: 143147, 1989.
20. K. R. Carver and J. W. Mink, Microstrip antenna technology,
IEEE Trans. Antennas Propag., AP-29: 224, 1981.
21. K.-L. Wong and S.-Y. Ke, Cylindrical-rectangular microstrip for
circular polarization, IEEE Trans. Antennas Propag., AP-41:
246249, 1993.
22. T. Kashiwa, T. Onishi, and I. Fukai, Analysis of microstrip antennas on a curved surface using the conformal grids FD-TD method,
IEEE Trans. Antennas Propag., AP-42: 423427, 1994.
23. J. M. Jin et al., Calculation of radiation patterns of microstrip
antennas on cylindrical bodies of arbitrary cross section, IEEE
Trans. Propag., AP-45: 126132, 1997.
24. J. R. Descardeci and A. J. Giarola, Microstrip antenna on a conical surface, IEEE Trans. Antennas Propag., AP-40: 460463,
1992.
25. H. S. Jones, Design of dielectric-loaded circumferential slot antennas of arbitrary size for conical and cylindrical bodies, Report
HDL-TR-1684, Hary Diamond Laboratories, Adelphi, MD, 1974.
26. D. H. Schaubert, A. R. Sindons, and F. G. Farrar, The spiral slot:
a unique microstrip antenna, Proc. 1978 Antenna Applic. Symp.,
Univ. of Illinois Allerton Conf., Hary Diamond Laboratories, October 1978.
211
27. A. D. Krall et al., The omni microstrip antenna: a new small
antenna, IEEE Trans. Antennas Propag., AP-27: 850853,
1979.
28. D. H. Schaubert, H. S. Jones, and F. Reggia, Conformal dielectric-filled edge-slot antennas with inductive-post turning, IEEE
Trans. Antennas Propag., AP-27: 713716, 1979.
29. D. L. Sengupta and L. F. Martins-Camelo, Theory of dielectricfilled edge-slot antennas, IEEE Trans. Antennas Propag., AP-28:
481490, 1980.
30. K. F. Munson, Conformal microstrip antennas and microstrip
phased arrays, IEEE Trans. Antennas Propag., AP-22: 7478,
1974.
31. R. F. Munson, Omnidirectional microstrip arrays, pp. 7-19, 7-21
of Ref. 3.
32. S. B. D. A. Fonseca and A. J. Giarola, Analysis of microstrip
wraparound antennas using dyadic Greens functions, IEEE
Trans. Antennas Propag., AP-31: 248253, 1983.
33. A. Das, S. K. Das, and M.-S. Narasimhan, Radiation characteristics of wraparound microstrip antenna on spherical body, IEEE
Trans. Antennas Propag., AP-39: 10311034, 1991.
34. I. Jayakumar et al., A conformal cylindrical microstrip array for
producing omnidirectional radiation pattern, IEEE Trans. Antennas Propag., AP-34: 12581261, 1986.
35. G. G. Sanford, Conformal microstrip phased array for aircraft
tests with ATS-6, IEEE Trans. Antennas Propag., AP-26: 642
646, 1978.
36. A. K. Bhattacharyya and R. Garg, Input impedance of annular
ring microstrip antenna using circuit theory approach, IEEE
Trans. Antennas Propag., AP-33: 369374, 1985.
37. I. Saha-Misra and S. K. Chowdhury, Concentric microstrip ring
antenna: theory and experiment, J. Electromagn. Wave Applications, 10: 439450, 1996.
38. M. W. Nurnberger and J. L. Volakis, A new planar feed for slot
spiral antennas, IEEE Trans. Antennas Propag., AP-44: 130
131, 1996.
39. H. E. Schrank, Basic theoretical aspects of spherical phased
arrays, in A. A. Oliner and G. H. Knitted (eds.), Phased Array
Antennas, Dedham, MA: Artech House, pp. 323327, 1972.
40. D. L. Sengupta, T. M. Smith, and R. W. Larson, Radiation characteristics of spherical array of circularly polarized elements,
IEEE Trans. Antennas Propag., AP-16: 17, 1968.
41. D. L. Sengupta, J. E. Ferris, and T. M. Smith, Experimental
study of a spherical array of circularly polarized elements, Proc.
IEEE, 56: 20482051, 1968.
42. T. S. Chu, On the use of uniform circular arrays to obtain omnidirectional patterns, IEEE Trans. Antennas Propag., AP-7: 436
438, 1959.
43. W. F. Croswell and C. M. Knop, On the use of an array of circumferential slots on a large cylinder as a omnidirectional antenna,
IEEE Trans Antennas Propag., AP-14: 394396, 1966.
44. W. F. Croswell and C. R. Cockrell, An omnidirectional microwave
antenna for use on spacecraft, IEEE Trans. Antennas Propag.,
AP-17: 459466, 1969.
45. P. Shelton, Application of hybrid matrices to various multimode
and multibeam antenna systems, IEEE Washington Chapter P
GAP Meet., March 1965.
46. B. Sheleg, A matrix-fed circular array for continuous scanning,
Proc. IEEE, 56: 20162027, 1968.
47. Q. Balzano and T. B. Dowling, Mutual coupling analysis of arrays
of apertures on cones, IEEE Trans. Antennas Propag., AP-22:
9297, 1974.
48. A. D. Munger et al., Conical array studies, IEEE Trans. Antennas
Propag., AP-22: 3542, 1974.
DIPAK L. SENGUPTA
University of Detroit Mercy
CONICAL ANTENNAS
221
P(r, , )
CONICAL ANTENNAS
1
H e j r
H =
r sin 0
(1)
E = H
(2)
where 2/ (: wavelength) is the free-space wave number and 0 / 0 is the free-space wave impedance. The
characteristic impedance K of the biconical transmission line
is given by the ratio of the transmission voltage (i.e., the integral of E along the cone-meridian) to the conduction current
Figure 1. Infinite biconical conductor fed by a -gap generator.
along the cone as follows:
K = ln cot
(3)
Figure 2 shows the characteristic impedance K versus the
half-cone angle .
BICONICAL ANTENNAS
Figure 3 shows the geometry of the biconical antenna. The
conducting antenna surfaces are defined by the conical surfaces and , and the two spherical end surfaces
at r a. The analytical procedure of the biconical antenna
will be outlined below. In region I, the electric and magnetic
fields are represented as a sum of the outward- and inwardtraveling TEM principal modes and an infinite number of
complementary (higher) transverse magnetic (TM) modes. In
region II, the fields are represented by an infinite series of
complementary radiating modes. Boundary conditions on the
aperture indicated by the dashed lines in Fig. 3 and the end
800
Characteristic impedance K ( )
This article describes characteristics such as input impedance, radiation pattern, and directivity gain of conical antennas consisting of solid conducting cones, conducting conical
plates, or their modifications. Conical conducting structures
on which simple antenna elements such as dipole and a slot
are mounted are also described. An important general feature
of conical antennas is their lack of sensitivity to frequency
variation, that is, their broadband characteristics. Note that
a conical horn antenna is described in HORN ANTENNAS.
The history of the practical use of conical antennas is long.
Sir Oliver Lodge constructed a biconical antenna in 1897 and
made a wireless communication experiment, while a single
cone antenna on the ground and a fan (flat triangular) antenna were used by Marconi and others. The history of the
theory of conical antennas is also long. The spherical coordinate is one of the few coordinates for which the method of
variable separation in electromagnetic field problems can be
applied. The conical surface is defined by a constant polar
angle in the spherical coordinate system. For this reason conical antennas and biconical antennas, in particular,
have been extensively investigated by Schelkunoff (1,2),
Smith (3), Tai (4,5), Papas and King (6), and many others.
In the limit as 0 and , the biconical antenna is reduced to a vanishingly thin linear antenna that is very sensitive to frequency (that is, of narrow bandwidth). As the cone
angle increases, the antenna shows broadband characteristics
that make it useful for practical applications.
Figure 1 shows an infinitely long, symmetric biconical conductor with a half-cone angle and an infinitesimally small
feeding gap ( gap). The antenna is assumed to be located in
free space. This structure can support the transverse electromagnetic (TEM) transmission line mode, that is, the outwardpropagating principal spherical wave mode expressed by
700
600
K = 120 In cot
500
400
300
200
100
0
10
20
30
40
50
60
Half-cone angle (deg)
70
80
Figure 2. Characteristic impedance of a biconical transmission line.
J. Webster (ed.), Wiley Encyclopedia of Electrical and Electronics Engineering. Copyright # 1999 John Wiley & Sons, Inc.
222
CONICAL ANTENNAS
400
2 = 5
Resistance ( )
300
2
h
10
200
20
30
40
100
40 50 60
90
30
20
5
50
60
70
80
90
70
10
20
5
30
40
50
60
80
90
0
0
Figure 3. Symmetric biconical antenna.
50
100
150
200
Antenna length h (deg)
250
300
Figure 5. Measured input resistance of a conical unipole versus
length in electrical degrees showing broadband characteristics with
increasing cone-angle (from 10).
driven by a coaxial line has been numerically analyzed by using the finite difference time domain method (9).
When the upper half-cone of the biconical antenna is
mounted on an infinite conducting plane (ground plane), the
antenna forms a conical unipole having one half of the input
impedance of the biconical antenna. Figures 5 and 6 [Brown
and Woodward, Jr. (10)] show respectively the measured input resistance and reactance of a conical unipole having flat
caps instead of spherical caps. It is clear that the antenna
tends to have a constant input resistance and a small reactance around zero versus frequency, showing broadband
characteristics as the cone-angle is increased.
The radiation pattern of the biconical antenna has been
computed by Papas and King (6) and by Bevensee (11). Figure
7 shows the far-zone electric field pattern (6) for the cone
angle of 30. It is found that the patterns are not much
different from those of a straight wire antenna.
Theoretical analysis of biconical antennas loaded with
and/or immersed in dielectric, lossy, and ferromagnetic mate200
2
h
100
Resistance ( )
surfaces of the cone at r a are used to obtain an infinite set
of linear algebraic equations from which the amplitudes of
the complementary modes and the principal mode reflected
back at the aperture are determined.
The input admittance of the biconical antenna is represented by the equivalent transmission line circuit as shown
in Fig. 4, where K is the characteristic impedance given by
Eq. (3). The terminal admittance Yt represents the effect of
the truncation of the biconical transmission line at r a, that
is, the transformation of the outward-traveling TEM mode
into the complementary modes in both regions and the reflected TEM mode, which eventually determines the input admittance of the biconical antenna Yi.
Schelkunoff (2) has formulated the above boundary value
problem rigorously and has discussed in detail special cases
of a vanishingly thin antenna and a very wide-angle cone, or
a spherical antenna with a very narrow equatorial gap. Tai
(4) has obtained the exact analytical solution of the terminal
admittance of the vanishingly thin antenna, which has been
found to be identical to the expression obtained ingeniously
by Schelkunoff. Tai (5) has also made an important contribution to the development of the theory for biconical antennas
by applying Schwingers variational method. He has given the
first order variational numerical solution for the specific wide
cone-angles 39.23, 57.43, and 66.06.
The recent development of computers has made feasible
the numerical solution of Schelkunoff s formulation. However, it is still not easy to solve the infinite set of linear determining equations with reasonable accuracy because of slow
convergence of the infinite series when the cone-angle decreases. For example, 15 or more modes for 5 (7), and 13
modes for 5 (8) are necessary for computation of the
input impedance. A conical monopole above an image plane
2 = 5
10
20
30
40
50
60
90
90
90
70
80
60
50
70 40
80
60
50
40
30
20 10
30
5 90
60
50
40
20
30
100
10
20
10
5
200
Yi
Yt
a
Figure 4. Equivalent circuit of a symmetric biconical antenna.
300
50
100
150
200
Antenna length h (deg)
250
300
Figure 6. Measured input reactance of a conical unipole versus
length in electrical degrees showing broadband characteristics with
increasing cone-angle (from 10).
CONICAL ANTENNAS
60
60
ka = 2
ka = 1.5
ka 1
(a)
0.25 0.5 0.75 1.0 1.25
223
70
70
80
80
90
(a)
0.25 0.5 0.75 1.0 1.25
90
(b)
60
ka = 3
ka = 2.5
60
ka = 4
ka = 3.5
70
80
0.25 0.5 0.75 1.0 1.25
90
(c)
70
80
0.25 0.5 0.75 1.0 1.25
90
(d)
60
60
ka = 6
ka = 5.5
ka = 5
ka = 4.5
0.25 0.5 0.75 1.0 1.25
70
70
80
80
90
(e)
0.25 0.5 0.75 1.0 1.25
(f)
rials has been provided by Schelkunoff (2), Tai (4), Polk (12),
and others. These topics are reviewed by Wait (13). The theory of an asymmetric biconical antenna was also discussed by
Schelkunoff (2). The variational approach by Tai was extended to a semi-infinite asymmetric conical antenna consisting of an infinite cone and a finite cone (14).
To reduce wind resistance and/or weight, a solid biconical
antenna can be replaced by a skeletal conducting wire structure using several radial rods (15,16). It has been found, however, by analysis using the moment method that a considerable number of wires (e.g., 16) is required to approximate the
solid biconical antenna.
ANTENNAS ON CONICAL STRUCTURE
In practice, a conical structure on which simple antenna elements such as dipole, disk, cone, slot, or patch are mounted
to excite a cone is often used. The infinite cone excited with
an axial dipole at the tip has shown (14) that the strong radiation occurs along the small-angle cone unless the dipole
length is about a half-wavelength. Figure 8 shows a finite
wide-angle cone excited with a quarter-wavelength long and
a half-wavelength long dipole (17). Note that the maximum
radiation can be directed toward the horizontal plane by
proper choice of a and 0 at a desired frequency. A cone excited with an axially symmetric circumferential slot close to
the tip (18) shows radiation characteristics similar to those of
a dipole-excited cone, since a small circumferential slot (mag-
90
Figure 7. Far-zone electric field patterns of a biconical antenna, 30. Patterns do not change
very much from ka 0 to ka .
netic current loop) is equivalent to a small axial electric dipole. When the circumferential slot is not too close to the tip,
that is, apart by 2.5 wavelength (2 30), the radiation pattern shows a rather complicated lobe structure (18). The cone
excited with a circular disk at the tip is called a discone antenna (19). This antenna is fed with a coaxial cable whose
inner conductor terminates on the center of the disk and
whose outer conductor terminates on the tip of the cone. The
radiation pattern of the discone is similar to that of a dipole
antenna, but its input impedance bandwidth is exceedingly
broad compared with an ordinary dipole antenna.
Figure 9 shows the measured input VSWR (for 57.6 cable) versus frequency of the finite conical antenna with different exciting elements at its tip (17). The cone excited with a
conical element (an asymmetric biconical antenna) indicates
very broadband characteristics, particularly in lower-frequency regions, without affecting the radiation pattern.
TRIANGULAR (BOW-TIE) ANTENNAS
A triangular plate antenna above a conducting ground plane
and a bow-tie antenna are shown in Figs. 10(a) and 10(b).
These antennas also possess broadband characteristics,
though not as broad as a solid conical antenna. The theoretical characteristics of the bow-tie antenna have been obtained
numerically (20) by using the method of finite difference time
domain (FDTD). Figures 11(a) and 11(b) show the calculated
input impedance. The input impedance of the triangular plate
224
CONICAL ANTENNAS
o
a
2
2
(a)
h
2h
(b)
(a)
(b)
Figure 10. (a) Triangular plate antenna and (b) bow-tie antenna.
(c)
Figure 8. Far-zone electric field pattern of a tip-excited conical antenna, a 50. (a) 0 90, l /4; (b) 0 103.8, l /4; (c) 0
103.8, l /2. Radiation beam can be directed to horizon by slanting
a cone downward (from 17).
7
0.635 cm
6
Thin
polystyrene
disk
Cone excited with
a cylinder element
Disk excited with a
cylindrical element
(dia of disk = dia
of cone base)
VSWR
5
0.238 cm
L = 7.45 cm
L = 0.835 cm
Cone excited with
a conical element
0.90 cm
0.835 cm
0.238 cm
2
Figure 9. VSWR versus frequency of a
disk and a cone excited with a cylindrical
element or a conical element showing that
a cone excited with a conical element is
most broadband (from 17).
Frequency (GHz)
10
11
12
13
CONICAL ANTENNAS
225
90
60
120
0.8
0.6
30
700
= 15
= 45
600
0.4
= 30
= 60
0.2
Resistance ( )
500
0
400
300
200
210
330
100
0
0.5
1
1.5
Length of bow-tie antenna 2h ( )
2h = 0.5
2h =
2h = 1.5
(a)
(a) x-y plane
400
= 15
= 45
Resistance ( )
200
300
270
= 30
= 60
30
30
0.8
0.6
60
0
0.4
200
0.2
90
400
600
0
0.5
1
1.5
Length of bow-tie antenna 2h ( )
(b)
Figure 11. Input impedance of a bow-tie antenna showing the broadband characteristics with increasing flare angle (from 20).
120
2h = 0.5
2h =
2h = 1.5
150
180
(b) x-z plane
Figure 12. Far-zone electric field pattern of a bow-tie antenna,
30 (from 20).
226
CONSTRUCTIVE LEARNING AND STRUCTURAL LEARNING
12. C. Polk, Resonance and supergain effects in small ferromagnetically or dielectrically loaded biconical antennas, IRE Trans. Antennas Propag. 7 (Special Supplement): 414423, 1959.
= 15
= 45
Directivity gain (dBi)
= 30
= 60
4
2
0
2
4
0.2
0.2
0.2
0.2
0.2
0.2
0.2
Length of bow-tie antenna 2h ()
0.2
Figure 13. Directivity gain of a bow-tie antenna in x-direction
(from 20).
antenna above the ground plane is half of that of the bowtie antenna.
The far-zone electric field patterns in the x-y plane and in
the x-z plane are shown in Figs. 12(a) and 12(b), respectively.
Note that the radiation is enhanced in the direction perpendicular to the antenna plate for the antenna length 2h ,
because the radiation from the antenna surface current is
added in phase in that direction. The theoretical directivity
gain of the bow-tie antenna in the direction of the x-axis is
shown in dBi in Fig. 13 versus the antenna length 2h/ for
various cone angles (20). It is noted here that the bow-tie antenna can also be simulated by several radial wire rods as the
solid biconical antenna.
BIBLIOGRAPHY
1. S. A. Schelkunoff, Principal and complementary waves in antennas. Proc. IRE, 34 (1): 2332, 1946.
2. S. A. Schelkunoff, Advanced Antenna Theory, New York: Wiley,
1952.
3. P. D. P. Smith, The conical dipole of wide angle, J. Appl. Phys.,
19 (1): 1123, 1948.
4. C. T. Tai, On the theory of biconical antennas, J. Appl. Phys., 19
(12): 11551160, 1948.
5. C. T. Tai, Application of variational principle to biconical antennas, J. Appl. Phys., 20 (11): 10761084, 1949.
6. C. H. Papas and R. W. P. King, Radiation from wide-angle conical
antennas fed by a coaxial line, Proc. IRE, 39 (1): 4951, 1951.
7. S. A. Saoudy and M. Hamid, Input admittance of a biconical antenna with wide feed gap, IEEE Trans. Antennas Propag. 38 (11):
17841790, 1990.
8. V. Badii, K. Tomiyama, and D. M. Grimes, Biconical transmitting
antennas, a numerical analysis, Appl. Comput. Electromagn. Soc.
J. 5 (1): 6293, 1990.
9. J. G. Maloney, G. S. Smith, and W. R. Scott, Jr., Accurate computation of the radiation from simple antennas using the finite-difference time-domain method, IEEE Trans. Antennas Propag., 38
(7): 10591068, 1990.
10. G. H. Brown and O. M. Woodward, Jr., Experimentally determined radiation characteristics of conical and triangular antennas, RCA Rev., 13 (4): 425452, 1952.
11. R. M. Bevensee, Handbook of Conical Antennas and Scatterers,
New York: Gordon and Breach, 1973.
13. J. R. Wait, Electromagnetic radiation from conical structures, In
R. E. Collin and F. J. Zucker (eds.), Antenna Theory, New York:
McGraw-Hill, 1969.
14. S. Adachi, A theoretical analysis of semi-infinite conical antennas, IEEE Trans. Antennas Propag., 8: 534547, 1960.
15. C. E. Smith, C. M. Butler, and K. R. Umashanker, Characteristics of a wire biconical antenna, Microw. J., 22 (9): 3740, 1979.
16. O. Givati and A. P. C. Fourie, Analysis of skeletal wire conical
antennas, IEEE Trans. Antennas Propag., 44: 844858, 1996.
17. S. Adachi, R. G. Kouyoumjian, and R. G. Van Sickle, The finite
conical antenna, IEEE Trans. Antennas Propag., 7 (Special Supplement): S406S411, 1959.
18. L. L. Bailin and S. Silver, Exterior electromagnetic boundary
problem for sphere and cones, IRE Trans. Antennas Propag. 4 (1):
516, 1956, and Corrections, 4 (3): 313, 1957.
19. A. G. Kandoian, Three new antenna types and their applications,
Proc. IRE, 70W75W, 1946.
20. Private communication from Y. He, T. Uno, and S. Adachi, 1997.
SABURO ADACHI
Tohoku Institute of Technology
DIELECTRIC-LOADED ANTENNAS
279
DIELECTRIC-LOADED ANTENNAS
A transmit antenna converts the energy of a guided wave in
a transmission line into the radiated wave in an unbounded
medium. The receive antenna does the reverse. The transmission lines such as waveguides, coaxial lines, and microstrip
lines use conductors mostly to confine and guide the energy,
but antennas use them to radiate it. Because the radiated
energy is in an unbounded region, phase control is often used
to direct the radiation in the desired direction. Dielectrics
play an important role in this process, and this article discusses a few representative cases. An important antenna parameter is its directivity, which is the measure of its control
over the energy flow. To increase the directivity the antenna
size must be increased, and the influence of dielectrics on
their performance changes considerably. Thus, in this article,
the use of dielectrics in antenna applications is divided into
two categories: large high-gain antenna applications and
small low-gain antenna applications.
In high-gain antenna applications, reflectors and lenses
are used extensively (1). They operate principally on the basis
of their geometry. Consequently, they are relatively low cost,
reliable, and wideband. Reflectors are usually made of good
conductors, and thus have lower loss, and because of their
J. Webster (ed.), Wiley Encyclopedia of Electrical and Electronics Engineering. Copyright # 1999 John Wiley & Sons, Inc.
280
DIELECTRIC-LOADED ANTENNAS
high strength they can be made light. However, reflectors suffer from limited scan capability. Lenses, on the other hand,
because of transparency, have more degrees of freedomthat
is, two reflecting surfaces and the relative permittivity or refractive index. They also do not suffer from aperture blockage.
However, lenses have disadvantages in large volume and
weight.
In microwave antenna applications, lenses have numerous
and diverse applications, but in most cases they are large
with respect to the wavelength. Thus, physical and geometrical optics apply, and most of the lens design techniques can be
adopted from optics to microwave applications. The aperture
theory and synthesis techniques can also be used effectively
to facilitate designs. In addition, the use of optical ray path
in lens design makes the solution frequency-independent. In
practice, however, the lens size in microwave frequencies is
finite with respect to the wavelength, and the feed antenna is
frequency-sensitive. Thus, the performance of the lens antenna also becomes frequency-dependent.
Natural dielectrics at microwave frequencies have reflective indices larger than unity, and for collimation they need
convex surfaces. However, artificial media using guiding
structures, such as waveguides, are equivalent to dielectrics
with refractive index less than unity, and they result in concave lenses. They are usually dispersive, resulting in variation of the refractive index with frequency, and have narrower operating bandwidths.
In small antennas, dielectrics are used often to improve
the radiation efficiency and polarization of other antennas,
such as waveguides and horns. This is important in telecommunication applications where polarization control is required to implement frequency reuse and minimize interference, especially in satellite and wireless communications.
Horn antennas and reflector feeds are examples that incorporate dielectrics or lens loading to improve performance (2).
Another area of important dielectric use is insulated antennas, in biological applications and remote sensing with
buried or submerged antennas. The use of dielectric loading
eliminates direct radio-frequency (RF) energy leak into the
lossy environments, and it ensures radiative coupling into the
target objects. Often a full wave analysis is needed to provide
a proper understanding of resonance property and coupling
mechanism to the surrounding media.
Finally, the antenna miniaturization depends primarily on
the dielectric loading. Low-loss dielectrics with medium to
high relative permittivities are now available and are used
increasingly to reduce the antenna size. A number of important areas include dielectric-loaded waveguides and horns, dielectric resonator antennas, and microstrip antennas. By aperture loading of waveguides and small horns, excellent
pattern symmetry and low cross-polarization can be obtained,
which are essential features of reflector and lens feeds. In addition, the dielectric loading reduces their size and makes
them useful candidates for multiple beam applications, using
reflectors and lenses. Miniaturization of the antenna is also
an important requirement in wireless communications. Microstrip patch or slot antennas with high relative permittivity
substrates play an important role in this area, and their derivatives are used in most applications.
DIELECTRIC LENS ANTENNAS
In optical terms a lens produces an image of a source point
at the image point. These points could be located anywhere
Feed antenna
n>1
Concave lens
n = r
Feed antenna
n>1
Concave lens
Figure 1. Geometry of lens antennas.
in space. For an antenna, this property means that the
source and image points are focused at each other and the
lens has two focal points. In turn, these focal points signify
locations in space, where rays emanating from the lens
arrive at equal phases. This property provides a mathematical relationship for describing the lens operation and,
hence, its design.
To simplify the mathematics, the lens configuration is assumed to be rotationally symmetric, and the focal points are
placed on its axis. A further simplification can be made for
antenna applications, where the image point moves to infinity. That is, the lens focuses a nearby source point, on its axis,
to another axial point at infinity. In such a case, all rays leaving the lens travel parallel to its axis, and their phase fronts
are planes normal to the lens axis. This is shown in Fig. 1,
where r is the relative permittivity of the lens material and
n r is its refractive index.
To design the lens, one needs to determine the geometry of
its two faces, front and back, or the coordinates x1, y1, and x2,
y2 of points P1 and P2 (Fig. 2). There are four unknowns to be
determined. The equality of the phase on the phase fronts
requires that the electrical length between the focal points
and the phase fronts be independent of the path lengths. This
provides one equation. Two other equations can be obtained
from the ray optics at the lens interface points P1 and P2,
namely, Fermats principle of minimum path lengths. This
enforces the well-known Snells law of refraction at the lens
surface points. An additional relationship must be generated
from the required lens properties, to enable a unique solution
for the lens design.
To enforce the invariance of the ray path length, the central ray passing through points A, B, and C is selected as the
DIELECTRIC-LOADED ANTENNAS
Equations (2), (5), and (6) are three fundamental equations
used to design the required lens. Without another relationship, x1 may be selected as the independent variable. Then
others (i.e, x2, y1, and y2) become dependent variables to be
determined in terms of x1. The solutions give the lens profiles
in rectangular coordinates. If the lens profiles in polar coordinates are required, Eqs. (2), (5) and (6) can be obtained in
terms of r1, 1, and r2, 2, the polar coordinates of P1 and P2.
Differentiating Eqs. (2) in terms of 1 and 2 gives
Phase
front
y
P1(x1, y1)
(r1, 1)
r3
P2(x2, y2)
(r2, 2)
L1
P3
r1
r2
2
L0
A
(x3, 0)
281
dr1
nr1 r2 sin(2 1 )
=
d1
r3 n[r2 cos(2 1 ) r1 ]
(7)
dr2
nr1 r2 sin(2 1 ) + r2 r3 sin 2
=
d2
r3 sin 2 n[r2 r1 cos(2 1 )]
(8)
and
F
S1
S2
Figure 2. Geometry of a lens indicating ray and surface coordinates.
where use is made of the following polar coordinate relationships:
x1 = r1 cos 1
reference, and its length from S to C is compared with that of
the ray passing through points P1, P2, and P3. This provides
the following equation:
SP1 + nP1 P2 + P2 P3 = SA + nAB + BC
(1)
or
r1 + nr3 + L1 = F + nT + L0
(2)
where in terms of P1 and P2 coordinates each length is given
by
L1 = (x3 x2 )
(3)
L0 = x3 (F + T )
and F and T are the lens focal length and axial thickness and
are therefore constant lengths defining the lens.
Enforcing Fermats principle at points P1 and P2 results in
differentiation of the path length in Eq. (1) in terms of its
variables x1, y1 and x2, y2 and setting it to zero. This provides
the slope of the lens surface profiles at each point P1 and
P2.
At point P1, one obtains
d
d
d
[r + nr3 + L1 ] =
[F + nT + L0 ] =
L
dx1 1
dx1
dx1 0
(4)
where F and T are constants. After simplification, one
obtains
dy1
x r (x2 x1 ) nr1
= 1 3
dx1
(y2 y1 ) nr1 y1 r3
(5)
At point P2, a similar differentiation in terms of x2 gives
dy2
(x x1 )n r3
= 2
dx2
(y2 y1 )n
x2 = r2 cos 2
(9)
y2 = r2 sin 2
r3 = |r1 r2 | = [r21 + r22 2r1 r2 cos(2 1 )]1/2
Solutions of Eqs. (7) and (8) give the lens profiles in polar
coordinates, which are often more compact in form. Also, for
some simple lens configurations they result in well-known
and easily recognizable parametric equations of the conic sections, generalizing the solution.
Examples, Simple Lenses
r1 = (x21 + y21 )1/2
r3 = [(x2 x1 )2 + (y2 y1 )2 ]1/2
y1 = r1 sin 1
(6)
The lens design becomes considerably easier if one of its surfaces is predetermined. This eliminates one of the differential
equations, as the surface profile is already known. The planar
and spherical surfaces are among the simpler surfaces to select. The planar surfaces are normal to the lens axis. Such
selections give simple profile equations. The planar surface is
described by a constant x coordinate, and the spherical one is
described by a constant polar coordinate r. These simplifications also assist in solutions of the other lens profile, for
which an analytic solution can also be determined. Since either of the lens profiles can be predetermined as planar or
spherical, four possible solutions exist. Only two, however, result in simple conical sections.
If the second surface S2 is assumed to be planar and normal to the lens axis, the rays arriving from the right-hand
side, parallel to the lens axis x, enter the lens unaffected and
change direction only after the first lens surface S1. Then they
focus at S. That is, only the S1 surface of the lens collimates
the beam. Looking from the left side, spherical rays originating from the focal point S enter the lens S1 and become parallel to its axis. Thus, after leaving the lens at S2, since they
are normal to S2, their direction remains unchanged. In this
case, the active surface S1 of the lens is a hyperbola in a cylindrical lens but is hyperboloid in a rotationally symmetric lens.
If the surface S1 is spherical, it becomes inactive, since the
focal point is a point source and rays emanating from it constitute spherical waves. Thus, when S1 is predetermined as a
282
DIELECTRIC-LOADED ANTENNAS
and
1 max = cos1
P1(x1, y1)
(r1, 1)
r1
= tan1
1
n
D
2
(16)
(F + T )
Equation (16) shows that, for a given dielectric, the lens aperture angular size is limited by its refractive index n. In other
words, with common dielectrics there is a limit on the compactness of the lens. That is, the focal length F cannot be
reduced beyond the limit specified by Eq. (16).
S
F
T
S1
S2
Lens with Planar S1. In this case, both lens surfaces contribute to the beam collimation. Its surface can be determined
similar to case (a) by enforcing x1 F and infinite slope for
S1 (Fig. 4). The results are (3)
Hyperboloid
Figure 3. Geometry of lens with a planar surface S2.
x1 = F
spherical surface, they enter the lens unaffected. Their collimation is done entirely by the lenss second surface S2. Its
surface is again a conic section and its cross section is elliptic.
In the other two cases, both surfaces S1 and S2 of the lens
participate in beam collimation and consequently are enterdependent and more complex.
{[(n 1)T [F 2 + y21 ]1/2 ][(n2 1)y21 + n2 F 2 ]1/2
+n2 F[F 2 + y21 ]1/2 }
x2 =
[n2 (F 2 + y21 )1/2 [(n2 1)y21 + n2 F 2 ]1/2 ]
(x2 F )
y2 = y1 1 +
[(n2 1)y21 + n2 F 2 ]1/2
Lens with Planar S2. On S2, x is constant and slope is infinite
(Fig. 3), and the surface is defined by
T=
x2 = F + T
(10)
y2 = y1
A consequence of this is L1 L0 in Eq. (2); when we use Eqs.
(10), Eq. (2) becomes
r1 + nr3 = F
(11)
1
(n 1)1 [(4F 2 + D2 )1/2 2F]
2
Note that since the beam collimation is due to both surfaces,
the coordinates of S2 are now dependent on those of S1.
Lens with One Spherical Surface. When S1 is a spherical surface, all spherical waves originating at the focal point S pass
through it unaffected. The second surface S2 collimates the
beam. The geometry is shown in Fig. 5, and S2 is an ellipse
given
r2 =
which, when using Eq. (10), becomes a function of x1 and y1.
It can be solved directly to yield the profile of S1 as
y21 (n2 1)(x1 F )2 = 2(n 1)F (x1 F )
(12)
y2 =
(13)
They represent rectangular and polar equations of a hyperbola, which is the lens profile on S1. They can be used also to
determine the lens thickness on the axis. For this, one can
use two extreme rays passing through its tip and the axis.
The equality of the electrical lengths gives
F + nT = r1 (1 max ) =
D
2
2
x2 + (n 1)R
n
+ (F + T )
(14)
P1(F, y1)
(r1, 1)
T = (n + 1)1
(n + 1)D
+ F2
4(n 1)
2
1/2
x2
P2(x2, y2)
(r2, 2)
x
D
T
S1
1/2
2
1
F
A solution of this equation gives the lens thickness T as
(18)
1/2
2
(n 1)R
n cos 2
where R F T and other parameters are defined in Fig. 5.
Its equation in rectangular coordinates has the form
or, in polar coordinates,
(n 1)F
r1 =
n cos 1 1
(17)
S2
(15)
Figure 4. Geometry of lens with a planar surface S1.
DIELECTRIC-LOADED ANTENNAS
283
Elliptic lense with spherical surface S1
y
Spherical
P1(n1, y1)
(F, 1)
1
A()
=
A(1 )
F
P2(x2, y2)
(r2, 2)
(n cos 1 )3
(n 1)2 (n cos 1 1)
1/2
(22)
An inspection of these equations shows that in Eq. (21) the
amplitude ratio decreases with 1. That is, after leaving the
1
lens
the field is concentrated near its axis. The amplitude, in
x
S
fact, drops to zero at the angle 1 max, given by Eq. (16). ThereD
F
T
fore, this lens enhances the field taper of the source and is a
good candidate for low sidelobe applications, but its aperture
S1
S2
efficiency will be low. In contrast, the amplitude ratio in Eq.
(22) increases with 1. That is, this lens corrects the amplitude taper of the source and enhances the aperture efficiency
Figure 5. Geometry of lens with a spherical surface S1.
but, in the process, raises the sidelobe levels. Thus, it may
be used in applications where the aperture efficiency is more
critical than the sidelobe levels.
and
For most common dielectrics the refractive index is n
1
1.6, that is, r 2.55. For these materials the limit of the
1
2
4 1/2
T = (n 1) [2F (4F D ) ]
aperture angle is 1 max 51.3. Within this limit the ampli2
1
(19)
tude
ratios of Eqs. (21) and (22), normalized to axial values
2 max = cos1
are shown in Table 1. The amplitude tapering of hyperbolic
n
lenses is clearly evident. At 35 lens adds another 10 dB to
The last equation again sets a limit for the peak angular ap- the aperture field taper, and beyond 40 the lens is practically
erture of the lens for a given dielectric material.
useless. For large-angle lens applications, higher dielectric
When the surface S2 is assumed spherical, then both lens constant materials must be used. Table 1 also shows the amsurfaces participate in collimating the beam. The inner sur- plitude enhancement of elliptic lens. A 35 lens improves the
face S1 can be obtained from (3)
aperture field uniformity by as much as 6.3 dB. It increases
rapidly thereafter and becomes about 10 dB and 20 dB imn2 [r22 + r21 2r1 r2 cos(1 2 )] = [(n 1)T + r2 cos 2 r1 ]2
provements for lens angles of 45 and 50, respectively. These
2
amplitude
enhancements, however, must be accepted as theon r1 sin(1 2 ) = sin 2 [(n 1)T + r2 cos 2 r1 ]
retical
limits,
since at these wide angles the lens surface re
1/2
4(n 1)F 2 (n 3)D2
flectivity
will
reduce
the practically attainable levels. Surface
T=
4(n 1)(n 3)2
matching layers must be used to minimize the reflections.
F
(20) General Lens Design
+
n3
n
EFFECT OF LENS ON AMPLITUDE DISTRIBUTION
The lens equations, Eqs. (1) to (6), were based on the ray path
analysis, or in antenna terms the phase relationships. The
amplitude distributions were not considered. In practical applications, however, the amplitude distributions are also important and will influence the aperture efficiency of the lens,
sidelobe levels, and cross-polarization. To state it briefly, a
uniform aperture distribution gives the highest directivity,
but has high sidelobes because of its high edge illumination.
Sidelobes can be reduced by tapering the field toward the
edge. Excessive tapering, however, rapidly reduces the lens
directivity. It is therefore useful to know the influence of the
lens on the field amplitude as well.
Assume that A() is the angular dependence of the wave
amplitude radiating from the focal points; and assume that
A(), with r sin , is the amplitude distribution of the
collimated beam. Then, using the conservation of power and
neglecting the reflection at the lens surface, the following amplitude relationships can be obtained (1).
Hyperbolic lens with planar S2
1
A()
=
A(1 )
F
(n cos 1 1)3
(n 1)2 (n cos 1 )
1/2
(21)
In the general lens of Fig. 2, both surfaces are profiled and
participate in collimating the beam. Thus, a more versatile
lens can be obtained. However, Eqs. (1) to (6) showed that
there are at least four unknown coordinates (x1, y1, x2, y2) to
be determined. However, the optical relationships provided
only three equations, which are not sufficient to uniquely determine the coordinates of both surfaces S1 and S2. Another
relationship must be generated, which may be imposed on the
amplitude distribution A(), to control the directivity or sidelobes. Alternatively, one may impose conditions on the aperture phase errors. An important case is the reduction of phase
errors due to the source lateral defocusing. This will allow
beam scanning without excessive degradation in efficiency
and sidelobe levels. In most cases, however, the problem is
too complex for an analytic solution and a numerical approach
must be used.
ABERRATIONS
The term aberration, which originated in optics, refers to the
imperfection of lens in reproduction of the original image. In
antenna theory, the performance is measured in terms of the
aperture amplitude and phase distributions. The phase distribution, however, is the most critical parameter and influences
the far field significantly. It is therefore used in evaluating
284
DIELECTRIC-LOADED ANTENNAS
Table 1. Amplitude Distributions for the Hyperbolic and Elliptic Lenses of Figs. 3 and 5
Amplitude Ratio,
A()
A(1)
Hyperbolic lens equation
[Eq. (21)]
Elliptic lens equation
[Eq. (22)]
Ray Angle, 1 (deg)
Relative value
dB
Relative value
dB
10
20
30
35
40
45
50
1.0
0.0
1.0
0.0
0.928
0.65
1.060
0.51
0.733
2.70
1.26
2.01
0.466
6.64
1.69
4.55
0.328
9.70
2.06
6.29
0.196
14.17
2.67
8.54
0.084
21.5
3.17
10.03
0.008
41.75
9.25
19.33
n 1.6
r 2.55
1 max 51.3
the performance of aperture antennas such as lenses and reflectors. With a perfect lens and a point source at its focus, the
phase error should not exist. However, there are fabrication
tolerances, and misalignments can occur that will contribute
to aberrations. Even without such imperfections, lens antennas can suffer from aberrations. Practical lens feeds are horn
antennas and small arrays. Both have finite sizes and deviate
from the point source (2). This means that part of the feed
aperture falls outside the focal point, and rays emanating
from them do not satisfy the optical relationships. Thus, on
the lens aperture the phase distribution is not uniform. Similar situations also occur when the feed is moved off axis laterally to scan the beam. Again, aperture phase error occurs due
to the path length differences. A somewhat different situation
arises when the feed is moved axially, front or back. In this
case the phase error is symmetric, because all the rays leaving the source with equal angles travel equal distances and
arrive at the aperture at an equal radial distance from the
axisthat is, on a circular ring. However, the length of the
ray increases, or decreases, with radial distance on the aperture. The phase error is, therefore, quadratic on the aperture
and reduces the aperture efficiency, while raising the sidelobes.
The general aberration (i.e., the lens aperture phase error)
can depend implicitly on both feed and lens coordinates and
can be difficult to comprehend. However, like all other phaseerror-related problems, it can also be represented as the path
length difference with a reference ray. For rotationally symmetric rays, the natural reference is the axial ray. The path
length difference can then be obtained by a Taylor-type
expansion of the general ray length in terms of the axial one.
For small aberrations the first few terms in the expansion
will be sufficient to describe the length accurately. In terms of
the aperture polar coordinates and the expansion becomes
ing a Fourier transformation or diffraction integral. For onedimensional errors (i.e., x and o) the effect can be
understood easily and has been investigated by Silver (1). The
first term is linear, and in a Fourier integral it shifts the
transform variable. It thus causes a tilt of the beam, but the
gain remains the same. Using Silvers notation, if f(x) is the
aperture distribution and g(u) is the far field (i.e., its Fourier
transform with a linear phase error), one finds with no phase
error
L(, ) = Laxial + cos + 2 [1 + cos2 ] + 3 cos +
(23)
A quadratic phase error is symmetric on the aperture and
does not tilt the beam, but reduces its gain. For small values
of , it can be calculated analytically (1) and is given by
where , , and are constants indicating the magnitude of
each phase error. The leading term is linear in and , then
becomes quadratic, cubic, and so on, and the magnitude of
each depends on the nature of imperfection causing the phase
error. The even terms are caused either by an axial defocusing or by an axially symmetric error. The odd terms can be
due to a lateral displacement of the feed or can be due to
asymmetric errors.
The effects of each error can be investigated by its introduction in the aperture field and determining the far field us-
g0 (u) =
a
2
1
1
f (x) exp[ jux] dx
(24)
and with phase error
g(u) =
a
2
1
1
f (x) exp[ j(ux x)] dx = g0 (u )
(25)
where u (a/ )sin and a is the aperture length. Equation
(25) shows that the beam peak is moved from the 0 direction to 0, calculated by
u =0
or
0 = sin1
a
a 1
f (x) exp[ j(ux x2 )] dx
2 1
a
= [g0 (u) + jg0 (u)]
2
g(u) =
(26)
(27)
where g0(u) is the second derivative of g0(u). Due to this phase
error the gain decreases progressively with increasing , and
eventually the beam bifurcates and maxima appear on either
DIELECTRIC-LOADED ANTENNAS
degradation than others, it is desirable to eliminate it, especially that it manifests mostly in beam scanning. Feed lateral
displacements to scan the beam can readily cause coma lobes.
Fortunately, a number of lens surface modifications have
been found to reduce the effects of this error (3).
Gain decibels
= /2
10
285
= /4
ZONED LENSES
20
No phase error
30
3
u = ( a/ ) sin
Figure 6. Effect of quadratic phase error on the far-field pattern.
side of the axis. It also raises the sidelobe levels. Figure 6
shows typical pattern degradation due to this error.
The next important phase error is the cubic one, which has
odd power dependence on the aperture coordinate. This error
not only tilts the beam, but also reduces the gain and asymmetrically affects the sidelobes, raising them on one side
while reducing them on the opposite side. Its effect is therefore a combination of that of the linear and quadratic phase
errors. For small errors its far field is given by (1)
a 1
f (x) exp[ j(ux x3 )] dx
2 1
a [g (u) + g (u)]
=
0
2 0
g(u) =
(28)
where g0 (u) is the third derivative of g0(u). For a few small
phase errors the far fields of this phase error are shown in
Fig. 7. They show clearly the beam tilt, the gain loss, and the
rising of the sidelobes toward the beam tilt. They are known
as coma lobes, after the corresponding aberration in optics.
Also, because this phase error causes more severe pattern
So far, the equations used for lens design equalized the ray
path lengths. The frequency of operation, or its wavelength,
did not enter into the equations. Thus, in principle, these
lenses should function at all frequencies. However, the directivity of a lens depends on its aperture size D, and these
lenses are often used for high gain applications. This results
in large lens sizes in wavelength; and at microwave frequencies, in large physical sizes, both the aperture diameter D and
thickness T. It can, therefore, become excessively heavy and
difficult to use. However, the thickness of the lens can be several wavelengths, and thus it can be reduced along the ray
path in multiple wavelengths without altering the relative
phase change. The process starts at the edge where the thickness is zero. Moving down toward the axis, the thickness increases progressively until it becomes one wavelength. This
thickness can be made zero without altering the phase. The
process can be continued K times until one arrives at the lens
axis. In practice, one must maintain a small thickness tm to
provide adequate mechanical strength, the value of which will
depend on the lens size, the material strength, and application type.
With zoned lenses, and neglecting tm because the thickness
does not exceed one electrical wavelength, its thickness is limited to /(n 1). Including the minimum thickness tm, the
total thickness is limited to tm l/(n 1) regardless of the
number of zones. The path lengths in wavelength, however,
are not equal. With K zones, the ray path at the edge will be
longer by a length equal to (K 1). This causes the frequency dependence of lens operation, limiting its bandwidths.
Enforcing the commonly used Silvers criterion for this aperture phase error (1) (i.e., less than 0.125), the useful bandwidth of a lens with K zones can be calculated from (1)
Bandwidth
=
25
%
K 1
(29)
0
No phase error
Gain decibels
= /4
10
= /2
20
30
3
u = ( a/ ) sin
Figure 7. Effect of cubic phase error on the far-field pattern.
Equation (29) is valid for small variations of and uniform
aperture distributions. For taper distributions the effects of
phase errors is smaller and the actual bandwidth can exceed
that of Eq. (29).
Zoning the lens can cause one additional, severe problem
due to shadowing. Two adjacent rays from the focus can
travel through two separate zones, resulting in a dark ring
zone on the aperture. This occurs in the transmit mode, and
it causes the loss of directivity and increased sidelobe levels.
In the receive mode, the energy falling on the shadow zones
never reaches the lens focus and diffracts into the space,
again causing reduction of gain along with increased noise
temperature. Figure 8 shows the geometry of a three-zone
lens and shadowing due to R1 and R2 rays.
Zoning without shadowing is also possible, but should be
done on the nonrefracting surface of the lens. In a hyperbolic
lens, this should be done on the planar back surface. Shadow-
286
DIELECTRIC-LOADED ANTENNAS
Shadow rings
R2
R1
S
Focus
K=1 K=2 K= 3
F
tm
Figure 8. Geometry of a zoned lens with shadowing effects.
ing will be eliminated, but phase errors still occur at the transition lines due to diffraction effects.
used, the reflection effects on the feed can be minimized by
lateral defocusing of the feed, or retuning of the feed over a
narrow bandwidth.
LENSES WITH n 1
Lens equations (1) to (6) were developed without specifying
the value of the refractive index, and therefore they are valid
for n 1 cases as well. However, the lens surface becomes
inverted. For instance, the hyperbolic lens equation [Eq. (13)]
for n 1 modifies to
r1 =
(1 n)F
1 n cos 1
(30)
and the lens surface becomes elliptical, concave toward the
focus, similar to Fig. 1(b). On the inner region a minimum
thickness t is required to provide mechanical strength. Zoning
is also possible and will cause shadowing when incorporated
on the active refracting surface. The bandwidth limitations
due to n remains the same as the dielectric lenses with n
1. However, the lens media for n 1 such as metal plates
and waveguides are usually frequency-sensitive and exhibit
narrower bandwidths.
REFLECTION FROM LENS SURFACE
Because the wave impedance in air and the dielectric medium
of lens are different, reflections occur for all the rays. The
reflection coefficient depends both on the wave polarization
and the angle of incidencethat is, the angle of ray with the
local normal on the lens surface. Neither can be avoided. With
a linearly polarized wave, the relative polarization, with respect to the plane of incidence, changes from perpendicular to
parallel as the ray direction rotates on the lens surface. However, their reflection coefficient behaves differently. For perpendicular polarization, it increases progressively with the incidence angle. For parallel polarization, it decreases initially;
and after vanishing at the Brewster angle, it increases rapidly. Consequently, incidence angles must be kept small, less
than 30, to minimize the polarization effects on the lens aperture distribution.
The surface reflection effects can be reduced when warranted, but this requires utilizing an impedance matching
layer between the lens and free space. At normal and small
angles of incidence, the refractive index of the matching layer
can be found using a quarter-wavelength transformer rule. It
is the geometric means of the refractive index of the lens dielectric and that of air. In practice, a different dielectric material may be used as the matching layer, or it may be synthesized by preferentially removing a fraction of the dielectric
material from the lens surface, such as drilling /4 holes or
cutting grooves (4). However, care must be taken to determine
their polarization effects.
The surface reflections also influence the impedance mismatch at its feed. The problem is most severe in cases where
the lens surface is coincident with one of the equiphase surfacesthat is, the wave front. Then, the entire reflected wave
travels back to the feed, the degree of which depends on the
lens refractive index. Since at normal incidence the reflection
coefficient is R (n 1)/(n 1), the reflected power is unacceptably large for all common dielectrics, and a matching surface should be used. In the event a matching layer cannot be
CONSTRAINED LENSES
The function of a lens is to modify the phase front of an incident wave, say from spherical to planar. In practice, this may
be accomplished by means other than the dielectric lenses. In
the most general case, the lens surfaces consist of a plurality
of receiving and radiating elements, interconnected by processing elements. The received signals of one surface are modified in amplitude and phase and are reradiated from the elements of the next surface. In passive designs, the
interconnection is due to transmission lines, such as parallel
plates, waveguides, and even coaxial lines. The design process
is similar to that of the dielectric lenses and is governed by
the path-length equation. Snells law, however, is not satisfied
at all surfaces, and the problem of surface reflection and
transmission must be solved through the use of the wave
equation. Nevertheless, lenses can be designed with surfaces
similar to that of dielectric lenses, but with inverted curvature (3).
The simplest case uses parallel plates, with spacing a, between one and one-half wavelength. When the electric field is
parallel to the plates, a non-TEM waveguide mode is excited
and has a wavelength p given in terms of the free space
wavelength by
p =
1
2a
2 1/2
(31)
which can be used to define an equivalent refractive index as
2
<1
(32)
= 1
n=
p
2a
In cylindrical lenses, when the plates and electric field are
normal to the cylinder axis, Snells law of refraction governs
DIELECTRIC-LOADED ANTENNAS
the transition between the lens and outside media. But, when
they are parallel to the cylinder axis, the incident rays are
constrained to pass between the plates and Snells law is not
satisfied (1).
An example of the rotationally symmetric constrained lens
is the planarelliptic surface lens of Eq. (30). It is usually
zoned to reduce its size and weight (4). Other useful transmission media are the rectangular and square waveguides, operating in TE10 or TE01 modes. The waveguide dimensions
must be such that only these modes can propagate and
higher-order modes are suppressed. The square waveguide
can be used for circularly polarized applications; otherwise, it
must be avoided to reduce cross-polarization.
INHOMOGENEOUS LENSES
In the lenses so far studied, the refractive index n was constant and the shape was profiled to satisfy the ray path condition. On the other hand, if the lens shape is kept fixed, then
another parameter, such as the refractive index, must be allowed to change to help in collimating the beam. This is
achieved in a family of lenses, with the most important ones
being spherical in shape, such as Luneberg lens, Maxells fisheye, and Eaton lenses. Their spherical shape provides a perfect three-dimensional symmetry, useful in applications, such
as wide-angle scanning. They also have only a radial inhomogeneity, making them both physically and electrically symmetric.
Luneberg Lenses
The Luneberg lens refers to a family of lenses with two axial
foci. They can be both outside the lens or one inside and the
other outside. The most useful case, however, is the lens with
one focus on its surface, while the second one is at infinity.
That is, an axial point on the lens surface is focused to an
axial point at infinity, on the opposite side of the lens. The
refractive index of this lens is given by
r 2 1/2
n(r) = 2
a
287
Limits of solid
angular coverage
determined by
boundaries of cap
reflectors
Spherical cap
reflector
Lens
Figure 10. Passive Luneberg lens reflector.
cally significant. The refractive index values and variations
are in reasonable range, and the lens can be synthesized. Second, the unity of its refractive index on the surface eliminates
the impedance mismatch and, consequently, the surface reflections. The geometry and ray paths of this lens are shown
in Fig. 9, with a feed horn on its surface. Scanning the feed
on its surface scans the radiated beam, without alteration.
The scan limit is set only by the mechanical limitation of the
feed horn motion. With a spherical conducting cap on its surface the lens also acts as perfect reflectorthat is, a backscatter (Fig. 10). The main difficulty with this lens is its fabrication problems. Multilayer shells are normally used to
synthesize the refractive index inhomogeneity. Figure 11
shows one case where 10 layers are used to construct an 18
in. diameter lens. While the approximation to continuously
variable refractive index is reasonable, the wave scattering at
the layer transitions reduces the lens efficiency.
With the above refractive index, the Luneberg lens performance is ideal at the geometrical optics limitsthat is, when
(33)
where a is the lens radius and r is the radial distance of a
point inside the lens. At the origin, the refractive index is
n(o) 2, and on its surface it becomes unity. Both are practi-
Lens
Feed antenna
Figure 9. Typical ray paths in a Luneberg lens.
Figure 11. Multilayer spherical shell construction of a Luneberg
lens.
288
DIELECTRIC-LOADED ANTENNAS
60
2.4
r = 1.5
50
2.2
40
2.0
Normalized phase (deg)
Dielectric constant
1.8
1.6
d
a
1.4
c
1.2
r = 2.0
30
20
r = 2.5
10
r = 3.0
0
r = 3.5
10
r = 4.0
20
30
0.2
0.6
0.8
0.4
Radial distance (r/a)
1.0
40
Figure 12. Refractive index of modified Luneberg lenses.
60
the lens diameter in wavelength is large. At microwave frequencies, the wavelength is large and the lens diameter in
wavelength may not be large. Its performance (i.e., directivity) and sidelobe levels deteriorate rapidly. In such cases, the
refractive index profile can be modified to improve its performance. This can be done by determining the excitation efficiencies of various spherical modes and calculating its far
field and directivity (5). The new dielectric permittivity profile
is defined as
r = n2 = 2B A2
r 2
(34)
r = 6.0
50
where the constant parameters A and B are determined to
maximize the gain. Three different cases are identified and
investigated. Their refractive index profiles are shown in Fig.
12. Cases (b) (A 1, B 1.1) and (c) give larger refractive
indexes and are expected to perform better at lower frequencies. This is investigated using the spherical harmonies, and
the results for the directivity, sidelobe levels, and beamwidths
are shown in Table 2.
Constant n Spherical Lens
The difficulty with fabrication of the inhomogeneous lenses
encouraged investigators to search for quasi-ideal spherical
0.1 0.2 0.3 0.4 0.5 0.6 0.7
Normalized effective radius
0.8
Figure 13. Phase across aperture of a constant n spherical lens.
lenses with constant refractive index. An interesting case is a
lens with r 3, studied earlier by Bekefi and Farnell (6) and
recently by Mason (7). With a Huygen source feed at its surface, the computed phase distribution across its aperture, for
different relative permittivities are shown in Fig. 13. For r
3, the phase error is below 10, across about 60% of the aperture. It remains within acceptable range for gain calculating
over at least 70% of the aperture, resulting in excellent gain
performance over a wide range of frequencies. The only drawback seems to be the excitation of internal modes at their
resonance. Their effect reduces with the loss tangent of the
dielectric material.
DIELECTRIC LOADED HORNS
Horn antennas are among the most useful and versatile antennas. They have a relatively simple shape and are easy to
fabricate and use. They are used as test antennas, as feeds
for reflector and lens antennas, or independently as communi-
Table 2. Performance Parameters of Modified Luneberg Lens
Luneberg Lenses
Modified Luneberg Lenses
AB1
Diameter
in
2
4
6
8
10
A1
A2 B
B Value
Gain
(dB)
Beam
Width
(degrees)
First SideLobe Level
(dB)
Gain
(dB)
Beam
Width
(degrees)
First SideLobe Level
(dB)
Gain
(dB)
Beam
Width
(degrees)
First SideLobe Level
(dB)
1.4
1.16
1.1
1.075
1.04
14.79
20.761
24.34
26.90
28.78
30.2
15.1
9.8
7.3
5.8
14.41
16.05
16.9
17.1
16.35
17.56
22.70
25.75
27.98
29.35
23.5
13.0
9.0
6.7
5.5
17.15
16.9
16.97
17.01
15.97
16.85
22.0
25.17
27.56
29.26
24.0
13.25
9.1
7.0
5.6
14.79
16.1
16.4
16.6
16.81
yyy
;;;
;;;
yyy
;;;
yyy
;;;
yyy
;;;
yyy
DIELECTRIC-LOADED ANTENNAS
cation antennas. Because of their diverse applications, their
electrical specifications vary considerably. As test antennas,
they are used as gain standards and required to have good
polarization isolation in the principal E and H planes. Rectangular horns are commonly used for this application to simplify
the polarization definition and gain calculation. As a feed for
reflector and lens antennas, the requirements are significantly different. While having a finite aperture size, they
must behave as a point source, have small side and back lobes
to minimize power spillovers, and have negligible cross-polarization in the entire radiation zone. To achieve such stringent
requirements, their design must be precise and an accurate
solution must be known to access their performance. This is
more so with circular horns, which, consequently, have found
more widespread applications as feeds than rectangular ones.
The electromagnetic analysis, however, has shown that conventional smooth wall horns cannot provide radiation patterns with acceptable polarization purity and low spillover.
Corrugated horns are developed for these applications, but
are costly and narrowband. Dielectric loading of the horn has
been shown to improve the performance as well, and in certain applications it may be used to replace corrugated ones.
In applications where horn antennas are used as independent communication antennas, the gain and aperture efficiency may be the fundamental parameters to optimize. However, to obtain high gains, the horn aperture size must
increase, which also increases the aperture phase errors. The
phase errors can be kept low by using small cone angles, but
this increases the horn size. Thus a convenient solution involves the use of a lens at the horn aperture to reduce or
eliminate the phase errors, by collimating the beam. Consequently, compact high gain horns can be designed with controlled aperture phase and amplitude distributions, to improve the aperture efficiency and horn gain. Alternatively,
lenses can be used to suitably modify the aperture distribution in both amplitude and phase to shape the radiation patterns.
In this section, initially the dielectric loaded and lens corrected horns will be discussed. Then the use of dielectric in
small antennas such as waveguides, microstrip antennas, and
dipoles to improve their operation in specific applications will
be considered.
Dielectric Loading
Historically, dielectric-cone loading inside smooth wall conical
horns was used by Clarricoats et al. (8) and Lier (9) to simulate the effect of corrugations. Corrugated horns, with quarter-wavelength corrugation depths, can support the hybrid
HE11 mode. This mode radiates with low cross-polarization
and can be designed to have negligible sidelobes. Introduction
of the cone dielectric, with an air gap as shown in Fig. 14
inside a smooth wall horn, was shown also to support hybrid
modes and improve the performance. Clarricoats et al. (8)
used low-dielectric-constant materials, such as foams with a
relative permittivity of 1.13. But, in Liers work (9), solid-dielectric cones with a relative permittivity of 2.5 were used,
again showing good performance. Both investigators also analyzed these dielectric loaded horns using modal expansions,
and they studied the effects on the air gap, horn permittivity,
aperture diameter, flare angle, and the throat region. It was
found that the air gap size is strongly dependent on the aper-
289
(a)
(b)
Figure 14. Geometry of a dielectric loaded horn.
ture diameter, and both are dependent on the dielectric permittivity. The air gap size generally increases with the horn
diameter, and for a given diameter there is a minimum relative permittivity of dielectric to support the hybrid mode to
minimize the cross-polarization. Both flare angle and the
throat region have similar influences. Large flare angles and
an asymmetric throat region design excite higher-order
modes and thus increase cross-polarization.
A variation of the conical dielectric loaded horn is shown
in Fig. 15. Its wall is profiled. A large flare angle near its
throat reduces its axial length and results in a compact horn.
Then, its small flare angle near the aperture improves its
cross-polarization. The profile is described by the following
equation:
r(z) = rth + 3 r 1
2z
3L
z 2
L
(35)
r = rap rth
yyy
;;;
;;;
yyy
;;;
yyy
where rap and rth are the horn radii at its aperture and throat.
A profile horn of this type was designed and optimized. Its
Figure 15. Geometry of a dielectric loaded profile horn.
290
DIELECTRIC-LOADED ANTENNAS
Table 3. Performance of Dielectric Loaded Linear and
Profiled Horns
Parameter
Linear
Horn
Profiled
Horn
3 dB beam width (degrees)
10 dB beam width (degrees)
Directivity (dB)
Efficiency (percent)
Peak cross-polar (dB)
VSWR
14.8
26.9
22.1
61.8
32.2
1.04
13.7
24.8
22.5
68.1
36.0
1.03
Rth 1.14 cm
rup 27.7 cm
L 30.9 cm
r 1.13
Air gap 1.2 cm
max
D
T
(a)
max
performance is compared with the linear horn in Table 3. Its
cross-polarization is improved by 4 dB. The effect of length
reduction on the performance of the above profile horn is also
shown in Table 4. It shows that the performance remains
steady and comparable to a linear horn, for length reductions
by as much as 22%.
Lens Corrected Horns
F
(b)
r
In high-gain horns, the aperture diameter in wavelength is
large, and the horn length can be excessive, unless its flare
angle is made large. However, the combination of large aperture size and large flare angle can cause severe aperture
phase error. This problem can be remedied by using a lens at
the horn aperture. Figure 16 shows three possible options.
These simple lenses and others including zoned lenses may
be used, and they would correct the horn aperture phase distributions. But each lens will have different influences on the
aperture amplitude distribution. The case of the first two
lenses were investigated earlier, and Table 1 showed their
effect on the amplitude distribution. Type a increases the amplitude taper according to Eq. (21) and will reduce both sidelobes and the aperture efficiency. Type b will compensate for
the amplitude taper, and according to Eq. (22) the lens permittivity can be used to control the aperture distribution and,
hence, the horn efficiency and the pattern sidelobes. For type
c, an analytic expression is not available and a numerical procedure must be used. However, as was indicated earlier in
this lens, both surfaces help collimating the beam. But its
second surface is similar to type b lens, and its influence on
the aperture distirbution will be similar as well. With a hybrid mode horn, corrugated or dielectric loaded, the resulting
aperture distributions for different lens relative permittivity
Table 4. Performance of Profile Horn with Length Reduction
Length
(cm)
Peak
Cross-Polar
(dB)
3 dB
Beamwidth
(deg)
Efficiency
%
30.9
27.5
24.0
15.0
36.0
36.8
31.6
27.6
13.7
13.8
14.0
16.1
68.1
64.4
57.1
32.0
T
(c)
Figure 16. Lens types for loading horn aperture.
are shown in Fig. 17. It shows that for r around 1.22 the
aperture amplitude distribution is nearly uniform.
DIELECTRIC LOADED WAVEGUIDES
Waveguides have small aperture size and are not as efficient
radiators as horns. Part of the energy leaks out and induces
current on the outside wall, which radiates side and backward, causing large back lobes. The wave impedances of
waveguide modes are also different from the free-space intrinsic impedance and strong reflections can occur on the aperture, causing poor input impedance match. These problems
can be partly overcome by flaring the waveguide at its aperture. However, a similar and even better performance can be
obtained by loading the waveguide by a short section of a dielectric. The constant, size, and shape of a dielectric provide
several parameters that can be used to shape the radiation
patterns and tailor them to desired specifications. Table 5
shows the results for three different end loadings, along with
the type of performance variations one could achieve (2). Two
other examples are shown in Figs. 18 and 19, with combinations of dielectric and cavity loadings (2). In Fig. 18, the end
geometry is optimized for nearly perfect pattern symmetry,
with negligible cross-polarization. Figure 20 shows its copolar
and cross-polar radiation patterns. In Fig. 19, again the com-
yyy
;;;
;;;
yyy
DIELECTRIC-LOADED ANTENNAS
18
291
37.5 mm
1.1
r = 2.6
12
26 mm
61 mm
32 mm
1.2
6
dB
Figure 18. Geometry of a dielectric and cavity loaded waveguide
feed.
20
10
30
(deg)
6
1.23
Horn alone
1.5
14
Figure 17. Aperture amplitude distribution for a lens corrected
horn, 30 semiflare angle hybrid mode horn, type c lens.
bination was optimized for a heavily shaped radiation pattern
with again negligible cross-polarization in the forward direction. It is an ideal feed for deep parabolic reflectors with
small f /D 0.25. It provides high aperture efficiency of 81%
due to its front pattern null, very low cross-polarization, and
extremely low noise temperatures due to small f /D, the focallength-to-diameter ratio.
MICROSTRIP AND DIELECTRIC RESONATORS
Microstrip antennas are discussed in a separate article, and
they usually consist of a conducting patch separated from
a ground plane by a dielectric substrate. They are low profile and increasingly popular antennas for practically any
type of applications. Their radiation patterns, however, are
asymmetric with unequal E- and H-plane patterns. But,
with careful optimization, the pattern symmetry can be
achieved to minimize cross-polarization. Figure 21 shows a
case of stacked patches with a side choke for equalizing
the principal plane pattern, low back radiation, and crosspolarization. Similar performance can also be obtained using a dielectric resonator in lieu of a microstrip patch. The
dimensions of the dielectric resonator are related to the
wavelength by
d 2 1/2
1.841
16 +
d=
4n
1.841h
(36)
The excited mode is the TM110 mode, and it gives radiation
similar to that of a microstrip patch. In Fig. 22, the resonator
and the cavity are optimized for symmetric pattern in the
principal planes to reduce the cross-polarization. They are
shown in Fig. 23, with excellent symmetry. Both the microstrip and resonator antennas can be used as efficient reflectors and lens feeds with high aperture efficiency and low
cross-polarization.
Table 5. Performance of Dielectric Loaded Waveguide with Shaped Dielectrics
Geometry
Half-Beamwidths
Peak
Cross-Polarization
0 90 (dB)
Gain
(dB)
E Plane
H Plane
E Plane
H Plane
33.95
8.28
36.82
36.18
71.47
72.51
24.74
8.11
37.21
38.32
73.42
71.35
24.43
13.47
19.43
20.25
33.13
35.17
3 dB
10 dB
0.519
60
0.1
0.6
b
60
0.6
c
0.619
D = 0.6, r = 2.5
292
DIELECTRIC-LOADED ANTENNAS
E plane
H plane
Cross-plane
10
16
24
Relative power (dB)
Relative power one way (dB)
20
32
30
40
36
72
108
144
180
(deg)
Figure 19. Geometry and radiation pattern of a shaped dielectric
and cavity loaded waveguide feed.
40
60
120
180
(deg )
Practically all antennas have conducting parts, but in certain
families of antennas, especially small resonant ones, the conduction current radiates directly. Typical examples are the
wire antennas and microstrip antennas that are often halfwavelength resonators. In wire antennas, the current is excited by the applied voltage directly on the wire, which radiates in the surrounding space. In microstrip antennas, the
currents are both on the patch and on its ground plane, which
are separated by a dielectric substrate. Because of this, only
the patch current is exposed to the surrounding medium.
However, in either case, the physical constants of the medium
is excessively lossy: It can short-circuit the antenna current
and prevent its operation. In practice, this problem can occur
in remote sensing and biological applications. In the former
case, the antennas may be buried underground or may be
submerged in sea and ocean waters, which have high electrical conductivities. In the latter case, the antennas are implanted into various type tissues in the body that can have
excessively high conductivities. In such cases, to ensure the
antenna operation, the conduction currents must be insulated
from the surrounding conducting medium. A simple but effective method is to use a thin dielectric coating on the antenna
conductor carrying the radiating currents. The coating will
provide insulation between the conducting antenna and the
medium, thereby eliminating the conduction current. The excitation energy will then transfer into the poynting vector,
leaving the antenna.
The behavior of the insulated antennas in a medium of
complex permittivity differs considerably from that in free
space, and it should be analyzed carefully. For instance,
consider a conventional dipole of length 2h, as shown in
Fig. 24. The wire is a good conductor and has a diameter
Figure 20. Radiation patterns of the waveguide feed of Fig. 18.
of 2a, insulated by a cylindrical dielectric region of diameter 2b and propagation constant k1, located in an infinite
exterior region of k2. With a thin-wire approximation, the
dipole current can be represented by a sinusoidal distribution of the form (10). The time factor is assumed as
exp( jt):
I(z) =
jV sin kL (h |z|)
2Zca cos kL h
(37)
0
Relative power one way (dB)
INSULATED ANTENNAS
16
24
32
40
180
135
90
45
0
(deg)
45
90
135
180
Figure 21. Geometry and radiation patterns of a stacked microstrip feed.
DIELECTRIC-LOADED ANTENNAS
293
2b
3t
2a
K1 K2
d
h
+
v
D
Figure 22. Geometry of a dielectric resonator antenna.
where
1/2
H0(2) (k2 b)
k L = k 1 1 +
b
k2 bH1(2) (k2 b)ln
a
Zca =
Figure 24. Geometry of an insulated dipole antenna.
(38)
and so on. Thus, the insulation parameters should be selected
appropriately to minimize the dependence of k1 and k2.
b
1 k L
ln
2k1
a
(39)
0
k1
(40)
1 =
k1 = [1
1 ]
MEDICAL AND BIOLOGICAL ANTENNAS
1/2
(41)
(2)
And H(2)
0 and H1 are Hankel functions of zero and first order.
Note that with a perfect insulation dielectric, k1 is real; but
k2 is complex due to the presence of Hankel functions in Eq.
(38). It reduces to k1 when b, the radius of the insulation,
becomes infinitely large. In view of Eq. (38) the dipole current
distribution, input impedance, radiation resistance, and resonance frequency can depend strongly on the radius b, propagation constant k1, and the propagation constant of the exterior region, k2. The latter may not be fully known, or constant,
during the application due to variations in moisture content,
0
E plane
H plane
Cross-plane
Relative power (dB)
10
Another area that insulated antennas play an important role
in is the biological and medical applications. They can be noninvasive (i.e., not penetrating the body) or invasive. In either
case, the property of insulated antennas can be significantly
different from those in free space. Thus, care must be taken
in their design and analysis to ensure adequate power transfer to the right tissue. Noninvasive radiators are often dielectric loaded waveguides and horns, discussed in the previous
section. The dielectric loading in this case is used to improve
impedance matching and coupling to the body. Their design
is not significantly different from other dielectric loaded
waveguides, except that the end shaping must prevent hot
spots and improve penetration.
Microstrip antennas and arrays are another type of radiators suitable for noninvasive applications. However, their resonance property and power coupling to the body can be sensitive to the extent and nature of contact to the skin. Dielectric
coating over the radiating patch or slot can insulate the antenna and minimize the bodys influence. This is due to the
fact that in microstrip antennas the resonance depends on
the effective dielectric constant, and not the actual substrate
permittivity. With single-layer substrates of thickness h, this
effective permittivity, for a conductor line width of W, is given
by
20
eff =
30
40
180
90
0
(deg)
90
180
Figure 23. Radiation patterns of the dielectric resonator antenna.
r + 1
+
2
1
2
1+
12h
w
1/2
(42)
However, it can change significantly by introducing a higher
permittivity layer over the substrate. Consequently, in biological applications, where the tissue relative permittivity can be
excessively high, due to the water content having r 80, the
nature of the proximity or contact with body can alter eff significantly (11). Because microstrip antennas are narrowband,
or at best not wideband, the efficiency of their radiation and
coupling to the body can be deteriorated. The effect can be
294
DIELECTRIC MEASUREMENT
Figure 25. Implantable radiator types. (a) Needle radiator, (b)
sleeve antenna.
reduced by introducing a superstrate layer over the microstrip antenna to control the relative permittivity variations.
Invasive-type radiators can produce more uniform and
controllable heating patterns, but they require implantation
in the tissue. The most convenient types are the insulated
needle radiator, basically the end of the coaxial line. However, this type of antenna can generate strong currents on
the outer coaxial conductor and cause tissue heating behind
the antenna. An improvement can be obtained by introduction of a quarter-wavelength choke over the coaxial conductor to form a sleeve antenna. Their analysis and sensitivity
study can be carried out similar to the insulated dipole
antennas. Figure 25 shows the geometry of needle and
sleeve antennas.
BIBLIOGRAPHY
1. S. Silver, Microwave Antenna Theory and Design, London: P.
Peregrinus, 1984.
2. A. D. Oliver et al., Microwave Horns and Feeds, London: P. Peregrinus, 1994.
3. Y. T. Lo and S. W. Lee, Antenna Handbook, Theory Applications
and Design, New York: Van Nostrand Reinhold, 1988, chap.
16.
4. R. C. Johnson and H. Jasik, Antenna Engineering Handbook, 2nd
ed., New York: McGraw-Hill, 1984.
5. M. Barakat and L. Shafai, Studies on certain modified Luneberg
lenses, IEE Proc., 130, part H (5): 363368, 1983.
6. G. Bekefi and G. W. Farnell, A homogeneous dielectric sphere as
a microwave lens, Can. J. Phys., 34: 790803, 1956.
7. V. B. Mason, The electromagnetic radiation from simple sources
in the presence of a homogeneous dielectric sphere, PhD dissertation, Univ. of Michigan, 1972.
8. P. J. B. Clarricoats, A. D. Oliver, and M. Rizk, A dielectric loaded
conical feed with low cross-polar radiation, Proc. URSI Symp. EM
Theory, Spain, 1983, pp. 351354.
9. E. Lier, A dielectric hybrid mode antenna feed, a simple alternative to the corrugated horn, IEEE Trans. Antennas Propag., AP34: 2129, 1986.
10. R. W. P. King et al., The insulated monopole: Admittance and
junction affect, IEEE Trans. Antennas Propag., AP-23: 172
177, 1975.
11. I. J. Bahl and S. S. Stuchly, Analysis of a microstrip covered with
a lossy dielectric, IEEE Trans. Microw. Theory Tech., MTT-28:
104109, 1980.
L. SHAFAI
University of Manitoba
DIPOLE ANTENNAS
Innitesimal Dipole (Hertzian Dipole)
INTRODUCTION
An innitesimal dipole (Length L << wavelength ) is a
linear element that is assumed to be short enough that
the current (I) is constant along its length. This element
is also called a Hertzian dipole. The electric and magnetic
eld components of the Hertzian dipole are (2)
Dipole antennas and their associated arrays are among the
most common antennas used for communication systems
and measurement of electric and magnetic elds. This article describes the basic nature of these antennas and some
of their applications. Variations such as biconical and bowtie antennas, slot dipoles, folded dipoles, sleeve dipoles, and
shunt-fed dipoles are also described.
DIPOLE ANTENNAS
The simplest type of wire antenna is the dipole. A dipole antenna is most commonly a linear metallic wire or rod with
a feed point at the center as shown in Fig. 1. Most often,
a dipole antenna has two symmetrical radiating arms. Because of the symmetry of dipoles relative to the xy plane
containing the feed point, the resultant radiation is independent of (rotationally symmetric about the z-axis) and
is shown in Fig. 2. The shape of the lobe(s) depends on
the length of the antenna, as will be discussed in the next
section. In general, however, dipole antennas are used for
applications where the radiation is desired in the xy plane
of the antenna.
Figure 1. Dipole antenna.
Figure 2. Radiation pattern for an innitesimal (or Hertzian)
dipole. (From Reference 1, IEEE 1992.)
1
jko
1
ILsin(
+ 2 )e jko r
4
r
r
jo IL
jko
1 jko r
E=
cos( 2 + 3 )e
r
2ko
r
r
2
jo IL
k
1
jko
sin( o + 2 + 3 )e jko r
4ko
r
r
r
H=
(1)
(2)
where o = (o /o )1/2 is the intrinsic impedance (= 377 )
for free space and ko = (o o )1/2 is the propagation constant (= /c, where c is the velocity of light). The elds
are observed to decay rapidly (1/r3 variation, where r is the
distance from the antenna, very near the antenna, and less
rapidly (1/r variation) farther away. The elds with terms
1/r2 and 1/r3 (the induction terms) provide energy that is
stored near the antenna. This energy can be used for applications where heat is desired near the antenna such as for
cardiac ablation or hyperthermia. The elds with 1/r variation (the radiation terms) provide energy propagation away
from the antenna. These elds are used for communication
applications. The distance away from the antenna where
the induction and radiation terms are equal is r = /2.
When r < /2, this is the near eld of the antenna, and
the induction terms dominate. When r > /2, this is the
far eld, and the radiation terms dominate. In the far eld,
the wave propagation is in the transverse electromagnetic
(TEM) mode, which is characteristic of far-eld radiation
from nite structures.
The far-zone radiated elds of the Hertzian dipole follow from equations (1) and (2) by retaining the 1/r varying
terms:
H=
j
ILsine jko r
4r
(3)
E=
jo
ILsine jko r
4r
(4)
As expected for TEM wave propagation, the E and H elds
are perpendicular to each other and to the outward propagation in the r direction. Also the ratio of E/H = o =
(o /o )1/2 , which is the intrinsic impedance of free space.
The radiation pattern of the Hertzian dipole is shown
in Fig. 2 and exhibits the classic symmetry expected of
dipole antennas, being both independent of and symmetric about the xy plane through the center (feed point) of
the dipole. The magnitude of the total radiated power is
Prad = 40 2 Io2 (L/)2 , where Io is the current on the dipole.
From equations 3 and 4, it is interesting to note that, even
for this constant-current innitesimal dipole, the radiated
power density is proportional to sin2 . Hence, it is maxi
mum for = 90 (i.e., in the xy plane normal to the orientation of the dipole) and zero for the directions along the
length of the dipole ( = 0 and 180 ). The latter property
for zero radiation along the length of the dipole is observed
for all linear antennas regardless of length, which follows
from the fact that a linear antenna of nite length may be
J. Webster (ed.), Wiley Encyclopedia of Electrical and Electronics Engineering. Copyright 2007 John Wiley & Sons, Inc.
Dipole Antennas
Figure 3. (a) Current distributions. (From Reference 3.) (b) Radiation patterns for a 1.25 dipole antenna.
considered to be composed of a set of innitesimal dipoles.
As the innitesimal dipoles do not create E and H elds or
radiated power density for the = 0 and 180 directions,
the sum of many of them also does not have power in these
directions.
Linear Dipole Antennas
The geometry of a linear dipole antenna of length-2h is
shown in Fig. 1. The current distribution is sinusoidal and
is given by
I(z) =
I(0)
sink(h |z|) f or h < z < h
sinkh
(5)
where I(0) is the current at the feed point of the antenna,
h is the half length of the antenna, and k = ()1/2 is
the propagation constant in the material surrounding the
dipole. The current distributions for several lengths of
dipole antennas are shown in Fig. 3.
The electric and magnetic elds around the dipole are
calculated by modeling the antenna as a series of Hertzian
dipoles having different current strengths shown by the
patterns in Fig. 1 and by integrating the elds from each
of these elements. The resultant elds far from the antenna
at a distance r are
E=
jI(0)
F ()e j(tkro )
2ro sin(kh)
(6)
H=
jI(0)
F ()e j(tkro )
2ro sin(kh)
(7)
and
where = ()1/2 , and the -dependence of the radiated
elds F() is called the pattern factor and is given by the
following:
F () =
cos(khcos()) coskh
sin
(8)
The radiated power density (radiation pattern) is given by
P() =
E
E
I 2 (0)
F 2 ()
r =
2
2
8 r2 sin2 (kh)
(9)
Figure 4. Variation of feed-point Ra and reactance Xa for an endfed monopole above ground as a function of height h/. Values for
dipoles are double that of monopoles. (From Reference 4, p 543.)
Using = o = 120, this can also be expressed in terms of
the total radiated power W(= I 2 (0)Ra /2) and the feed-point
resistance Ra as follows:
P() =
30 W F 2 ()
ro2 Ra sin2 kh
(10)
The normalized radiation patterns are shown in Fig. 3b for
several different lengths of dipoles and monopoles above
ground. The radiation patterns for the dipole and monopole
are the same, except that the monopole does not radiate in
the lower half plane and therefore radiates twice the power
in the upper half plane. Mathematically, this is accounted
for, because the feed-point resistance Ra of the monopole
is half that of the dipole, as will be discussed later in this
section.
The directivity of a dipole antenna is the power density
in the direction of maximum radiation (at a specied distance ro ) normalized by the power density of an isotropic
radiator. Directivity is given by
D=
Pmax
=
Po
F 2 ()max
1
2
0
F 2 ()sind
120 F 2 ()|max
Ra sin2 kh
(11)
where Po = W/(4r2 ) is the power density of an isotropic
radiator.
A graph of the variation of the feed-point resistance and
reactance of a monopole antenna above ground is given in
Fig. 4 as a function of length h/, where is the free-space
wavelength at the radiation frequency (4). Note that the
reactance Xa depends on the conductor radius a, whereas
the feed-point resistance Ra is relatively independent of
conductor radius a for thin antennas (a/ 1). The input
Dipole Antennas
Figure 5. Input Impedance of a cylindrical dipole antenna with
h/a = 60. From [E. Hallen, Admittance Diagrams for Antennas and
the Relation between Antenna Theories, Cruft Laboratory, Harvard University, Technical Report 46, June 1948].
resistance Ra of a center-fed dipole antenna of length 2h is
twice that of an end-fed monopole of length h.
The ohmic losses of a dipole antenna (given by I2 (0)
Rohmic /2) are small, particularly for h/ > 0.1. The resultant
antenna radiation efciencies [given by Ra /(Ra + Rohmic )]
are on the order of 9099%.
Physical dipoles act slightly differently than predicted,
because they have some nite thickness, and the ends of
the wire capacitively couple to air. This effectively makes
the antenna electrically longer than its physical length by
2% to 9%. For a half-wave dipole (length = 2h = /2), for
instance, the physical length must be slightly shortened to
create a resonant length antenna (Xa = 0). Table 1 shows
the wire lengths required to produce a resonant half-wave
dipole. This shortening factor varies from 2% to 9%, depending on the thickness of the dipole.
As a dipole antenna is a physically resonant structure,
its feed-point impedance (particularly the reactance Xa )
varies greatly with frequency. The input impedance of a
dipole antenna is shown in Fig. 5. Thus, these antennas
have a fairly narrow bandwidth. The VSWR of a dipole
antenna as a function of frequency and wire thickness is
shown in Fig. 6 for an antenna that would be half-wave
resonant at 300 MHz. Using a measure of useable bandwidth that the measured VSWR should be less than 2:1,
this antenna has bandwidths of 310262 = 48 MHz for
the thicker wire and 304280 = 24 MHz for the thinner
wire. As fractions of the design frequency (300 MHz), the
bandwidths are 16% and 8%, respectively.
Specialized Dipole Antennas
Slot Dipole. A slot dipole antenna is a dual to the linear
dipole antenna. A slot antenna is produced by creating a
thin slot of length h in a conducting metal sheet and by
feeding it in the center of the slot. The radiation pattern
of a slot antenna is identical to that of the linear dipole of
the same length (see Fig. 3) except that orientations of the
E and H are interchanged, which means that the power
pattern is the same, even though the orientation of the
elds is different. Also the feed-point impedance Zs of a
slot antenna is related to that of the dual linear antenna
Figure 6. VSWR of a dipole antenna as a function of frequency
and wire thickness. (From Reference 6.)
Figure 7. Biconical dipole antenna and variations: (a) biconical
dipole antenna, (b) at bow-tie antenna, and (c) wire version of
biconical dipole antenna.
by the following equation:
Zs =
2
4Za
(12)
where Zs is the impedance of the slot and Za is the
impedance of the dual linear antenna. Cavity-backed slot
antennas with integrated matching networks are among
the smallest (physical size/electrical size) antennas available (7).
Biconical Dipoles. A biconical dipole, such as shown in
Fig. 7a, is commonly used for broadband applications. Typical are angles are between 30 and 60 . The exact are
angle is not critical, so it is generally chosen so that the
impedance of the dipole nearly matches the impedance of
the feed line to which it is connected. The impedance of the
biconical dipole varies as a function of wavelength and are
angle, with a relatively at impedance response for wide
are angles. Hence, this antenna is broader band than a
simple linear dipole. Some variations of this method of using aring to increase bandwidth are the at bow-tie antenna (which may be built on a printed circuit board) and
the wire version of the biconical antenna shown in Fig. 7b
and c, respectively.
Folded Dipole Antennas. A folded dipole antenna is
shown in Fig. 8. The dipole is created by joining two cylin-
Dipole Antennas
Table 1. Wire Lengths Required to Produce a Resonant Half-Wave Dipole for a Wire Diameter of 2a and a Length L
Length-to-diameter ratio, L / (2a)
Percent shortening required
Resonant length
Dipole thickness class
2
5
9
0.49
0.475
0.455
Very thin
Thin
Thick
5000
50
10
Table 2. Relationships Between Monopole and Dipole Antennas
Monopole above ground-length = h
Radiation pattern
Feed-point reactance Ra
Feed-point reactance Xa
Directivity Da
Corresponding dipole of twice length L = 2h
Same as that for the dipole but only for angle 0 90
1
Ra,d (2h)
2
1
Xa mono pole| = Xa,d (2h)
2
1
Da mono pole| = 2Da,d (2h)
2
Ra mono pole| =
Ra,d : function of lengthL = 2h (see Fig. 2)
Xa,d : function of lengthL = 2h (see Fig. 2)
Da,d : function of lengthL = 2h
Figure 8. Folded dipole antenna.
Figure 9. Shunt-fed dipoles: (a) delta match and (b) T-match.
drical dipoles at the ends and driving the entire structure by a transmission line (often a two-wire transmission
line) at the center of one arm as shown. The feed-point
impedance of a folded dipole of two identical-diameter arms
is four times as large as for an unfolded dipole of the same
length, which can actually be advantageous, because the
feed-point resistance may now be comparable with the
characteristic impedance Zo of the transmission or feed
line. The reactance of the antenna may easily be compensated by using a lumped element with a reactance that
is negative of the reactance at the terminals of the folded
dipole antenna or else by using a foreshortened antenna
length to resonant length arms so that Xa = 0 (see Table
1).
Shunt-Fed Dipoles. Matching networks of reactive elements are generally required to match the feed-point
impedance (Ra + j Xa ) of center-fed dipoles to transmission
lines. Typically these lines have characteristic impedance
on the order of 300 to 600 , and a thin half wave dipole
has impedance Z = 73 + j 42 . To alleviate the need for
matching networks, the dipoles are at times shunt-fed at
symmetric locations off the center point as shown in Fig.
9. This procedure using either the delta match (Fig. 9a) or
the T-match (Fig. 9b) is often used for half-wave dipoles
(2h = /2) with A and B dimensions that are typically on
the order of 0.10 to 0.15 .
Figure 10. Sleeve dipole antenna: (a) physical model and (b)
equivalent electrical model.
Sleeve Dipole. The sleeve dipole antenna and its equivalent electrical model are shown in Fig. 10. In practice,
this antenna is built from a coaxial line with the outside
conductor and insulation stripped away from the center
conductor, which is left protruding. The outer conductor is
connected to the ground plane, and the image produced by
the ground plane creates an equivalent sleeve dipole antenna. These dipoles are useful, because they have a broad
band VSWR over nearly an octave of bandwidth.
Dipole Antenna Arrays
Dipole antennas and arrays of dipole antennas are used
for short wave (3 MHz to 30 MHz) and for VHF and UHF
(30 MHz to 900 MHz) radio and TV broadcasting. If directional communication is desired such as for short-wave
radio transmission via the ionosphere, a phased array of
horizontal dipoles may be used mounted above a ground
plane. The spacing is chosen to send the major lobe of radiation toward the sky at a suitable angle to reect off the
ionosphere and to provide broadcast coverage over the desired service area.
For VHF and UHF radio and TV broadcasting over a
360 azimuthal angle, colinearly mounted vertical dipoles
that are excited in-phase with each other are often used.
Two examples of this are shown in Fig. 11a and b. An example variation of this is a three- to eight-bay turnstile
Dipole Antennas
Figure 11. Colinearly mounted vertical dipoles for VHF and
UHF radio and TV broadcasting: (a) pole-mounted array of colinear dipoles and (b) vertical dipoles spaced around a pole.
Figure 13. Log-periodic dipole array: (a) geometry of a logperiodic array showing how the phase-reversal feed system for
this antenna is constructed. (From Reference 8.) (b) Equivalent
antenna model of the log-periodic array. Mc Graw-Hill 1993.
and its adjacent shorter neighbor is given by
=
Figure 12. Variation on colinearly mounted vertical dipoles: (a)
turnstile antenna used for TV broadcasting and (b) one of two
perpendicular slot antennas that comprise each turnstile.
antenna used for TV broadcasting shown in Fig. 12a. Each
turnstile is made of two perpendicular slot antennas as
shown in Fig. 12b.
Log-Periodic Antennas. For broadband applications, logperiodic antennas are commonly used as both transmitting
and receiving antennas. The bandwidth is easily controlled
by adjusting the relative lengths of the longest and shortest
elements in the array. The geometry of a log-periodic array
is shown in Fig. 13a, which shows how the phase-reversal
feed system for this antenna is constructed. The equivalent
antenna model of this array is shown in Fig. 13b. The elements of the array are dipole antennas that increase in
both length and spacing according to the formula:
where = f n / f n+1 is the ratio of the resonant frequencies
fn and fn+1 of the adjacent dipole
=
Rn+1
dn+1
=
Rn
dn
(13)
elements. As lengths and spacings are interrelated, the
choice of one initial value controls the design of the remaining elements. The spacing between one half-wave dipole
dn
(1 )
cot
=
2Ln
4
(14)
Log-periodic arrays are generally constructed with
small values of [10 45 (3)] and large values of
[0.95 0.7 (3)] that essentially gives a traveling wave
propagating to the left in the backre direction, away from
the antenna array. The nature of this array is that only the
elements that are approximately a half-wavelength long
radiate, and as they are radiating to the left, the smaller
elements do not interfere with them. This effect is accomplished by the phase-reversal of the feeds. An array that
is built without the phase-reversal radiates in the end-re
direction. The interference of the longer elements to the
right of radiating elements results in spurious reections
and erratic impedance behavior, which is known as the end
effect.
An effective way to further increase the bandwidth of
a log-periodic antenna is to change from dipole elements
to elements with individual broader bandwidths, similar
to changing from a dipole antenna to a biconical antenna.
This effect is accomplished for log-periodic arrays by using
a conguration of wires such as shown in Fig. 14, where
each element is a sawtooth element and therefore has
broader bandwidth than the individual dipole elements.
Broadband Dipole Curtain Arrays. A broadband dipole
curtain such as shown in Fig. 15 is commonly used for highpower (100 kW to 500 kW), high-frequency ionospheric
broadcasting and short-wave broadcasting stations. The
curtain is composed of several dipoles, usually a halfwavelength long, mounted horizontally or vertically in a
rectangular or square array, often backed by a reecting
Dipole Antennas
Figure 16. YagiUda array.
Figure 14. LPA with sawtooth wire elements for increased bandwidth.
Figure 15. A broadband dipole curtain.
plane or wire mesh. This array has several desirable features, including high gain, broad bandwidth, independent
control of horizontal and vertical radiation patterns, ease of
matching (low VSWR), and the ability to efciently broadcast efciently. Using a phased-feed system, this array allows beam steering of the radiation pattern in both the
azimuthal and the elevation planes, providing a very high
degree of exibility.
YagiUda Dipole Array. YagiUda arrays are commonly
used as general-purpose antennas from 3 MHz to 3000
MHz, in particular, as home TV antennas. They are inexpensive, have reasonable bandwidth, and have gains up
to 17 dBi or more if multiple arrays are used. They have
unidirectional beams with moderate side lobes (8).
A typical YagiUda array is shown in Fig. 16. This array
is a simple end-re array of dipole antennas where only one
element is driven and the rest are parasitic. The parasitic
elements operate as either reectors or directors. In general (8), the longest antenna, which is about /2 in length, is
the main reector and is generally spaced /4 in back of the
driven dipole. The feed element is commonly a folded dipole
antenna 0.45 to 0.49 long. Adding directors, which are
generally 0.4 to 0.45 long, to the front of the driven
element increases the gain of the array. The directors are
not always of the same length, diameter, or spacing. Common arrays have 6 to 12 directors and at most 2 reectors.
Additional improvements in gain by adding more elements
are limited; however, arrays have been designed with 30 to
40 elements (3). A gain (relative to isotropic) of 5 to 9 per
wavelength of array length is typical for YagiUda arrays,
for an overall gain of 50 to 54 (14.8 to 17.3 dB).
The YagiUda array is characterized by a main lobe
of radiation in the direction of the director elements and
small side lobes. The beamwidth is small, generally 3060
(3). Typical E and H plane patterns of a YagiUda array are
shown in Fig. 17. Typically, the performance of a Yagi-Uda
array is computed using numerical techniques (10). For
the simple case where all elements are approximately the
same size, the electric eld pattern can be computed from
the array factors of the various elements.
The input impedance of a YagiUda array is often small.
For example, for a 15-element array with reector length =
0.5 , director spacing = 0.34 , and director length = 0.406
, the input impedance is 12, 22, 32, 50, or 62 for reector
spacings of 0.10, 0.13, 0.15, 0.18, and 0.25 , respectively.
This small inpedance can make matching to typical transmission lines (50, 75, or 300 ) difcult. Folded dipoles used
for the driven element are therefore used to boost the input
impedance by a factor of four or more.
Extensive studies of the design of YagiUda arrays have
been made (11, 12), and tables are provided to optimize the
YagiUda array for a desired gain.
Crossed Dipoles for Circular Polarization. For applications that require a circularly polarized antenna such as
TV and FM broadcasts and space communications, at least
two dipoles, each of which has a linear polarization, must be
combined in an array, often referred to as crossed dipoles.
Several possible congurations are shown in Fig. 18. In a
crossed dipole conguration, dipoles are mounted perpendicular to each other for circular polarization or at other
angles for elliptical polarization. Currents are fed 90 out
of phase between the two dipoles. These currents can also
be used as probes for sensing vector elds to isolate individual components of the electric eld. Adaptations of the
crossed dipole are shown in ca and b. Dipole arrays such
as the YagiUda can also be combined to provide circular
polarization, as shown in Fig. 13c.
Modern Applications of Dipole Antennas
Printed Dipole Antennas. Monopoles and dipoles are not
limited to wire devices. Printed monopoles (13) and dipoles
(14) have also been designed and are in many cases easier to fabricate than a wire-type device. Figure 19 shows
one such printed monopole used for dual-band wireless local area network coverage in the Industrial, Scientic, and
Medical (ISM) bands of 2.42.4835 GHz and 5.155.825
GHz, and Figure 20 shows a printed dipole used for measurement of electric elds for evaluation of cell phone compliance with ratio-frequency (RF) exposure guidelines.
Dipole Antennas for Medical Imaging. Confocal imaging
for breast cancer detection is an exciting application of antenna arrays in medical imaging. This method typically
uses a single antenna scanned in a at array pattern above
the breast or a cylindrical array of very small broadband
antennas (15). For planar imaging, the patient lies face up,
and the antenna is physically scanned in a plane above the
Dipole Antennas
Figure 17. Typical E and H plane patterns of a YagiUda array. Total number of elements = 27, number of directors = 25, number of
reectors = 1, number of driven elements = 1, total length of reector = 0.5 , total length of feeder = 0.47 , total length of each director =
0.406 , spacing between reector and feeder = 0.125 , spacing between adjacent directors = 0.34 , and radius of wires = 0.003 . (From
Reference 9.) IEEE 1969.
breast (1618). For cylindrical imaging, the patient lies face
down, with the breast extending into the cylindrical array
through a hole in the table (19, 20). Matching uid surrounding the breast, similar to that used for microwave
tomography, is suggested in this case. Both methods provide similar results (20). One antenna in the array transmits an ultra-wideband (UWB) pulse, which propagates
into the breast, where it is reected off signicant electrical discontinuities and is received in parallel by the other
antennas in the array. Knowing the physical spacing between the array elements, the different delays among the
transmit antenna, scattering point, and receiving antenna
can be calculated geometrically. The received pulses representing a specic point in space can then be time delayed
appropriately for each antenna, added up, and integrated
to indicate the magnitude of the scattered energy from that
point in space. This process is effectively correlating the
signals received from that point at all antennas.
The antennas used for confocal imaging must be ultrawideband and small enough to t within the relatively
small array area. Resolution of less than 1 cm requires
a bandwidth of at least 5 GHz. The lossy nature of tissue attenuates high-frequency signals, limiting the upper
frequency to about 10 GHz. Initially, resistively loaded
bow-tie antennas were suggested for the planar conguration 1618,21,22, whereas dipole antennas were suggested
for the cylindrical system (19, 20). Resistively loaded Vee
dipoles have also been proposed (23). In the cylindrical conguration, multiple antennas are present in the array, although they are not simultaneously active. In the planar
system, a single antenna is scanned over the surface, creating a synthetic antenna aperture. To overcome the inherent inefciency of resistively loaded antennas, a modied
ridged horn antenna operating from 1 to 11 GHz has been
introduced (23). Most antennas are designed to observe
copolarized reections from the breast; however, using two
resistively loaded bow-tie antennas in the shape of a Maltese cross shown in Reference 24 has also been proposed to
pick up the cross-polarized reections (18). Cross-polarized
reections from simple tumor models were also examined
in References 24 and 25.
The antenna shown in Figure 21 (24) consists of two
cross-polarized bow-tie antenna elements, an octagonal
cavity behind the bow-tie elements, and a metal ange attached to the cavity. The broadband bow-ties have are
angles of 45 . They are 1.67 cm long, which is a halfwavelength at 3 GHz in fat (similar to breast). The octagonal cavity blocks waves radiated away from the breast.
The cavity is approximated as a circular waveguide lled
with fat material for matching and size reduction. The rst
cutoff frequency is set to be 2 GHz for 24 GHz operation.
The cavity length is a quarter-wavelength, which is 11 mm
at 3 GHz. The ange consists of an inner and outer component and is designed to block unwanted waves such as
surface waves. The antenna performance does not change
signicantly when the ange size is varied between 10 and
6.25 cm; therefore, the width of the outer ange is set to be
6.25 cm. The inner ange is designed to prevent possible
electric eld overshoot at the inner corners of the opening
of the octagonal cavity or at the ends of the bow-tie elements. A slotline bow-tie antenna has also been proposed
in Reference 26.
Vector Antennas. Loop antennas have received considerable attention lately in the development of compact wireless communication systems when integrated with other
antennas such as dipoles. This use of loop antennas leads to
the possible development of compact Multiple Input Multiple Output (MIMO) systems, which can nd applications
not only in communication systems, but also in the direction of arrival estimation, sensor networks, and imaging.
Although, traditionally, the increase in channel capacity in
Dipole Antennas
Figure 20. Miniature printed dipole antenna. (From Reference
14, IEEE 1997.
Figure 21. Cross-polarized antenna for confocal imaging. The
properties of the substance inside the cavity and the medium outside the antenna are similar to fat (r = 9; = 0:2 S/m). (From
Reference 24.) IEEE 2005.
Figure 18. Cross-dipole applications for circular or elliptical
polarization. (From Reference 8.) (a) Two shunt-fed slanted Vdipoles, (b) series-fed slanted dipoles, and (c) Circularly polarized
YagiUda array.
Figure 22. Vector antenna.
Figure 19. Geometry and dimensions of the proposed dual ISMband antenna. (From Reference (13), IEEE 2005.)
a communication system has been achieved through spatial array of antennas, similar increases in channel capacity have been obtained through the use of vector antennas consisting of colocated loops and dipoles, which can
respond to more than one component of the electromagnetic eld. One example of such antennas is given in Reference 27, where three- and four-element vector antennas,
consisting of one loop and two or three dipoles, respectively,
are employed in a MIMO system. Several research efforts
that address the characterization of systems of colocated
loops and dipoles can be found in the literature (2732).
Dipole Antennas
In Reference 27, for example, it is shown experimentally
that, in a rich multipath, scattering environment systems
with three- and four-element vector antennas at both the
transmitter and the receiver support three and four times
more information, respectively, as compared with conventional systems consisting of sensors with single antennas.
Fig. 22 shows an example of a three-element vector antenna. In this particular system, as the loop must have a
constant current distribution to retain the radiation characteristics of a magnetic dipole, the loop was realized by
means of four pie-shaped sectors fed in phase at their
corners (33). This arrangement ensured that the currents
were directed in opposite directions along adjacent feed
lines, thus effectively nullifying any spurious radiation.
Ultra-wideband versions of such systems have also been
developed for imaging purposes.
BIBLIOGRAPHY
1. Balanis, C. A. Antenna Theory: A Review. Proc. IEEE 1992, 80.
2. Johnson, R. C. Antenna Engineering Handbook, 3rd ed.;
McGraw-Hill: New York, 1993.
3. Balanis, C. Antenna Theory, Analysis and Design, 3rd ed.; Wiley: New York, 2005.
4. Jordan, E. C.; Balmain, K. G. Electromagnetic Waves and Radiating Systems, 2nd ed.; Prentice-Hall: Englewood Cliffs, NJ,
1968.
5. Jordan, E. C.; Balmain, K. G. Electromagnetic Waves and Radiating Systems; Prentice Hall: Englewood Cliffs, NJ, 1968.
6. Stutzman, W. L.; Thiele, G. A. Antenna Theory and Design;
Wiley: New York, 1991.
7. Hong, W.; Behdad, N.; Sarabandi, K. Size Reduction of CavityBacked Slot Antennas. IEEE Trans. Antennas Propagat 1996,
54, pp 14611466.
8. Johnson, R. C. Antenna Engineering Handbook, 3rd ed.;
McGraw-Hill: New York, 1993.
9. Theile, G. A. Analysis of Yagi-Uda Antennas. IEEE Trans. Antennas Propagat. 1969, 17.
10. Thiele, G. A. Analysis of Yagi-Uda type antennas. IEEE Trans.
Antennas Propagat. 1966, 14, pp 648649.
11. Cheng, D. K.; Chen, C. A. Optimum Spacings for Yagi-Uda
Arrays. IEEE Trans. Antennas Propagat. 1973, 21, pp 615
623.
12. Chen, C. A.; Cheng, D. K. Optimum Element Lengths for YagiUda Arrays. IEEE Trans. Antennas Propagat. 1975, 23, pp
815.
13. Chen, I.-F.; Peng, C.-M.; Liang, S.-C. Single Layer Printed
Monopole Antenna for Dual ISM-Band Operation. IEEE
Trans. Antennas Propagat. 2005, 53, pp 12701273.
14. Bassen, H. I. Electric Field Probes for Cellular Phone Dosimetry. Proc. 19th International Congress IEEE/EMBS Society;
Chicago, IL, Oct. 30Nov. 2, 1997.
15. Camart, J.-C. et al.Coaxial Antenna Array for 915 MHz Interstitial Hyperthermia: Design and ModelizationPower Deposition and Heating PatternPhased Array. IEEE Trans. Microw. Theory Tech. 1992, 40, pp 22432250.
16. Fear, E. C.; Hagness, S. C.; Meaney, P. M.; Okieniewski, M.;
Stuchly, M. Enhancing Breast Cancer Detection using Near
Field Imaging. IEEE Microw. Mag. 2002, pp 4856.
17. Hagness, S. C.; Taove, A.; Bridges, J. E. Two-Dimensional
FDTD Analysis of a Pulsed Microwave Confocal System for
Breast Cancer Detection: Fixed-Focus and Antenna-Array
Sensors. IEEE Trans. Biomed. Eng. 1998, 45, pp 14701479.
18. Hagness, S. C.; Taove, A.; Bridges, J. E. Three-Dimensional
FDTD Analysis of a Pulsed Microwave Confocal System for
Breast Cancer Detection: Design of an Antenna-Array Element. IEEE Trans. Antennas Propagat. 1999, 47, pp 783
791.
19. Li, X.; Hagness, S. C. A Confocal Microwave Imaging Algorithm for Breast Cancer Detection. IEEE Microw. Wireless
Comp. Lett. 2001, 11, pp 130132.
20. Fear, E.; Stuchly, M. Microwave System for Breast Tumor Detection. IEEE Microw. Guided Wave Lett. 1999, 9, pp 470
472.
21. Yun, X.; Fear, E. C.; Johnston, R. H. Compact Antenna for
Radar-Based Breast Cancer Detection. IEEE Trans. Antennas
Propagat. 2005, 53, pp 23742380.
22. Hernandez-Lopez, M. A.; Pantoja, M.; Fernandez, M.; Garcia,
S.; Bretones, A.; Martin, R.; Gomez, R. Design of an UltraBroadband V Antenna for Microwave Detection of Breast Tumors. Microw. Opt. Tech. Lett. 2002, 34, pp 164166.
23. Hagness, S. C.; Taove, A.; Bridges, J. E. Wideband Ultralow
Reverberation Antenna for Biological Sensing. Electron, Lett.
1997, 33, pp 15941595.
24. Fear, E. C.; Stuchly, M. A. Microwave Detection of Breast
Cancer. IEEE Trans. Microwave Theory Tech. 2000, 48, pp
18541863.
25. Li, X.; Hagness, S. C.; Choi, M. K.; Choi, D. W. W. Numerical and Experimental Investigation of an Ultrawideband
Ridged Pyramidal Horn Antenna with Curved Launching
Plane for Pulse Radiation. IEEE Antennas Wireless Propagat.
Lett. 2003, 2, pp 259262.
26. Yun, X.; Fear, E. C.; Johnston, R. H. Radar-Based Microwave
Imaging for Breast Cancer Detection: Tumor Sensing with
Cross-Polarized Reections. IEEE Antennas Propagat. Soc.
Symp. Dig. 2004, 3, pp 24322435.
27. Konanur, A.; Gosalia, K.; Krishnamurthy, S.; Hughes, B.; Lazzi,
G. Compact MIMO Systems Employing Vector Antennas for
Increased Wireless Channel Capacity. IEEE Trans. Microw.
Theory Tech. 2005, pp 18371844.
28. Andrews, M. R.; Mitra, P. P.; de Carvalho, R. Tripling the Capacity of Wireless Communications using Electromagnetic Polarization. Nature 2001, 409, pp 316318.
29. Poon, A. S. Y.; Brodersen, R. W.; Tse, D. N. C. Degrees of Freedom in Multiple Antenna Channels: A Signal Space Approach.
IEEE Trans. Inform. Theory 2005, 51, pp 523536.
30. Svantesson,T.; Jensen, M. A.;Wallace, J. W. Analysis of Electromagnetic Field Polarizations in Multiantenna Systems. IEEE
Trans. Wireless Commun. 2004, 3, pp 641646.
31. Andersen, J.; Getu, B. The MIMO CubeA Compact MIMO
Antenna. Proce. 5th Wireless Personal Multimedia Communications Int. Symp. 2002, 1, pp. 112114.
32. Stancil, D. D.; Berson, A.; Hof, J. P. V.; Negi, R.; Sheth, S.; Patel, P. Doubling Wireless Capacity using Copolarized, Colocated Electric and Magnetic Dipoles. Electron. Lett. 2002, 38,
pp 746747.
10
Dipole Antennas
33. Kandonian, A. G. Three New Antenna Types and Their Applications. Waves Electrons 1946.
Reading List
ITT Handbook. Reference Data for Radio Engineers, 5th ed.;
Howard W. Sams: Indianapolis, In, 1973.
Gandhi, O. P.; Chen, J. Y. Electromagnetic Absorption in Human Head from Experimental 6-GHz Hand-Held Transceivers.
IEEE Trans. Electromagn. Compat. 1995, 37, pp 547558.
Tai, C. T. On the Theory of Biconical Antennas. J. Appl. Phys. 1948,
pp 11551160.
King, R. W. P. Theory of Linear Antennas; Harvard University
Press: Cambridge, MA, 1956.
Ma, M. T.; Spies, K. P. A Simplied Formulation Computation for
Conical Monopole Antennas. U.S. Dept. of Commerce Report OT
74-53, 1974.
Hicks, D. E. CB Radio Antennas; Howard W. Sams: Indianapolis,
IN, 1967.
The A.R.R.L. Antenna Book; The American Radio Relay League:
West Hartford, CT, 1956.
Johnson, R. C.; Jasik, H. Antenna Engineering Handbook, 2nd ed.;
McGraw-Hill: New York, 1984.
Lazzi, G., et al. Comparison of FDTD-Computed and Radiation
Patterns of Commercial Mobile Telephones in the Human
Head. IEEE Trans. Antennas Propagat. 1998, 46.
Luebbers, R., et al. FDTD Calculation of Radiation Patterns,
Impedance and Gain for a Monopole Antenna on a Conducting Box. IEEE Trans. Antennas Propagat. 1992, 40, pp 1577
1583.
Jensen, M. A.; Rahmat-Samii, Y. EM Interaction of Handset Antennas and a Human in Personal Communications. Proc. IEEE
1995, 83, pp 717.
Okoniewski, M.; Stuchly, M. A. A Study of the Handset Antenna
and Human Body Interaction. IEEE Trans. Microw. Theory
Tech. 1996, 44, pp 18551864.
Lazzi, G. et al. Comparison of FDTD-Computed and Radiation Patterns of Commercial Mobile Telephones in the Human Head.
IEEE Trans. Antennas Propagat. 1998, 46.
Gandhi, O. P.; Lazzi, G.; Furse, C. M. Electromagnetic Absorption
in the Human Head and Neck for Mobile Telephones at 835
and 1900 MHz. IEEE Trans. Microw. Theory Tech. 1996, 44, pp
18841897.
Special Issue of IEEE Trans. Microw. Theory Tech. 1986, MTT-34.
Durney, C. H.; Iskander, M. F. Antenna Handbook, Lo, Y. T., Lee, S.
W. Eds.; 1993.
Sneed, P. K.; Phillips, T. L. Combining Hyperthermia and Radiation: How Benecial? Oncology, 1991, 5, pp 99108.
Vernon, C. C.; Hand, J. W.; Field, S. B., et al. Radiotherapy With or
Without Hyperthermia in the Treatment of Supercial Localized Breast Cancer: Results from Five Randomized Controlled
Trials. Int. J. Radiat. Oncol. Biol. Phys. 1996, 35, pp 731744.
Turner, P. F. Interstitial Equal-Phased Arrays for EM Hyperthermia. IEEE Trans. Microw. Theory Tech. 1986, 34, pp 572
578.
Furse, C. M.; Iskander, M. F. Three-Dimensional Electromagnetic
Power Deposition in Tumors using Interstitial Antenna Arrays.
IEEE Trans. Biomed. Eng. 1989, 36, pp 977986.
Nevels, R. D.; Arndt, G. D.; Raffoul, G. W.; Carl, J. R.; Pacico, A.
Microwave Catheter Design. IEEE Trans. Biomed. Eng. 1998,
45, pp 885890.
Manry, C.; Broschat, S. L.; Chou, C.-K.; McDougall, J. A. An Eccentrically Coated Asymmetric Antenna Applicator for Intracavity
Hyperthermia Treatment of Cancer. IEEE Trans. Biomed. Eng.
1992, 39, pp 935942.
Fear, E. C.; Stuchly, M. A. Microwave Breast Tumor Detection: Antenna Design and Characterization. IEEE Antennas Propagat.
Symp. Dig. 2000, 2, pp 10761079.
Shannon, C. J.; Fear, E. C.; Okoniewski, M. Dielectric-Filled Slotline Bowtie Antenna for Breast Cancer Detection. Electron.
Lett. 2001, 41.
Sill, J. M.; Fear, E. C. Tissue Sensing Adaptive Radar for Breast
Cancer Detection: A Study of Immersion Liquid. Electron. Lett.
2005, 41, pp 113115.
Sill, J. M.; Fear, E. C. Tissue Sensing Adaptive Radar for Breast
Cancer Detection: Preliminary Experimental Results. Proc.
IEEE MTT-S Int. Microw. Symp. Dig.; Long Beach, CA, June
2005.
Sill, J. M.; Fear, E. C. Tissue Sensing Adaptive Radar for Breast
Cancer DetectionExperimental Investigation of Simple Tumor Models. IEEE Trans. Microw. Theory Tech. 2005, 53, pp
33123319.
Wu, T.; King, R. The Cylindrical Antenna with Nonreecting Resistive Loading. IEEE Trans. Antennas Propagat. 1965, AP-13,
pp 369373.
Wu, T.; King, R. Corrections to The Cylindrical Antenna with Nonreecting Resistive Loading. IEEE Trans. Antennas Propagat.
1965, AP-13, p 998.
Labonte, S., et al. Monopole Antennas for Microwave Catheter
Absorption. IEEE Trans. Microw. Theory Tech. 1996, 44, pp
18321840.
Wolff, E. A.. Antenna Analysis; Wiley: New York, 1966.
Werner, D. H. An Exact Integration Procedure for Vector Potentials of Thin Circular Loop Antennas. IEEE Trans. Antennas
Propagat. 1996, 44, pp 157165.
Smith, G. S. Loop Antennas. In Antenna Engineering Handbook;
McGraw-Hill: New York, 1984.
Storer, J. E. Impedance of Thin-Wire Loop Antennas. AIEE Trans.
1956, 75.
Greatbatch, W.; Holmes, C. F. History of Implantable Devices.
IEEE Eng. Med. Biol. 1991, pp 3842.
Woolons, D. J. To Beat or Not to Beat: The History and Development of Heart Pacemakers. IEE J. Eng. Sci. Educ. 1995, 4, pp
259268.
Allan, R. Medtronic Sets the Pace with Implantable Electronics.
Electron. Design 2003, 51, pp 5256.
Spelman, F. A. The Past, Present, and Future of Cochlear Prostheses. IEEE Eng. Med. Biol. 1999, pp 2733.
Rauschecker, J. P.; Shannon, R. V. Sending Sound to the Brain.
Science 2002, 295, pp 10251029.
Weiland, J. D.; Humayun, M. S. A Biomimetic Retinal Stimulating
Array. IEEE Eng. Med. Biol. Mag. 2005, 24, pp 1421.
Walter, P.; Kisvarday, Z. F.; Gortz, M.; Alteheld, N.; Rossler, G.;
Stieglitz, T.; Eysel, U. T. Cortical Activation via an Implanted
Wireless Retinal Prosthesis. Invest. Ophthalmol. Visual Sci.
2005, 46, pp 17801785.
Margalit, E.; Maia, M.; Weiland, J. D.; Greenberg, R. J.; Fujii, G.
Y.; Torres, G.; Piyathaisere, D. V.; OHearn, T. M.; Liu, W.; Lazzi,
G.; Dagnelie, G.; Scribner, D. A.; de Juan, E.; Humayun, M. S.
Retinal Prosthesis for the Blind. Surv. Ophtalmol. 2002, 47, pp
335356.
Dipole Antennas
Normann, R. A.; Maynard, E. M.; Guilloty, K. S.; Warren, D. J.
Cortical Implants for the Blind. IEEE Spectrum 1996, pp 54
59.
Zrenner, E. Will Retinal Implants Restore Vision? Science 2002,
295, pp 10221025.
Nicolelis, M. A. L. Brain-Machine Interfaces to Restore Function
and Probe Neural Circuits. Nature Rev. Neurosci. 2003, 4, pp
417422.
Chapin, J. K.; Moxon, K. A. Eds. Neural Prostheses for Restoration
of Sensory and Motor Function. CRC Press: Boca Raton, FL,
2000.
Ghovanloo, M.; Lazzi, G. Transcutaneous Magnetic Coupling of
Power and Data (invited paper). In Wiley Encyclopedia of
Biomedical Engineering; Akay, M. Editor, Wiley (to appear,
2006).
CYNTHIA M. FURSE
OM P. GANDHI
GIANLUCA LAZZI
University of Utah
North Carolina State University
11
DIRECTIVE ANTENNAS
589
Direction of wave
Figure 1. Three sources of waves giving a directed wave front. The
phase of the middle and left sources have been delayed.
DIRECTIVE ANTENNAS
Radio antennas that are directive have many advantages
over stationary antennas. They can observe in one direction
while ignoring much larger signals coming in from the
sides. In addition, as transmitting antennas, they can direct
a small amount of radiation directly to the proper receiver.
Conventional directive antennas use large mirrors for short
waves, such as are used in radar. For longer-wave antennas, they use reflector and director elements aligned with
the antenna. The basic problem with such arrays is that
they are mechanically steered, which is slow and difficult
to do with large antennas, especially in the presence of
wind, ice, or snow.
One way of having a large steerable antenna with no moving parts is to use a phased array. A phased array is based
on Huygens principle, which states that a wave front can be
determined at a point in time by constructing a surface tangent to a collection of secondary waves. Thus, if one has a
large number of small antennas located on a line, and if the
antennas are excited in phase, the resultant wave front is
normal to the line. If each small antenna is excited with a
small phase shift relative to the next antenna, the wave propagates at an angle to the line. Conversely, if the array is used
for reception, the small antenna signals, combined with the
individual phase shifts, allow the array to scan in azimuth.
The principle can be improved by using a two-dimensional
array located on a plane to scan in altitude as well as azimuth. In fact, by properly adjusting the phase shifts, such an
antenna can be located on a curved surface, such as the nose
or wing of an aircraft. The basic principle is shown in Fig. 1.
Since the subject of directive antennas is so large, Professor
Herb Neff, UTK Emeritus (an antenna specialist), recommends using Refs. 1 and 2 for reference.
Newer antennas being developed include the agile mirror
(3,4) designed by Dr. Wallace M. Manheimer of the U.S. Naval Research Laboratory. In this concept, a sheet of ionized
air or other gas forms a reflecting surface. Since the sheet is
not a mechanical body, it can be tilted and re-formed in a
very short span of time. Thus, the direction of transmission
or reception can be varied extremely rapidly. A magnetic field
is used to help direct the sheet electron beam that forms the
plasma. A second advantage of such an antenna is that it vanishes when the electrical discharge is terminated. This is a
great advantage for stealth technology, because a mechanical
antenna presents a large scattering cross section to radar signals near the antenna resonant frequency. One disadvantage
of the agile mirror is that it must be formed in a gas at
reduced pressure.
A second new type of antenna is the stealth antenna,
which is being developed by the Patriot Scientific Corporation
in San Diego, California. The idea, originally developed by
Dr. Igor Alexeff at the University of Tennessee (5), uses glow
discharge tubes to comprise elements of an antenna. When
the tubes are energized, the antenna is a complete conducting
structure. When the tubes are deenergized, the antenna becomes either a large number of separate, nonresonant conducting components or just a nonconducting structure of glass
tubing. By selectively energizing various tubes, the antenna
also can be directed.
BIBLIOGRAPHY
1. C. A. Balanis, Antenna Theory: Analysis and Design, New York:
Harper & Row, 1982.
2. R. S. Elliott, Antenna Theory and Design, Englewood Cliffs, NJ:
Prentice-Hall, 1981.
3. J. Matthew et al., Phys. Rev. Lett., 77: 1982, 1996.
4. R. A. Meger et al., Phys. Plasmas, 2 (6): 2532, 1995.
5. W. L. Kang, M. Rader, and I. Alexeff, A conceptual study of
stealth plasma antenna, IEEE Conf. Plasma Sci., Boston, MA,
1996, p. 261.
IGOR ALEXEFF
MARK RADER
WENG LOCK KANG
The University of Tennessee
J. Webster (ed.), Wiley Encyclopedia of Electrical and Electronics Engineering. Copyright # 1999 John Wiley & Sons, Inc.
752
DIVERSITY RECEPTION
DIVERSITY RECEPTION
Diversity reception is a radio communication technique that
improves system performance during periods of adverse propagation conditions by providing more than one transmission
channel (or branch) to deliver the signal intelligence to a specified destination. Generally, the goal is to increase the transmission link availability sufficiently to meet prescribed system performance criteria and provide acceptable service. To
take advantage of the multiple channels and increase the link
performance, the capability must be provided either to select
among the available signals, or else to combine the signals.
(An approximate analogy is found in telephony, where many
paths are generally available to the Public Switched Telephone Network to complete a call from one telephone to another, greatly increasing the probability of completing any
given call.)
Usually, the objective in diversity reception is to reduce
performance degradations caused by signal fading, such as
multipath fading in mobile and terrestrial point-to-point systems, or signal attenuation caused by rainfall on the propagation path (rain attenuation) in earth-space (satellite) systems.
Significant performance improvements can also be achieved
with respect to other path impairments, such as unwanted
signal depolarization (as encountered in dual-polarization frequency-reuse communication systems), angle-of-arrival variations, and cochannel interference. In many scenarios, link
performance can of course be enhanced by simply increasing
the transmitter power (assuming this approach is cost-effective), but this option is often precluded by regulations established to limit intersystem interference.
The propagation medium is presumed to cause occasional
degradations in a single transmission channel that are sufficiently severe to justify the expense and complexity of implementing diversity reception. A sufficient understanding of the
propagation environment is essential for designing effective
diversity reception systems. For example, when multiple
earth terminals are installed in earth-space telecommunication systems to reduce rain attenuation outages that would
be experienced on a single path, the minimum site separation
for the diversity terminals is dictated mainly by characteristics of the rain environment, although performance elements
of the earth terminals, such as antenna gain and link fade
margin, are also quite important.
For diversity reception to be effective, impairments on the
separate channels are preferably independent, or at least sufficiently decorrelated that simultaneous severe signal degradations are rare. If the time-varying propagation effects on
the individual channels are highly-correlated, the probability
of simultaneous signal impairments is large, and the benefits
offered by diversity reception will be small. (Anticorrelation
of impairments is even more advantageous than zero correla-
tion, but this condition is not generally attained.) If the propagation medium is spatially uniform or is not time-varying,
there is little reason to use diversity. Diversity reception is
not generally intended to counter slowly-varying macroscopic
(bulk) changes in the propagation environment, as such
changes tend to affect all available channels more-or-less
equally. A general condition to be met in diversity systems is
that the individual diversity signals should have similar
mean received power levels (within 10 dB or so). Otherwise,
the link performance is dominated by the strong signal(s),
with little gain derived from the other channels.
Providing redundant transmission channels to deliver
identical information can be expensive, and inevitably increases the equipment complexity. For example, diversity installations that employ spatially separated antennas must be
connected by a communication link, such as a microwave link
or optical cable, to allow combining of, or switching among,
the signals from the diversity branches. In some installations,
the diversity terminals may be separated by many kilometers.
Furthermore, conditioning of the multiple signals is typically
required to support selection or combining of the signals without losing information (hitless switching), and a decision
criterion or algorithm must be devised to control any diversity
operation. There must usually be an expectation of substantial performance benefits to justify implementation of diversity reception.
A well-designed diversity reception system can yield impressive enhancements in system performance during impaired propagation conditions. There are two standard measures used to quantify the benefits provided by diversity
reception. One measure, diversity gain, specifies the reduction
in single-path impairment level (signal fading in decibels,
usually) achieved with diversity reception for a given operating time percentage (of the year or worst month). Diversity gain equals the decrease in signal-to-noise ratio, SNR
(dB), that is required to meet a given performance criterion,
relative to the SNR that would be required without diversity.
The other measure, called diversity improvement (or diversity
improvement factor or diversity advantage) is defined in the
orthogonal sense as the ratio of the nondiversity and diversity
probabilities of exceeding a specified impairment level. Both
measures are used in this article.
APPLICATIONS OF DIVERSITY RECEPTION
Diversity reception has long been recognized as a viable impairment-mitigation technique in telecommunication systems
(1,2), and applied in a variety of modes to practical systems
(next section). Diversity reception has been used in high-frequency (HF) communications since the 1920s, when spaced
receive antennas were found to yield partially decorrelated
fading signals that could be used to improve path availability.
Diversity operation is virtually mandatory in modern troposcatter communication systems, for which 4-channel (quadruple) diversity operation is common, implemented with dual
spaced antennas at both ends of the link, each capable of
cross-polarized reception or some other form of antenna-pattern diversity (3).
Vertical space diversity (often combined with angle diversity) improves the performance of terrestrial microwave links
(4), and has been evaluated as a countermeasure against low-
J. Webster (ed.), Wiley Encyclopedia of Electrical and Electronics Engineering. Copyright # 1999 John Wiley & Sons, Inc.
DIVERSITY RECEPTION
TYPES OF DIVERSITY RECEPTION
Several major classes of diversity reception are used in communication systems. These methods include space, angle, polarization, antenna-pattern, field-component, frequency, time,
and RAKE diversity (see following subsections). Space diversity generically applies to methods that exploit the spatial
characteristics of the propagating field and its interaction
with the propagation environment. Frequency diversity requires access to an alternate, lesser-impaired frequency band
to which communication traffic may be switched when the
normal channel is impaired. Time diversity relies on retransmission of information that is received with inadequate unintelligibility, as in packet-switched communication systems
whereby packets containing errors are identified at the destination and requested to be resent. In wideband systems that
permit individual multipath components (echoes) within the
delay-spread spectrum to be resolved, a RAKE diversity system, so named because it uses a comb of signal components
that mimic the appearance of a garden rake, can combine the
components to increase the signal power available to the receiver. Time diversity and RAKE diversity are generally used
only for digital systems. (In general, diversity techniques are
easier to implement in digital than in analog systems.)
Frequency diversity imposes a spectrum utilization penalty, since spare capacity must be reserved for access during
adverse propagation conditions. Likewise, time diversity involves a penalty in information throughput, as some intelligence must be retransmitted, and overhead bits are required
to identify information packets and control the diversity operation. For these reasons, these methods are often inefficient
from the standpoint of spectrum utilization (12).
Space Diversity
Space diversity, probably the most common and easily visualized form of diversity, relies on the provision of two or more
spatially separate propagation paths, typically by installing
more than one receive antenna (or equivalently, more than
one transmit antenna, then called diversity transmission) at
10.0
5.0
Time percentage attenuation is exceeded
angle refractive fading on satellite links (5). Protection
against severe frequency-selective (notch) fading is achieved
by reserving an alternate frequency-diversity channel to protect several other channels that suffer notch fading (6). (In
general, however, frequency diversity is considered wasteful
of spectrum, and not recommended for many applications.)
Space diversity using separated base-station antennas has
proven valuable for mobile and cellular radio systems (7).
Site diversity reception is used on earth-satellite links at
small path elevation angles to decrease the effects of severe
low-angle fading (8), as well as to improve performance during rain impairments for high-reliability earth terminal installations (9). Antennas with small horizontal separations
can be used to decorrelate tropospheric scintillation fading on
earthspace paths (10). Since at frequencies above about 10
GHz, rain impairments are often severe for significant percentages of the time at many locations, site diversity may find
wide application in earthsatellite systems at Ka-band (11),
especially to protect feeder links that carry information from
a central hub earth terminal to a satellite for eventual distribution to user terminals such as mobiles.
753
2.0
1.0
0.5
0.2
Path 1
0.1
0.05
Path 2
Diversity
0.02
0.01
4
6
8
10
12
11.6 GHz rain attenuation (dB)
14
Figure 1. Annual statistics of 11.6 GHz rain attenuation for two
earthspace paths, compared to diversity statistics obtained by always selecting the smaller single-path attenuation. (Copyright 1979
COMSAT Corp. All rights reserved by COMSAT Corp. Used by permission.)
one end of the link. Planning for the multiple paths is governed by the primary path impairment that the diversity configuration must overcome, such as rain attenuation, refractive
fading on satellite paths at low elevation angles, or multipath
fading on terrestrial line-of-sight links. Space diversity has
been investigated in particular for application to the mobilepropagation environment. The benefits can be very large for
the frequent and severe signal fading encountered while communicating with a terminal in motion.
In terrestrial line-of-sight telecommunications, space diversity generally refers to the use of multiple antennas displaced vertically on a tower, mainly to overcome refractivelyinduced fading caused by unwanted multipath propagation on
such links. Other examples include path diversity (access to
adjacent microwave routes in terrestrial fixed telecommunications) and site diversity (deployment of multiple earth terminals in earthspace telecommunications), both generally used
to reduce the effects of severe rainfall attenuation at frequencies above about 10 GHz.
Figure 1 shows 11.6 GHz cumulative rain attenuation distributions derived from measurements with two terminals
separated by 35 km on propagation paths at an elevation
angle of 18 (13). The curves labeled Path 1 and Path 2
are the rain attenuation distributions for the individual
paths. The Diversity curve is the joint cumulative rain attenuation distribution obtained via computer simulation by
always selecting the lesser-faded signal for each concurrent
pair of rain attenuation samples:
AJ (t) = min{A1 (t), A2 (t)}
(1)
DIVERSITY RECEPTION
12
20 GHz
16 GHz
30 GHz
Diversity gain (dB)
10
14 dB
12 dB
10 dB
8 dB
4
6 dB
2
0
4 dB
2 dB
0
10
15
20
25
Site separation (km)
30
Single-path attenuation
15.3 GHz
35
Figure 2. Measured Ka-band dual-site diversity gains (symbols) vs.
site separation, along with curve fits (lines) parameterized in terms
of single-path attenuation. ( 1976 IEEE.)
The increase in path availability achievable by (perfect) diversity switching is observed to be considerable, although independent fading statistics (equivalent to the product of the two
single-path probability distributions) are almost never observed for dual diversity paths (14) because of mesoscale
(widespread) rainfall effects. Practical diversity-switching
systems are unlikely to achieve the degree of improvement
indicated in Figure 1 since the switching algorithm would
purposely be designed to avoid switching under conditions for
which the resulting performance gain is not required to meet
system performance objectives.
Space diversity gains are large during periods of heavy
rainfall since it is highly probable that there is considerable
spatial variability in the rain intensity. For terrestrial and
earthspace links, separating diversity antennas by several
kilometers greatly reduces the probability of simultaneous
large fades on the different paths. Antenna separation is the
dominant parameter, as illustrated in Figure 2 for several
earthspace experiments at frequencies from 15 to 30 GHz
(15). The majority of the available diversity gain is in fact
achieved with separations smaller than 10 to 15 km. At temperate latitudes, diversity gains measured during the spring
and summer (thunderstorm) seasons are generally significantly greater than those observed during the fall and winter (16).
The International Telecommunication Union (ITU) has
compiled results of many earthspace site diversity experiments, mainly in the 10 GHz to 20 GHz range, to derive an
average representation of site diversity improvement that
gives the decrease in unavailable time percentage attained by
dual-site diversity for a specified impairment level (17). The
improvement is plotted in Figure 3 for site separations of
0 km (no improvement) up to 50 km. Results in this figure
are approximate, since weaker influences such as frequency
and path elevation angle are not explicitly taken into account.
The figure indicates that performance gains achieved by increasing the site separation tapers off for separation greater
than about 1520 km. As an example, the curve for a separation of 30 km indicates that a single-path unavailability of
p1 0.01% (about one hour per year) can be reduced by two
orders of magnitude (to p2 0.0001%) by the addition of a
diversity terminal 30 km distant.
An empirical formula (18) has been derived to predict the
dual-site diversity gain, G (dB), in terms of site separation D
(km), frequency f (GHz), path elevation angle (deg), and the
angle (deg) between the path azimuth and the baseline between the two sites (defined so that 90), as
G = a(1 ebD ) e0.025 f (1 + 0.006 ) (1 + 0.002 )
(2)
where a 0.78A 1.94(1 e0.11A) and b 0.59(1 e0.1A),
and A (dB) is the single-path rain attenuation exceeded for a
specified time percentage. Site separation is the dominant factor in this expression for site diversity gain. Three-site diversity has also been examined, but the additional gain achieved
by adding a third terminal is generally marginal (16).
Another form of space diversity envisaged for earth-space
communications is orbital diversity, in which multiple signal
paths are established by providing access to two or more satellites that are within view of an earth terminal antenna.
(The diversity antennas are in space, instead of on the
ground.) Figure 4 shows attenuation time series measured
during a fade event for an orbital diversity configuration using two satellites at frequencies of 19.8 GHz (Satellite 1)
and 18.7 GHz (Satellite 2), with a geostationary orbital angular separation of 32 (19). The peaks in attenuation on the
two paths occur at different times, and the benefits of switching between the two paths in this event would have been substantial.
Orbital diversity requires that at least two satellites be
available, and that spare capacity be reserved or that communication traffic be suitably prioritized so that higher-priority
traffic may be switched upon demand. Since the diversity
paths terminate at an earth station situated in that region of
the atmosphere where rainfall occurs, the diversity gain may
be small when the rain region surrounds the earth station.
1
0
2
5
10
20
30
50
Time percentage for two sites
754
10
1
5
2
km
km
km
km
km
km
km
10 2
5
2
10
3
5
2
10 4
10 1
10 2
10 1
Time percentage for a single site
Figure 3. Relation between annual time percentages that a given
earthspace path attenuation is exceeded single-path and dual-site
diversity. ( ITU. Adapted from Recommendation ITU-R P.618-5.
Used with permission of ITU. Author solely responsible for this presentation.)
DIVERSITY RECEPTION
10
Attenuation (dB)
8
Satellite 2
(18.7 GHz)
6
4
Satellite 1
(19.8 GHz)
2
0
11:10
11:30
Time (GMT)
11:50
Figure 4. Concurrent 20 GHz path attenuation time series for two
earthspace paths with geostationary orbital angular separation of
32, demonstrating orbital diversity for rain attenuation. ( 1995
IEEE.)
Protection of large frequency bandwidths by transferring traffic from one satellite to another (and accommodating the corresponding network reconfiguration) appears difficult.
Angle Diversity
The general technique of configuring antennas to detect signals that have propagated along different paths, and which
appear at the receive antenna with different directions of arrival, is called angle diversity. On terrestrial paths, vertically
separated space diversity antennas may be mounted with
slightly different pointing angles with respect to the normal
line-of-sight direction to the transmit antenna to enhance reception during periods of angle-of-arrival variations caused by
refractive structures near the earth surface. Vertical beam
separation is preferable to horizontal separation in this application because variation in the refractive structure of the troposphere is more pronounced in the vertical direction. Another approach, used in troposcatter systems, is to employ a
large antenna with several feeds configured to yield somewhat different pointing directions to create some degree of
decorrelation among the corresponding propagation paths.
In troposcatter systems, the optimal vertical separation between the beams is about one antenna beamwidth, although
the range of the optimum is rather broad (3). For terrestrial
microwave paths, more complicated procedures are required
to determine the diversity gain provided by angle diversity,
which is related to the average angle of arrival as determined
by the average value of the vertical refractivity gradient for
the location of interest (4).
Polarization Diversity
For environments in which the polarization properties of a
signal are altered during propagation, orthogonally-polarized
transmission channels may become sufficiently decorrelated
for polarization diversity to be effective. Polarization diversity
is the reception of a signal on two mutually orthogonal polarizations, with or without transmission in the same two polarizations (20). If both polarizations are transmitted, a 3 dB
penalty in transmit power per channel is imposed with respect to single-polarization transmission, as the power must
755
be split between the channels. In contrast to the unwanted
path-induced signal cross-polarization that afflicts dual-polarization frequency-reuse communication systems, signal depolarization is essential in polarization diversity systems. Several polarization-dependent reflections are usually required to
depolarize the incident wave adequately.
Although the technique is limited to two diversity channels
(one polarization and its orthogonal state), no additional frequency spectrum is needed, and a single dual-polarized antenna can be deployed instead of separate space diversity antennas, with positive cost and real estate consequences in
mobile communication systems. The capability to receive both
polarizations with adequate isolation between the channels
must be incorporated into the receiver. Polarization diversity
reception may be particularly beneficial for mobile handheld
terminals, especially to compensate for the random orientations that such a terminal can assume in everyday use (21).
Antenna Pattern Diversity
Antenna pattern diversity, basically the same concept as angle
diversity and polarization diversity, can introduce decorrelation among received signals by, in effect, sampling arriving
wavefronts in different ways (6). This type of diversity can be
accomplished by using adjacent antennas of different types,
for example. Performance improvements with this method are
primarily attributed to the sensitive dependence of frequencyselective (notch) fading on the amplitudes of the separate rays
that destructively interfere to cause such fading, as observed
in terrestrial microwave systems. With different antenna patterns, the probability of simultaneously experiencing the conditions for deep fades at both antennas is much reduced because the received signal amplitudes are less likely to be of
almost equal magnitude, a condition for nearly complete destructive interference.
Field-Component Diversity
The electromagnetic fields of propagating signals are comprised of both electric and magnetic components which may
be designated as Ez, Hx, and Hy, each of which contains the
intelligence of the transmitted signal. In some environments,
particularly mobile radio, these components are uncorrelated
upon arrival at the receive antenna, but the powers in the
field components obey the conservation relation (22)
2 2 2
E = H + H
z
x
y
(3)
If the individual components can be detected and combined,
conditions for diversity reception are met and the average received signal power can be augmented by summing these
components to achieve field-component diversity.
Three methods are envisaged to achieve the desired diversity action. Incoherent combining of the field components
yields a resultant, R,
R = Ez + Hx + Hy
(4a)
Coherent combining of the components is equivalent to
R = Ez + Hx + Hy
(4b)
756
DIVERSITY RECEPTION
A third energy-diversity approach is given by (22)
2 2 2
R = Ez + Hx + Hy
(4c)
which yields a resultant that is approximately constant with
time. Special energy-density antennas have been designed
for the implementation of field-component diversity (22).
Frequency Diversity
If the propagation characteristics of the medium depend significantly on the signal transmission frequency, then frequency diversity, the capability to select the transmission frequency to suit prevailing conditions, can be effective. In
severe multipath fading on terrestrial line-of-site links, for example, the fading is often very frequency selective, such that
deep but narrow notches appear in the receive channel (6). If
provision can be made to switch the information contained in
the channel experiencing the notch fading into a less impaired reserve channel, the probability of successfully transmitting the intelligence can be greatly increased.
The capability of frequency diversity to overcome severe
multipath fading on digital terrestrial links is demonstrated
by Figure 5, which displays bit-error rate (BER) data collected
for Channel 1 and Channel 2 of a 6 GHz, 42.5 km digital
radio link operating at 90 Mb/s during a 2.5 month period of
active multipath fading (23). The center frequencies of the
two channels were separated by 59 MHz. If a typical outage
criterion of 103 BER is specified, the observed diversity improvement factor is about 45, an impressive enhancement in
performance (approximately equivalent to the improvement
expected for a vertical space-diversity separation of 10 m).
The experimental data revealed that although power fading
in the two channels was highly correlated, the multipath dispersion was decorrelated. The latter finding was presumed to
account for the good performance of frequency diversity on
this link.
In another form of frequency diversity envisaged to overcome deep fading associated with rain attenuation, use is
made of the fact that the severity of rain attenuation in-
Duration (s) BER exceeded
104
Channel 1
103
creases rapidly with increasing frequency. If traffic on a highfrequency channel (say, at 14/12 or 30/20 GHz), where rain
attenuation is severe, can be switched to a lower-frequency
(such as 6/4 GHz), the probability of successful transmission
is much enhanced. Frequency diversity requires that spectrum be made available at the diversity frequency to carry
the information that would otherwise be lost in the impaired
channel, and that the traffic either be satisfactorily prioritized or spare capacity held in reserve, to allow the transfer
of protected channels to the alternate satellite when required.
Diversity in this form might be used to protect smaller terminals of modest bandwidth capacity, such as Very Small Aperture Terminals (VSATs), by providing a small spare capacity
on a satellite that can be accessed by terminals when experiencing propagation impairments. However, frequency diversity is less likely to be useful for large trunking earth stations, which generally continuously utilize all of the allocated
frequency spectrum in a given band to support many communication links, and for which alternate provision will be expensive.
Time Diversity
Time diversity refers to the exploitation of the time-varying
nature of the signal impairments to retransmit information
at suitable time intervals. A commonplace analogy is resending an unsuccessfully transmitted facsimile. Information
should be retransmitted at time intervals somewhat greater
than the reciprocal of the signal fading rate, to ensure adequate decorrelation between successive transmissions (1). If
the fading conditions are quite variable, adaptive adjustment
of the time interval may be necessary to attain efficient performance over the anticipated range of environmental situations.
In digital systems, such as packet-switched networks, the
bit stream can be reconstituted from successfully received
packets, possibly by interpolation between successfully-received packets or other sophisticated methods (24), allowing
for powerful implementation possibilities. Overhead capacity
must be provided both for network control and to identify and
process the information bits, so increased spectrum and
higher data rates may be needed to maintain sufficient information throughput. In time diversity systems, storage of the
communication information is required at both the transmitter (to permit retransmission) and receiver (to support bit manipulation and message reconstruction), which constitutes an
important disadvantage for many analog, real-time, and
wideband applications.
Channel 2
RAKE Diversity
102
Frequency
diversity
101
100
106
104
102
100
Bit error rate (BER)
Figure 5. Bit error rate (BER) statistics measured at 6 GHz on 42.5
km link for two channels separated by 59 MHz, compared to BER
statistics with one-for-one frequency diversity protection in severe
frequency-selective fading environment. ( 1985 IEEE.)
RAKE diversity is effective with wideband signals if the individual multipath components (echoes) can be separately identified and processed (1). RAKE diversity, also called
multipath diversity or path-delay diversity, takes advantage
of the existence of multipath components in its operation. It
can be viewed either as a variation of frequency diversity (as
in spread spectrum systems, where to reduce small-scale
multipath effects the transmission signal is spread by several
times the frequency width of the reciprocal of the delay
spread) or time diversity (where the incremental time delays
among components are used with a system time reference to
identify and process the individual multipath components).
DIVERSITY RECEPTION
RAKE diversity is considered applicable mainly for digital
systems, which can support elegant information-processing
techniques (25). A RAKE receiver should be matched to the
signal received after distortion by the environment, instead of
being matched to the transmitted signal. Therefore, channel
adaptivity may be required if the propagation environment is
quite variable.
DIVERSITY-RECEPTION PERFORMANCE ANALYSIS
Many investigations of diversity reception techniques have
been undertaken to quantify and model the corresponding
gain in channel performance to support reliable system design. Much of the analytical basis for diversity reception derives from studies related to communication with mobile terminals, for which the fading environment is generally quite
severe. The motion of mobile terminals creates numerous opportunities for destructive interference to occur among
multipath transmission components. Conditions for such interference can arise for every half wavelength of travel (perhaps every few centimeters). Methods to overcome such fading impairments are required. Diversity reception is a
prominent impairment-mitigation method, and classic detailed treatments of diversity techniques are available (1,2).
METHODS FOR SELECTING OR COMBINING CHANNELS
A fundamental requirement in diversity reception is a reliable method for either switching among, or combining, the
available transmission channels in order to enhance link performance. While various difficulties are encountered in practice, the basic concepts of switching and combining are amenable to theoretical analysis. In this section, the major classes
of linear diversity selection and combining are analyzed. Linear diversity combining applies to implementations based on
relatively straightforward linearly-weighted sums of the multiple received signals, the only method generally capable of
distortionless reception in analog systems (1). Diversity signal
processing schemes that are more-sophisticated can be envisioned for digital systems.
An obvious choice for diversity switching is simply always
to select the best available signal, as defined using some
criterion based on a quantity that can be measured or estimated (such as signal-to-noise ratio, bit error rate, signal-tointerference ratio) to enable the system to identify the best
signal. This approach is called selection diversity. Each channel must have its own receiver or some other detection device
to supply the information that permits selection of the best
signal. A simpler form of diversity switching is threshold selection, in which the available signals are sampled in sequence until one is determined to be above some minimum
acceptable threshold. That signal is then used for reception
until it falls below the specified threshold, at which point the
scanning process is repeated.
A more sophisticated approach is to combine the diversity
signals to achieve overall augmentation in signal level available at the receiver. Common methods include maximal ratio
combining and equal gain combining, which typically require
that signals be suitably conditioned to ensure that they sum
coherently. Other variations on these basic methods can be
757
envisaged, depending on the signal conditioning and processing to be performed.
There are two general classes of diversity combining,
called predetection and postdetection combining, depending
on whether the diversity decisions take place before or after
baseband detection. In predetection combining, the information required to make a decision regarding selection or combining of signals is acquired and applied prior to baseband
detection, so the diversity operation can take place anywhere
from the receive antenna down to the intermediate frequency
(IF) input to the baseband receiver. Postdetection combining
is implemented at baseband after the signal detection receiver, and typically implies that each diversity channel must
be supplied with its own receiver.
Essentially identical diversity performance is typically obtained with well-implemented predetection or postdetection
diversity combining for linear modulation formats. For nonlinear modulation techniques such as frequency modulation
(FM), coherent predetection combining of diversity signals
can in cases increase the signal resultant above the receiver
signal detection threshold, even if the individual signals are
below the detection threshold. Conversely, predetection
switching can cause unwanted switching transients in the
signal carrier, requiring some strategy to minimize the resulting degradation in receiver performance (2).
Selection Diversity. This subsection discusses the gain in
performance achievable by selecting one channel from among
M (1, 2, . . .) available diversity channels (M-channel selection diversity), and summarizes the approach found in classic
texts (1,2). In the severe fading environments often encountered in mobile radio systems, the signal fading signal is usually observed to obey Rayleigh fading statistics, which correspond to the sum of random multipath components in the
absence of a direct line-of-sight (LOS) signal. When a direct
LOS signal exists, as often is the case in microcellular and
some other systems, Rician statistics (corresponding to the
sum of a direct component and several random multipath
components) are applicable.
If the electric field available to the receive antenna is assumed to be the summation of in-phase and quadrature
terms, where each term is comprised of the N diversity components that arrive at the antenna, the in-phase and quadrature fields may be modeled as independent (uncorrelated)
zero-mean Gaussian random processes (2). The in-phase and
quadrature terms, designated as EI and EQ, each obey the
standard normal distribution with probability density given
by
2
1
p(x) =
ex /2b 0
2b0
(5)
where x is either EI or EQ and b0 is the mean power level. The
envelope of these two components is the modulus (magnitude)
of the electric field composed of the in-phase and quadrature
terms:
r=
EI2 + EQ2
(6)
758
DIVERSITY RECEPTION
The probability density for the envelope, r, formed by the sum
of many multipath components (without a direct LOS component) is the Rayleigh density function:
p(r) =
r r 2 /2b
0
e
b0
r0
(7)
where the density is zero for r 0.
If the signal in each diversity channel is assumed to obey
the Rayleigh distribution, the signal envelope for the ith
channel is given by Eq. (7), with r replaced by ri. Over one RF
cycle of the field (assumed sinusoidal), the instantaneous
mean signal power in the ith channel is simply ri2 /2. The noise
power in the ith channel may be designated as ni2. If each
channel is assumed to contain the same mean noise power,
this power is a constant, N, independent of the channel. The
instantaneous signal-to-noise ratio (SNR) in the ith channel
is thus simply the ratio of the local mean signal power and
the mean noise power in each channel:
i = SNR i =
r 2i
2N
(8)
The average signal power in the ith channel is b0 ri2 /2, so
the mean value of the channel signal-to-noise ratio is
SNR b0 /N. Replacement in Eq. (7) yields a representation
for the probability density function of the envelope in terms
of the SNR quantities i and :
p(i ) =
achieved in going from M 1 (no diversity) to M 2 (dual diversity).
The mean SNR can be calculated by integrating S over the
probability density function for the range of allowable values
(zero to infinity):
1 /
e i
(9)
The probability that the SNR in the ith channel does not exceed a particular threshold value of interest, S, where the
subscript S indicates selection diversity, is obtained by integrating the probability density over the domain of interest:
S
P(i S ) =
p(i )di = 1 e S /
(10)
S =
S
M
1
d
PM (S ) dS =
dS
k
k=1
(12)
The mean decibel signal-to-noise ratio, 10 log[S/] (dB), is
plotted as curve (a) in Figure 6 to illustrate the gain in average output SNR obtained by selection diversity with an increasing number of channels (2). (The other curves in this
figure are explained in the next two subsections.)
A disadvantage of pure selection diversity is that each
channel must be provided with a receiver, or at least some
detection device that can identify the best signal at some
stage of the detection process, which can be expensive to implement. In a variation called scanning or threshold selection
diversity, available channels are scanned until an acceptable
signal is found. The selected channel is used until the received signal falls below a specified threshold, where upon the
scanning process is repeated until an acceptable signal is
again obtained. This mode of operation is clearly nonoptimum, and can lead to rapid and unproductive switching when
all diversity signals are below threshold.
A similar technique for two-channel diversity, called
switch and stay, is to switch to the alternate channel when
the received signal falls below threshold and stay at the new
position, even if the signal is below threshold, until that signal is available and itself eventually goes below threshold.
The signal, R(t), resulting from such a process (2) is illustrated in Figure 7. If the individual envelopes r1(t) and r2(t)
are assumed to be independent Rayleigh-fading time series,
the probability density for each being above a specified fade
threshold, At, is given by Eq. (7), with r replaced by At. By
further assuming that the individual segments of each time
For M channels, the probability that the SNR values in all
branches are all concurrently less than the threshold value
t is just the M-channel product:
(b)
(11)
The probability distribution given in Eq. (11) can be computed
for various values of M to estimate the efficacy of M-branch
diversity selection. For M 1 (no diversity), the distribution
is again the Rayleigh representation, which predicts an impairment level that increases at the rate of 10 dB per probability decade (a straight line on normal probability paper).
For example, at a probability level of 99.99%, the expected
ratio of the local SNR to the average SNR is computed to be
40 dB, 30 dB at a probability level of 99.9%, etc., demonstrating the Rayleigh roll-off of 10 dB/decade. These fade
depths illustrate the severe fading encountered in some mobile environments. With dual-channel diversity (M 2), the
multipath fading level predicted for 99.99% reduces from 40
dB to 20 dB, and the fade depth at 99% reduces from 20 dB
to 10 dB, representing substantial improvements. Adding yet
more diversity channels yields useful performance enhancements, but by far the largest diversity gain increment is
8
Diversity gain (dB)
PM (t ) = (1 e S /
)M
10
(c)
6
(a)
4
3
4
5
6
7
8
Number of branches, M
10
Figure 6. Predicted diversity gain in Rayleigh-fading environment
with M diversity branches: (a) selection diversity; (b) maximal ratio
combining; (c) equal gain combining. ( 1994 IEEE.)
DIVERSITY RECEPTION
759
is the chi-square distribution, with corresponding probability
density function
r1(t)
r2(t)
R(t)
R0
p(R ) =
RM1 e R /
M (M 1)!
R 0
(16)
The cumulative distribution function, obtained by integrating
Eq. (16) from zero to R, is (2)
I
t0
II
t1
III
t0
t2
t0
t3
Figure 7. Signal envelope resulting from switching between two
Rayleigh-fading envelopes with switch-and-stay technique: r1(t) and
r2(t) are the two envelopes, R(t) is the composite switch-and-stay envelope, and R0 identifies the start of time segments defined by a signal
switch. ( 1994 IEEE.)
series (portions prior to switching from, and portions subsequent to switching back to, either envelope) are uncorrelated,
the density function of the composite (switched) carrier envelope, R(t), can be established, as in the previous case for a
single envelope. The composite probability density is found to
be comprised of two Rayleigh densities with different
weighting factors (2).
Maximal Ratio Combining. In maximal ratio combining, the
signals are phased and coherently summed, instead of being
selected one at a time. The complete scheme is to cophase the
M channels, then apply weights, wi, to the signals that are
proportional to the SNRi of the individual channels, and finally sum these signals. The resultant envelope, r, is the sum
of the weighted envelopes ri:
r=
M
wi ri
PM (R ) = 1 e R /
(17)
From Eq. (17), the diversity performance for maximal ratio
combining is found to be somewhat better than that for selection diversity with the same number of diversity channels.
Instead of the 10 dB diversity gain obtained with selection
diversity at a time percentage of 99%, maximal ratio combining provides an 11.5 dB gain. Maximal ratio combining in fact
provides the best performance that can be achieved with linear diversity combining techniques.
The average SNR, obtained in parallel with the first two
terms on the left of Eq. (15), is
R =
M
M
i =
= M
i=1
(18)
i=1
The average SNR (dB) is plotted as curve (b) in Figure 6 to
show the incremental diversity gain achieved by adding diversity branches to a maximal ratio combining system.
Equal Gain Combining. Equal gain combining is a simplified
form of signal combining in which the weights are all constant
and equal, and can be set so that wi 1. From Eq. (13), the
signal envelope of the combined signal is
(13)
r=
i=1
M
(R /
)k1
(k 1)!
k=1
M
ri
(19)
i=1
The channel noise contributions also scale in proportion to
the corresponding weights, wi. If the average noise powers
(before weighting) are all assumed equal to N, the total noise
power is
NT = N
M
w2i
(14)
i=1
In parallel with the analysis for selection diversity (but with
a change of subscript to R to indicate ratio combining), the
resultant SNR is R r2 /2NT.
If the wi are weighted proportionally to the instantaneous
channel SNRi i, then
M
M
M (r 2 + r 2 )
r 2i
I
Q
=
R =
i =
2N
2N
i=1
i=1
i=1
(15)
As in Eq. (5), the in-phase and quadrature components obey
zero-mean Gaussian probability densities with equal variances. The sum of the squares of independent standard normal variables, as given by the right-hand side of Eq. (15),
For equal noise power in all diversity channels, the corresponding output SNR is E r2 /2NM, a sum of Rayleigh variables, for which there is no general solution for the probability distribution function. Solutions generated numerically
reveal that the performance of equal gain combining is only
slightly worse than maximal ratio combining (usually by less
than 1 dB).
The general expression obtained for the average value of
the output SNR is given by (2)
E =
1 + (M 1)
4
(20)
where is the mean channel SNR. The output SNR (dB) is
shown as curve (c) in Figure 6 for comparison with the selection diversity and maximal ratio combining techniques. Maximal ratio combining provides the best performance, though it
is not very superior to equal gain combining. Maximal ratio
and equal gain combining are both better than selection diversity, but these performance gains are achieved with added
system complexity and cost.
760
DIVERSITY RECEPTION
The performance gains estimated above for different types of
diversity systems ignore several limitations that are confronted in practical applications. For example, impairments
on the individual diversity channels may not be completely
independent, and combining errors may introduce additional
degradations in diversity system performance. Estimates of
these degradations (1,2) are briefly illustrated here.
Imperfect Channel Decorrelation. Prior analyses implicitly
assumed independence of fading among the diversity channels. In many environments, complete decorrelation is not
achieved, and indeed is found to be unnecessary for successful
diversity operation. General limits can immediately be placed
on the behavior of the resulting statistical distributions: Complete independence of channel impairments yields results
identical to analyses in the previous section, while complete
correlation leads to Rayleigh fading statistics equivalent to a
single channel without diversity. To investigate the effects of
intermediate channel correlation, the complex correlation coefficient, , between the diversity signals must be taken into
account, where 2 approximates the correlation function between signal envelopes (7).
For selection diversity, analyzing more than two diversity
channels is difficult, but for dual channels the probability distribution is found to be (1)
P2 (S ) = 1 e S /
[1 Q(a, b) + Q(b, a)]
(21)
where Q(a, b) is expressed in terms of the zeroth-order modified Bessel function, I0, as
Q(a, b) =
e(a
2 +x 2 )/2
I0 (ax)x dx
(22)
with the parameters a and b given by
a=
2S
(1 + ||2 )
b=
2S
(1 ||2 )
(23)
Figure 8 displays the cumulative distribution functions computed with these expressions (2). The curve for 2 0 corresponds to zero correlation, as in prior analyses, while for 2
1, no diversity advantage is conferred by switching between
the two channels. However, substantial diversity gain is
achieved even when the correlation between the two signal
envelopes approaches 0.8, attesting to the efficacy of diversity
operation for this environment. Similar results are obtained
for other modes of diversity combining.
Time percentage SNR is exceeded
Operational Considerations in Diversity Systems
30
40
50
60
70
80
90
95
2 = 1.0
98
99
0.8
99.5
0.6
99.8
0.3
99.9
99.95
99.98
99.99
25
20
15
10
Mean SNR (dB)
Figure 8. Rayleigh-fading statistics for 2-channel selection diversity
assuming varying degrees of envelope correlation, 2. Fade distributions are referenced to the SNR for a single (nondiversity) channel.
( 1994 IEEE.)
in the control of the combiner. Degradations resulting from
imperfect correlation between a pilot signal to control operation of a maximal ratio combiner and the signals themselves
are summarized here to indicate the magnitude of the anticipated errors.
The output of an M-channel maximal ratio combiner that
relies on a pilot for the reference control information is found
to have the probability density (2)
M1
M1
1
2 M1 R /
pM (R ) = (1 )
e
n
n=0
n
(24)
1
R 2
(1 2 ) n!
The first term after the summation is the binomial distribution, and is the correlation coefficient between the pilot and
the adjacent signal channel. (Note that here correlation is desirable, unlike the case for envelope correlation.) The first moment (mean) of R is obtained by integrating with respect to
the probability density in Eq. (24):
R =
R pM (R )dR =
[1 + (M 1) 2 ]
(25)
0
Switching and Combining Errors. No diversity switching or
combining device is expected to operate perfectly, especially
since randomly-fading signals supply much of the information
used to control the switching or combining device. Errors introduced by imperfect operation degrade the performance of a
diversity system. As already noted, a maximal ratio combiner
must cophase and sum the diversity signals in proportion to
the SNR in each channel, necessitating SNR estimates for
each channel. In some systems, a continuous-wave pilot signal is transmitted adjacent to the communication band to
supply reference amplitude and phase information to assist
and is equivalent to the mean SNR. The probability distribution obtained by integrating over the density function is
R
PM (R ) =
pM (x) dx
0
M1
n
M1
(R /
)k
R /
=1e
2n (1 2 )Mn1
n
k!
n=0
k=0
(26)
which represents the statistics of the combiner output signal.
DIVERSITY RECEPTION
If the correlation between the pilot and channel signal is
perfect (2 1), representing perfect operation of the diversity
combiner, Eqs. (26) and (25) respectively give
PM (R ) = 1 e R /
M1
k=0
(R /
)k
k!
R = M
(27)
which are equivalent to Eqs. (17) and (18), respectively. If
there is no correlation between the pilot and channel signal
(2 0), the resulting expressions are
PM (R ) = 1 e R /
R =
(28)
showing that diversity operation provides no benefit for this
case.
To illustrate the impact of such errors on the overall performance of a maximal ratio combiner, Figure 9 displays fading statistics for 4-channel operation with varying degrees of
correlation (2). The performance penalty imposed by combiner-control errors is considerable, especially in the critical
deep-fading portion of the distribution (for example, compare
curves for 2 1.0 and 2 0.75). From Eq. (25), however,
the degradation in mean received signal power is modest. If
the correlation coefficient 2 decreases from 1 to 0.5, for instance, the mean SNR for the 4-channel combiner decreases
from 4 to 2.5 , a loss of only 2.0 dB.
10
20
30
40
50
60
70
90
95
2 = 0
5
0.7
99.5
1.0
0.5
99
99.8
99.9
99.95
99.98
99.99
30
20
Other System Considerations. Even if diversity operation offers substantial improvement in overall circuit availability,
other practical constraints may need to be taken into account.
For example, site diversity significantly improves availability
for earthsatellite paths subject to rain attenuation, as verified in Fig. 3. However, in large earth stations that utilize
wide bandwidths to serve many users, potential outages related to switching among available diversity signals is a severe problem to be avoided. Therefore, the entire receive band
for the diversity channel must be conditioned and synchronized with the main-station signal to support switching
among channels with no loss of information.
However, on the uplink to the satellite, such synchronization is extremely difficult due to variations in radio path
length (such as caused by satellite motion) between the earth
stations and the satellite. Therefore, uplink site diversity is
much less viable than downlink site diversity, except possibly
in packet-switched applications where lost packets can be recovered. A potential compromise solution for this case is to
protect the downlink path with site diversity, but implement
transmitter power control (27) to increase the availability of
the uplink path.
Figure 10 shows a 14/11 GHz earth-space site diversity
configuration (28), planned for the two sites represented in
Fig. 1. In this system, the entire downlink receive band (500
MHz) from the secondary station is transported to the main
station by the microwave Diversity Interconnect Link (DIL),
buffered and synchronized with the main station receive signal, and made available at the diversity switch. (Signal combining of the diversity signals is unlikely to be considered for
this wideband application because of the difficulty in matching phase variations across the two 500 MHz receive bands.)
Signal regeneration (demodulation and remodulation) is implemented in the DIL, not only to support frequency conversion, but also to preserve the quality of the transmissions. In
this design, the uplink signal is also made available at both
transmit sites, but this capability mainly increases the reliability of the overall system by enabling a redundant uplink
signal transmission capability.
RECENT DEVELOPMENTS
98
0.2
Time percentage SNR is exceeded
80
761
10
Mean SNR (dB)
10
Figure 9. Rayleigh-fading statistics for 4-channel maximal ratio diversity combining with combiner errors (specified by 2) between reference pilot and channel signal. Fade distributions are referenced to
the SNR for a single (nondiversity) channel. ( 1994 IEEE.)
Despite the rather well-developed state of diversity reception
concepts, the field remains quite active. Many recent developments are related to new service offerings such as nongeostationary (NGSO) satellite systems, digital cellular systems and
indoor mobile systems, and systems that often must operate
in severe propagation environments.
One novel consumer application is the installation of
space-diversity antennas in some automobiles to mitigate reflection multipath fading and improve urban FM radio reception (29). Antenna diversity has also been demonstrated for
vehicular reception of mobile-satellite transmissions. For reception at 1.5 GHz with a single terminal fade threshold of
10 dB, space diversity reception using two antennas separated by 3 m provided a diversity gain (fade reduction) of
4 dB when the major cause of fading was shadowing and
blockage by roadside trees (30).
Because of the importance of preserving links supporting
multiple users, site diversity is beginning to be implemented
to protect feederlink earth stations in some mobile-satellite
762
DIVERSITY RECEPTION
14 GHz
11 GHz
14 GHz
11 GHz
Diversity switch control
TX
UC
RX
11 GHz
4 GHz
RX
LNA
HPA
14 GHz
IPA
HYB
HPA
DC
IPA
6 GHz
TX
DC
LNA
UC
RX
TX
TX
UC
4
GHz
6
GHz
UC
Rx
Customer interface
Main site
Repeater
Diversity site
Figure 10. Configuration for 14/11 GHz earthspace site diversity, including microwave link
that interconnects the main and diversity stations to support selection diversity (TX/RX
transmit/receive; IPA/HPA intermediate/high-power amplifier; LNA low-noise amplifier;
UC/DC up/down converter; HYB hybrid). (Copyright 1979 COMSAT Corp. All rights reserved by COMSAT Corp. Used by permission.)
99
98
Time percentage fade is exceeded
systems (9). A novel proposed application of site diversity,
called wide area diversity, is to protect many VSAT terminals
connected to a metropolitan area network by switching traffic
among the VSATs as required to counteract impairments on
the separate earth-space paths (31).
Recently the orbital-diversity concept has been investigated for narrowband VSAT systems, where a small reserve
capacity can be made available on an alternate satellite as
protection for several VSAT links. This application does not
require the difficult switching of wideband signals between
satellites. Interestingly, early experimental tests (32) indicate
that the diversity gain in snow events was superior to that
for rain events, but rain will likely represent the more important impairment.
Yet another variation of orbital diversity (also called path
or satellite diversity in this context) is planned for some
NGSO satellite configurations, such as low earth orbit (LEO)
constellations, intended to communicate with ground-based
terminals (especially handheld terminals). The primary path
impairments are shadowing and blockage by terrain and surface objects (trees, buildings, etc.) as the NGSO satellites
change position with respect to a user terminal. In such constellations, more than one satellite may often be potentially
accessible from a given location on the earth, providing the
capability to switch among the separate, independently-fading paths to create a diversity configuration (33).
Highest satellite
Best satellite
2 best satellites
3 best satellites
4 best satellites
95
90
80
70
60
50
40
30
20
10
5
2
1
5
10
15
20
Signal fade (dB)
25
30
35
Figure 11. Urban shadowing/blockage fading statistics simulated
for NGSO satellite constellation. Highest satellite is the one with
greatest elevation angle; best satellite is the one with least-faded
path; other curves assume coherent combining of signals from 2, 3,
or 4 satellite paths, respectively. ( 1997 IEEE.)
DIVERSITY RECEPTION
763
BIBLIOGRAPHY
Diversity gain (dB)
5
4
1. M. Schwartz, W. R. Bennett, and S. Stein, Communication Systems and Techniques (in particular chaps. 10 and 11), Piscataway,
NJ: IEEE Press, 1996.
2. W. C. Jakes (ed.), Microwave Mobile Communications, Piscataway, NJ: IEEE Press, 1994.
3. G. Roda, Troposcatter Radio Links, Norwood, MA: Artech
House, 1988.
4. Recommendation ITU-R P.530-7, Propagation data and prediction methods required for the design of terrestrial line-of-sight
systems, International Telecommunication Union, Geneva: ITUR Recommendations 1997 P Series, Part 2: 271295, 1998.
1
0
10
15
20
25
Cross coupling (dB)
30
Figure 12. Polarization diversity gain (90% probability level) as a
function of antenna cross-polarization discrimination (XPD), simulated for worst-case fading with selection diversity in urban environment. ( 1997 Horizon House Publications, Inc. Used by permission.)
5. D. L. Bryant and J. E. Allnutt, Use of closely-spaced height diversity antennas to alleviate the effects of low angle non-absorptive
fading on satellite slant paths, Electron. Lett., 26: 480481, 1990.
6. S. H. Lin, T. C. Lee, and M. F. Gardina, Diversity protections
for digital radiosummary of ten-year experiments and studies,
IEEE Commun. Magazine, 26 (2): 5164, 1988.
7. R. G. Vaughan and J. Bach Andersen, Antenna diversity in mobile communications, IEEE Trans. Veh. Technol., VT-36: 149
172, 1988.
There will be many instances when at least one of the satellite paths is free of obstruction. If the most favorable path
can be selected as required, or the multiple signals can be
combined, path performance is enhanced. Figure 11 shows
the cumulative fading distributions estimated by simulating
the path availability achieved by equal-gain combining of
available signals from multiple satellites for an urban
shadowing/blockage environment (33). The substantial benefits of path diversity for this environment are apparent. In
such systems using path diversity, the RAKE receiver technique may be implemented, not only to overcome shadowing
and blockage impairments, but also to enable smooth handoff
among the available satellite beams (34), which might be required every few minutes in LEO systems, since satellite motion is rapid with respect to earth-based terminals.
Polarization diversity has found renewed interest in mobile telephony as a means to improve performance while
avoiding the need for additional spectrum or a second spacediversity antenna, especially for handheld terminals (35) in
which the antenna orientation is quite variable. In a study of
polarization-diversity reception based on measurements in an
urban area (36), the achievable diversity gain was related to
the cross-polarization discrimination, XPD, of the received
signals. Figure 12 shows diversity-gain results at the 90%
probability level for selection combining with worst-case reception (assuming the two signals arrive 180 out of phase),
referenced to the 45 branch of two linearly-polarized elements oriented at 45. The results indicate that the diversity gain achievable with polarization diversity in this application is only a dB or so less than typical space diversity gains
achieved with antenna separations of the order of 20 wavelengths. In this environment, better performance was obtained with linear polarizations oriented at 45 than for vertical and horizontal polarizations, as the former maintained
more nearly equal mean signal levels in the two diversity
channels.
Additional diversity reception applications will inevitably
arise in response to future developments in telecommunication systems and technology.
8. V. Mimis and A. Smalley, Low elevation angle site diversity satellite communications for the Canadian arctic, Conf. Record, IEEE
Int. Conf. Commun., Philadelphia, 2: 4A.4.14A.4.5, 1982.
9. C. Kittiver and E. Sigler, Propagation considerations in the
American Mobile Satellite System design, Proc. 17th NASA Propag. Experimenters Meeting, Pasadena, CA: JPL Publ., 9321:
171181, 1993.
10. J. C. Cardosa, A. Safaai-Jazi, and W. L. Stutzman, Microscale
diversity in satellite communications, IEEE Trans. Antennas Propag., 41: 801804, 1993.
11. A. Bosisio et al., Analysis and application of short-distance site
diversity techniques for 20/30 GHz communication links, Conf.
Record, IEEE Global Telecommun. Conf. : Singapore, 1: 749
753, 1995.
12. D. C. Cox, Universal digital portable radio communications, Proc.
IEEE, 75: 436477, 1987.
13. D. V. Rogers and G. Hyde, Diversity measurements of 11.6-GHz
rain attenuation at Etam and Lenox, West Virginia, COMSAT
Tech. Rev., 9: 243254, 1979.
14. D. V. Rogers, Diversity and single-site radiometric measurements
of 12-GHz rain attenuation in different climates, IEE Conf. Publ.,
No. 195, Part 2: 118123, 1981.
15. D. B. Hodge, An empirical relationship for path diversity gain,
IEEE Trans. Antennas Propag., AP-24: 250251, 1976.
16. J. Goldhirsh et al., Three-site space-diversity experiment at 20
GHz using ACTS in the eastern United States, Proc. IEEE, 85:
970980, 1997.
17. Recommendation ITU-R P.618-5, Propagation data and prediction methods required for the design of earth-space telecommunication systems, International Telecommunication Union, Geneva:
ITU-R Recommendations 1997 P Series, Part 2: 305323, 1998.
18. D. B. Hodge, An improved model for diversity gain on earth
space propagation paths, Radio Sci., 17: 13931399, 1982.
19. E. Matricciani and M. Mauri, Italsat-Olympus 20-GHz orbital diversity experiment at Spino dAdda, IEEE Trans. Antennas Propag., 43: 105108, 1995.
20. R. G. Vaughan, Polarization diversity in mobile communications,
IEEE Trans. Veh. Technol., 39: 177186, 1990.
21. P. C. F. Eggers, J. Toftgard, and A. M. Opera, Antenna systems
for base station diversity in urban small and micro cells, IEEE J.
Sel. Areas Commun., 11: 10461057, 1993.
764
DIVERSITY RECEPTION
22. W. C. Y. Lee, Mobile Communications Engineering, New York:
McGraw-Hill, 1982.
23. P. L. Dirner and S. H. Lin, Measured frequency diversity improvement for digital radio, IEEE Trans. Commun., COM-33:
106109, 1985.
24. W. C. Wong et al., Time diversity with adaptive error detection
to combat Rayleigh fading in digital mobile radio, IEEE Trans.
Commun., COM-31: 378387, 1983.
25. G. J. R. Povey, P. M. Grant, and R. D. Pringle, A decision-directed
spread-spectrum RAKE receiver for fast-fading mobile channels,
IEEE Trans. Veh. Technol., 45: 491502, 1996.
26. A. Mehrotra, Cellular Radio Performance Engineering, Norwood,
MA: Artech House, 1994.
27. J. E. Allnutt, Satellite-to-Ground Radiowave Propagation (Sect.
7.5.2), London, UK: Peter Peregrinus, 1989.
28. L. F. Gray and M. P. Brown, Transmission planning for the first
U.S. Standard C (14/11 GHz) INTELSAT earth station, COMSAT Tech. Rev., 9: 6190, 1979.
29. H. K. Lindenmeier, Antennas on land vehicles for improved communications, NATO AGARD Lecture Series, No. 165: 7-17-19,
1989.
30. W. J. Vogel, J. Goldhirsh, and Y. Hase, Land-mobile satellite fade
measurements in Australia, J. Spacecraft Rockets, 29: 123128,
1992.
31. K. Hobson, R. Heron, and T. Spracklen, Wide area diversity and
wide area networks, Proc. Olympus Utilization Conf., European
Space Agency Publ. WPP-60: 227232, 1993.
32. T. Hatsuda et al., Comparison of multiple satellites diversity
characteristics between rain and snow attenuations for 14/12
GHz band, Digest, IEEE Antennas Propag. Soc. Int. Symp., Montreal, 4: 25562559, 1997.
33. R. Akturan and W. J. Vogel, Path diversity for LEO satellite-PCS
in the urban environment, IEEE Trans. Antennas Propag., 45:
11071116, 1997.
34. J. Schindall, Concept and implementation of the Globalstar mobile satellite system, Proc. 4th Int. Mobile Satellite Conf., Ottawa,
Canada, A11A16, 1995.
35. F. Lotse et al., Base station polarization diversity reception in
macrocellular systems at 1800 MHz, Proc. IEEE 46th Veh. Tech.
Conf., Atlanta, 3: 16431646, 1996.
36. C. Beckman and U. Wahlberg, Antenna systems for polarization
diversity, Microwave J., 40 (5): 330334, 1997.
DAVID V. ROGERS
Industry Canada
ELECTROMAGNETIC WAVE SCATTERING
487
ELECTROMAGNETIC WAVE SCATTERING
Electromagnetic wave scattering is the reradiation of electromagnetic energy that results when an electromagnetic field
encounters an abrupt change in electrical parameters. Typically, this occurs when an electromagnetic field is incident on
a structure or scattering object. When the original electromagnetic field crosses the boundary between two regions of
different material, each with different electrical properties,
the field will change as it enters the second region. Sources,
including conduction, displacement, and polarization currents, will be induced at the discontinuity between the two
regions. These currents act as sources of electromagnetic radiation, much like the sources of the original incident electromagnetic field. This reradiation is called electromagnetic scattering, because it scatters the incident electromagnetic field
from its original propagation path.
TYPES OF ELECTROMAGNETIC SCATTERING
Fundamentally, there are three types of electromagnetic scattering mechanisms: reflection, refraction, and diffraction.
These scattering mechanisms can radiate specularly or diffusely. Specular scattering means that electromagnetic reradiation travels in parallel rays. Diffuse scattering means the
spreading of the electromagnetic field as it propagates away
from the scattering object.
Specular Scattering
Of the three fundamental scattering mechanisms, the most
familiar are specular reflection and refraction. If any corners
or bends that exist at the boundary are very gradual compared to the wavelength of the incident field, then the boundary tends to cause specular scattering. Optical scattering is
often assumed to be specular, because most obstructing bodies are electrically large compared to optical wavelengths.
Specular scattering can be modeled with the specular law of
reflection and Snells law of refraction.
Specular Reflection. A familiar example of specular reflection is the common reflection of a visible image in a mirror,
since the dimensions of the mirror are huge compared to the
wavelength of visible light. The ratio of the reflected field to
the incident field strength is called the reflection coefficient .
Refraction. The energy that is not reflected from the electrically large boundary is transmitted through the boundary.
The ratio of the transmitted field strength to the incident field
strength is called the transmission coefficient T. In the process of propagating from one electrical medium to the next,
J. Webster (ed.), Wiley Encyclopedia of Electrical and Electronics Engineering. Copyright # 1999 John Wiley & Sons, Inc.
488
ELECTROMAGNETIC WAVE SCATTERING
the speed of propagation changes. This change in speed
causes a change in the propagation angle at the boundary.
This phenomenon is commonly seen with visible light at the
surface of a calm pool of water. The fact that an object extending from the air into the water appears bent is due to the
increase in the propagation velocity of light as it leaves the
water and enters the air. This phenomenon can be modeled
with Snells law of refraction.
Er
Hr
Ht
z
Diffuse Scattering
The laws for specular scattering are only valid for electrically
large scattering bodies. If the object causing the electromagnetic scattering is small compared to the wavelength of the
incident electromagnetic field, the induced currents would
tend to bend around the contour, creating diffuse scattering.
Unlike specular scattering, diffuse scattering results when
the electromagnetic energy spreads out as it radiates from the
scattering object. The smaller the object, the more the energy
will spread as it reradiates.
A simple example of diffuse scattering is an electromagnetic field incident upon a cylindrical conductor of small radius, where the electric field is parallel to the axis of the cylinder. At frequencies below the microwave region of the
electromagnetic spectrum, a thin copper wire will have an
electrically small radius. A first approximation is to assume
that the current that is induced by the incident field is uniformly distributed across the entire cross section of the wire.
The scattered or reradiated field from this wire is similar to
the field radiated from a wire antenna having the same linear
current distribution. However, the total field around the wire
is the superposition, or vector sum, of the scattered field and
the original incident field that would have existed without the
wire present. The scattered field, added to the incident field,
creates a pattern with constructive reinforcement in some directions and destructive cancellation in other directions. This
is the function of the passive elements found on the YagiUda
antenna, common in television and other VHF and UHF communications.
In the YagiUda antenna, only one set of elements is active. The active (or driven) element usually makes up a halfwavelength dipole antenna. The other elements are simply
conductive cylinders or wires that reradiate some of the energy incident on them from the active element. Depending on
the relative lengths of these passive elements, each of their
radiated fields will add to the incident field of the active element, to create an overall pattern of power flow (1) This focusing of energy is called antenna gain. Other forms of diffuse
scattering by electrically small bodies are not so intentional.
Et
2, 2, 2
Ei
1, 1, 1
Hi
Figure 1. Illustration of Snells law. A plane-wave electromagnetic
field obliquely incident onto a plane boundary, separating medium 1
(1, 1, 1) from medium 2 (2, 2, 2). A reflected field and a transmitted field scatter from this discontinuity in electrical constants.
ronment, a cell-phone user rarely has a direct line of sight to
the cell base station. Often, the communication link can only
be established due to energy reflecting off of a nearby building, or energy diffracted around a building or over a hill.
Since typically, these obstructions are electrically large, raytracing techniques, which incorporate the laws for specular
reflection and diffuse diffraction at edges, are often used to
model the propagation characteristics of the communication
channel.
THE LAWS OF SPECTRAL REFLECTION AND REFRACTION
In many applications, an electromagnetic field can be assumed to be a plane wave. A plane wave is a convenient approximation amounting to the assumption that the electromagnetic field does not vary over the plane perpendicular to
the direction of propagation. This approximation is similar to
assuming that over small geographic areas the Earth is flat.
For wave propagation, this assumption is valid for a small
observation area at a great distance from the source of a
spherically propagating wave.
Figure 1 shows a plane-wave electromagnetic field incident
on a boundary in the xy plane. The generalized electric field
will have components in the x, y, and z directions, that is,
E = (xxE
x + yyE
y + zzE
z )e j (sin i x+cos i z)
(1)
Diffuse Diffraction
Another form of scattering, which cannot be accounted for by
reflection or refraction, is diffraction. For electrically large
scattering bodies, diffraction appears to occur at geometrical
discontinuities such as edges and corners. A first approximation is that currents induced only at these discontinuities reradiate electromagnetic energy. Diffraction is the scattering
mechanism that accounts for radiation filling in the region
that would have been completely blocked (shadowed) by an
opaque obstruction.
Diffraction and reflection are important scattering mechanisms in communications. Whether in an urban or rural envi-
where is the phase constant or wave number in
radians per meter, is the permeability in henrys per meter,
and is the permittivity of the material in farads per meter.
Imposing the tangential boundary condition for the electric
field, the sum of the tangential components of the incident
and reflected fields must be equal to that of the transmitted
field (24),
xi + yyE
yi )e j 1
(xxE
sin i x
xr + yyE
yr )e j 1
+ (xxE
sin r x
= (xxE
xt + yyE
yt )e j 2
sin t x
(2)
ELECTROMAGNETIC WAVE SCATTERING
This equality can only be true for all x when the exponents,
or phases, are equal:
1 sin i x = 1 sin r x = 2 sin t x
(3)
Equation (3), proves the well-known specular law for reflection (5),
i = r
(4)
which simply states that the angle of reflection equals the
angle of incidence. Equation (3) also leads to Snells law of
refraction,
sin i
=
sin t
r
2 2
(5)
1 1
For most material, the permeability is the same as that of
free space, 0. Assuming 1 2, Eq. (5) reduces to
n1 sin 1 = n2 sin 2
where n r is the index of refraction and r is the relative
permittivity or dielectric constant.
489
2. Region 2 represents a scattering body of material consisting of a different permittivity , permeability , and conductivity , than those of the surrounding region 1. Region 2
is bounded by a surface S. Assume that an electric field, E1,
and its associated magnetic field H1, originate from a source
current density J0. This source could simply be the current
oscillating in a transmitting antenna. These fields propagate
undisturbed, through region 1, until they become incident
upon the scattering body of region 2. As the fields cross the
boundary between region 1 and region 2, they will be perturbed, that is, E2 and H2 in region 2 will generally not be
equal to the fields E1 and H1 propagating in region 1. This
abrupt change or discontinuity in electric and magnetic field
strength results in currents that are induced at the discontinuity. In general these currents will be distributed through
the volume of regions 1 and 2, depending on their electrical
constants.
According to the induction theorem, whenever there is a
discontinuity of the E and H fields crossing a boundary S between two media with different electrical constants, one can
assume that induced currents at S cause the discontinuities
in the fields. The induced current can be an electric current
sheet (68)
ELECTROMAGNETIC THEOREMS
Many electromagnetic scattering problems do not lend themselves to the simple application of the laws of reflection and
refraction. To develop more sophisticated analysis tools, a discussion of some basic electromagnetic theorems will be useful.
J s = n (H
H s H t ) = n H i
or a fictitious, but mathematically useful magnetic current
sheet
E s E t ) = n E i
M s = n (E
Uniqueness Theorem
Knowledge of the sources induced on the surface of a scattering body S enables unique solutions of the fields reradiated
by those induced sources. Conversely, the known fields allow
a unique calculation of the induced sources. The electric field
E and magnetic field H are uniquely determined if (6,7)
1. n
E, the tangential component of E, is specified on
S,
2. n
H, the tangential component of H, is specified on
S, and
3. n
E is specified on part of S, and n
H is specified
on the remaining part of S.
Induction Theorem
In general, sources, such as conduction, displacement, and polarization currents, are induced at electrical discontinuities in
the medium through which the incident field is propagating.
Figure 2 shows a typical discontinuity represented by region
J0
E 1, H 1
JS
MS
2, 2,
^
n
1, 1,
(6b)
The superscripts i, s, and t pertain to the incident, outwardly
scattered (reflected), and transmitted fields, respectively, and
is the normal unit vector pointing out of the scattering
n
body. If the scattering object is a perfect conductor, the transmitted fields vanish, leaving
J s = n H i = n H s
(7a)
M s = n E i = n E s
(7b)
and
The induction theorem alleviates the problem of knowing the
exact distribution of current densities throughout the volume
of the scattering body. The assumed currents exist only on
the boundary S between the two media. Furthermore, the induced currents can be calculated directly from knowledge of
the incident field that would have existed in the absence of
any scattering object.
Equivalence Principle
E 2, H 2
M0
(6a)
Figure 2. A scattering body. The incident field propagates from the
source current, J0, through region 1. As this field strikes region 2, the
currents Js and Ms are induced at the surface S of the scattering body.
If two different sources produce the same radiating field
within a region, these sources are equivalent (8). If both regions have the same electrical constants, only an inwardly
scattered (transmitted) field exists. It follows from Eq. (6)
that the fields that are incident on the boundary S can be
replaced by the equivalent current sheets
J s = n H s = n H i
(8a)
490
ELECTROMAGNETIC WAVE SCATTERING
and
Wavefront
M s = n E s = n E i
(8b)
where in this case, n
is pointing in the direction of the transmitted or scattered wave. The equivalence theorem is useful
for modeling radiation through apertures, such as a slot in a
conductive plane or a horn antenna.
z
r
GO shadow
DIFFRACTION
Diffraction is the scattering mechanism that neither reflects
off nor transmits through a structure. Even with opaque
structures, allowing no transmission, diffraction accounts for
radiation into the geometrical shadow region. This scattering
mechanism cannot be modeled with Snells law. To analyze
diffraction exactly would require more knowledge about the
induced current distribution around the scattering structure
than would typically be available. Therefore approximations
must be made to simplify the analysis. Two common approaches to analyzing diffraction are the use of geometrical
optics and physical optics.
Conducting
half plane
Figure 3. Half-screen diffraction using the GTD. The currents induced at the edge of the conducting half screen radiate into the GO
shadow region.
Figure 3 shows a plane wave incident on a perfectly absorbing
half screen. The diffraction coefficient can be quite involved,
even for this simple scattering structure. However, away from
the shadow boundary (15),
E =
d
Geometrical Optics
Geometrical optics (GO) is a ray-tracing technique that assumes that the electromagnetic energy travels in straight
parallel lines, or rays, that are perpendicular to the wavefront. These rays travel from the point of reradiation to the
observation point. While relatively easy to implement (9,10),
GO is an approximation that relies on some important assumptions, primarily that the wavelength of the electromagnetic field must approach zero. Clearly, GO is an asymptotic
technique only valid for sufficiently high frequencies, such
that the wavelength is small compared to the dimension of
the obstruction. Since GO assumes infinite frequency, it ignores the wave nature of the electromagnetic scattering field,
thus ignoring diffraction. The GO model creates an abrupt
change in energy at the transition from the illuminated region to the shadow region. The abrupt change in field
strength, without currents or charges to account for this discontinuity, violates boundary conditions. Therefore, GO provides only the crudest model, accounting only for reflection
and refraction, but not for diffraction.
The geometrical theory of diffraction (GTD) extends GO to
account for diffraction, by introducing a diffraction coefficient, D, analogous to for reflection and to T for transmission (1113). The total electric field ET around the obstruction
is
ET = Eg + Ed
(9)
where Eg is the electric field predicted by GO and is zero in
the shadow region. The diffracted field for a plane wave of
incidence is given by (14)
e j r
E0
E d (r) = DE
r
(10)
12
r j r
e
E0
x
(11)
In the shadow (x), the diffracted field given in Eq. (11) is
the only field present. In the region of GO illumination (x),
the magnitude of the diffracted field of Eq. (11) subtracts from
the incident field. Figure 4 illustrates the sum of the diffracted field and the GO incident field for (a) z 2 and (b)
z 20. Clearly, there is a discontinuity at the transition
between the GO illumination and shadow regions, around
x 0. This is an obvious limitation of the GTD, since there
should be a smooth transition. One crude solution would be
to simply draw a smooth curve connecting each side of the
discontinuity through the point x 0, E E0 /2. A more sophisticated method is the uniform theory of diffraction (UTD),
which is an extension of the GTD that forces a smooth transition between the GO illumination and shadow boundary (16).
Many common diffraction problems, such as hilltops and
buildings, can be modeled with this half-screen or knife-edge
approximation. However, the GTD still relies on several assumptions. The diffracted ray is assumed to depend entirely
on the incident ray and the characteristics of the discontinuity itself, such as an edge of a scattering structure (17). The
GTD is still a high-frequency asymptotic approximation, because it assumes that the structure is electrically large and
conductive (18). Furthermore, the GTD suffers from the unrealistic discontinuity problem at the GO illuminationshadow
boundary.
Physical Optics
The edge-diffraction problem of Fig. 3 can also be analyzed
using the concept of physical optics (PO), which relies on Huygens principle. Huygens principle states that each point of a
primary wavefront acts as a secondary point source. Each of
these secondary sources radiates a spherical wave (14). The
primary difference between PO and the GTD is that the GTD
assumes rays connect from the geometrical discontinuity to
ELECTROMAGNETIC WAVE SCATTERING
the observation point, while PO assumes that secondary
spherical waves radiate from the unobstructed primary wavefront. Figure 5 shows Huygens sources radiating into the GO
shadow region behind the absorbing half screen. The elementary electric field due to each secondary point source is
Wavefront
Secondary
sources
E 0 j (r+ )
E=
e
dE
r+
(12)
r
x0
where r is the distance from the observation point to the conducting half plane, and is the additional distance to the secondary sources. From Fig. 5,
(r + )2 = r2 + 2r + 2 = x2 + r2
GO shadow
Conducting
half plane
(13)
Clearly, the secondary sources closest to the half screen will
dominate the amplitude term in Eq. (12). Therefore, one can
make the assumption that r in the amplitude term, and
Figure 5. Half-screen diffraction using PO. The unblocked secondary
sources radiate into the GO shadow region, accounting for diffraction.
r2 2 for the phase term. Thus, Eq. (13) reduces to
Eg + Ed with E0 = 1 : z = 2 wavelengths
491
x2
2r
(14)
Electric field (V/m)
1.5
and Eq. (12) becomes
1
E = E 0 /r e j r
0.5
0
e j x
2 /2r
dx
(15)
x0
where r is a constant. Letting u 2/ r x and u0 2/ r x0,
Eq. (15) becomes
0.5
1
20
E=
15
10
5
0
5
x (wavelengths)
(a)
10
15
20
2r
E 0 e j r
e j u
2 /2
du
(16)
u0
The limits of integration can be split into two terms:
Eg + Ed with E0 = 1 : z = 20 wavelengths
E=
2r
E 0 e j r
e j u
2 /2
uo
du
e j u
2 /2
du
(17)
Electric field (V/m)
1.5
which has the form of Fresnel cosine and sine integrals.
Equation (17) can be written as
1
0.5
E=
2r
E 0 e j r
1
2
+ j 12 [C(u0 ) + jS(u0 )]
(18a)
0
0.5
1
20
15
10
10
15
20
x (wavelengths)
(b)
Figure 4. Simulation of half-screen diffraction by the GTD. The solid
straight line represents the GO incident field. The oscillating curve
is the diffracted field, calculated by the GTD, added to the GO field,
at a distance behind the screen of (a) z 2 wavelengths and (b) z
20 wavelengths.
where C(u0) and S(u0) are the Fresnel sine and cosine integrals respectively (1921). The solution to Eq. (18a) is similar
to the GTD solution for Fig. 3, with the exception that Eq.
(18a) for PO does not suffer the discontinuity of Eq. (11) for
GTD. In fact, Eq. (18a) has an analytic solution in the GO
illuminationshadow transition region. The total electric field
at x0 0 is
E (x0 = 0) =
2r
E 0 e j r
and has a magnitude of (E0 /2) /r.
1
2
+ j 12
(18b)
492
ELECTROMAGNETIC WAVE SCATTERING
DIFFRACTION THROUGH AN APERTURE
The equivalence principle can be combined with PO to analyze scattering through an aperture. Figure 6 shows an electromagnetic plane wave which is incident normally on an aperture in a conducting screen of infinite extent. While this
problem may not be realistic, it can make a good approximation for an aperture in an electrically large conductive plane.
From the equivalence principle, the reradiated field appears
to be generated by the current sheets described in Eq. (8).
Starting from Maxwells equations, the electric and magnetic
fields radiated from the electric and magnetic current sources
are (22)
e j R
dx dz
E = j
J
4
R
S
1
4
S
e j R
dx dz
M
R
4
M
S
1
4
e j R
dx dz
R
S
e j R
dx dz
J
R
E
H = jE
(19b)
where the primed symbols refer to the source rather than the
field. The equivalence principle allows the electric field in the
aperture to be replaced by the magnetic current sheet Ms over
a continuous conducting screen (7). The aperture is essentially shorted, which cancels Js. From image theory, it appears as though an identical image of Ms lay on the opposite
side of the screen. Since these two current sheets nearly coincide, the entire problem can be replaced with 2Ms at the aperture location, and no screen at all. Then Eq. (19b) becomes
e j R
dx dz
H = j
M
(20)
2
R
(21)
The distance R from each elemental source to the field point
can be obtained from the law of cosines,
R=
r2 + r2 2rr cos
(22)
where r cos x sin cos y sin sin . Equation (20)
would be difficult to integrate with a direct substitution of Eq.
(22). However, if the scattered field is observed in the far-field
region, R and r will be virtually parallel. The far-field limit is
usually taken to be
r
(19a)
and
H=j
Rather than solving the integrodifferential Eq. (19a), Amperes law can be used to obtain directly the scattered electric
field in the source-free region,
2D2
(23)
where D is the largest dimension of the aperture, in this case,
the length of the diagonal (19). The far-field assumption
allows for the approximation R r r cos in the phase,
and R r in the amplitude. Furthermore, in the far field,
E H, where / is the intrinsic impedance of the
surrounding medium. This eliminates the need to solve Eqs.
(19a) or (21). Since the incident plane wave is normal to the
aperture, it will not vary over the aperture. Therefore, it can
be brought out of the integral. Then the equation for the scattered magnetic field becomes
H = j
E
2r 0
a/2
a/2
b/2
e j (x
sin cos +y sin sin )
dx dz
b/2
(24)
While appearing messy, Eq. (24) is a straightforward integral. After integrating the two exponential terms, substituting the limits, and applying the identity
sin =
S
e j e j
j2
the scattered magnetic field in Eq. (24) becomes
x
H= j
abe j r
r
sin X sin Y
X
(25a)
where
R
X =
a sin cos
2
(25b)
Y=
b sin sin
2
(25c)
and
y
Figure 6. Coordinate system for the aperture diffraction problem.
R is the vector pointing from the differential element dx dy to the
field point, and is the resultant vector sum of r and r. The distance
from the origin to the secondary source in the aperture plane is r
x2 y2.
Figure 7 is a plot of Eq. (25a), with the amplitude normalized. The x dimension is a 6, the y dimension is b 3,
and the observation screen is z 100 from the aperture.
The normal incidence was chosen for this problem to illustrate the concept while keeping the mathematics simple.
ELECTROMAGNETIC WAVE SCATTERING
493
v
E( m )
1
0.8
0.6
0.4
0.2
0
10
5
y( )
10
0
5
x( )
5
10
5
10
However, Eq. (25) can be extended to oblique incidence by
modifying the current source Ms. Assuming that the source of
the incident field is far from the aperture, the amplitude will
not vary significantly across the aperture. However, the phase
of each differential element of Ms. will vary. The procedure is
the same as for this analysis, except that some angle terms
for the incident field will be added to X and Y in Eq. (25). The
integration then follows in a similar manner (22).
BABINETS PRINCIPLE
Scattering from a conductive plate can be modeled in a manner that virtually parallels the preceding solution to the aperture. In the case of scattering from a conductive plate, the
current sources are obtained using the induction theorem. In
fact, scattering through the aperture is the exact complement
to the scattering off of the conductive plate that was essentially cut out of the conductive screen to create the aperture.
If every electric parameter and the corresponding magnetic
parameter were swapped, the solutions would be identical.
Babinets principle originally stated that the sum of the intensities from an obstruction and its complement (i.e., a similarly
shaped aperture in an infinite screen) is equal to the intensity
that would have existed if no obstruction existed at all:
Sa + Sc = S0
(26)
While this relationship works for optics, it does not take
account of polarization. To apply Babinets principle to vector
fields, it must be modified to (23)
Ec
Ha
+
=1
Hi
Ei
(27)
The first term in Eq. (27) is the ratio of the field diffracted by
the aperture to the field with no screen present at all, and the
second term is the ratio of the field produced by the complementary screen to the conjugate source. The conjugate source
refers to the opposite field rotated by 90. In vector form, Eq.
(27) can be rewritten as
E c = E i H a
(28)
Figure 7. Normalized scattering pattern through the aperture. The dimensions of the aperture are 6 by 3 wavelengths;
the observation screen is 100 wavelengths from the plane of
the aperture. The scattering pattern is wider in the x direction, since the x dimension of the aperture is twice the y dimension.
which indicates that the electric field scattered from a conductive plate can be calculated from the field scattered from the
aperture, by subtracting the latter from the incident field (22).
SPECIAL CASES OF ELECTROMAGNETIC WAVE SCATTERING
Rayleigh Scattering
If the scattering object is much smaller than a wavelength,
its scattered energy varies inversely as the fourth power of
the wavelength (1,24,25). Therefore, for a given subwavelength object, higher frequencies will scatter more than lower
frequencies. This is the basis behind the concept of Rayleigh
scattering for small scatterers. In fact, Rayleigh scattering
answers the commonly asked question: Why is the sky blue?
Since the blue end of the visible spectrum has the shortest
wavelength, blue light scatters more than the rest of the visible spectrum from dust, water, and even air molecules. As the
scattering objects become larger, they fall into the category
called Mie scattering.
Radar Cross Section
Electromagnetic wave scattering is the basis by which radar
signals are returned to the radar receiver from a target. Since
the typical radar system employs a colocated transmit and
receive antenna, the source and observation points are the
same. This scenario is a specific case of electromagnetic wave
scattering, as previously discussed, and is known as monostatic scattering.
As the transmitted power Pt propagates through space, it
spreads over an increasing surface area, A, resulting in decreased power density St Pt /A. If Pt spreads spherically, as
with a point source or isotropic radiator, A 4d2, where d
is the distance from the transmitter. A target can intercept
part of the transmitted power and scatter it in various directions. The radar cross section (RCS) is the effective area of
the target that would return the monostatic power density
back to the source, if this target scattered the power isotropically (1,26,27). The RCS is related to the physical cross-sectional area of the target but also depends on factors such as
the frequency and polarization of the radar signal as well as
the targets shape, material, and orientation to the transmitter.
494
ELECTROMAGNETIC WAVES IN THE IONOSPHERE
SUMMARY
Since electromagnetic scattering perturbs the incident field,
it can create interference, both constructive and destructive.
The three main mechanisms of scattering are reflection, refraction, and diffraction. Reflection and refraction are the
most common, and since these scattering mechanisms tend to
be specular, they are easiest to analyze. Diffraction is much
more difficult to analyze, and is typically not as dominant as
the other two.
Various approximations can lead to solutions of diffraction
problems. The two main approximations covered were the
GTD, which takes advantage of ray tracing, and PO, which
relies on the Huygens secondary sources. Slight modifications
to the PO solution for the aperture problem can lead to solutions for backscattering and forward scattering from a conductive plate. This simple structure can serve as a building
block for more complicated structures, which can be modeled
as composites of conductive plates.
BIBLIOGRAPHY
1. J. D. Kraus, Antennas, 2nd ed., New York: McGraw-Hill, 1988.
2. C. R. Paul and S. A. Nasar, Introduction to Electromagnetic
Fields, New York: McGraw-Hill, 1987.
3. C. T. A. Johnk, Engineering Electromagnetic Fields and Waves,
New York: Wiley, 1988.
4. G. G. Skitek and S. V. Marshall, Electromagnetic Concepts and
Applications, Englewood Cliffs, NJ: Prentice-Hall, 1982.
5. P. A. Tipler, Physics, New York: Worth, 1976.
6. C. A. Balanis, Advanced Engineering Electromagnetics, New
York: Wiley, 1989.
7. E. C. Jordan and K. G. Balmain, Electromagnetic Waves and Radiating Systems, Englewood Cliffs, NJ: Prentice-Hall, 1968.
8. R. F. Harrington, Time-Harmonic Electromagnetic Fields, New
York: McGraw-Hill, 1961.
9. G. E. Corazza et al., A characterization of indoor space and frequency diversity by ray-tracing modeling, IEEE J. Selected Areas
Commun., 14 (3): 411419, 1996.
10. M. Kimpe, V. Bohossian, and H. Leib, Ray tracing for indoor radio channel estimation, IEEE Proc. 2nd Int. Conf. on Universal
Personal Commun. (ICUPC), October 1993, pp. 6468.
11. J. B. Keller, A geometric theory of diffraction, in L. M. Graves
(ed.), Calculus of Variations and its Applications, Proc. Symp.
Appl. Math., Vol. III, New York: McGraw-Hill, 1958, pp. 2752.
12. G. E. Athanasiadou, A. R. Nix, and J P. McGeehan, Indoor 3D
ray tracing predictions and their comparison with high resolution
wideband measurements, Proc. IEEE 46th Veh. Tech. Conf., April
1996, Vol. 1, pp. 3640.
13. O. Landron, M. J. Feuerstein, and T. S. Rappaport, A comparison
of theoretical and empirical reflection coefficients for typical exterior wall surfaces in a mobile radio environment, IEEE Trans.
Antennas Propag., 44: 341351, 1996.
14. R. H. Clarke and J. Brown, Diffraction Theory and Antennas, New
York: Wiley, 1980.
15. J. D. Kraus, Electromagnetics, 4th ed., New York: McGraw-Hill,
1992, pp. 622627.
16. R. G. Kouyoumjian and P. H. Pathak, A uniform geometrical theory of diffraction for an edge in a perfectly conducting surface,
Proc. IEEE, 62: 14481461, 1974.
17. J. B. Keller, Geometrical theory of diffraction, J. Opt. Soc. Amer.,
52 (2): 116130, 1962.
18. J. B. Keller and Albert Blank, Diffraction and reflection of pulses
by wedges and corners, in The Theory of Electromagnetic Waves,
New York: Interscience, 1951, pp. 139158.
19. C. A. Balanis, Antenna Theory, Analysis and Design, New York:
Harper & Row, 1982, pp. 9294, 448.
20. E. Hecht, Optics, 2nd ed., Reading, MA: Addison-Wesley, 1987,
p. 393.
21. M. Spiegel, Advanced Mathematics for Engineers and Scientists,
New York: McGraw-Hill, 1971.
22. R. L. Musselman, Analytic non-asymptotic model for diffraction
from large walls consisting of complex-shaped conductive scatterers, Univ. Colorado, Colorado Springs, 1997.
23. H. G. Booker, Slot aerials and their relation to complementary
wire aerials, J. Inst. Electr. Eng., Part III A, 1946, pp. 620626.
24. A. Ishimaru, Wave Propagation and Scattering in Random Media,
San Diego, CA: Academic Press, 1978.
25. H. C. van de Hulst, Light Scattering by Small Particles, New York:
Dover, 1981.
26. M. I. Skolnik, Introduction to Radar Systems, 2nd ed., New York:
McGraw-Hill, 1980.
27. S. A. Hovanessian, Radar System Design and Analysis, Norwood,
MA: Artech House, 1984.
RANDALL L. MUSSELMAN
United States Air Force Academy
HELICAL ANTENNAS
HELICAL ANTENNAS
The helix antenna has a long and fascinating history. It was
discovered in 1946 by John Kraus. Since then, new variations
have continued to arise, even up to the present day. The discovery itself is a very interesting story, which is told in John
Krauss book Antennas (1). It all started with an afternoon
lecture at Ohio State University. John Kraus listened as the
speaker described the wave-guiding helix used in travelingwave tubes; he wondered if this interesting helix could be
used as an antenna and, after the talk, he asked the wellknown lecturer. The speaker replied emphatically, No, Ive
tried it and it doesnt work. That very evening, John Kraus
went down to his basement, wound a seven-turn helical coil
one wavelength in circumference and fed it by means of a
coaxial line and ground plane [Fig. 1(a)]. He found that it
produced a sharp beam of circularly polarized radiation off
the open end of the helix. So the helix antenna was born,
despite the advice of experts.
plane parallel to the axis, for example, any plane that includes the axis.
Figure 1(a) shows the wire helix antenna. Parameters are
defined as follows:
D is the diameter of helix, which is equal to 2a.
C is the circumference.
S is the spacing between turns.
is the pitch angle.
L is the length of one turn.
N is the number of turns.
C, S, and L represent the respective distances in wavelengths such that C C/ , and so on. When one turn of the
helix is unrolled on a flat plane, the relationships between the
spacing S, circumference C, and turn length L and pitch angle
can be obtained from the triangle shown in Fig. 1(b) as follows:
S = C tan
Helical Curves
The helix was well-known in ancient Greece. Geminus described it in the first century B.C. and there are references to
earlier work on the helix. The cylindrical helix may be defined
by considering a right circular cylinder of radius a, whose axis
is the z axis. Using a right-handed cylindrical coordinate system (r, , z) the equations of the helix are
x = a cos
y = a sin
z = a tan
(1)
where is the pitch angle of the helix and 2a is the diameter.
The lines of the cylinder parallel to the z axis, that is, the
lines (r a, o) are considered to be the generators of
the cylinder. The cylinder is generated by rotating any generator about the z axis. The helix cuts the generators at a constant angle (/2) . It also projects as a sine curve on any
z
D
S
S
C = D
(b)
Ground
plane
Coax
(a)
Figure 1. The helical antenna. (a) A helix fed by a coaxial line and
a ground plane. (b) One turn of the helix unrolled on a flat plane.
661
= tan1
S
C
L=
C2 + S2
(2)
Thus we need only three independent parameters C, , N to
describe a helix. Note that when 0, S 0, and the helix
reduces to a planar loop. When 90, C 0 and the helix
becomes a straight line.
The round-wire helix such as Kraus first built consists of
a round wire of a radius whose axis is the helix curve of Eq.
(1). The tape helix is a conducting tape of width w, which is
wound around a cylinder or a thin cylindrical tube. Its centerline is the helix of Eq. (1). A single-wire helix is called the
monofilar helix. The double, or bifilar, helix is constructed by
adding an additional member, which is formed by replacing
with everywhere in Eq. (1). The quadrifilar helix antenna is formed by adding three members; in Eq. (1) is replaced with (/2), , (3/2). A left-handed helix can be
formed by using a left-handed coordinate system, or by replacing in Eq. (1) with 2 . Figure 1(a) shows a right-handed
helix. There are other helical curves. The conical helix lies on
the surface of a cone and cuts the radial lines of the cone, the
generators, at a constant angle. The spherical helix lies on the
surface of the sphere and cuts the generators, for example,
the longitude lines, at a constant angle.
There are many striking examples of the helix in nature,
from some of the smallest to some of the largest objects. Most
important of all is the deoxyribonucleic acid (DNA) molecule,
which is a double helix. The marks on a snail of the family
Helicidae resemble a spherical helix. The human ear has a
prominent helical ridge. The Heliconia is a family of herbs
with a helical shape. The helictite is similar to a stalagtite.
Some trees have helical bark. Finally, we have the largest
of the nebulae, the Helix Nebula. In addition, we see many
manmade forms of the helix around us, including automobile
springs, spiral staircases, inductance coils, transformer coils,
parking lot ramps, automobile antenna coils, and finally, the
lowly screw, which is a combination of conical and cylindrical helices.
MONOFILAR HELICAL ANTENNA
We first consider the characteristics of a monofilar, or unifilar, helical antenna. In this section we simply call it a helical
J. Webster (ed.), Wiley Encyclopedia of Electrical and Electronics Engineering. Copyright # 1999 John Wiley & Sons, Inc.
662
HELICAL ANTENNAS
antenna, implying that it is a monofilar helix. The helix commonly operates in two different modes, the normal mode and
the axial mode depending on the electrical size of the helix.
When the dimensions of the helix are small compared with a
wavelength (D , NS ), the maximum radiation is normal (or perpendicular) to the helix axis. This condition is
called the normal mode. When the helix circumference is on
the order of one wavelength, the maximum radiation is along
the helix axis. Thus, this type of operation is called the axial
mode. The axial mode helix is a broadband antenna. The radiation from this axial mode helix is close to circular polarization along the axis. There is also a backfire mode which is
discussed in a later section.
helix which consists of a single small loop of diameter D and
one short dipole of length S.
Assume that the complex amplitude of current is I and the
angular frequency is . The radiation electric field of the
small loop of diameter D has only an E component, given by
E = k2 IA
E = jIS
(a)
(b)
(c)
Figure 2. The normal mode helix. (a) Coordinate system. (b) Loop
and dipole model. (c) Beam pattern.
ejkr
sin
4r
(4)
The total radiation field for one turn is then given by
E = a E + a E = {a jS + a k2 A}I
ejkr
sin
4r
(5)
The normalized radiation field pattern f() of the normal
mode helix is
f ( ) = sin
(6)
which is the same as that of the Hertzian dipole and the small
loop and is shown in Fig. 2(c). The field is zero along the axis
(in the end-fire direction) and is maximum in the xy plane
( 90), which is normal to the helix axis.
Because E and E are 90 out-of-phase, as shown in Eqs.
(3) and (4), the radiated wave is elliptically polarized. The
axial ratio (AR) of the polarization ellipse of the far field is
obtained by dividing the magnitude of Eq. (4) by that of Eq.
(3):
AR =
(3)
where A D2 /4 is the area of the loop, k the propagation constant, and / the intrinsic impedance. The
far field of the short dipole, or the Hertzian dipole of length
S, has only an E component, given by
Normal Mode Helix
Lets consider a helix with its axis along the z axis, centered
at the origin [Fig. 2(a)]. The geometry of the helix reduces to
a loop when the pitch angle approaches zero and to a
straight wire when it approaches 90. Since the limiting geometries of the helix are a loop and a dipole, the far field
radiated by a small helix can be described by the radiation
fields of a small loop and a short dipole when dimensions are
small compared to a wavelength. The analysis of a small short
helix is facilitated by assuming that the helix consists of a
number of small loops and short dipoles connected in series
as in Fig. 2(b). The diameter of the loops is the same as the
helix diameter (D) and the length of the dipoles is approximately the same as the spacing (S) between turns of the helix.
Because the helix is small and short, the current distribution
is assumed to be uniform in magnitude and phase over the
entire length of the helix. For the same reason, the far-field
pattern will be independent of the number of turns and thus
can be obtained by considering the pattern of a single-turn
ejkr
sin
4r
|E |
=
|E |
S
S
2S
=
=
2
2A
(D)2
A
k
(7)
where we have used k 2/ , k . Because E and E
are 90 out-of-phase, the polarization ellipse becomes a circle
when E E , indicating circular polarization. Setting
AR 1 yields
C = D = 2S or C = 2S
(8)
Under this condition the radiation field is circularly polarized
in all directions except of course along the axis where the radiation is zero. The polarization ellipse of the radiation from
a helix of constant turn-length (L) changes progressively as
the pitch angle is varied. When tan1(S/C) 0 (the
helix reduces to a loop), AR 0, E 0, E a E ; thus the
wave is linearly polarized with horizontal (or perpendicular)
polarization. As increases, the polarization becomes elliptical with the major axis of the ellipse being horizontal. When
reaches a value such that condition (8) is satisfied, AR 1,
and the polarization is circular. With the help of Eq. (2), the
condition (8) leads to the following value of :
!
"
1 + L2
1 1 +
(9)
CP = sin
L
HELICAL ANTENNAS
As increases further, the polarization again becomes elliptical with the major axis being vertical. Finally, when 90
(the helix reduces to a dipole), AR , E 0, E aE;
thus the polarization is linear with vertical (or parallel) polarization. For small pitch angles ( 1), Eq. (9) is simplified to
CP = C /2
(10)
where CP is in radians. For small pitch angles, circular polarization can occur at frequencies such that the circumference
is very small compared to a wavelength (C 1).
From Eqs. (3) and (4), we note that the loop field E and
the dipole field E, respectively, are proportional to the second
and first powers of frequency. Correspondingly, radiation resistance of loop and dipole are proportional to the fourth and
second powers, respectively. Thus, as frequency decreases,
the dipole radiation predominates and the beam pattern is
linearly polarized. In this linearly polarized frequency range,
the normal mode helix has some interesting properties. Its
beam pattern is essentially that associated with the dipoles,
that is, a monopole of length NS above a ground plane. Its
impedance, however, is significantly affected by the loops.
The normal mode helix is limited by its size. It has the
same restrictions and limitations that apply to any electrically small antenna. But within those restrictions, it has certain advantages over a dipole antenna of the same height.
These include a lower frequency for resonance and a larger
radiation resistance, both because of the longer path of the
helical structure. While the dipole may require additional impedance-matching circuits to achieve resonance, the helix is
resonant without supplementary matching elements. Another
advantage over the dipole is that the helix is flexible and
more resilient. The higher radiation resistance, resonant
characteristic, and flexibility make the normal mode helix
suitable for small antennas used in mobile communications.
ground plane can be made from either solid metal or wire
mesh.
Analysis of Radiation Pattern. The axial mode helix has a
circumference of approximately one wavelength, so the current distribution would not be uniform, and we assume that
there is an outgoing current wave, traveling along the helical
conductor at phase velocity v pc (p is the phase velocity
relative to the speed of light c in free space). Then
I() = I0 e j
AF =
3
4
< C <
4
3
(11)
The bandwidth ratio, the ratio of the upper and lower frequencies is 4/3 3/4 or 1.78, which is close to a 2 : 1 bandwidth. The helix is usually fed axially or peripherally with
the inner conductor of the coaxial line connected to the helix
and the outer conductor attached to the ground plane. The
(12)
where is the distance measured along the helix from the
beginning of the turn closest to the ground plane, I0 the input
current, k/p the phase constant of the current wave.
When the total length of one turn is approximately a wavelength, the current distribution in Eq. (12) has opposite phase
(180 out-of-phase) on opposite sides of a turn, because they
are separated by about a half-wavelength. Also the helical coil
physically reverses current direction for opposite points. Thus
the currents at opposite points of a turn are essentially in
phase, giving rise to reinforcement in the far field along the
helix axis. We can find the radiation pattern by using the
principle of pattern multiplication because a helix with uniform cross section can be considered as an array of N identical
elements (or turns). We have a uniformly excited, equally
spaced array with spacing S, so the total pattern is the product of the pattern for one turn (the element pattern) and the
pattern for an array of N isotropic point sources (an array
factor). When the helix is long (say, NS ), the array factor
is much sharper than the element pattern and hence determines the shape of the total far-field pattern.
Array Factor. The array factor (AF) of a uniformly excited,
equally spaced, linear array of N elements is given by
Axial Mode Helix
A very useful mode of operation for the helical antenna is the
axial or endfire mode. In this mode the radiation pattern has
a single main beam along the axis of the helix (z direction),
that is, it is an endfire antenna. Experiments have shown
that the axial mode occurs when the circumference of the helix is approximately one wavelength and when the helix has
several turns. A primary component of current on the helix
is a wave traveling outward from the feed along the wire at
approximately the speed of light and the radiation is a beam
off the end of the helix. Because the electric field vector rotates around in a circular fashion as does the current on the
helix, we expect that the radiation field is circularly polarized
along the helix axis. One very important feature of the axial
mode helical antenna is its broadband character. As a rule of
thumb, the approximate bandwidth for the axial mode is
given as follows (2):
663
sin[(N/2)]
N sin(/2)
(13)
and
= kS cos +
(14)
where is an angle measured from the array axis (z axis) and
is the phase shift between adjacent elements. Explaining
the phase relationships of the axial mode helix is difficult.
Begin by finding the phase shift required for endfire operation, because we know that the radiation is endfire. For ordinary endfire, we find the conditions such that is zero at
0:
= kS 2m
m is an integer
(15)
where the term (2m) reflects the basic ambiguity of phase.
Next we find conditions for increased directivity Hansen
Woodyard (HW) (3) endfire, since this is an optimum form of
endfire:
= kS 2m /N
= 2 (kS + /N) m = 1
(16)
To be more accurate, ther term /N should be replaced by
2.94/N (3). However, the choice of /N hardly changes the
664
HELICAL ANTENNAS
radiation field (4) and is convenient for expressing other
quantities. We have tried the arbitrary choice (m 1), because this term 2 corresponds roughly to one turn of the
circumference at midband (C 1). Later it will be clear that
other choices are not possible solutions. Experiments show
(1,5) that the phase shift obtained is close to that of Eq. (16)
at midband (C 1). How does this happen? We have already
pointed out that the term (2) corresponds roughly to one
turn of the circumference. In addition, the length around one
turn of the helix is greater than a wavelength (L 1) at
midband. More importantly, the velocity of travel is less than
that of light (p 0.9) at midband (5). These two additional
contributions account for the minor terms (kS /N). Thus
the phase at midband is explained. However, the experimental data for tracks Eq. (16) fairly well over the entire bandwidth of the axial mode. How is this possible? We cannot alter
our choice of m; one choice must work for the entire frequency
range. And if p remained constant, Eq. (16) could not be satisfied over the entire band. Fortunately, p does vary quite a bit
(from 0.73 to 0.97) over the axial mode frequency range; the
result is that HW endfire described by Eq. (16) is tracked
quite well over most of the band, falling off a little toward the
high end. All in all, this is quite a remarkable story. The
phase, so to speak, locks in to HW endfire over the bandwidth of almost 2 : 1. When first reported by Kraus, this was
called an anomalous phase progression. It still continues to
mystify succeeding generations.
To summarize, the phase progression along the helix wire
is relatively simple; it corresponds roughly to that of the
speed of light along the wire. The phase progression in z,
which determines the phase difference between turns, follows the phase progression of the wire. Taking into account
the phase ambiguity (2m), we see that HW endfire is obtained at midband. The relative phase velocity p then changes
with frequency just enough to maintain the HW endfire. Another point worth noting is that we have not discussed backfire ( 180) radiation. It does, in fact, occur along with the
axial mode but is usually suppressed by the ground plane. It
will return, to our advantage, with the multifilar helix.
Assuming, then, the validity of Eq. (16),
=
1
1
2 S 2N + 1
=
kS + 2 1 +
=
+
(17)
L
L
2N
L
2N
L
k
p= =
S + (2N + 1)/2N
(18)
Circular Loop Radiation. In this section we consider the radiation from a circular loop carrying a current I(). The result
is useful in understanding the operation of the helix in both
the mono- and multifilar forms. In addition, it will yield an
approximate element factor for a single turn of the helix. The
loop of radius a is centered at the origin and lies in the xy
plane. The current distribution I() may be represented in
terms of a complex Fourier series representation:
I() =
In ejn
(20)
n=
Consider the typical term Inejn of the current distribution
I(). First we evaluate components An, An of the far-field
magnetic vector potential as follows. Directions , are associated with the field point rather than the source point.
ejkr
4r
jkr
An =
A n =
e
4r
In ejn cos( )ejka sin cos( ) a d
(19)
(21)
In ejn [sin( )] cos ejka sin cos( ) a d
(22)
To evaluate A , we introduce the change of variables
and change limits to obtain
An =
ejn ejkr In a
4r
2
0
ejn
e j + e j
2
e j(ka sin ) cos d
Next, we use the following integral expression for the Bessel
function of the first kind Jm(x):
ejx cos ejm d = 2 j m Jm (x)
A is then evaluated directly to obtain
An =
ejn (In a)ejkr j n+1
[Jn+1 (ka sin ) Jn1 (ka sin )] (23)
4r
En = jAn
Using p as obtained from Eq. (18) to calculate the array factor
yields patterns in good agreement with measured patterns.
The p value calculated from Eq. (18) also is in closer
agreement with measured values of p (1). Therefore, it appears that the HansenWoodyard increased directivity condition is a good approximation for helices radiating in the axial
mode. For a typical case where C , 14, N 10, we
find from Eq. (18) that S C tan 0.249, L 1.031,
and p 0.79. Thus the traveling current wave has a phase
velocity less than that of free space. Finally, substituting Eq.
(16) into Eq. (14) yields
= kS(cos 1) 2 +
N
Equations (13) and (19) provide the complete normalized
array pattern of the axial mode helical antenna. For the element pattern we will need an analysis of the radiation from
a circular loop, which is covered in the next section.
(24)
Using similar methods, we obtain the following evaluation of
A.
A n =
e jn (In a)ejkr j n+1 ( j cos )
[Jn+1 (ka sin )
4r
+ Jn1 (ka sin )]
E n = jA n
(25)
(26)
The total fields may of course be obtained by adding contributions of all Fourier modes.
Now lets evaluate the far fields along the z axis ( 0,
180) for each of the separate Fourier modes. We note that
Jn(0) 0 (n 0) and J0(0) 1. Evaluating the cases n 1,
HELICAL ANTENNAS
we find that, along the z axis,
E
= j
E
(27)
I( ) = I0 e j = I0 e j a
(28)
where k/p, a D/2 and is the angle measured from
the x axis. For accurate analysis of the element pattern, Eq.
(28) should be used to calculate the radiation integral. However, when the helix with several turns operates in the axial
mode (C 1), the array factor dominates the endfire beam
pattern and the element pattern provides minor corrections.
Thus it suffices to consider the radiation field of a planar loop
with C 1, instead of a three-dimensional one-turn helix. If
we also assume that p 1, then
Using the simple form of the current distribution in Eq. (29),
we can easily calculate the radiation fields for the element
pattern from Eqs. (23)(26) for C 1 (n 1):
E (, ) = C(r)[J0 (sin ) + J2 (sin )] e j
In other words, the modes n 1 representing traveling
waves yield circular polarization along the z axis. Note that
all other traveling-wave modes yield a null on axis. Of all the
Fourier modes, only n 1 radiate in the forward endfire or
backfire directions. For the helix, we define forward or backward radiation as radiation away from or toward the feed
point, respectively.
This result can also be seen by considering currents
around the loop for various modes. For n 1, each current
element is matched by its opposite across the loop that is in
the same direction such as to add along the z axis and to
rotate polarization as time progresses. All of the other modes
cancel along the axis. For even modes, each element is cancelled by its opposite across the loop. For odd modes, a group
of elements will cancel. For the general odd case n, any group
of n elements each separated by 180/n yields zero contribution. For n 5, for example, any group of five elements each
separated by 36 yields zero contribution.
Any currents on the cylindrical surface may be resolved
into - and z-directed currents. The z-directed currents do not
radiate along the axis. Thus, for currents of any direction,
only the n 1 Fourier modes can contribute to endfire or
backfire. These results will be useful when considering
multifilar helices.
The above discussion makes it easier to understand the
operation of the helical antenna. At low frequencies the zeroth mode (n 0) is strongly excited, because there is little
variation of phase around the cylinder on one turn. In addition, the impedance of the higher modes is highly reactive. As
frequency increases and C approaches unity, we have one
complete cycle around the cylinder on one turn, and we expect
the ej mode to be excited for a right-hand helix. The phase
velocity of the helix is lower than that associated with the
speed of light and the impedance of the mode n 1 is reasonable, and so the axial mode begins at approximately C
0.75. Similarly, as frequency increases we expect the mode
n 2 to appear; this mode would produce beam pattern
deterioration. The axial mode continues until about C 1.33.
Element Pattern of the Axial Mode Helix. For the element
pattern of one turn of the helix, the current distribution is
assumed to be
I( ) I0 ejka = I0 e j(2 /)a I0 e j
(29)
665
E (, ) = C(r)[J0 (sin ) J2 (sin )]( j cos ) e j
(30)
(31)
where C(r) gives the r dependence of the fields. Note that
ka 1 when C 1. If we plot the radiation patterns of E
and E using Eqs. (30) and (31) we obtain a figure-eight pattern for E with a null at 90, and a nearly omnidirectional
pattern for E (6). From the plot, it is interesting to note that
the normalized E can be approximated by cos. We also observe that E and E are 90 out of phase. In particular, when
0, E E , thus the radiation field is circularly polarized in the endfire direction. As one departs from 0, E
decreases more rapidly than does E , so the polarization becomes elliptical. Finally, it should be noted that Kraus (1) has
analyzed the element pattern, by using a single turn of a
three-dimensional helix with uniform traveling wave current.
Beam Patterns. The complete total far-field pattern is given
by the product of the array factor shown by Eq. (13) and the
element pattern in Eq. (30) or Eq. (31). However, the array
pattern is much sharper than the element patterns. Thus the
total E and E patterns are nearly the same, in spite of the
difference in the single-turn patterns. The main lobes of the
E and E patterns are very similar to the array pattern.
Therefore, for long helices (NS ), a calculation of only the
array factor is sufficient for an approximate pattern of any
field component of the helix.
The measured patterns of a six-turn helix with 14 as
a function of frequency are presented in Fig. 3. Patterns are
shown over a range of circumferences from approximately
0.66 to 1.35. The solid patterns are for the horizontally polarized component (E ) and the dashed for the vertically polarized (E). Both are adjusted to the same maximum. We observe that the endfire beam patterns are preserved over the
range of 0.73 C 1.22, indicating that the axial mode
helix is a broadband antenna.
Important Parameters. Four important parameters for practical design of an axial mode helical antenna are beamwidth
(BW), gain or directivity, input impedance, and axial ratio
(AR). They are all functions of the number of turns, the turn
spacing (or pitch angle), and the frequency. For a given number of turns, the behavior of the BW, gain, impedance, and
AR determines the useful bandwidth. The nominal center frequency of this bandwidth corresponds to a helix circumference of about 1.
Beam Width. Based on a large number of measurements
King and Wong (7) give the following quasiempirical formula
for the beamwidths:
HPBW (half-power beam width) =
K
B
[degrees]
C NS
(32)
where KB varies from 61 to 70, for 3/4 C 4/3, 12
15, and 8.6 N 10. Note that as N increases the beamwidth decreases. Figure 4(a) shows measured HPBW of a sixturn, 14 axial-mode helix as a function of the normalized cir-
666
HELICAL ANTENNAS
C = 0.66
275 MHz
C = 0.73
300 MHz
C = 0.85
350 MHz
C = 0.97 C = 1.09
400 MHz 450 MHz
C = 1.22
500 MHz
C = 1.35
550 MHz
= 14
n=6
Figure 3. Measured beam patterns of the monofilar axial mode helix. From Kraus (1). 1988
by McGraw-Hill, Inc. Reprinted with permission of the McGraw-Hill Companies.
cumference (C). We observe that HPBW changes slowly over
the range of approximately 0.7 C 1.25.
Gain. The gain of the axial mode helix can be approximately obtained (8) by
Half-power beam width
G = KGC2 NS
(33)
180
150
AR =
Axial ratio
2N + 1
2N
( = 0 )
(34)
100
If N is large, the axial ratio approaches unity and the polarization is nearly circular. For example, for a six-turn helix,
AR 13/12 1.08 according to Eq. (34). This axial ratio is
independent of frequency or circumference. In Fig. 4(b), the
measured values of the axial ratio for the six-turn, 14 axialmode helix are plotted as a function of the circumference
(C). We observe that AR is nearly 1 over the range of about
0.73 C 1.4. The sense of circular polarization is determined by the sense of the helix windings.
Input Impedance. The input impedance of the axial mode
helical antenna is nearly purely resistive. The empirical formulas for the input resistance are given (1) by
50
0
6
5
4
3
2
1
7
6
VSWR
where KG is the gain factor which depends on the design parameters. King and Wong (7) report that KG varies from 4.2
to 7.7. Experiments show that the gain is peak when C is
slightly larger than 1.
Axial Ratio. We have shown from the approximate analysis
described in a previous section that the radiation field is circularly polarized in the mainbeam direction ( 0), implying
AR 1. With a more accurate analysis including the effect
of relative phase velocity for increased directivity, Kraus (1)
obtains the axial ratio along the helix axis as follows:
+
X
5
4
.5 .75 1.0 C 1.5
3
2
1
.5 .6
.7
.8
.9 1.0 1.1 1.2 1.3 1.4 1.5 1.6
Circumference (C )
Figure 4. Measured performance of the monofilar axial mode helix.
(a) Beamwidth. (b) Axial ratio. (c) VSWR. From Kraus (1). 1988 by
McGraw-Hill, Inc. Reprinted with permission of the McGraw-Hill
Companies.
Rin = 140C
(35)
within 20% for the case of axial feed, and
Rin = 150/ C
(36)
within 10% for the case of peripheral feed. Both relations are
valid when 0.8 C 1.2, 12 14 and N 4. With a
suitable matching section, Rin can be made any desired value
from 50 to 150. In the inset of Fig. 4(c), trends of input
resistance R and reactance X are shown as a function of the
relative frequency or circumference. Note that R is relatively
constant and X is very small for 0.7 C 1.5. Figure 4(c)
also shows the voltage standing wave ratio (VSWR) measured
on a 53 coaxial line. We observe that the VSWR nearly re-
;;
;;
;;
;;
;;
;
;;;;;
;;;;
mains constant (approximately 1), and equivalently the input
impedance of the helix remains unchanged, over the range of
about 0.7 C 1.6.
Ends
open
HELICAL ANTENNAS
667
Ends
shorted
Broadband Characteristics. Considering all the characteristics of beam pattern, input impedance, and polarization as a
function of circumference, we find that the performance of the
axial mode helix is satisfactory over the range of about
0.75 C 1.25 within the restrictions given on and N.
Thus the bandwidth, defined by the ratio of upper and lower
frequencies, is almost an octave. The broadband characteristics of the helix can be explained by the natural adjustment
of the phase velocity. As the helix size C, or equivalently the
frequency, varies over rather wide range, the phase velocity
adjusts itself automatically such that the fields from each
turn add nearly in phase in the axial direction.
Tape 4
Tape 3
Variations and Applications of the Helical Antenna
A slight taper on the end of the helix (9,10) reduces the axial
ratio at the expense of a slight reduction in gain. Axial ratio
is improved both on and off axis. A taper is also used at the
input to improve impedance characteristics. A circular-cavity
backing is sometimes used to reduce the back radiation and
increase the forward gain. Dielectric-tube support has been
used with the helix antenna. This lowers the frequency for
the onset of axial-mode operation and has an effect on the
terminal impedance. A solid dielectric core has also been used
with the helix (the polyrod helix). A helix with an inner concentric metal core has been used as a TV transmitter (11).
The antenna utilizes higher order Fourier modes such as
ej2 , ej5 which radiate sidefire rather than endfire. This is
particularly useful with towers and masts whose circumference is much larger than a wavelength. An array of helices is
stacked along the mast to produce the required beam pattern.
The helical antenna has often been used as an element in
various types of arrays. Large planar arrays of helices have
been used in radio astronomy (12). An array of axial-mode
helices has been used for global positioning system (GPS) satellite transmitters (1). Helices are also used as feeds for parabolic dishes. Applications of the helix are legion.
MULTIFILAR HELIX ANTENNAS
The helical antenna described in the preceding section may be
termed the monofilar helical antenna. It is constructed from a
single wire, or tape, and fed from a single source. In this section we consider the multifilar helix antenna, which consists
of a number of wires or tapes, each of which may be fed from
a separate source. The wires may be interleaved as shown in
Fig. 5(a) for the quadrifilar helix. The excitations in all cases
discussed is of the form ej . Other excitations are certainly
possible but have not been thoroughly studied. Many different
forms of the multifilar helix have been used, including bifilar,
quadrifilar, and octofilar helices. There are two distinct
classes of multifilar helices which have been used, namely,
the broadband forward-fire axial mode multifilar helix and
the narrowband backfire multifilar helix. For both, the quadrifilar helix has been widely studied and used.
We have already discussed the Fourier modes due to a loop
of circumferential current. The rigorous consideration of the
entire geometry of the helix yields modes with the same
Ground
plane
Tape 2
Tape 1
DG
(a)
Feed region
(phases 0, 90, 180, 270)
(b)
Figure 5. Quadrifilar helical antenna. (a) Endfire. (b) Backfire.
dependence but yields the phase progression in z as well.
Samuel Sensiper (13) carried out this rigorous analysis to determine the real propagation constants of the normal and
axial modes. He also determined some of the characteristics
of the multifilar helix. Later, Paul Klock (14) found an additional mode with a complex propagation constant. This mode
starts with backfire then splits and scans forward as a conical
beam as frequency increases. It operates simultaneously with
the axial mode but is usually suppressed by the ground plane.
Early experimenters of the multifilar helix showed some improvements over the monofilar helix but did not always recognize that larger bandwidth could be obtained nor how to obtain it, as explained in the following section.
The Axial Mode Quadrifilar Helix
Gerst and Worden (15) invented the broadband axial mode
multifilar helix. He pointed out that the frequency range of
the axial mode could be extended by (1) adding more wires,
(2) using ej excitation to maintain that mode and suppress
others, and (3) increasing the pitch angle . The technique is
readily understood by considering the bifilar helix. First, the
two wires are fed 180 out of phase. Consider a cross section
perpendicular to the helix axis. We find, in any cross section,
two wires 180 apart in space and phase. Only the odd Fourier modes (ejn , n odd) are excited. The mode ej2 , which may
be a culprit in the pattern breakup of the monofilar helix, is
suppressed. The mode ej3 is not suppressed and the bandwidth for the axial mode approaches 3 : 1. A pitch angle of
about 2530 is required. Now consider the quadrifilar helix.
A cross section through the right-hand helix of Fig. 5(a) displays four wires symmetrically arranged around the periphery, with phases 0, 90, 180, and 270. This ej excitation
suppresses all of the even modes, as in the bifilar helix, and
mode ej3 as well. ej5 is not suppressed. The bandwidth ap-
668
HELICAL ANTENNAS
270
270
180
180
90
90
(a)
(b)
270
270
180
180
90
90
(c)
(d)
270
270
180
180
90
90
(e)
(f)
270
270
180
Figure 6. Beam patterns of the axial
mode quadrifilar helix antenna. C 0.44,
0.52, 0.72, 1.1, 1.6, 1.8, 2.1, 2.7 in (a)
through (h), respectively. From A. T. Adams and C. Lumjiak (17). 1971 IEEE.
Reprinted with permission of IEEE.
90
C min < C < C max
C min
cos
=
1 + sin
90
(g)
proaches 5 : 1. A pitch angle of approximately 40 is required.
In general, with the M-filar helix, the mode ej is excited
around the periphery. All other modes up to ejM are suppressed. The bandwidth lies between M and M 1 : 1.
Gerst and Worden determined that the frequency range of
the multifilar axial mode helix may be approximated as follows (15):
(37a)
(37b)
180
(h)
C max is the lesser of C max 1 and C max 2:
C max 1 =
C max 2 =
cos
1 sin
(37c)
M
cot
2
(37d)
where M is the number of wires. Equation (37) can be used to
predict the bandwidth of the axial mode unifilar or multifilar
helix. For example, consider the monofilar helix with 14,
whose beam patterns and other characteristics are given in
Figs. 3 and 4. The figures indicate that the bandwidth is ap-
HELICAL ANTENNAS
proximately 0.75 C 1.25 as noted previously. In comparison, Eq. (37) predicts that Cmin 0.78 and Cmax 1.28 for an
approximate bandwidth of 0.78 C 1.28. The two bandwidth ratios are very close. Equation (37) has been applied to
bifilar, quadrifilar, and octofilar axial beam helices, and the
results agree well with experiments as shown in Refs. 1517.
Figure 6 shows the beam patterns of a quadrifilar helix antenna with a ground plane [see Fig. 5(a)]. The pitch angle
is 35, diameter D is 3, antenna length is 24, ground plane
diameter DG is 10, and the tape width is 1/2. The feed system (16) provides four outputs that are phased 0, 90, 180,
and 270, each output being connected to one of the four wires
of the quadrifilar helix antenna. Equations (37b,c) yield
Cmin 0.52 and Cmax 1.92 for 35 and M 4. Thus the
bandwidth of the antenna is given by 0.52 C 1.92 for a
bandwidth ratio of 3.7 : 1. The progression of the beam patterns may be described as follows. At a frequency somewhat
below the lower limit, backfire operation begins, as evidenced
by the strong backlobe at C 0.44. The backlobe decreases
rapidly as we approach the lower limit. The axial ratio also
decreases rapidly and is less than 2 : 1 at the lower limit.
Other antenna characteristics such as VSWR are also acceptable (17). The axial mode then predominates over the 3.5 : 1
bandwidth. The beam pattern narrows steadily and the directivity increases with frequency. The upper limit occurs at
C 1.92, at a frequency just above that of Figure 6(f). Above
the upper limit, beam-splitting occurs at C 2.10 and complete pattern breakup at C 2.70. Thus, with the multifilar
helix, the bandwidth of the axial mode is extended to both
lower and higher frequencies as predicted in Eq. (37).
The beam pattern at C 1.80 exhibits the undesirable
characteristic of high side lobes. This is caused by the backfire
operation, which changes from backfire through sidefire towards endfire as frequency increases. It is sometimes called a
scanning mode. The quadrifilar helix may also be used in a
counterwound version with both right- and left-hand windings. The on-axis polarization is linear rather than circular.
The backfire mode is much more effectively suppressed in this
version, and the sidelobe levels are much lower. Bandwidths
are between 4 : 1 and 5 : 1 as shown in (16). Gerst and Worden
(15) describe a 53 pitch angle counterwound octofilar helix
with 9 : 1 bandwidth. The multifilar helix antenna does not
radiate in a normal mode at low frequencies because of the
phase excitation 0, 90, 180, and 270 of the windings. The
excitation ej is a supergain excitation at low frequencies
(C 1). Details on the axial multifilar antenna are given in
(1517) and related references.
mono- and quadrifilar helix the backfire mode shows up just
before the onset of the axial mode. The backfire mode exists
along with the axial mode, but is suppressed by the presence
of the ground plane. The backfire mode is favored over the
forward endfire as pitch angle increases. Pitch angles in the
4050 range with no ground plane show both forward and
backward radiation. With no ground plane, the radiation is
primarily backfire. For the right-hand helices of Figs. 1 and
5, the endfire radiation is right-handed circularly polarized
and the backfire radiation is left-handed circularly polarized.
The backfire bifilar helix was first studied by Patton (18),
who carried out extensive theoretical and experimental work.
He showed the range of backfire beam patterns which were
obtained with the monofilar and bifilar helices. Later, Charles
Kilgus (19) showed that beam pattern improvements could be
obtained with the quadrifilar helix. He investigated in detail
the shaped-conical and cardioid patterns which are obtainable
with the backfire quadrifilar helix.
Figure 5(b) shows the backfire quadrifilar helix. It consists
of two bifilar antennas fed 90 out of phase to produce the
ej excitation. No ground plane is required and the ends of
the helices may be shorted together. The backfire helix is used
with pitch angles as high as 60 and 70. With these high
pitch angles, the backfire mode can operate at low frequencies, yielding a small cross section for the antenna.
A typical shaped conical beam is shown in Fig. 7. The beam
is very wide for broad sector coverage with a dip or a local
maximum (not shown) at the center ( 180). The dip is
appropriate for satellite coverage, because it can be chosen to
yield uniform signal strength at the receiver as the satellite
passes. Kilgus (19) shows numerous beam patterns for different designs. The beamwidths vary from 100 to 180. Directivities of up to 7 dB are observed.
The backfire quadrifilar helix has characteristics that
make it especially suitable for many satellite, spacecraft, and
navigational applications. It has been used as a transmitter
and receiver in satellite communication systems and as a receiver for GPS applications (20). It has also been considered
for cellular phones and new GPS applications (21).
= 0
The Backfire Quadrifilar Helix
In the analysis of circular loops we noted that, with ej , circularly polarized radiation occurs at both 0, 180. To distinguish between these two directions, we need additional information about the helix. A rigorous analysis of the infinite
monofilar helix by Paul Klock (14) shows that there are two
modes operating simultaneously in the axial mode region.
Both involve ej excitation for right-hand helices and are circularly polarized on axis. One is the axial mode and the other
is a backfire mode which starts at backfire and scans forward
as a conical beam as frequency increases. We may, for purposes of discussion, combine the backfire and forwardscanned operations into a single backfire designation. In the
669
90
180
Figure 7. Beam pattern of the backfire quadrifilar helix.
670
HELICOPTER NIGHT PILOTAGE
SUMMARY
The helical antenna, first discovered in 1946 by John Kraus,
has evolved into many different forms with many different
applications. The normal mode helix has some advantages for
low-frequency applications. The broadband, circularly polarized, axial mode helix radiates forward endfire along its axis.
It has been the most widely used of all forms of the helix. The
quadrifilar axial mode helix extends the bandwidth further.
The backfire quadrifilar helix radiates a broad sector coverage
suitable for satellite applications.
BIBLIOGRAPHY
1. J. D. Kraus, Antennas, 2nd ed., New York: McGraw-Hill, 1988,
pp. 265339.
2. W. L. Stutzman and G. A. Thiele, Antenna Theory and Design,
2nd ed., New York: Wiley, 1998, pp. 231239.
3. W. W. Hansen and J. R. Woodyard, A new principle in directional
antenna design, IRE Proc., 26: 333345, 1938.
4. C. A. Balanis, Antenna Theory: Analysis and Design, 2nd ed., New
York: Wiley, 1997, pp. 271276.
5. H. Nakano, Helical and Spiral Antennas: A Numerical Approach,
New York: John Wiley and Sons, 1987, pp. 123195.
6. R. S. Elliott, Antenna Theory and Design, Englewood Cliffs, NJ:
Prentice-Hall, 1981, pp. 7178.
7. H. E. King and J. L. Wong, Characteristics of 1 to 8 wavelength
uniform helical antennas, IEEE Trans. Antennas Propagat.,
AP28: 291296, 1980.
8. E. A. Wolf, Antenna Analysis, New York: Wiley, 1967, pp.
437444.
9. R. C. Johnson and H. Jasik, Antenna Engineering Handbook, 2nd
ed. New York: McGraw-Hill, 1984, pp. 13-113-23.
10. J. L. Wong and H. E. King, Broadband quasi-taper helical antennas. IEEE Trans. Antennas Propagat., AP-27: 7278, 1979.
11. L. O. Krause, Sidefire helix UHF-TV transmitting antenna, Electronics, 24: 107109, Aug. 1951.
12. J. D. Kraus, Radio Astronomy, 2nd ed., Powell, OH: CygnusQuasar, 1986.
13. S. Sensiper, Electromagnetic wave propagation on helical structures, Proc. IRE, 43: 149161, 1955; also Ph.D. Thesis, M.I.T.,
1951.
14. P. W. Klock, A Study of Wave Propagation of Helices, Ph.D. Thesis,
University of Illinois, Urbana-Champaign, 1963.
15. C. Gerst and R. A. Worden, Helix antennas take turn for better,
Electronics, 39: 100110, Aug. 1966.
16. A. T. Adams et al., The quadrifilar helix antenna, IEEE Trans.
Antennas Propagat., AP-22: 173178, 1974.
17. A. T. Adams and C. Lumjiak, Optimization of the quadrifilar helix antenna, IEEE Trans. Antennas Propagat., AP-19: 547548,
1971.
18. W. T. Patton, The Backfire Helical Antenna, Ph.D. Thesis, University of Illinois, Urbana-Champaign, 1963.
19. C. C. Kilgus, Shaped-conical radiation pattern performance of the
backfire quadrifilar helix, IEEE Trans. Antennas Propagat., AP23: 392397, 1975.
20. J. M. Tranquilla and S. R. Best, A study of the quadrifilar helix
antenna for global positioning system (GPS) applications, IEEE
Trans. Antennas Propagat., AP-38: 15451550, 1990.
21. N. Padros et al., Comparative study of high-performance GPS receiving antenna designs. IEEE Trans. Antennas Propagat., AP45: 698706, 1997.
ARLON T. ADAMS
JAY K. LEE
Syracuse University
J. Webster (ed.), Wiley Encyclopedia of Electrical and Electronics Engineering
c 1999 John Wiley & Sons, Inc.
Copyright
HORN ANTENNAS
A horn antenna is a length of conducting tube that is flared at one end and used for the transmission and
reception of electromagnetic waves. For an efficient transition between guided and radiated waves the horn
dimensions must be comparable to the wavelength. Consequently, horns are used mostly at centimeter and
millimeter wavelengths. At lower or higher frequencies they are inconveniently large or small, respectively.
They are most popular at microwave frequencies (3 GHz to 30 GHz) and higher frequencies (30 GHz to 300
GHz), as antennas of moderate directivity or as feeds for reflectors or elements of arrays.
Since acoustic horns have been in use since prehistoric times, the design of horns as musical instruments
was a highly developed art well before the appearance of the first electromagnetic horns. This occurred shortly
after Hertz first demonstrated the existence of electromagnetic waves in 1888. Experimenters placed their spark
gap sources in hollow copper tubes [Figs. 1(a) and 5(a)]. These tubes acted as high-pass filters for microwave and
millimeter-wave radiation from the open end. In London in 1897, Chunder Bose used rectangular conducting
tubes with collecting funnels, or pyramidal horns [Fig. 1(d)], in his demonstrations at 5 mm and 25 mm
wavelengths [see Ramsay (1)]. Thus the electromagnetic horn antenna was introduced, but this early beginning
of microwave invention closed with Marconis demonstration that longer wavelengths could be received at
greater distances. Horns were too large to be practical at those wavelengths, and it was almost 40 years before
microwave horns reappeared with the need for directive antennas for communications and radar. Horns alone
were often not sufficiently directive; but combined in an array or with a lens [Fig. 4(a)], or more often a parabolic
reflector [Figs. 4(b) and 4(c)], highly directive antenna beams are obtained.
Radiating Waveguides and Horns
Horns are normally fed by waveguides supporting only the dominant waveguide mode. For a rectangular
waveguide [Fig. 1(a)] with TE01 mode propagation only, these dimensions in wavelengths are /2 < a < and
b a/2. Open-ended waveguides have broad radiation patterns, so when they are used as a feed for a reflector,
there is substantial spillover, or radiation missing the reflector and radiation directly backwards from the
feed. To increase the directivity of a radiating waveguide and its efficiency as a reflector feed, for example, its
aperture dimensions must be enlarged, because the beamwidth of an aperture of width a is proportional
to /a radians.
This waveguide enlargement by a flare characterizes horns. The aperture fields of a horn are spherical
waves originating at the horn apex (Fig. 2). The path from the horn apex to the aperture plane at a distance x
from the aperture center of a horn of slant length is
HORN ANTENNAS
Fig. 1. (a) Open-ended rectangular waveguide. (b) E-plane sectoral horn. (c) H-plane sectoral horn. (d) Pyramidal horn.
when x cos . Thus the phase variation in radians across the aperture for small flare angles is approximately kx2 /(2), where k = 2/ is the propagation constant. This quadratic phase variation increases
with increasing flare angle, thus reducing directivity increase due to the enlarged aperture dimension. It is
HORN ANTENNAS
Fig. 2. Effect of horn flare on the aperture field phase of a horn.
convenient to quantify aperture phase variation by the parameter
which is the approximate difference in wavelengths between the distance from the apex to the edge (x = d/2)
and the center (x = 0) of the aperture. The radiation patterns of Figs. 3(a) and 3(b) [from Love (2)] show the
effect of increasing s on the E- and H-plane radiation patterns of sectoral and pyramidal horns. The main beam
is broadened, the pattern nulls are filled, and the sidelobe levels raised over those for an in-phase aperture
field (s = 0). With large flare angles, radiation from the extremities of the aperture can be so out of phase with
that from the center that the horn directivity decreases with increasing aperture width.
The adverse effects of the flare can be compensated by a lens in the aperture [Fig. 4(a)]; but because that
adds to the weight and cost and because bandwidth limitations are introduced by matching the lens surfaces
to reduce reflections, it is not usually done except in some millimeter wave applications. Instead a combination
of aperture width and flare length in wavelengths is chosen which provides maximum axial directivity or
minimum beamwidth. This is an optimum horn design. To achieve higher directivity or narrower beamwidth
for a given aperture width, a longer horn is required.
Sectoral horns [Figs. 1(b) and (c)] are rectangular waveguides flared in one dimension only. The incident
waveguide mode becomes a radial cylindrical mode in the flared region of the horn. Since radiation pattern
beamwidths are inversely proportional to aperture dimensions in wavelengths, sectoral horns have beams
which are narrow in the plane containing the broad dimension. Such fan-shaped beams may be useful for
illuminating elongated parabolic reflectors or parabolic cylinder reflectors.
A pyramidal horn [Fig. 1(d)] is flared in both waveguide dimensions and thus is more adaptable both
as a reflector feed and on its own. The forward radiation pattern may be calculated quite accurately from
Kirchhoff diffraction theory for all but small horns. The TE01 rectangular waveguide mode yields an aperture
field uniform in one dimension (in the E plane) and cosinusoidal in the other (the H plane). A comparison of
Figs. 3(a) and 3(b) shows that this results in higher sidelobes in the E plane and, for a square aperture, a
narrower beam. Pyramidal horns are relatively easily constructed, and for all but small horns their axial gain
can be predicted accurately. Consequently they are used as gain standards at microwave frequencies; that is,
HORN ANTENNAS
Fig. 3. Universal radiation patterns of sectoral and pyramidal horns flared in the (a) E plane and (b) H plane. The
parameter s equals b2 /8E in (a) and a2 /8H in (b), and 2s/ is the maximum phase difference between the fields at the
c 1984 McGraw-Hill, Inc., from Love (2)].
center and the edge of the aperture [
they are used to experimentally establish the gain of other microwave antennas by comparing their response
with the same illuminating field.
Most of the preceding remarks on open-ended rectangular waveguides and pyramidal horns apply also
to open-ended circular waveguides and conical horns [Figs. 5(a) and (b)]. For propagation of the lowest-order
mode (TE11 ), only in a circular waveguide the interior diameter must be 0.59 < a < 0.77. This mode has a
uniform aperture field in the E plane and an approximately cosinusoidal distribution in the orthogonal H
plane. This appears, modified by a quadratic phase variation introduced by the flare, in the aperture field of the
horn. Consequently the E-plane radiation pattern of the horn is narrower, but with higher sidelobes than the
H-plane pattern and the radiated beam is elliptical in cross section. In addition, cross-polarized fields appear
in pattern lobes outside the principal planes.
Horn Feeds for Reectors
Many refinements to horns arise from their use as efficient feeds for parabolic reflectors, particularly in satellite
and space communications and radio astronomy. The phase center, where a horns far radiation field appears
HORN ANTENNAS
Fig. 4. (a) Horn aperture field phase correction by a lens. (b) Parabolic reflector fed by a horn. (c) Horn reflector antenna.
f is the focal length of the lens or reflector.
to originate, must be placed at the focus of the reflector [Fig. 4(b)]. This phase center is within the horn on the
horn axis and depends on the flare angle and aperture distribution. For both rectangular and conical horns
the position of the phase center is not the same in the E and H planes, or planes containing the electric and
magnetic field vectors, respectively. A phase center can be calculated from the average of the positions of the
HORN ANTENNAS
Fig. 5. (a) Open-ended circular waveguide. (b) Conical horn. (c) Corrugated horn. (d) Circular waveguide with corrugated
flange.
E- and H-plane phase centers, or it can be determined from the position of the feed which maximizes the gain
of the reflector antenna.
For efficient reflector aperture illumination the feed horn radiation pattern should approximately match
the shape of the reflectors aperture, and illuminate it essentially uniformly and with minimal spillover, or
radiation missing the reflector. Pyramidal horns may seem suitable for rectangular apertures because their
HORN ANTENNAS
beams are rectangular in cross section, and conical horns may seem a natural choice for a circular aperture;
but efficient aperture illumination is not obtained in either case, because their principal plane patterns differ.
Both horns have high E-plane pattern sidelobes and low H-plane sidelobes. A dual (TE11 /TM11 ) mode conical
horn provides equal E- and H-plane beamwidths and equally low sidelobes and is an efficient feed for a
circular aperture over a narrow frequency band [Love (3), p. 195]. (Reference 3 contains reprints of most earlier
significant articles on horn antennas.) A broadband solution achieves an axisymmetric beam with annular
corrugations on the interior surfaces of a conical horn [Fig. 5(c)] [Love (3), pp. 248, 277]. These produce a
horn aperture field distribution which is approximately cosinusoidal across the conical horn aperture in all
directions and hence an axisymmetric radiation pattern with low sidelobes. Such corrugations in the E-plane
interior walls only of a pyramidal horn will produce a nearly cosinusoidal E-plane aperture distribution, and
consequently similar E-plane and H-plane radiation patterns for a square horn aperture.
A feed for a small circular aperture reflector which is more easily constructed than a corrugated conical
horn, but with a less axisymmetric radiation pattern, is an open-ended circular waveguide ringed by a recessed
disc of approximately quarter-wavelength-deep corrugations [Fig. 5(d)]. These corrugations suppress back radiation from the feed and thus improve the aperture illumination over that of a simple open circular waveguide
[Love (3), pp. 181, 226]. Combined with dual mode excitation, this arrangement provides an efficient field for
a paraboloidal reflector.
Radiation from Apertures
The far-field radiation pattern of an aperture can be calculated exactly from the Fourier transform of the
tangential fields in the entire aperture plane. Either electric or magnetic aperture fields may be used, but for
apertures in space a combination of the two gives the best results from the usual assumption that aperture plane
fields are confined to the aperture and negligible outside it. This aperture field is assumed to be the undisturbed
incident field from the waveguide. For apertures with dimensions larger than several wavelengths, a further
simplifying assumption usually made is that the aperture electric and magnetic fields are related as in free
space.
Rectangular Apertures. With the above assumptions, at a distance much larger than the aperture
dimensions, the radiated electric field intensity of a linearly polarized aperture field Ex (x, y, 0) in the coordinates
of Fig. 6(a) is
Here
and
HORN ANTENNAS
Fig. 6. Coordinates for radiation from (a) rectangular and (b) circular apertures.
is a vector defining the angular behavior of the radiation polarization for an aperture in space. For an aperture
in a conducting plane, it is more accurate to use
which, since it is based on the aperture plane electric fields only, fully satisfies the assumption of a vanishing tangential field in the aperture plane outside the aperture. Consequently, radiation fields of open-ended
waveguides and small horns can be calculated accurately from Eq. (3) with Eq. (6) if they are mounted in a
conducting plane. Clearly, Eqs. (5) and (6) differ significantly only for large angles off the beam axis.
If the aperture field is separable in the aperture coordinatesthat is, in Eq. (3), Ex (x, y, 0) = E0 E1 (x)E2 (y),
where E1 (x) and E2 (y) are field distributions normalized to E0 then the double integral is the product of two
single integrals:
where
define the radiation field.
HORN ANTENNAS
Open-Ended Waveguides
Rectangular Waveguides. With the TE01 waveguide mode the aperture field
in Eq. (7) yields the following for Eqs. (8) and (9):
This defines the radiation pattern in the forward hemisphere /2 < < /2, 0 < < 2. If the aperture is in
space, then Eq. (5) is used for A(r, , ), but this is not an accurate solution since the aperture dimensions are
not large. Rectangular waveguides mounted in conducting planes use Eq. (6) for A(r, , ) in Eq. (7), which then
accurately provides the far field. The pattern has a single broad lobe with no sidelobes. For large apertures,
plots of the normalized E-plane ( = 0) and H-plane ( = /2) patterns of Eq. (7) appear in Figs. 3(a) and 3(b)
for those of a horn with no flare (s = 0), but without the factor (1 + cos )/2 from Eq. (5) or cos from Eq. (6).
Circular Waveguides. The dominant TE11 mode field in circular waveguide produces an aperture field
distribution which in the aperture coordinates , of Fig. 6(b) is
where J 1 is the Bessel function of the first kind and order, J 1 is its derivative with respect to its argument kc ,
and kc a/2 = 1.841 is the first root of J 1 . E0 is the electric field at the aperture center ( = 0). Since Eq. (13) is
not linearly polarized, its use in Eq. (3) provides only part of the total radiated far field. This total field is
in which k = k sin .
In the E and H planes ( = 0 and /2) the cross-polarized fields cancel and the patterns shown in Fig. 14(a)
are similar to those of Eqs. (11) and (12), respectively, but with slightly broader beams and lower sidelobes for
the same aperture dimensions. As with rectangular waveguides, open-ended circular waveguide apertures are
10
HORN ANTENNAS
insufficiently large for Eq. (14) to represent all the radiated fields accurately. In the principal planes ( = 0,
/2) it can give a reasonable approximation for the copolarized fields but fails to accurately represent the crosspolarized field patterns in = /4. This is evident from a comparison of numerical results from approximate
and exact solutions [Collin (4), p. 233].
Pyramidal and Sectoral Horns
Radiation Patterns. A pyramidal horn fed by a rectangular waveguide supporting the TE01 mode has
an incident electric field in the aperture of Fig. 6(a) which is approximately the mode distribution modified by
a quadratic phase variation in the two aperture dimensions; that is,
With Eq. (15), Eq. (3) becomes
where Eq. (5) is used for A(r, , ) and
The E-plane ( = 0) and H-plane ( = /2) radiation patterns are, respectively,
These integrals can be reduced to the Fresnel integrals
HORN ANTENNAS
11
which are tabulated and for which computer subroutines are available. For example,
with
Figure 3(a) shows plots of the magnitude of Eq. (22) for various values of the E-plane flare parameter s = b2 /8E ,
while Fig. 3(b) shows corresponding plots of |I2 (k sin )/I2 (0)| for the H-plane flare parameter s = a2 /8H . For
no flare (s = 0) the patterns are those of a large open-ended rectangular waveguide supporting only the TE01
mode. The effect of the flare is to broaden the main beam, raise the sidelobes, and fill pattern nulls. For larger
values of s there is enhanced pattern beam broadening and eventually a splitting of the main beam on its axis.
These curves also represent the radiation patterns of the E- and H-plane sectoral horns of Figs. 1(b)
and 1(c). For an E-plane sectoral horn (H ) the E-plane pattern is given by Eq. (19) and the H-plane
pattern is given approximately by Eq. (12). For an H-plane sectoral horn (E ) the E-plane pattern is given
approximately by Eq. (11) and the H-plane pattern is given by Eq. (20).
In comparing Figs. 3(a) and 3(b) it is evident that E-plane beamwidths of a square aperture are narrower
than H-plane beamwidths. For horns of moderate flare angle and optimum horns the E-plane half-power
beamwidth is 0.89 /b radians and the H-plane half-power beamwidth 1.22 /a radians. E-plane patterns have
minimum sidelobes of 13.3 dB below peak power while H-plane pattern minimum sidelobes levels are 23.1
dB.
The universal patterns of Figs. 3(a) and (b) can also be used to predict the approximate near-field radiation
patterns of horns by including the quadratic phase error which is a first-order effect of finite range r. This is
done by including
in Eq. (15). Then the near-field principal plane patterns of a pyramidal horn are given by Eqs. (17) and (18)
with E , H replaced by
and
12
HORN ANTENNAS
These near-field effects are analogous to decreasing the flare length of a horn with a fixed aperture width. The
main beam broadens, nulls are filled in, and sidelobes rise.
Limitations and Extensions. Results from Eq. (16) do not apply to small horns and are limited to the
forward direction ( < 90 ). They are most accurate on and around the beam axis ( = 0), becoming progressively
less accurate as increases. The simplest method for extending the analysis is by the uniform geometrical theory
of diffraction [e.g., Love (3), p. 66], which provides the edge-diffracted fields in the lateral and rear directions
which receive no direct illumination from the aperture. Only the edges normal to the plane of the pattern
contribute significantly to the E-plane pattern but the rear H-plane pattern requires contributions from all
four aperture edges and so is difficult to calculate this way.
While the geometry of the pyramidal horn defies rigorous analysis, numerical methods have been used
with some success for open waveguides and small horns. For larger horns, this approach becomes computationally intensive, but some results from Liu et al. (5) are shown in Fig. 7 and compared with measurements and
approximate computations. Their numerical computations, along with measurements of the aperture fields by
Nye and Liang (6) show that higher-order modes need to be added to the dominant mode field of Eq. (15) and
that the parabolic phase approximation of Eq. (1) improves as the aperture size increases.
Gain. Pyramidal horns are used as gain standards at microwave frequencies because they can be
accurately constructed and their axial directive gain reliably predicted from a relatively simple formula. The
ratio of axial far-field power density to the average radiated power density from Eq. (16) yields
where G0 = 32 ab/(2 ) is the gain of an in-phase uniform and cosinusoidal aperture distribution. The reduction
of this gain due to the phase variation introduced by the E-plane flare of the horn is
where the Fresnel integrals and their argument are defined by Eqs. (21) and (23). Similarly the gain reduction
factor due to the H-plane flare of the horn is
where
A plot of RE and RH in decibels as a function of the parameter 2d2 /, where d is the appropriate aperture
dimension b or a and is the slant length E or H , respectively, is shown in Fig. 8. Calculation of the gain
from Eq. (28) is accurate to about 0.1 dB for 22 dB standard gain pyramidal hornsthat is, optimum horns
with dimensions of at least 5. For 18 dB gain horns the accuracy is about 0.2 dB and for 15 dB horns
0.5 dB. Since optimum gain pyramidal horns have an aperture efficiency of approximately 50%, the gain is
approximately
HORN ANTENNAS
13
Fig. 7. Calculated and measured (a) E-plane and (b) H-plane radiation patterns of a pyramidal horn of dimensions
c 1993 IEEE, from Liu et al. (6)].
a = 4.12, b = 3.06, E = 10.52, and H = 9.70 [
For an E-plane sectoral horn H and RH (v, w) 1 the axial gain is then GE = G0 RE (u), an inaccurate
formula because aperture dimension a is less than a wavelength. A result which includes the fact that aperture
electric and magnetic fields are not related by free-space conditions and that interaction occurs across the
narrow aperture of the horn is
14
HORN ANTENNAS
Fig. 8. E- and H-plane flare and near field gain reduction factors RE and RH of pyramidal and sectoral horns in decibels
c 1981 IEE, from Jull (11)].
[
where
and
is the guide wavelength. The accuracy of Eq. (33) is comparable to that of Eq. (28) for the horns of similar b
dimension.
The gain of an H-plane sectoral horn, obtained by letting E so that RE (u) 1, is GH = G0 RH (v, w).
It probably is reasonably accurate, but there appears to be no experimental evidence available to verify it.
The near-field gain of pyramidal and sectoral horns can be calculated from the above expressions by
replacing E and H by Eqs. (26) and (27), respectively.
Conical Horns
The aperture field of a conical horn fed by a circular waveguide supporting the TE11 mode is approximately
HORN ANTENNAS
15
Fig. 9. Copolar and cross-polar radiation patterns for a conical horn with dimensions a = 4 and = 23. E plane,
c 1994 IEE, from Olver et al. (7)].
H plane, cross-polarization [
where E( , ) is given by Eq. (13) and is the slant length of the horn. Numerical calculation of the radiation
patterns is necessary. In the example of Fig. 9 [Olver et al. (7)] with a flare angle = 5 and aperture width
a = 4, the E-plane ( = 0) pattern is narrower than the H-plane ( = /2) pattern as in square rectangular
horns. The cross-polar ( = /4) radiation pattern peak level is 18.7 dB relative to the copolar pattern peak
levels, a level typical of conical horn apertures larger than about 2. Smaller conical horns can have more
axisymmetric patterns. E- and H-plane patterns have equal beamwidths for an aperture diameter a = 0.96,
and cross-polarized fields cancel for a = 1.15. This makes small conical horns efficient as reflector feeds and
as array elements with high polarization purity.
Larger conical horns are similar to rectangular horns in their lack of axial pattern symmetry. Optimum
gain conical horns have an aperture efficiency of about 54% and half-power beamwidths in the E and H planes
of 1.05/a and 1.22/a radians, respectively, for aperture diameters of more than a few wavelengths.
Multimode and Corrugated Horns
Lack of axisymmetric radiation patterns make rectangular and conical horns inefficient reflector feeds. Conical horns also have high cross-polarization levels, which are undesirable in a reflector feed. Multimode and
corrugated horns were developed largely to overcome these deficiencies. In a dual-mode horn in Love (3, p. 195)
16
HORN ANTENNAS
c 1984 McGraw-Hill,
Fig. 10. Excitation of axisymmetric linearly polarized aperture fields in a stepped conical horn [
Inc. from Love (2)].
this is done by exciting the TM11 mode, which propagates for waveguide diameters a > 1.22, in addition to
the TE11 mode, which propagates for a > 0.59. The electric field configuration of these modes in a waveguide
cross section is shown in Figs. 10(a) and (b). Added in phase and in the right proportion, cross-polarized and
aperture perimeter fields cancel, while the copolar fields around the aperture centre add, yielding the aperture
HORN ANTENNAS
17
field configuration of Fig. 10(c). These mixed mode fields are linearly polarized and tapered approximately
cosinusoidally radially across the aperture. This yields the essentially linearly polarized and axisymmetric
radiation patterns desired.
Partial conversion of TE11 to TM11 fields can be effected by a step discontinuity in the circular waveguide
feed, as in Fig. 10(d), or by a circular iris or dielectric ring in the horn. The TM11 /TE11 amplitude ratio depends
on the ratio of waveguide diameters, and the relative phase of the modes depends on the length of largerdiameter circular waveguide and the horn. This limits the bandwidth of the horn to about 5%. A multimode
square-pyramidal horn has similar low sidelobe level E- and H-plane radiation patterns due to an essentially
cosinusoidal aperture distribution in the E-plane as well as the H-plane [Love (2)]. This can be achieved by
excitation of a hybrid TE21 /TM21 mode by an E-plane step discontinuity or by changes in the E-plane flare.
Corrugated horns have aperture fields similar to those of Fig. 10(c) and consequently similar radiation
patterns, but without the frequency bandwidth limitations of the above horn. This is achieved by introducing
annular corrugations in the interior walls of a conical horn. There must be sufficient corrugations per wavelength (at least three) that the annular electric field E is essentially zero on the interior walls. The corrugations
make H also vanish. This requires corrugation depths such that short circuits at the bottom of the grooves
appear as open circuits at the top, suppressing axial current flow on the interior walls of the horn. This groove
depth is /4 on a plane corrugated surface or a curved surface of large radius. For a curved surface of smaller
radius, such as near the throat of the horn, the slot depths need to be increased; for example, for a surface
radius of 2, the depth required is 0.3. Usually slots are normal to the conical surface in wide flare horns but
are often perpendicular to the horn axis with small flares. To provide a gradual transition from the TE11 mode
in the waveguide to a hybrid HE11 mode in the aperture, the depth of the first corrugation in the throat should
be about 0.5 so that the surface there resembles that of a conducting cone interior. Propagation in corrugated
conical horns can be accurately calculated numerically by mode-matching techniques. The aperture field is
approximately
where kc a/2 is 2.405, the first zero of the zero-order Bessel function J 0 , is the slant length of the horn, and A is
a constant. This aperture field is similar to that of Fig. 10(c), and the resulting E and H patterns are similarly
equal down to about 25 dB. Some universal patterns are shown in Fig. 11. Cross-polarization fields are also
about 30 dB from the axial values, but are now over a bandwidth of 2:1 or more.
Broadband axisymmetric patterns with low cross-polarization make corrugated horns particularly attractive as feeds for reflectors. Low cross-polarization allows the use of dual polarization to double the capacity
of the system. Another notable feature for this application is that the position of the E- and H-plane pattern
phase centers coincide. Figure 12 shows the distance of the phase center from the horn apex, divided by the
slant length, of small-flare-angle conical [Milligan (8)] and corrugated [Thomas (9)]] horns for values of the
phase parameter s given by Eq. (2). For a conical horn the E-plane phase center is significantly further from
the aperture than the H-plane phase center. Thus if a conical horn is used to feed a parabolic reflector, the best
location for the feed is approximately midway between the E- and H-plane phase centers. With a corrugated
horn, such a compromise is not required so it is inherently more efficient.
Corrugated horns may have wide flare angles, and their aperture size for optimum gain decreases correspondingly. For example, with a semiflare angle of 20 the optimum aperture diameter is about 8, whereas
for a semiflare angle of 70 it is 2. Wide-flare corrugated horns are sometimes called scalar horns because of
their low cross-polarization levels.
18
HORN ANTENNAS
c 1984
Fig. 11. Universal patterns of small-flare-angle corrugated horns as a function of the parameter s = a2 /8 [
McGraw-Hill, Inc. from Love (2)].
Prole Horns
Most corrugated horns are conical with a constant flare angle. Figure 13 shows a profile conical horn in which
the flare angle varies along its length. This arrangement provides a horn shorter than a conical corrugated
horn of similar beamwidth, with a better impedance match due to the curved profile at the throat and an
essentially in-phase aperture field distribution due to the profile at the aperture. Consequently the aperture
efficiency is higher than that of conical corrugated horns. The phase center of the horn is near the aperture
center and remains nearly fixed over a wide frequency band. Radiation patterns of a profile horn similar to
that of Fig. 13 are shown in Fig. 14 [Gonzalo et al. (10)]. The patterns are similar to those of a Gaussian beam,
such as is radiated from the end of an optical fiber supporting the HE11 mode. The performance of this small
horn as a feed seems close to ideal, but larger profile horns may exhibit higher sidelobe levels due to excitation
of the HE12 mode at the aperture.
Horn Impedance
Antennas must be well matched to their transmission lines to ensure a low level of reflection in microwave
communication systems. The impedance behavior of a horn depends on the mismatch at the waveguide/horn
junction and at its aperture. For an E-plane sectoral horn, reflections from these discontinuities are comparable
in magnitude; and since they interfere, the total reflection coefficient oscillates with frequency and the input
voltage standing wave ratio (VSWR) may vary from 1.05 at high frequencies to 1.5 at the lowest frequency. With
HORN ANTENNAS
19
Fig. 12. Normalized distance of the phase center from the apex of conical ( E plane, H plane) and corrugated ()
horns [Data from Milligan (8) and Thomas (9)].
c 1994, IEE, from Olver et al. (7)].
Fig. 13. A profile corrugated horn [
H-plane sectoral horns, aperture reflection is much stronger than junction reflection so their VSWR increases
almost monotonically with decreasing frequency. An inductive iris in the waveguide near the E-plane horn
junction can match its discontinuity. A capacitive iris may be similarly used for an H-plane sectoral horn.
Aperture reflections in these horns may be matched with dielectric covers.
Pyramidal horns of sufficient size and optimum design tend to be inherently well-matched to their waveguide feeds because the E- and H-plane aperture and flare discontinuities partially cancel. For example, a 22
dB gain horn has a VSWR of about 1.04 while an 18 dB horn has a VSWR of less than 1.1.
Conical horns fed by circular waveguides supporting the dominant TE11 mode have an impedance behavior
similar to that of pyramidal horns of comparable size fed by rectangular waveguides. The waveguide/horn
discontinuities of both horns may be matched by an iris placed in the waveguide near the junction. A broader
20
HORN ANTENNAS
Fig. 14. (a) Far-field radiation patterns of TE11 mode and (b) radiation patterns of a profile corrugated horn of aperture
c 1997, IEEE, from
a = 15.8 mm and length L = 26.7 mm at 30 GHz. E plane,H plane cross-polarization [
Gonzalo et al. (10)].
bandwidth match is provided by a curved transition between the interior walls of the waveguide and the horn.
Broadband reduction of aperture reflection may be similarly reduced by a curved surface of a few wavelengths
radius. Such aperture-matched horns also have lower sidelobe levels and less back radiation in their E-plane
patterns than do conventional pyramidal and conical horns. Their H-plane flare patterns are affected little by
such aperture matching because the electric field vanishes at the relevant edges.
For dual-mode and corrugated horns there are also negligible fields at the aperture edges and hence little
diffraction there. Corrugated horns with initial groove depths near the throat of about a half-wavelength and
which gradually decrease to a quarter-wavelength near the aperture, as in Fig. 13, are well-matched at both
throat and aperture. For most well-designed corrugated horns a VSWR of less than 1.25 is possible over a
HORN ANTENNAS
21
frequency range of about 1.5:1. Dual-mode horns using a step discontinuity as in Fig. 10(d) may have a VSWR
of 1.2 to 1.4. If an iris is required for a match, the frequency bandwidth will of course be limited. Conical and
pyramidal horns using flare-angle changes to generate the higher-order modes can have VSWRs less than 1.03
and require no matching devices.
BIBLIOGRAPHY
1. J. F. Ramsay, Microwave antenna and waveguide techniques before 1900, Proc. IRE, 46: 405415, 1958.
2. A. W. Love, Horn antennas, in R. C. Johnson and H. Jasik (eds.), Antenna Engineering Handbook, 2nd ed., New York:
McGraw-Hill, 1984, chap. 15.
3. A. W. Love (ed.), Electromagnetic Horn Antennas, Piscataway, NJ: IEEE Press, 1976.
4. R. E. Collin, Antennas and Radiowave Propagation, New York: McGraw-Hill, 1985.
5. K. Liu et al., Analysis of pyramidal horn antennas using moment methods, IEEE Trans. Antennas Propag., 41: 1379
1389, 1993.
6. J. F. Nye, W. Liang, Theory and measurement of the field of a pyramidal horn, IEEE Trans. Antennas Propag., 44:
14881498, 1996.
7. A. D. Olver et al., Microwave Horns and Feeds, vol. 39, London: IEE Electromagnetic Waves Series, 1994.
8. T. Milligan, Modern Antenna Design, New York: McGraw-Hill, 1985, chap. 7.
9. B. M. Thomas, Design of corrugated horns, IEEE Trans. Antennas Propag., 26: 367372, 1978.
10. R. Gonzalo, J. Teniente, C. del Rio, Very short and efficient feeder design from monomode waveguide, IEEE Antennas
Propag. Soc. Int. Symp. Dig., Montreal, 1997, pp. 468470.
11. E. V. Jull, Aperture Antennas and Diffraction Theory, vol. 10, London: IEE Electromagnetic Waves Series, 1981.
EDWARD V. JULL
University of British Columbia
J. Webster (ed.), Wiley Encyclopedia of Electrical and Electronics Engineering
c 1999 John Wiley & Sons, Inc.
Copyright
LEAKY-WAVE ANTENNAS
Leaky-wave antennas (LWAs) constitute a type of radiators whose behavior can be described by an electromagnetic wave (leaky wave) that propagates in guiding structures that do not completely confine the field, thus
allowing a continuous loss of power to the external environment (leakage).
According to IEEE Standard 145-1983, a leaky-wave antenna is an antenna that couples power in small
increments per unit length either continuously or discretely, from a traveling wave structure to free space.
Distinctive Features And Applications
LWAs (1,2,3) belong to the class of traveling-wave line antennas, for which the illumination is produced by
a wave that propagates along a guiding structure (4). Compared with the wavelength, a LWA is long in the
propagation direction z, while its cross section is usually of the order of the wavelength [see the reference
examples of Fig. 1(a) and 1(b)].
LWAs radiate along their lengths, and in general their relevant behavior can be derived by considering
a linear aperture in which the illumination of the current distribution, according to the equivalence principle,
is directly linked to the tangential electromagnetic field. Usually LWAs are excited from one input of the
open guiding structure with a traveling wave, which propagates mainly in one longitudinal direction (e.g., +z)
and is attenuated due to the power leakage towards the exterior region, thus leaving a negligible field at the
termination of the guide. In a harmonic regime [with an exp(jt) time dependence], this wave is characterized
by a complex propagation constant of the type kz = z jz (5,6,7,8,9,10), where z is the phase constant and
z is the attenuation constant of the leaky wave (when only power loss due to radiation is taken into account,
z is also called the leakage constant).
Usually the radiation pattern of a single LWA has a typical fan shape: in the elevation (or zenith) plane a
narrow beam is achievable with a pointing direction that varies with frequency, whilst in the cross (or azimuth)
plane the beam is usually wider as a result of the smaller transverse aperture. Depending on the desired
application, a suitable longitudinal variation of the aperture distribution, usually reached by modulating
geometrical parameters (tapering), allows good control of the radiation pattern (sidelobe behavior, etc.). In
some cases, in order to obtain beam shaping or physical matching with curved surfaces of the mounting, LWAs
can be designed with certain amounts of curvature along their lengths.
The scanning properties in the elevation plane (pointing angle variable with frequency) are related to
the type of waveguide employed, which can be of either uniform [Fig. 1(a)] or periodic [Fig. 1(b)] type (1,2,3,4).
LWAs derived from waveguides that are longitudinally uniform (i.e., the structure maintains continuously the
same transverse geometry) typically allow angular scanning over one quadrant, from around broadside towards
one endfire (the forward one, which is concordant with the wave propagation direction). LWAs derived from
waveguides that are longitudinally periodic (i.e., the structure is periodically loaded with proper discontinuities,
at intervals that are usually short with respect to the wavelength) allow wider angular scanning, both in the
1
LEAKY-WAVE ANTENNAS
Fig. 1. Basic structures of LWAs: (a) uniform LWAs: geometry representing a partially open metallic waveguide; (b)
periodic LWAs: geometry representing a strip-loaded dielectric-rod waveguide; (c) topology of LWA arrays.
forward and in the backward quadrant. However, for various reasons, there are limitations on such scanning
ranges for both types of structures (as shown in a subsequent section).
When a pencil beam is to be used for two-dimensional (2-D) scanning, both in elevation and cross planes
(zenith and azimuth), a phased array of juxtaposed LWAs may be employed, thus enlarging the equivalent
aperture transversely (11,12) [see the reference example of Fig. 1(c)]. LWA arrays are therefore constituted
by a linear configuration of sources (i.e., 1-D elements), instead of the planar ones of standard arrays (i.e.,
2-D elements). For LWA arrays pointed-beam scanning is achievable by varying both the frequency for the
elevation plane and the phase shift for the cross plane.
Since LWAs are derived from partially open waveguides, they present a number of distinctive features
as radiators: large power-handling capability, particularly for structures utilizing closed metallic waveguides;
reduction of bulk, due to the usually small profiles in the cross sections; capability of designing a wide variety
of aperture distributions and consequent flexibility for the beam shaping; possible use as wide-band radiators,
allowing large angular scanning by varying the frequency (instead of using mechanical or other electronic
means); achievement of very narrow beams with good polarization purity; simplicity of feeding and economy
for 2-D scannable pencil-beam arrays (multiple linear phase shifters instead of one planar one); etc.
LWAs are mainly used in the microwave and millimeter-wave region, though investigations have been
performed also for both lower- and higher-frequency applications. The first studies on LWAs were presented
during the forties, basically for aerospace applications (radar, etc.); since then, a very wide range of solutions
LEAKY-WAVE ANTENNAS
for LWAs has been proposed in response to changing requirements and constraints. The applicability of this
type of antennas has also been widened, including various problems of traffic control, remote sensing, wireless
communications, and so forth (13,14).
Operation Principles
Existence and excitation of leaky waves in open guides. A leaky wave has a complex longitudinal
wave-number kz that can be derived by solving, as a function of the physical parameters (frequency and
geometry of an open waveguiding structure), the characteristic equation (or dispersion relation), which is of
the general type
where k0 = (0 0 ) 2 is the vacuum wave-number. [The eigensolutions of Eq. (1) can alternatively be viewed as
the pole singularities of the spectral Greens function of the structure, since the expression D(kz ,k0 ) is generally
present in its denominator (5,6,7,8,9,10).]
As is well known, for lossless closed waveguides the dispersion relation in Eq. (1) generally presents an
infinite discrete set of eigensolutions giving the guided modes (proper modes) which individually satisfy all
the relevant boundary conditions. Any field excited by a source in a closed guide can be expanded in terms
of the complete set of the infinite discrete eigensolutions of Eq. (1). In conventional guides, the longitudinal
1
wave numbers kz are either real [propagating waves above their cutoff, with kz = z < k = k0 (r ) 2 ; for air-filled
structures, one has z < k0 ] or imaginary (attenuating waves below their cutoff, with kz = jz ). (We recall that
in reciprocal structures such eigensolutions exist in pairs: kz = kz , describing waves traveling in opposite z
directions.)
In lossless open waveguides (e.g., dielectric guides), instead, only a finite number of propagating modes can
exist as eigensolutions of Eq. (1) satisfying all the boundary conditions (particularly, the radiation condition):
these are the so-called bound surface waves (each one exists only above its cutoff, with kz 2 > k0 2 , or kz = z >
k0 ). In addition to this, for a complete representation of the field, which is no longer confined in a closed section,
a continuous spectrum of modes must be introduced: it describes the radiated field as an integral over a set of
plane waves having a continuous range of wave numbers (e.g., such that < kz 2 < k0 2 , or 0 kz = z k0 and
j < kz = jz < j0) (5,6,7,8,9,10). Any field excited by a source in an open guide can therefore be expanded
in a spectral representation, that is, in terms of a finite set of proper modes and an integral contribution of the
continuous spectrum. (Again, in reciprocal structures the solutions exist in pairs: kz .)
On the other side, it is seen that the characteristic equation in Eq. (1) for open guides presents additional
discrete solutions that are nonspectral in that they correspond to fields that violate the radiation condition (they
attenuate along the propagation direction but increase exponentially in a transverse direction away from the
structure) and are not included in the spectral representation of the field. These roots are also called improper,
since they are located on the improper sheet of the Riemann surfaces for kz (5,6,7,8,9,10).
In many practical circumstances, for describing the radiative effects of the open structures in the presence of a source, the evaluation of the field through the spectral representation (i.e., including the integral
contributions of the continuous spectrum) can be very difficult and cumbersome. It is found that the field can
generally be evaluated in an alternative and rapidly convergent way, through the so-called steepest descent
or saddle point representation (5,6,7,8,9,10): by properly modifying the integration path of the continuous
spectrum involved in the spectral representation, the relevant integral evaluation can be simplified by introducing contributions of improper eigensolutions. The advantage of such a steepest-descent representation is
that, for guiding structures designed to operate as LWAs, the radiation field can be evaluated accurately in
LEAKY-WAVE ANTENNAS
a much simpler fashion by considering just the residual contribution due to the presence of one complex and
generally improper mode, that is, the leaky wave, which can therefore be viewed as a simple recasting of the
continuous spectrum. In fact, it is seen that in practical cases the remaining part of the continuous spectrum
in the steepest-descent representation (the so-called space wave) can be made to give negligible contributions
to the description of the LWAs radiation (5,6,7,8,9,10).
The existence of a leaky mode, which describes a wave propagating with a certain amount of attenuation
through the aperture of a traveling-wave structure, is a necessary condition to ensure its influence on the
illumination function. Furthermore, the leaky wave should be dominant, in the sense that its excitation has to
be strong and therefore its amplitude should be large in comparison with those of other possible guided modes
and of the space wave. [A deep discussion of the physical nature of the complex waves and its influence on the
evaluation of the field in open structures cannot be given within the limits of this article, but many detailed
studies of it can be found in the literature (1,2,3,4,5,6,7,8,9,10).]
From another point of view it can be seen that, when properly excited by a source at a finite section, a leaky
wave, even though improper, assumes its physical validity within an angular sector close to the equivalent
aperture of the open guiding structure, and the corresponding field distribution is able to furnish a fundamental
contribution to the evaluation of the near field (5,6,7,8,9,10). Since the far field is expressed as a simple Fourier
transform of the field at the aperture, a leaky wave can furnish a highly convergent and efficient evaluation of
the radiation of LWAs, as an extremely advantageous alternative to a continuous-spectrum evaluation.
In an open lossless structure the leaky-wave solutions that are of the type kz = z jz describe power
flowing away from the structure. Actually, in usual reciprocal structures these roots of the dispersion equation
in Eq. (1) appear in pairs of the type kz = kz . This means that, in general, both direct and reverse waves
exist, propagating in the +z and z directions (i.e., the kz =z jz and kz = z + jz solutions); as mentioned,
in standard applications the source is chosen to excite just one of the two opposite traveling waves. In addition,
complex conjugate solutions (kz = z + jz and kz = z jz ) also occur for describing leakage phenomena
related to power entering the guiding structure from the external region, because the field increases along the
propagation direction; again, in standard applications the source is placed inside the guiding structure and
leakage occurs only through one wave, which carries out power. It should be also mentioned that, in some cases
(e.g., the nonreciprocal structures, as discussed in a subsequent section), complex waves of the type kz = z
jz and its conjugate kz may exist, which describe effects of power leakage as well and also possess a spectral,
or proper, nature.
Basic characterization of leaky-wave antennas. LWAs present the advantage of a rather simple
characterization of their basic properties, with consequent straightforward approaches for their analysis and
synthesis. As mentioned, their characterization is reduced to the evaluation of a dominant complex eigensolution kz = z jz that can be supported and strongly excited in a specific open structure.
The characteristic dispersion of the real and imaginary parts of the longitudinal wave number of a leaky
wave is presented in Fig. 2 for two different types of structures; specifically, the normalized parameters z /k0 and
z /k0 are plotted against frequency f . The radiation region of LWA structures lies largely inside the frequency
range where the wave becomes fast (z /k0 < 1) and power can therefore leak out from the guiding structure
towards the outside air region in the typical form of a TEM-like mode: in fact, z /k0 < 1 is in general the
so-called condition for leakage of a complex wave that can radiate into an external air region.
The valid frequency range for LWA applications is actually where, as the frequency decreases, z /k0
diminishes monotonically from unity towards rather low values; in this region, to have an efficient directive
beam, z /k0 should assume rather small values (e.g., typically z /k0 can be of the order from about 10 1 to 10 3 ).
As frequency decreases further, a sudden rise of z /k0 is generally found, which signifies the predominance of
reactive phenomena over radiative ones, while z /k0 may present a flat zone with approximately constant low
values before showing a further steep rise as the frequency goes to zero; in these ranges, radiative effects can
no longer be represented by the leaky wave, and the structures usually cannot work well as radiators (2,7,8,12).
LEAKY-WAVE ANTENNAS
Fig. 2. Typical dispersion behavior of the leaky-mode complex wave number (normalized phase constant z /k0 and leakage
constant z /k0 versus frequency f ): (a) for an open planar waveguide (microstrip); (b) for a partially closed metallic waveguide
(slit ridge).
The case of Fig. 2(a) is representative in particular of leakage in unperturbed intrinsically open waveguiding structures (e.g., dielectric and printed lines). In this case, the leaky waves are complex roots of the same
characteristic equation D(kz ,k0 ) = 0 that furnishes the guided (or bound) modes of the structure. Each leaky
mode typically appears as a mathematical continuation of a related guided mode; at higher frequencies the
LEAKY-WAVE ANTENNAS
mode is generally slow and bound (real and proper), while at lower frequencies it becomes fast (complex and
improper) and can radiate (z /k0 < 1 and z = 0).
It is worth noting here that in open planar structures a different type of leakage can occur as well, which
is associated with surface waves (TE- or TM-like modes) propagating in the substrates (12), instead of the
standard space wave (TEM-like mode) that carries power out into the air. While the latter is able to describe
useful far-field radiation in LWA applications, the former usually describes power that leaks out transversely
in a layered structure and accounts for loss and interference effects in the planar circuits (see also a subsequent
section for comments on this phenomenon).
The case of Fig. 2(b) is representative of leakage in closed metallic waveguides (in this case, filled with
dielectric) that are perturbed with a suitable aperture to allow a certain amount of power loss to the environment. In this case the leaky waves are the complex roots of a characteristic equation that appears as a small
perturbation of that one for the closed waveguide, in such a way that the leakage mechanism also is properly characterized, having similar phase properties and the additional presence of an attenuation (leakage)
constant (see next section). In fact, from Fig. 2(b) it is seen that the behavior of z /k0 resembles that of the
closed structure until the cutoff region is approached, where reactive effects become predominant. Also, for
an open metal guide that is air-filled, the wave is always fast at any frequency (the condition for leakage is
always satisfied), while for an open metal guide that is partly filled with a dielectric the wave is fast at lower
frequencies and becomes slow at higher frequencies, where it can no longer radiate (z /k0 > 1), even in the
presence of nonzero z .
Evaluation of the leaky-wave phase and leakage constants. The evaluation of the complex eigensolutions for nonclosed waveguides depends on the physical parameters involved (frequency and geometry)
and is generally achievable with numerical methods. Among them, the transverse resonance technique (TRT)
(15,16) is one of the most efficient approaches for either rigorous or approximate (according to the antenna
topology) evaluations. It first requires the introduction of a suitable equivalent transmission-line network,
which describes the transverse geometry of the structure. Then, a numerically solvable transcendental equation in terms of transverse eigenvalues kt and of physical parameters is usually achievable by imposing a
resonance condition for the equivalent circuit. The complex eigenvalue kz is derived from the additional link
to the longitudinal problem given by a separation condition for the eigenvalues (e.g., in air, k0 2 =2 0 0 = kt 2
+ kz 2 ). Where the separation condition holds rigorously also for the variables in the transverse plane (e.g., kt 2
= kx 2 + ky 2 ), the TRT in general gives the exact characteristic equation for the geometry. Even if the problem
is not rigorously separable in the transverse plane, an equivalent characteristic equation can be achieved by
the TRT, which often gives solutions with good accuracy for many practical purposes. To this aim, the network
elements that give the electromagnetic description of the various parts of the transverse geometry (such as
radiation impedance, step discontinuities, T junctions, etc.) have to be derived in convenient forms, which are
found to be simple and usually accurate (3,4,12,17,18).
When separation of variables does not strictly hold, other numerical methods can be employed to accurately determine the complex eigensolutions for open waveguides. The most appropriate choice depends on
several factors related to the computational features of the methods, the geometry of the open structures,
etc. (15,16). Among the various possible approaches, integral-equation techniques can work particularly well.
In particular, spectral-domain approaches appear well suited for the derivation of the eigensolutions in structures of the printed type (15). Different formulations based on surface integral equations (the boundary-element
method, etc.) (16) can also furnish satisfactory numerical results in terms of efficiency, versatility, and accuracy for the analysis of quite general classes of arbitrarily shaped open waveguides, allowing the rigorous
determination of the longitudinal wave number for complex modes as well (19,20).
Interpretation of the behavior of a leaky-wave antenna. As described, LWAs are described by a
fast wave that propagates through an equivalent aperture, losing power towards free space, with a leakage
amount that is usually rather limited to allow a sufficiently directive beam. The simplest LWA geometry for this
purpose is that of a closed metallic waveguide in which a suitably small aperture is introduced longitudinally
LEAKY-WAVE ANTENNAS
in order to achieve a continuous power loss along its length, as shown in Fig. 3(a) for a rectangular guide with
a slit cut in a side wall. This structure, besides having historical importance as the first proposed LWA in
1940 (1,2,3,4), can be taken as a reference structure for explaining the basic behavior of LWAs in terms of a
waveguide description.
For such a structure, a leaky wave can be considered as excited by a standard incident mode for the closed
rectangular waveguide, that is, the dominant TE10 , which travels in the +z direction with a known phase
constant 0z for a fixed choice of the physical parameters (geometry and frequency). For a sufficiently small
geometry perturbation due to the slit, the phase constant is changed just slightly to a value represented by
z , and a low leakage rate z then occurs, which, as mentioned, accounts for the longitudinal attenuation due
to the field that is no longer confined and flows also in the outside region. The propagating field inside the
waveguide and in the proximity of its aperture is therefore described by the complex longitudinal wave number
kz = z jz , whose value depends on the physical parameters [the qualitative behavior is like that shown in
Fig. 2(b), and a quantitative description is given in this final part of this subsection].
In this case the leakage is assumed along +z (z > 0 and z > 0), and on supposing that the vertical field
variations are almost negligible (ky 0), it is easily seen, from the general separation condition for waveguides
(k0 2 = 2 0 0 = kt 2 + kz 2 kx 2 + kz 2 ), that the horizontal wave-number is also complex:
since one has x x = z z . Therefore a plane wave of inhomogeneous type exists, having a complex propagation
vector k of the type
where the phase vector is directed at an angle that describes the emission of power from the guide, and
the attenuation (leakage) vector is perpendicular to , and represents attenuation along z and amplification
along x. Consequently, the field has a spatial dependence of the type
Therefore, this plane wave travels at an angle = sin 1 (z /||) with respect to broadside, carrying
out power, and its amplitude increases transversely as expected in a leaky wave. It should be noted that
the direction angle of the leaky wave is usually expressed under the approximate form sin 1 (z /k0 ),
directly derivable from the previous relation, since in general the leakage constant is numerically negligible
with respect to the phase constant. The nature of the propagation vector is sketched in Fig. 3(b), while the
distribution of equiphase and equiamplitude surfaces with respect to the decreasing power flow along the guide
is represented in Fig. 3(c). It should be recalled that, even though the leaky wave has an improper nature, the
field generated from a source located at a finite distance along z still satisfies the radiation condition, since
the field increases transversely only in a limited sector given by angles less than the value describing the
direction of power leakage (5,6,7,8,9).
The leakage mechanism and the relationship for the beam peak direction as a function of frequency can
be interpreted in a very simple way, also referring to the description of the propagation of the dominant mode
LEAKY-WAVE ANTENNAS
Fig. 3. LWA achieved with a side-wall slit in a rectangular waveguide: (a) geometry of the structure; (b) nature of the
propagation vector of the inhomogeneous plane leaky wave (phase and attenuation vectors); (c) equiphase and equiamplitude planes of the leaky wave with the appropriate leakage along the guide; (d) interpretation of leakage in terms of
superposition of homogeneous plane waves at an angle with respect to the longitudinal direction (closed and partially open
structures); (e) equivalent transverse-resonance network, resonance conditions, and network parameters for the numerical
evaluation of the leaky-wave complex wave numbers as a function of the physical parameters involved.
of our structure in terms of superposition of two uniform plane waves incident at an angle on the lateral walls
of the guide, as represented in Fig. 3(d). With the introduction of a small aperture in one wall, at each reflection
a certain amount of power (related to the aperture dimension and to the other parameters involved through z )
leaves the guide at an angle sin 1 (z /k0 ). The pointing angle is therefore near broadside for frequencies
close to the cutoff of the closed guide and tends towards endfire as frequency increases.
A quantitative description of this LWA is easily achieved with a simple analysis of the complex eigenvalue,
derivable through a modification of the dominant mode by employing a TRT (3,4,21). To this end, one requires
LEAKY-WAVE ANTENNAS
a characterization of the slit aperture in the side wall as a circuit element in the equivalent transmission
line. For the evaluation of such discontinuities a great deal of work has been done in the past decades, mainly
through variational methods (4,12,15,16,17,21). The description of the radiative and reactive effects of the slit
in the side wall of the rectangular guide can be represented by a lumped element (e.g., an admittance Y R = GR
+ jBR ) as a function of geometry and frequency. The transverse network and the relationships that allow the
derivation of the characteristic equation having complex eigensolutions are briefly reported in Fig. 3(e). The
relevant solution in the complex plane for the perturbed dominant mode describes the leaky-wave behavior.
Design Techniques
Basic radiation features. The basic design principles of LWAs are generally derivable from the knowledge of the desired beam width and of the pointing direction. In LWAs these quantities can be linked in a
straightforward way to the complex longitudinal wave-number.
The direction M of the beam maximum, in fact, is mainly related to the phase constant for the chosen frequency. It has been seen that the value of the normalized phase constant z /k0 predicts with a good
approximation the direction of the maximum radiation M according to the simple relationship
Since z has dispersive behavior, as is typical of waveguiding structures, scanning of the beam is achieved
by changing the frequency: typically, as the frequency is increased from the cutoff, the pointing angle varies
from around broadside (M = 0 ), towards forward endfire (M = 90 ). (Comments on the limitations of the
scanning capability of LWAs will be presented in the next subsection.)
As regards the beam width, we recall that the leakage constant z expresses the rate of power loss along
the line due to the radiation, thus influencing primarily the effective dimension of the equivalent aperture for
the line source. In fact, the more z increases, the more the actual illumination length decreases (and the less
the beam width is focused).
A basic relation between the leakage constant and the antenna length L derives from the specification
of the radiation efficiency , expressible in LWAs as = [P(0) P(L)]/P(0), where P(0) is the input power
delivered to the structure and P(L) is the output power left at the termination. The relation between efficiency,
leakage rate, and length is generally dependent on the desired radiation pattern and therefore on the aperture
distribution: referring to a uniform-section LWA, where z is independent of z, one has = 1 exp(2z L).
It should also be noted that, for narrow-beam applications, very high efficiency should require an extreme
prolongation of the line source; in practice, in LWAs it is typical to radiate around 90% or at most 95% of the
input power, and the remaining power at the termination is absorbed by a matched load to avoid a back lobe
of radiation due to the reflected wave. (Other considerations related to the efficiency and the illumination will
be discussed in a subsequent subsection.)
Once the efficiency is chosen, a fixed relation therefore exists between the length in wavelengths (L/0 )
and the normalized leakage constant z /k0 . For a uniform-section LWA, an inverse proportionality between L
and z is found:
10
LEAKY-WAVE ANTENNAS
where c is related to the desired efficiency (e.g., for 90% efficiency c = 0.185). For a nonuniform section, since
z depends on z, the relation between efficiency, length, and leakage rate depends on the chosen illumination
and is more complicated.
In order to achieve narrow beams in the elevation angle, the effective longitudinal aperture has to be sufficiently wide (usually several wavelengths), and this implies a rather low leakage rate. The half-power (3 dB)
beam width is directly related to the antenna length given in free-space wavelengths (L/0 ) through an
approximate relationship, which takes into account also the contribution of the scan angle (3):
where the proportionality factor a is dependent on the aperture distribution: it has the smallest value for a
constant aperture distribution (a 0.88) and increases for tapered distributions (typically being more than
unity) (3). From the previous expression, it is seen that, since cos M kt /k0 , the beam width is also expressible
as 2/(kt L). This means that the beam width is, as a first approximation, practically constant when the
beam is scanned away from broadside by varying the frequency for air-filled LWAs (where kt is independent of
frequency), but it changes for dielectric-filled LWAs (where kt depends on frequency).
The effective aperture is reduced for a fixed antenna length as the beam approaches endfire (where the
previous expression becomes inaccurate), and tends in practice to increase. It can be seen that for an
ideal semiinfinite uniform structure (i.e. an antenna aperture from z = 0 to z = L ), the beam width
is determined by the leakage rate only, since in this case it can be found that 2z /kt . Moreover, in this
situation the radiation pattern depends only on z and z and does not present sidelobes:
For finite antenna lengths, sidelobes are produced and the expression for R() is more involved. In general
the specifications on the sidelobe level are related to the choice of the aperture distribution, whose Fourier
transform allows the derivation of the radiation pattern (see subsequent section on this subject).
Uniform and periodic leaky-wave antennas: scanning properties. It is seen that the beams for
LWAs based on partially open air-filled metallic waveguides scan, in theory, an angular region from around
broadside (z /k0 0) towards one endfire (z /k0 1).
In practice, around broadside the structure works near the cutoff region of the closed waveguide, where
reactive effects due to mismatching are increasingly important. The leaky-wave values for z /k0 cannot then
be extremely low, and at the same time z /k0 tends to increase too much, adversely affecting the possibility of
focusing radiation at broadside.
As regards the behavior at endfire it is seen that, since z /k0 tends to unity asymptotically as the frequency
increases, in the unimodal range (where these structures are usually employed) the beam cannot closely
approximate endfire radiation in an air-filled LWA. A way of improving the angular scanning is to fill these
structures with dielectric materials. Thus, since in this case the normalized phase constant approaches the
1
square root of the relative permittivity as the frequency is increased (z /k0 r 2 ), the value z /k0 = 1 can
actually be approached in a much more restricted frequency range. It should however be noted that for such
dielectric-filled structures the beam width may change strongly as a function of frequency and therefore as the
pointing angle varies [see comments on Eq. (7)].
Moreover, it should be noted that in many leaky structures (such as dielectric and printed ones), as the
frequency is increased, the leaky-mode solution changes into a guided-mode solution through a complicated
transition region (2,22,23,24). In this frequency range, also called the spectral gap, the contribution of the
LEAKY-WAVE ANTENNAS
11
leaky wave to the field tends progressively to decrease, and generally the structure does not work well as a
LWA. The representation of the electromagnetic field in this situation becomes more involved (the leaky-wave
contribution is in general no longer strongly representative of the aperture field, and the space wave assumes
increasing importance), and additional analyses are required, whose discussion is beyond the limits of this
work.
As mentioned, while the uniform LWAs usually radiate only in the forward quadrant, within the limits
specified above, the LWAs derived from periodically modulated slow-wave guides can radiate all the way to
backward endfire in the lower frequency range.
The design principles for periodic LWAs are for the most part similar to those for uniform LWAs (1,2,3,4).
The main difference lies in the characterization of the fast wave, which is now associated with a Floquet spatial
harmonic of the periodic guide (1,2,3,4,10,24). One can see that if a uniform guide is considered whose operating
mode is slow (z /k0 > 1; e.g., a dielectric waveguide), and a longitudinally periodic discontinuity is properly
added (e.g., an array of metal strips or notches, etc., placed at suitable distances p), such periodicity furnishes
a field expressible as an infinite number of space harmonics (zn p = z0 p + 2n), where z0 is the phase
constant of the fundamental harmonic, which is slightly different from the original value z of the unperturbed
guide. With proper choices of the physical parameters, it is in general possible to make only one harmonic fast
(typically, the n = 1 one), so that it can radiate as a leaky wave (with an additional attenuation constant z ).
In this case, the phase constant of this fast harmonic can assume both positive and negative values (1
< z /k0 < 1), as a function of the parameters involved: in particular, as the frequency is increased, the beam
inclines from backward endfire towards the broadside. In general, however, periodic LWAs have difficulties in
working well in the broadside region, since in general for periodic structures there exists an open stopband
(24) where the attenuation constant rapidly increases, resulting in a widening beam width with mismatching
effects.
As the frequency is further increased after attaining broadside, the beam is then scanned also in the
forward quadrant. In periodic LWAs, depending on the choice of the design parameters, additional limitations
on the forward scanning behavior may exist when a second harmonic starts to radiate before the first harmonic
reaches its endfire, thus limiting the single-beam scanning range (2,3,24).
Leaky-wave arrays for pencil-beam radiation. If an increase of directivity in the cross plane is
desired, a simple improvement in LWAs based on long radiating slots can be achieved by a physical enlargement
of the transverse aperture (e.g., with a flared transition to enlarge the effective cross aperture). As said before,
a more efficient way to increase directivity in the cross plane is to use a number of radiators placed side by
side at suitable lateral distances, thus constituting a linear array: it is then possible to achieve radiation with
a focused pencil beam. In addition, if properly phased, these arrays of LWAs allow 2-D scanning of the beam:
in the elevation plane, as is typical for LWAs, the scanning is achievable by varying the frequency, while in the
cross plane the scanning is achievable with phase shifters that vary the phase differences among the single line
sources. As noted, in LWAs only a unidimensional set of phase shifters is therefore necessary, yielding structural
simplicity and economic advantage in comparison with the usual radiators requiring a two-dimensional set of
shifters for the scanning. Additional desirable features of such arrays are in general the absence of grating
lobes and blind spots and the achievement of good polarization properties.
For the analysis of such LW arrays, an efficient method is that based on the unit-cell approach (11,12). In
this way, it is possible to derive the behavior of the whole structure by referring to a single radiator, taking into
account the effects on it due to the presence of all the others. In the equivalent network this is achievable by
changing only the description of the radiation termination for a periodic array environment (infinite number of
linear elements); in particular, an active admittance can be calculated, which describes the external radiating
region as a function of the geometry and of the scan angle. More sophisticated techniques also allow accurate
analyses of arrays by taking into account the mutual couplings for a finite number of elements (11).
Procedures for pattern shaping. Most of the basic considerations concerning the beam direction
and the beam width have been described in connection with a general line source, whose characteristics are
12
LEAKY-WAVE ANTENNAS
supposed to be longitudinally constant (uniform section, so that z and z are independent of z). However,
among the basic requirements on the radiation pattern, in addition to the specification for the maximum of the
beam direction and for its half-power width, the sidelobe behavior also has great importance. More generally,
it is desired to examine the properties of the source in comparison with a desired radiation pattern. Since
LWAs can be viewed as aperture antennas with a current distribution having a certain illumination A(z), it is
possible to obtain the far field through a standard relationship:
The radiation pattern for E is expressed in terms of a Fourier transform of the line-source complex current
distribution on the aperture multiplied by the pattern of the element current G (e.g., a magnetic dipole).
It is easily seen that if the LWAs geometry is kept longitudinally constant, the amplitude distribution
has always an exponential decay of the type exp( z z). As is known, this behavior furnishes a quite poor
radiation pattern for the sidelobes, which are rather high (around 13 dB). It follows that, in conjunction
with the choice of a fixed illumination function A(z) giving a desired sidelobe behavior (cosine squared, cosine,
triangular, Taylor, etc.), the leakage rate has to be modulated along the main direction z of the line source; in
practice this is achievable by properly modifying the cross section of the structure along z, a procedure usually
known as tapering. Considering that, for a smoothly tapered antenna, the power radiated per unit length from
the antenna aperture is simply related to the aperture distribution [viz., dP(z)/dz = 2z (z)P(z) = c|A(z)|2 ], it
is possible to obtain a useful analytical expression for z (z) as a function of the amplitude A(z), the line-source
length L, and the efficiency (3,4,25):
From Eq. (10) it is also seen that if high efficiency (close to unity) is desired, z has to be increased to
extremely high values near the terminal section (as mentioned, in common practice efficiency does not exceed
90% to 95%).
In general, in the tapering procedure the longitudinal modification of the geometry should be made in an
appropriate way in order to affect only the leakage constant, keeping the phase constant the same (in pencilbeam applications, z should not depend on z, in order to have the correct pointing angle for each elementary
current contribution on the aperture).
The pattern-shaping procedure requires therefore the knowledge of the phase and leakage constants as
functions of the geometrical and physical parameters of the chosen structure, and this is achievable, as was
said, by finding the suitable complex eigensolution with numerical methods. Since the pattern shaping requires
a proper z distribution with z constant, the procedure is greatly simplified if it is possible to find geometrical
parameters through which the leakage and phase constants are varied as independently as possible. This
possibility is related to the topology of the waveguiding structure.
An example of tapering is sketched in Fig. 4 for a leaky structure, the so-called stepped LWA [Fig. 4(a)],
recently proposed for high-performance applications with well-controlled radiation patterns (26). On the basis
of a preliminary analysis of the effects of the geometry on z and z , it is possible to find the appropriate
longitudinal variation of parameters to obtain the desired z distribution with a fixed z . In Fig. 4(b) the
detailed behavior of the modulation in the height of the lateral steps is shown as a function of z for a desired
illumination (cosine type). A first specification of the imbalance of the steps, with their mean value kept
LEAKY-WAVE ANTENNAS
13
constant (dashed profile), modifies appropriately the longitudinal distribution of the leakage constant, leaving
the phase constant almost unchanged. A second change is advisable to compensate the phase nonlinearity,
which can have rather disturbing effects on the radiation patterns; in this topology it is possible to vary the
steps mean value slightly, with the previously fixed imbalance, to obtain the final valid profile (solid curves).
The corresponding radiation patterns are then illustrated in Fig. 4(c) and 4(d), for the single-part and the
double-part tapering procedures: Fig. 4(c) is a rather distorted pattern related to the nonoptimized tapering
(dashed profile), while Fig. 4(d) is a correct cosine-type pattern related to the optimized tapering (solid profile).
The tapering procedure can be performed numerically in an easy way from a TRT network representation of
the structure. The typical scanning behavior of these kinds of antennas is finally illustrated in Fig. 4(e) for a
beam pointed by varying the frequency.
It is seen that, for the control of the aperture distribution, the longitudinal variation of the cross section
occurs in a continuous and quite smooth fashion, so that it can be assumed that for each z it is possible to
calculate the local value of kz as if the structures were infinitely long. Therefore, the first step in the design
procedure requires the evaluation of the z and z values as a function of the physical parameters of a specific
structure, in particular the dimensions and the frequency. After this basic analysis, which gives the phase and
leakage constants for a specified structure, the general synthesis procedure starts from the desired radiation
characteristics. Thus, from the angle of the beam maximum one finds the value of z /k0 at a desired operation
frequency; this value has to be kept longitudinally constant. From the beam width, it is also fixed the relative
length of the antenna. From the efficiency and the sidelobe specifications, the aperture illumination A(z) is
chosen, and then it is possible to calculate the leakage distribution z (z) from Eq. (10). Once the longitudinal
phase and leakage constants are determined, with the help of the analysis technique the required LWA
geometry is completely achievable.
Overview Of Specic Structures
Here we give a brief overview of the main LWA configurations with their basic behavioral features. More
detailed information can be obtained from the references presented for the various topologies.
Partially open metallic waveguides.
Slitted-side-wall rectangular-guide, channel-guide, holey and slotted, and circular-guide LWAs. The
first-proposed LWA exploits the leakage due to the perturbed dominant mode in a laterally cut rectangular
guide [Fig. 5(a)]. The resulting magnetic current is longitudinal, corresponding to the vertical electric field on
the aperture. A thorough analysis (4,22) has led to simple analytical expressions for the characterization of the
equivalent network, giving the complex wave number kz [see, e.g., Fig. 3(e)]. Taper is achievable by suitably
varying the width of the slit. Possible modifications of this basic geometry have been proposed; in particular,
it is possible to use a square section supporting a higher mode or to completely open the side wall (also filling
the structure with a dielectric), as investigated in the so-called channel-guide antenna (22) [Fig. 5(b)].
One of the main drawbacks of all these configurations having a side cut is related to the leakage constants,
which in general cannot be reduced below a certain limit. Spurious solutions due to channel-guide (27) and
slot-type (28) modes are also found to adversely affect the radiation performance.
Instead of long slits, the introduction of small holes (29) or slots (30) spaced at distances less than the
guide wavelength gives the possibility of reducing the leakage [as in the holey guide shown in Fig. 5(c)]. The
structure can be viewed as quasiuniform, and the described design principles are still valid. Holes and slots
can suitably be placed on the guide top wall as well, even in complicated configurations (transverse slots, series
of closely spaced planar holes, etc.) (1,2,3,4). These structures may present some difficulties in the control of
radiation patterns and in practical manufacture.
The main principles of LWAs based on partially open metal waveguides can be applied to many other
structures. Studies have been performed by several authors, for instance, on slitted circular waveguides [Fig.
14
LEAKY-WAVE ANTENNAS
Fig. 4. Example of the tapering procedure for a LWA to achieve a required aperture distribution for pattern shaping: (a)
Reference structure of a stepped rectangular-guide LWA, relevant parameters, and equivalent network. (b) Longitudinal
modulation of the lateral steps (bl , br , versus z) for a cosine illumination function for a microwave application. The leakage
mainly depends on the imbalance b = (b1 br )/(b1 + br ) of the steps, while the phase mainly depends on the their mean
height bm = (bl + br )/2. The dashed curves of bl , br versus z are obtained with a single-part tapering procedure, that is,
only varying b with constant bm (thus, variations on the phase constant are introduced). The solid curves are due to a
double-part tapering procedure, where phase errors are compensated by suitably varying bm : this is the correct profile for
achieving a radiation pattern with a cosine distribution [see part (c)]. (c) Distorted normalized radiation pattern R (dB)
according to the dashed curves. (d) Correct radiation pattern according to the solid curves for cosine illumination of the
stepped LWA. (e) Typical scanning properties for the pointed beam as a function of the frequency (stepped LWA).
5(d)] (31). In this case, the leakage is derivable from the perturbed dominant TE11 or other higher modes. Again,
practical limits may be encountered on the control of the leakage amount and on finding suitable parameters
for effective tapering.
Inductive-grid and asymmetrical trough-guide LWAs. The inductive-grid LWA [Fig. 6(a)] consists of a
parallel-plate guide operating in its first higher mode (TE1 ), with a cover made of closely spaced transverse
LEAKY-WAVE ANTENNAS
15
Fig. 5. (a) Slitted-side-wall rectangular-guide LWA; (b) channel-guide LWA; (c) holey- and slotted-side-wall LWAs; (d)
slitted circular-guide LWA.
metal strips similar to an array (32). This geometry allows for sidelobe control by varying the strip width to
control the leakage, and the side-plate height to adjust the phase. Though such a structure can work well,
delicate manufacturing problems can occur that make the geometry not very practical.
Another possibility based on a metallic guide, with wider apertures, is given by the trough waveguide
[(Fig. 6(b)] (33), which basically consists of a bisected strip-line structure operating in the first higher mode,
which becomes leaky when a suitable geometrical asymmetry (as a metal insert) is introduced with respect
to the central horizontal plane. It has been found that the structure offers flexible performance, though the
leakage properties can again be difficult to control by tapering.
Top-wall slitted and stub-loaded rectangular-guide and groove-guide LWAs. Reduced leakage is
achievable by slitting the top wall of the rectangular guide, decreasing the current modification due to the
cut [Fig. 7(a)]. By shifting the cut with respect to the central vertical plane, it is possible to modify the leakage
rate; investigations have also been performed with tapered meander profiles for sidelobe control (34). It is
found that the needed alterations of the phase constant for a tapered profile strongly affect the accuracy of the
radiation pattern, which can be severely distorted. Furthermore, this geometry does not easily allow for compensation of phase errors with an additional independent geometrical parameter. Moreover, the introduction
16
LEAKY-WAVE ANTENNAS
Fig. 6. (a) Inductive-grid LWA; (b) asymmetrical trough-guide LWA.
of a simple cut in the top wall of the guide can compromise the polarization purity, due to the effects of
higher-order modes at the aperture.
A way of improving the polarization purity in the basic geometry of a top-wall slitted rectangular guide
is to use an aperture parallel-plate stub, able to reduce the contribution of the higher modes at the aperture,
which are below cutoff, while the dominant leaky wave travels unattenuated as a TEM-like mode at an angle
(35,36,37) [Fig. 7(b)]. Wide metal flanges, simulating an open half space on the upper aperture, can increase
the directivity of this type of LWA. Attention has to be paid to possible channel-guide effects for large values
of the stub height. The topology might allow a printed-type version of this type of LWA, by filling the guiding
structure with a dielectric medium (3,12). It is also worth noting that the geometry under investigation can be
viewed equivalently as an asymmetrical horizontally bisected groove guide. Instead of shifting the stub with
respect to the central vertical plane, another possibility with such geometry is to create an asymmetry through
a small metallic insert [Fig. 7(c)]. Theoretical and experimental investigations have demonstrated the utility
of these structures, though the limits described above are still present when tapering techniques are required
for pattern shaping.
Slitted asymmetrical ridge and stepped rectangular-guide LWAs. The dominant mode of ridge waveguides can be made leaky with a symmetrical stub-loaded aperture if some asymmetry in the height of the
lateral arms is produced [Fig. 8(a)]. The characteristics of this topology show that the average value of the
heights (mean height) of the lateral arms of the ridge acts mainly on the phase constant, while the difference
between the arms heights (imbalance) acts mainly on the leakage constant (38). A tapering procedure that
first adjusts the leakage constant to set the imbalance and then compensates the phase constant through slight
changes in the mean height should be very simple and effective for pattern shaping.
The recently proposed geometry of the stepped LWA (26,39,40) [Figs. 4(a) and 8(b)] maintains the basic
advantages of the ridge structures with additional general desirable features (increased geometrical flexibility,
LEAKY-WAVE ANTENNAS
17
Fig. 7. (a) Top-wall-slitted rectangular-guide LWA; (b) stub-loaded rectangular-guide LWA; (c) strip-loaded groove-guide
LWA.
compactness, low profiles for aerospace applications, etc.). The structure has been investigated theoretically as
concerns the possibility of suitably adjusting the lateral steps for good independent control of phase and leakage. Simple design procedures show the actual capability of synthesizing arbitrary illumination functions with
greatly reduced sidelobes. It has been found that a two-part tapering procedure (first varying the imbalance
longitudinally and then compensating with slight modifications of the mean height) is generally sufficient for
appropriate sidelobe control. (An example of modulation by variation of the lateral steps has been presented
in Fig. 4.) Practical aspects have also been examined, including the effects of ohmic losses (25) and of radomes
for environmental protection (41). Experimental analyses have been performed, too, on both nontapered and
tapered prototypes (cosine illumination, etc.), for both microwaves and millimeter waves. Measured characteristics show very good agreement with theory and excellent radiation performance (precise pattern shaping,
high efficiency and power handling, good polarization purity, etc.) (39,40). Many practical advantages make
this structure one of the most desirable candidates among LWAs based on metal guides.
Printed and dielectric waveguides.
Printed lines: microstrip LWAs. The possibility of using LWAs also in printed circuitry (12) has received
in recent decades an interest that is probably destined to increase in the near future due to the wide use of
planar technology for light, compact, and low-cost microwave integrated circuits (MICs). Among the various
18
LEAKY-WAVE ANTENNAS
Fig. 8. (a) Slitted asymmetrical ridge LWA; (b) stepped rectangular-guide LWA.
printed waveguides that can act as leaky-wave radiators (coplanar guides, slot and strip lines, etc.) (12,42),
we particularly mention structures derivable from lengths of microstrip. Many different configurations can
be employed with microstrips acting as traveling-wave radiators. A first class is based on modulating the
dominant mode of the structure with periodic loadings, such as resonant patches or slots [Fig. 9(a)], and also
by varying the line shape periodically with different meander contours [Fig. 9(b)] (43). Different solutions have
been tested, but the theory of this topic seems to deserve further study.
A different configuration involves the use of uniform structures acting on higher-order modes that can
become leaky for certain values of the parameters involved [Fig. 9(c)]. The analysis of the complex propagation
characteristics of the microstrip line shows in fact that, in addition to the dominant quasi-TEM mode, the
higher-order modes generally become leaky in suitable frequency ranges (12,42) [see Fig. 2(a)]. In particular, it
is seen that the first higher mode EH1 can be excited with a proper odd-type source (the midplane of symmetry
is a perfect electric conductor) and, as the frequency is raised, it starts to leak power. In general, for the planar
structures, leakage can occur in two forms: the surface-wave leakage (power that is carried away through the
TE and/or TM surface modes of the layered structure), and the space-wave leakage (power that is carried away
through the TEM mode of free space). It is found that, for suitable choices of the parameters with an appropriate
excitation, the EH1 mode can represent rather accurately the radiation of the microstrip in a certain frequency
range (12) [see, e.g., Fig. 2(a)]. The coupling between the feeding and the radiating line must be well evaluated,
and simplified equivalent networks can be convenient for this purpose (45).
Radiation performance of printed-circuit LWAs (as concerns power handling, polarization, efficiency,
pattern shaping, etc.) is often less versatile and satisfactory than that of LWAs derived from metal guides
(12). From a practical point of view, difficulties are found particularly in adjusting the phase and leakage
constants independently through the physical parameters. Recent investigations involve simple possibilities
of tapering, also with microstrip topologies, utilizing other higher-order modes (46). Uniform-type microstrip
LWAs have also been investigated in array configurations for 2-D pencil-beam scanning (12,46,47).
LEAKY-WAVE ANTENNAS
19
Fig. 9. (a) Periodically loaded microstrip LWAs. (b) Periodic meander microstrip LWAs. (c) Uniform higher-mode microstrip
LWA: space-wave radiation can be associated, e.g., with the strip current distribution of the EH1 mode, which is leaky in a
suitable frequency range [see Fig. 2(a)].
Nonradiative dielectric guide LWAs. The nonradiative dielectric (NRD) waveguide is a recently proposed
guiding structure for millimeter-wave applications (48) [Fig. 10(a)]. It is a hybrid metaldielectric guide having
the same shape as the H guide proposed for microwave applications: it consists of a dielectric rod inserted
between metal plates placed at a distance apart that is less than the free-space wavelength. In this way, any
discontinuity that preserves the central horizontal-plane symmetry gives only reactive contributions, reducing
interference and radiation effects in integrated circuits. A number of passive and active components have been
realized with this topology (49,50), and integrated antennas have also been proposed (51,52,53,54). Usually,
NRD LWAs should employ some asymmetry in the basic geometry in order to make the operating modes leaky.
A first possible choice (51) [Fig. 10(b)] is to shorten the plates so that the bound operating mode (LSM01 )
(23,48,49,50) presents a nonnegligible amplitude contribution at the equivalent aperture and can give rise to
an outgoing leaky wave in the fast-wave range. Another possible choice (12,52) [Fig. 10(c)] is to insert some
geometrical asymmetry with respect to the central plane (typically an air gap between dielectric and metal), so
that a field having a net electric component perpendicular to the plates can be excited, and power can leak out
in the form of a TEM-like mode traveling at an angle in the parallel-plate region towards the outside. Various
analyses and design procedures have been developed for these configurations (51,52,53,54) in conjunction with
measurements on prototypes, both in single and in array configurations (12,53,54).
Dielectric-rod, image, insular, inset-guide, etc. LWAs. As was said, in basic dielectric guides a periodic
loading is required in order to isolate a suitable fast-wave space harmonic from the intrinsically slow-wave
structure. The reference structures can be various types of dielectric waveguides, such as circular or rectangular
20
LEAKY-WAVE ANTENNAS
Fig. 10. (a) Nonradiative dielectric (NRD) waveguide; (b) shortened NRD LWA; (c) asymmetrical NRD LWA.
dielectric-rod [Fig. 11(a)] (55), image (56), insular, and inset guides (57), [Fig. 11(b)]. The most usual periodic
perturbation is represented by a grating of grooves (58) or metal strips (55,59), usually placed in or on the top
surface of the guide [Fig. 11(c)]; also lateral metal patches can be used in hybrid forms (dielectricmicrostrip)
[Fig. 11(d)] (60). When sidelobe control is required, the taper is realized on the periodic perturbation (e.g.,
with grooves or strips slightly changing their dimensions longitudinally). Various studies have been made to
characterize the theoretical performance of these radiators (61); also, practical aspects have been analyzed,
such as the proper feeding elements in order to avoid spurious radiation and the reduction of the beam width
in the cross plane with flared horns (57) [Fig. 11(e)]. All these topologies are good candidates, particularly for
high-frequency applications (millimeter and submillimeter waves), where the use of dielectric instead of metal
for guidance can reduce the loss effects.
Layered dielectric-guide LWAs. It has been observed that LWAs based on single dielectric layers [Fig.
12(a)], with a ground plane on one side, usually present quite high leakage values, with consequent weak
capability of focusing radiation. A significant improvement is achievable by using additional dielectric layers [Fig. 12(b)]; in particular, interesting analyses have been performed on substratesuperstrate layered
LEAKY-WAVE ANTENNAS
21
Fig. 11. (a) Circular and rectangular dielectric rods; (b) image, insular, and inset dielectric waveguides; (c) periodically
loaded dielectric LWAs; (d) hybrid dielectric microstrip (insular guide with patches) LWA; (e) dielectric LWA with a flared
horn to reduce the cross-plane beam width.
structures (62,63,64). By properly dimensioning the heights and the dielectric constants (usually the substrate
has lower permittivity than the superstrate), it is possible to excite with a simple element (dipole or slot) a
leaky wave giving a conical (due to the symmetries of the topology), highly directive beam (62,63). Recently, this
basic substratesuperstrate topology has been arranged to allow for a very focused pencil beam with a limited
22
LEAKY-WAVE ANTENNAS
Fig. 12. (a) Dielectric-slab waveguide. (b) Layered-dielectric LWA based on a substrate superstrate structure with a
dipole excitation. (c) High-gain LW arrays of widely spaced elements in a substrate superstrate structure: linear and
planar configurations (for the latter case, a top view is shown for a microstrip feeding network of widely spaced slot
elements on the ground plane of the substrate superstrate structure).
number of radiating elements in the form of a widely spaced array, exploiting an interaction between leaky
and Floquet modes [Fig. 12(c)] (65). The advantages of such configurations are in the possibility of achieving,
through very simple design procedures, good radiative performance (high directivity, absence of grating lobes,
etc.) with an array of a few spaced 1-D or 2-D elements, thus reducing the cost of the beamforming network
and suitably exploiting the greater interspace available (dual-polarization applications, etc.).
Nonreciprocal structures: gyrotropic LWAs. In general the use of complex media can give additional
potentialities to LWAs. In particular, the possibility of achieving a unique widely scannable directional beam
with a symmetrical source, in addition to other advantageous features, has recently been investigated for basic
nonreciprocal structures employing gyrotropic media (66,67). A reference structure is considered, represented
by a transversely magnetized grounded ferrite slab with a simple current line source placed symmetrically
LEAKY-WAVE ANTENNAS
23
Fig. 13. (a) Nonreciprocal gyrotropic waveguide based on a transversely magnetized grounded ferrite slab with a line
current source; (b) bisected parallel-plate ferrite-loaded LWA.
inside the layer and in the direction of the bias field [Fig. 13(a)]. In practice, realizations of this topology can
be obtained with a bisected parallel-plate structure [Fig. 13(b)]. It has been found that further improvements
of the radiation performance are achievable with multilayered gyrotropic configurations.
In these nonreciprocal structures the analysis is still possible by means of a suitable extension of the TRT
(68). By varying the frequency or the bias field, even in a uniform structure, it is possible to have only one
pointed beam that scans continuously from one endfire to the other, passing through broadside. It has been
demonstrated that this unusual behavior is linked to two branches of a dominant anomalous mode of the
nonreciprocal waveguide (66): one branch is a complex improper solution (a standard leaky mode, responsible
for forward radiation); the other one is a complex proper solution, responsible for backward radiation.
Practical Considerations And Measurements
Feed, losses, manufacture, etc. Feeding LWAs is usually quite simple. In particular, for LWAs consisting of metal guides, the feed is represented by a continuous transition from the closed structure supporting
a suitable guided mode to the related open one supporting the perturbed (leaky) mode (1,2,3,4). Tapered transitions from the closed to the open structures can be realized to reduce the discontinuity effects and the possible
excitation of spurious modes that could arise from abrupt transitions. At the output termination, the introduction of a matched load further decreases the contribution of the remaining power that, if reflected, would give
rise to a back lobe, in a direction symmetrical to the main beam with respect to the broadside (40).
The use of dielectric structures can present special difficulties in feeding, in particular in planar configurations. Attention has to be paid to avoiding the excitation of additional guided and leaky modes, and also to
obtaining good excitation of the desired leaky wave. Neglect of these aspects can lead to deterioration of the
efficiency and of the radiation performance. For planar guides, such as microstrip or layered dielectrics, local
coupling elements (such as slots or dipoles) are usually employed to excite the leaky mode from an input line
24
LEAKY-WAVE ANTENNAS
towards the radiating line. Proper locations of matched loads or terminations again have to be arranged in
order to reduce the amount of power not properly radiated.
Ohmic losses usually do not greatly affect the radiative performance (efficiency, etc.) of LWAs, since the
attenuation due to the leakage of radiated power is generally more important than the attenuation due to
dissipated power in the nonideal guiding structure (25). However, as the frequency increases, power loss can
be excessive, particularly for LWAs based on closed metal guides. Therefore, for millimeter-wave applications
the choice of open guides with dielectrics and limited use of metal is often advisable.
The general simplicity of LWA structures makes their manufacture usually easy to accomplish, though
various construction problems can arise, depending on the chosen topology and the frequency range. Simple
structures are particularly desirable at millimeter wavelengths, due to the reduced dimensions. On the other
hand, too simplified shapes often cannot allow good control of the radiation performance. In particular, a
delicate aspect concerns the usually small longitudinal modifications of the geometry related to tapering for
sidelobe control. In this case, accurate analyses on the fabrication imprecisions and tolerances have a particular
importance in order to not overwhelm the required geometrical variations for tapering, thus degrading the
improvements of the pattern shaping (40).
Measurement techniques. The radiation properties of LWAs can be tested experimentally through
different types of measurements, most of them applicable to aperture antennas (1,2,3,4,40,69,70). Some basic
parameters, such as efficiency, mismatching effects, etc., can be measured directly through the transmission
and/or reflection scattering parameters with a network analyzer. Radiation patterns and directivity properties
as a function of the observation angles ( and in the zenith and azimuth planes, respectively) can be measured
for various values of the frequency, with different techniques, at the aperture, in the radiating near field (Fresnel
region), and in the far field (Fraunhofer region) (40).
Measurements at the aperture are quite easy to perform, in particular for LWAs incorporating partially
open metal guides. The basic parameters to be determined in LWAs, from which a complete knowledge of the
radiative characteristics is achieved, are the phase and the leakage constants. A measurement of the field in
close proximity to the aperture can be achieved with a small pickup element (e.g., an electric dipole probe placed
parallel to the aperture electric field). The amplitude and phase of the signal received by the probe are thus
measurable through a network analyzer, with possible compensations related to the mutual coupling between
the current distribution at the aperture and the current probe element (69). With numerical processing of the
data, the distributions of the leakage and phase constants are easily calculated. Far-field patterns are also
immediately calculable from these aperture measurements (40).
Alternative measurements can be performed directly on the radiation pattern in standard fashions. If
for instance an anechoic chamber is available whose dimensions do not allow immediate evaluations in the
far-field region (r > 2D2 /), measurements for achieving the radiation pattern are also possible by calculating
the LWAs performance at intermediate distances typical of the Fresnel region (radiating near field). The
LWAs field can be measured through another reference antenna (e.g., a calibrated horn), and rotating supports
are useful in order to facilitate the measurements as a function of the observation angles and frequency.
From these measurements at intermediate distance, by exploiting the equivalence principle, it is possible to
easily reconstruct the far-field radiation pattern. When either the dimensions of the anechoic chamber are
sufficiently wide or a controlled free-space region is available for simulating a far-field (Fraunhofer) region,
measurements of the radiation patterns are immediately achievable, again with a reference antenna and
movable structures. Experiments on all the types of measurements, recently performed for a class of LWAs
(40), show good agreement among the various results.
LEAKY-WAVE ANTENNAS
25
BIBLIOGRAPHY
1. F. J. Zucker Surface- and leaky-wave antennas, in H. Jasik (ed.), Antenna Engineering Handbook, 1st ed., New York:
McGraw-Hill, 1961, Chap. 16.
2. R. E. Collin F. J. Zucker (eds.) Antenna Theory, New York: McGraw-Hill, 1969, Chap. 19 (by A. Hessel) and Chap. 20
(by T. Tamir).
3. A. A. Oliner Leaky-wave antennas, in R. C. Johnson (ed.), Antenna Engineering Handbook, 3rd ed., New York: McGrawHill, 1993, Chap. 10.
4. C. H. Walter Traveling Wave Antennas 1st ed., New York: McGraw-Hill, 1965; 2nd ed., Los Altos, CA: Peninsula
Publishing, 1990.
5. N. Marcuvitz On field representations in terms of leaky modes or eigenmodes, IRE Trans. Antennas Propag., AP-4:
192194, July 1956.
6. S. Barone A. Hessel Leaky wave contributions to the field of a line source above a dielectric slab, part II, Microwave
Res. Inst. Report R698-58, Polytechnic Institute of Brooklyn, 1958.
7. T. Tamir A. A. Oliner Guided complex waves, parts, I, II, Proc. Inst. Elec. Eng., 110: 310324, 325334, Feb. 1963.
8. T. Tamir Inhomogeneous wave types at planar interfaces: IIIleaky waves, Optik, 38: 269297, 1973.
9. L. B. Felsen N. Marcuvitz Radiation and Scattering of Waves, Englewood Cliffs, NJ: Prentice-Hall, 1973.
10. R. E. Collin Field Theory of Guided Waves, New York: IEEE Press, 2nd ed., 1991.
11. R. C. Hansen, Ed. Microwave Scanning Antennas 1st ed., New York: Academic Press, 1966; 2nd ed., Los Altos, CA:
Peninsula Publishing, 1985.
12. A. A. Oliner Scannable millimeter wave arrays, Final Report on RADC Contract No. F19628-84-K-0025, Polytechnic
University, New York, Sept. 1988.
13. T. Itoh Millimeter-wave leaky-wave antennas, Proc. Int. Workshop Millimeter Waves, Italy, Apr. 1996, pp. 5878.
14. I. Ohtera Diverging/focusing of electromagnetic waves by utilizing the curved leakywave structure: application to
broad-beam antenna for radiating within specified wide angle, IEEE Trans. Antennas Propag., AP-47: 14701475,
1999.
15. T. Itoh (ed.) Numerical Techniques for Microwave and Millimeter-Wave Passive Structures, New York: Wiley, 1989,
Chap. 3 (by J. R. Mosig), Chap. 5 (by T. Uwaro and T. Itoh), and Chap. 11 (by R. Sorrentino).
16. R. Sorrentino (ed.) Numerical Methods for Passive Microwave and Millimeter Wave Structures, New York: IEEE Press,
1989.
17. N. Marcuvitz Waveguide Handbook, New York: McGraw-Hill, 1951.
18. P. Lampariello A. A. Oliner New equivalent networks with simple closed-form expressions for open and slit-coupled
E-plane tee junctions, IEEE Trans. Microw. Theory Tech., MTT-41: 839847, 1993.
19. C. Di Nallo F. Frezza A. Galli Full-wave modal analysis of arbitrarily-shaped dielectric waveguides through an efficient
boundary-element-method formulation, IEEE, Trans. Microw. Theory Tech., MTT-43: 29822990, 1995.
20. C. Di Nallo F. Frezza A. Galli Sviluppo di una procedura BEM per la caratterizzazione di structure guidanti arbitrarie,
Proc. XII Riunione Nazionale di Elettromagnetismo, Cetraro Italy, Sept.-Oct. 1998, pp. 277280.
21. L. O. Goldstone A. A. Oliner Leaky-wave antennaspart I: rectangular waveguides, IRE Trans. Antennas Propag.,
AP-7: 307319, Oct. 1959.
22. P. Lampariello F. Frezza A. A. Oliner The transition region between bound-wave and leaky-wave ranges for a partially
dielectric-loaded open guiding structure, IEEE Trans. Microw. Theory Tech., MTT-38: 18311836, 1990.
23. C. Di Nallo et al. Properties of NRD-guide and H-guide higher-order modes: physical and nonphysical ranges, IEEE
Trans. Microw. Theory Tech., MTT-42: 24292434, 1994.
24. S. Majumder et al. The nature of the spectral gap for leaky waves on a periodic strip-grating structure, IEEE Trans.
Microw. Theory Tech., MTT-45: 22962307, 1997.
25. C. Di Nallo et al. Rigorous evaluation of ohmic-loss effects for accurate design of traveling-wave antennas, J. Electromagn. Wave Appl. 12: 3958, Jan. 1998.
26. C. Di Nallo et al. Stepped leaky-wave antennas for microwave and millimeter-wave applications, Ann. Telecommun.,
52: 202208, Mar. 1997.
27. H. Shigesawa et al. Coupling between different leaky-mode types in stub-loaded leaky waveguides, IEEE Trans. Microw.
Theory Tech., MTT-42: 15481560, 1994.
26
LEAKY-WAVE ANTENNAS
28. P. J. B. Clarricoats P. E. Green A. A. Oliner Slot-mode propagation in rectangular waveguide, Electron. Lett., 2: 307308,
Aug. 1966.
29. J. N. Hines J. R. Upson A wide aperture tapered-depth scanning antenna, Report 667-7, Ohio State Univ. Res. Found.,
Dec. 1957.
30. R. S. Elliott Serrated waveguidepart I: theory, IRE Trans. Antennas Propag., AP-5: 270275, July 1957.
31. V. H. Rumsey Traveling wave slot antennas, J. Appl. Phys., 24: 13581365, Nov. 1953.
32. R. C. Honey A flush-mounted leaky wave antenna with predictable patterns, IRE Trans. Antennas Propag., AP-7:
320329, Oct. 1959.
33. W. Rotman A. A. Oliner Asymmetrical trough waveguide antennas, IRE Trans. Antennas Propag., AP-7: 153162, Apr.
1959.
34. F. L. Whetten C. A. Balanis Meandering long slot leaky-wave waveguide antennas, IEEE Trans. Antennas Propag.,
AP-39: 15531560, 1991.
35. P. Lampariello et al. A versatile leaky-wave antenna based on stub-loaded rectangular waveguide: part Itheory, IEEE
Trans. Antennas Propag., AP-46: 10321041, 1998.
36. F. Frezza et al. A versatile leaky-wave antenna based on stub-loaded rectangular waveguide: part IIeffects of flanges
and finite stub length, IEEE Trans. Antennas Propag., AP-46: 10421046, 1998.
37. M. Tsuji et al. A versatile leaky-wave antenna based on stub-loaded rectangular waveguide: part IIImeasurements,
IEEE Trans. Antennas Propag., AP-46: 10471055, 1998.
38. F. Frezza M. Guglielmi P. Lampariello Millimetre-wave leaky-wave antennas based on slitted asymmetric ridge waveguides, IEE Proc. Microw. Antennas Propag., 141: 175180, March 1994.
39. C. Di Nallo et al. Theoretical and experimental investigations on the stepped leaky-wave antennas, 1997 IEEE AP-S
Int. Symp. Dig., July 1997, pp. 14461449.
40. C. Di Nallo et al. Strutture lineari e/o planari del tipo pointed beam a scansione di frequenza, Final Report on Alenia
Contract No. 1324002617/D/00, La Sapienza University of Rome, Italy, Oct. 1997.
41. C. Di Nallo et al. Analysis of the propagation and leakage effects for various classes of traveling-wave sources in the
presence of covering dielectric layers, 1997 IEEE MTT-S Int. Microwave Symp. Dig., June 1997, pp. 605608.
42. H. Shigesawa M. Tsuji A. A. Oliner New improper real and complex solutions for printed-circuit transmission lines and
their influence on physical effects, Radio Sci., 31: 16391649, Nov.Dec. 1996.
43. J. R. James P. S. Hall Handbook of Microstrip Antennas, London: Peregrinus, 1989.
44. F. Mesa C. Di Nallo D. R. Jackson The theory of surface-wave and space-wave leaky-mode excitation on microstrip
lines, IEEE Trans. Microw. Theory Tech., MTT-47: 207215, 1999.
45. P. Burghignoli et al. An unconventional circuit model for an efficient description of impedance and radiation features
in printed-circuit leaky-wave structures, 1999 IEEE MTT-S Int. Microwave Symp. Dig., June 1999, pp. 179182.
46. P. Baccarelli et al. Improving performance in leaky-wave antennas based on uniform microstrip lines, Proc. 29th
European Microwave Conf., Oct. 1999, pp. 126129.
47. C. N. Hu C. K. C. Tzuang Microstrip leaky-mode antenna array, IEEE Trans. Antennas Propag., AP-45: 16981699,
1997.
48. T. Yoneyama Nonradiative dielectric waveguide, in Infrared and Millimeter-Waves, K. J. Button (ed.) Vol. 11, New York:
Academic Press, 1984, pp. 6198.
49. T. Yoneyama Millimeter wave integrated circuits using nonradiative dielectric waveguides, Proc. Yagi Symp. on Advanced Technol., 1990, pp. 5766.
50. A. Galli Componenti e circuiti a microonde basati su risonatori dielettrici in guida NRD, Doctorate Thesis, La Sapienza
University of Rome, Feb. 1994.
51. A. Sanchez A. A. Oliner A new leaky waveguide for millimeter waves using nonradiative dielectric (NRD) waveguide
part I: accurate theory, IEEE Trans. Microw. Theory Tech., MTT-35: 737747, 1987.
52. A. A. Oliner S. T. Peng K. M. Sheng Leakage from a gap in NRD guide, 1985 IEEE MTT-S Int. Microwave Symp. Dig.,
June 1985, pp. 619622.
53. Q. Han A. A. Oliner A. Sanchez A new leaky waveguide for millimeter waves using nonradiative dielectric (NRD)
waveguidepart II: comparison with experiments, IEEE Trans. Microw. Theory Tech., MTT-35: 748752, 1987.
54. J. A. G. Malherbe An array of coupled nonradiative dielectric waveguide radiators, IEEE Trans. Antennas Propag.,
AP-46: 11211125, 1998.
LEAKY-WAVE ANTENNAS
27
55. M. Ghomi et al. Radiation characteristics of uniform and nonuniform dielectric leaky-wave antennas, IEEE Trans.
Antennas Propag., AP-41: 11771186, 1998.
56. T. Itoh B. Adelseck Trapped image guide for millimeter-wave circuits, IEEE Trans. Microw. Theory Tech., MTT-28:
14331436, 1980.
57. T. N. Trinh R. Mittra R. J. Paleta Horn image-guide leaky-wave antenna, IEEE Trans. Microw. Theory Tech., MTT-29:
13101314, 1981.
58. F. Schwering S. T. Peng Design of dielectric grating antennas for millimeter-wave applications, IEEE Trans. Microw.
Theory Tech., MTT-31: 199209, 1983.
59. S. Kobayashi et al. Dielectric-rod leaky-wave antennas for millimeter-wave applications, IEEE Trans. Antennas Propag.,
AP-29: 822824, 1981.
60. A. Henderson A. E. England J. R. James New low-loss millimeter-wave hybrid microstrip antenna array, Proc. 11th
European Microwave Conf., Sept. 1981.
61. M. Guglielmi A. A. Oliner Multimode network description of a planar periodic metal-strip grating at a dielectric
interfacepart I: rigorous network formulations, IEEE Trans. Microw. Theory Tech., AP-37: 534541, 1989.
62. D. R. Jackson N. G. Alexopoulos Gain enhancement methods for printed circuit antennas, IEEE Trans. Antennas
Propag., AP-33: 976987, 1985.
63. D. R. Jackson A. A. Oliner A leaky-wave analysis of the high-gain printed antenna configuration, IEEE Trans. Antennas
Propag., AP-36: 905910, 1988.
64. H. Ostner J. Detlefsen D. R. Jackson Radiation from one-dimensional dielectric leaky-wave antennas, IEEE Trans.
Antennas Propag., AP-43: 331339, 1995.
65. L. Borselli et al. Arrays with widely-spaced high-gain planar elements, 1998 IEEE AP-S Int. Symp. Dig., July 1998, pp.
14461449.
66. P. Baccarelli et al. The role of complex waves of proper type in radiative effects of nonreciprocal structures, 1997 IEEE
MTT-S Int. Microwave Symp. Dig., June 1997, pp. 491494.
67. P. Baccarelli et al. Improvements of the radiation performance in leaky-wave antennas with gyrotropic media, 1999
IEEE AP-S Int. Symp. Dig., July 1999, pp. 982985.
68. C. Di Nallo et al. A convenient transmission-line formulation for wave propagation in typical ferrite structures, IEEE
Trans. Magn., MAG-32: 32283236, 1996.
69. D. T. Paris W. M. Leach E. B. Joy Basic theory of probe-compensated near-field measurements, IEEE Trans. Antennas
Propag., AP-26: 373379, 1978.
70. T. S. Laverghetta Handbook of Microwave Testing, Dedham, MA: Artech House, 1981.
ALESSANDRO GALLI
FABRIZIO FREZZA
PAOLO LAMPARIELLO
La Sapienza University of Rome
388
LINEAR ANTENNAS
nications. This radiating piece of straight wire without curvature is the linear antenna. A simple example of a linear antenna is a two-wire transmission line carrying equal currents
in opposite directions and hence no resultant radiation. A
two-wire transmission line may be bent to create an efficient
radiator such as a dipole. The linear antennas have been
treated in numerous references. Some of them are in Refs.
113. We will describe key features of linear antennas in
this chapter.
SOME RELEVANT TERMS
Before we proceed to discuss linear antennnas, we need to
define and discuss certain terms in accordance with the Institute of Electrical and Electronics Engineers (IEEE) standard
definitions of antenna terminology.
Power Radiated, Radiation Intensity and Radiation Resistance
Electromagnetic waves, by virtue of their transverse nature,
propagate in a direction perpendicular to the plane containing
the electric field E and magnetic field H. The instantaneous
Poynting vector P, which is a measure of the power density
associated with the electromagnetic wave, is given by
P =EH
(1)
where P, E, and H are instantaneous Poynting vector in
watts per square meter, electric field in V/m, and magnetic
field in amps per meter.
The total power P crossing a sphere enclosing the source
(antenna/scatterer) at its center is obtained by integrating
the power density over the sphere and is given by
P = W n dS = W da
(2)
where W is the instantaneous power crossing the sphere per
unit area held perpendicular to the direction of the flow, n is
the positive outwardly drawn at the point of incidence, and
dS is the unit area arbitrarily oriented at the point of incidence. With exp( jt) variation assumed, the average power
density is given by the time-average Poynting vector Pav:
W av (u, v, w) =
1
E H)
Re(E
2
(3)
The average radiated power is given by
P av =
1
2
E H ) da
Re(E
(4)
The radiation intensity U is defined by the product of power
density Prad and the square of the far-field range (r) and is
expressed as
LINEAR ANTENNAS
Historically, using a piece of radiating straight wire as an
aerial, or antenna, was a natural choice for wireless commu-
U = r2 Prad
(5)
The radiation resistance (Rr) is defined as the positive resistance across which the real power radiated (Prad) can be
J. Webster (ed.), Wiley Encyclopedia of Electrical and Electronics Engineering. Copyright # 1999 John Wiley & Sons, Inc.
LINEAR ANTENNAS
thought of as being dissipated. The relationship among Pr,
Rr, the input resistance, and the current I is
Pr
Rr = 2
I
389
The mathematical expressions for D and D are
(6)
D =
4U
Prad + Prad
(13a)
D =
4U
Prad + Prad
(13b)
The input resistance of an antenna is a sum of radiation
resistance plus the positive resistance due to ohmic losses.
Antenna Gain and Radiation Efficiency
Radiation Intensity, Directivity, and Gain
The antenna radiates real power in the far zone in space over
a solid angle of 4 radians. The radiation intensity U(, ),
the real power radiated per unit solid angle, is a product of
the radiation intensity Prad, the real power per unit solid area
on the surface, multiplied by the square (r2) of the distance
and is given by
U (, ) = r2 Prad (, )
U d =
S
=0
2
=0
U sin d d
(8)
An isotropic source, such as an ideal point source, radiates
uniformly in all directions and is independent of and , and
the radiation intensity U0 is related to the real power radiated
by the simple formula:
U0 =
Prad
4
(9)
The directivity is a measure of how efficiently the antenna is
directing the radiation in space, according to the 1983 IEEE
standard (14). The directivity D, a dimensionlesss quantity,
of an antenna is given by
D=
U
4U
=
U0
Prad
(10)
The directivity is dependent on the direction. If the direction
is not specified, the default is the direction of maximum radiation intensity.
The dimensionless maximum directivity Dmax, denoted by
D0, is expressed as
D0 =
4Umax
Umax
=
U
Prad
(11)
Many practical antennas work with dual polarizations in mutually perpendicular directions, and then the directivity is defined in that particular direction; the total maximum directivity is a sum of directivities in mutually perpendicular
directions and is expressed as
D0 = D + D
G=
The radiation intensity in the maximum
direction of radiation (U0 )
The radiation intensity of a lossless isotropic
source with the same input
(14)
(7)
The total power can be estimated by integrating the radiation
over a large sphere enclosing the antenna over 4 radians:
Prad =
An antenna is a passive device, but it can be designed to radiate more energy in a desired direction. The gain (G) of an
antenna is defined as
(12)
All practical antennas have losses, and therefore efficiencies
of practical antennas are less than 100%. The antenna efficiency () is defined as the ratio of the real power radiated
in space by the antenna to the real power input at its feed
terminals:
Radiation efficiency () =
Real power radiated by the test antenna (Prad )
Total real input at the antenna feed terminals (Pin )
(15)
The antenna efficiency is related to the directivity D and
the gain G through the relationship
G = D
(16)
The Vector and Scalar Potentials and Field Calculations Using
Potentials. Most of the time a direct solution of Maxwells
equations subject to the boundary conditions for a practical
problem becomes difficult. Therefore, it is customary to use
intermediatory (or auxiliary) functions, called potential functions, to obtain solutions of electromagnetic problems. There
are four such functions; two of them are scalar (one electric
and one magnetic) and two of them are vector (one electric
and one magnetic) potentials.
The magnetic vector potential A is related to the magnetic
flux density through the relation B A and the electric
scalar potential V is related to E and A through the relation
E V A.
The steps to determine the fields at any point due to the
linear antenna are as follows: (a) Define the current distribution on the dipole, (b) find expressions for the four potentials,
and (c) transfer the cartesian components of the magnetic vector potentials to those in spherical polar coordinates; (d) once
the magnetic vector potential is determined, the magnetic
field at any point is obtained, and (e) what remains to be done
is to use Maxwells equation to determine the electric fields
at any point from the magnetic field obtained.
Before we proceed to determine radiated fields, let us discuss the four potentials for this example. The magnetic current Im is equal to zero since the wire carries a filamentary
electric current and hence the electric vector potential F is
390
LINEAR ANTENNAS
For the infinitesimal dipole (Fig. 1), the current on the infinitesimal dipole is given by
J e (x , y , z ) = zJ
J 0
(18)
where
l/2
l/2
x y 0, since the length of the dipole is infinitesimal
and of length dl
R [(x x )2 (y y )2 (z z )2] (x2 y2 z2)
r(let)
x
(a)
With these, the magnetic vector potential A is given by
A(x, y, z) = z
Er
0
I dl
exp( jkr) dz = z 0 0
exp( jkr)
4r
4r
for r = 0 (excluding the source)
(19)
The components of A are given by
r
0 I0 dl
exp( jkr) cos
4r
I dl
A = Az sin = 0 0
exp( jkr) sin
4r
A = 0
Ar = Az cos =
(20a)
(20b)
(20c)
Due to symmetry of the radiating dipole, we have
/
0;
thus we obtain
(b)
Figure 1. (a) The infinitesimal dipole and (b) its coordinate system.
This figure geometrically shows how the field at any observation point
from an infinitesimal dipole, which is a building block, can be estimated.
H=
Ar
1
(rA)
4r r
(21)
The expressions for magnetic fields are given by
zero since it is a function of magnetic current only. In this
situation, the magnetic vector potential A is given by
A= 0
4
+dl/2
dl/2
exp( jkR)
dz
J (x , y , z )
R
Hr = H = 0
H = j
(17)
where (x , y , z ) are source coordinates, (x, y, z) are the field
coordinates, R is the distance between the observation point
and any point on the source (Fig. 1). Jz is the z-directed electric current element, and the linear path C is along the length
of the source.
1
k0 I0 dl
1+
sin exp( jkr)
4r
jkr
(22)
The electric field can be found from a curl relationship,
namely,
E=
1
H
j
(23)
This gives the three longitudinal and transverse electric field
components as
THE INFINITESIMAL, OR HERTZIAN, DIPOLE
Before we do the analysis for a practical antenna, namely a
linear antenna, let us establish the analysis procedure for an
infinitesimal, elementary, or Hertzian dipole. These are building blocks for more complex antenna systems. Since the dipole is infinitesimal, the current is assumed to be constant.
0 I0 dl
1+
2r2
k I sin
E = j0 0 0
4r
E = 0
Er =
1
cos exp( jkr)
jkr
1
1
1+
exp( jkr)
jkr
(kr)2
(24a)
(24b)
(24c)
LINEAR ANTENNAS
Near and Far Fields
391
and
The near-field region are at a close enough distance such that
kr 1.
exp( jkr)
cos
2k0 r3
exp( jkr)
E
sin
= j0 I0 dl
4k0 r3
exp( jkr)
sin
H
= I0 dl
4r2
E = Hr = H = 0
Er = j0 I0 dl
r2
|E (r, , )|2
20
U=
(30)
(25a)
The maximum directivity D0 turns out to be equal to 1.5.
(25b)
Radiation Resistance
(25c)
The radiation resistance is obtained by dividing total power
radiated by the lossless antenna by I02 /2 and is given by
(25d)
Several observations are in order. Er and E have (1/r2) variation as distance and therefore decays very fast. These are induction components and die down rapidly with distance. The
electric field components Er and E are in time phase, but the
magnetic field component H is in time quadrature with
them. Therefore, there is no time-average power flow associated with them. Hence, the average power radiated will be
zero, and the Poynting vector is imaginary. This can easily be
verified by integrating the average power density over a
sphere in the near region.
The space surrounding the antenna can be divided into
three regions, namely, induction, near-field (Fresnel), and farfield regions. The induction region has 1/r3 space variation,
the near field has 1/r2 variation, and the far field has a 1/r
variation with distance r.
Rr = 80 2
dl 2
(31)
THE THIN LINEAR ANTENNA
This section deals with the analysis and properties of a finitelength dipole. The wire is considered to be thin such that
tangential currents can be neglected and the current can be
considered as only linear. The thin linear antenna and its
geometry are shown in Fig. 2. The boundary conditions of the
current are that the currents are zero at the two ends and
maximum at the center. There is experimental evidence that
the current distribution is sinusoidal. The current distribu-
P(r, , )
Far Field
The far-field expression can be obtained with kr 1 and by
extracting the (1/r) term and is given by
exp( jkr)
sin
4r
Er = E = Hr = H = 0
E = j0 k0 I0 dl
H = jk0 I0 dl sin
exp( jkr)
4r
E
= m
H
z
y
(26b)
l/2
(26c)
The intrinsic impedance Zm of the medium is defined as the
ratio of the tangential electric and magnetic fields and is
given by
Zm =
l/2
(26a)
dz
=
x
(a)
P(r, , )
(27)
dz
Intermediate Field Region
For expressions for field components in the intermediate region (kr 1), the reader is referred to any standard text on
antennas (1).
l/2
l/2
z
z cos
Directivity
The radiation intensity U is given by
U = r Wav
2
(28)
where
Wav =
1
E H)
Re(E
2
(29)
x
(b)
Figure 2. (a) Thin linear antenna and (b) its coordinate system. This
figure geometrically shows how the field at any observation point can
be formulated using the basic building block, namely the infinitesimal dipole.
392
LINEAR ANTENNAS
tions are for a dipole l and for length varying from /2 to .
Thick dipoles will be treated in a subsequent section.
/4
The Current Distribution
/4
I0
zI
0 sin[k(l/2 z )],
0 z l/2
zI
0 sin[k(l/2 z )],
l/2 z 0
(32)
0.707I0
/4
I
/4
The current distribution on the thin dipole is given by
I x (x = 0, y = 0, z )
/4
/4
t=0
t = T/8
t = T/8
(a)
(b)
(c)
/4
0.707I0
/4
I0
/4
The current distributions on the linear dipoles for different
lengths are shown in Fig. 3, and Fig. 4 shows the current
distributions on a half-wave dipole at different times.
Fields and Radiation Patterns. To determine the field due to
the dipole, it can be subdivided into small segments. The field
at any point is a superposition of the contributions from each
of the segments. Since the wire is very thin, we have x 0
and y 0. The electric and magnetic field components due
to the elementary infinitesimal dipole segment of length dz
/4
t = 3T/8
t = T/2
(d)
(e)
Figure 4. Current distribution on a /2 wire antenna for different
times. The current, which is alternating, changes with time. This
figure shows the current changes on a half-wavelength wire antenna
at different time instants.
at an arbitrary point are given by
exp( jkR)
sin dz
4R
dEr
= dE = dHr = dH = 0
dE0 = j0 k0 Iz (x , y , z )
exp( jkR)
sin dz
dH
= jk0 Iz (x , y , z )
4R
Iin
I0 = Iin
where
R=
l<<
x2 + y2 + (z z )2 =
(33a)
(33b)
(33c)
(r2 2rz cos + z2 )
with
(a)
l = /2
r 2 = x2 + y2 + z 2 ,
z = r cos
(b)
The expression for R can be expanded binomially as
R = r z cos + higher-order terms decaying
very fast with r z
Iin
I0
Iin
I0
/2<l<
(c)
<l<3 /2
(d)
Figure 3. Current distribution on dipoles of different lengths. Different physical lengths of the dipole support different current distributions with varying number of half sinusoids. This is because at the
two ends of the wire, the boundary condition that the current must
be zero has to be satisfied.
(34)
The total field due to the dipole is obtained by integrating
over the whole length. Omitting the straightforward steps, it
turns out that the field components E and H are given by
j I exp( jkr) cos(kl/2 cos ) cos(kl/2)
(35a)
E
= 0 0
2
r
sin
H = E /0
(35b)
where 0 is the intrinsic impedance of free space.
To save space we will not describe the derivation of power
radiated which involves Ci(x), Si(x), and Cin(x) functions.
The Radiation Resistance. The resistance can be shown to
be given by
Rr =
{C + ln(kl) C(kl) + 12 sin(kl) [S(2kl) 2S(kl)]
2
+ 12 cos(kl)[C + ln(kl/2) + C(2kl) 2C(kl)]}
LINEAR ANTENNAS
where C(X) and S(X) are well-known functions constituting
Fresnel integrals (see Appendix IV in Ref. 1).
4. Define a set of testing or weighting functions wm, m
1, 2, . . ., N, in the range. Taking the inner product of
Eq. (40) with each wm and obtain
N
THE METHOD OF MOMENTS SOLUTION
For many practical antennas and scatterers including linear
antennas, the desired current distribution is obtained by numerically solving the integral equations. The Method of Moments (MOM) is a technique to convert an integral equation
to a matrix equation and hence solve the linear system by
standard matrix inversion techniques. The MOM is very well
documented in the literature (15), and only the basic steps
will be briefly discussed below.
The magnetic field integral equation (MFIE) for the unknown current density can be rewritten as an inhomogeneous
equation in operator form as follows:
nH
L(Js ) = 2n
H(r)
(36)
where the right-hand side is a known quantity and L(Js) is
an integrodifferential linear operator defined as
1
J s ) = J s n
L(J
2
where m 1, 2, . . ., N; is the inner product, the
product of the two functions integrated over the domain.
5. Express the set of algebraic equations given by Eq. (40)
in the matrix form:
[Imn ][n ] = [gm ]
where the matrix
W1 , L( f 1 )
W , L( f )
2
2
[Imn ] =
WN , L( f N )
J s (r ) G ds
1. Expand f as
n f n
(39)
n=1
where the ns are the unknown coefficients, and the
f ns are known functions of x known as expansion, or
basis, functions.
2. Using Eq. (38) in Eq. (37), we get
n L( f n ) = g
W1 , L( f 2 )
W2 , L( f 2 )
WN , L( f N )
...
...
...
W1 , L( f 1 )
W2 , L( f 2 )
WN , L( f N )
(43b)
(40)
3. Define a suitable inner product f, g defined in the
range L of x:
f (x)g(x) dx
(41)
(44a)
W1 , g
W , g
2
[gN ] = . . .
...
(44b)
WN , g
Moment Method Solution for Radiation from Thin Wire
Two types of volume integral equations are used for the linear
antenna and wire scatterer problem. These are the integral
equation of Hallen type and the integral equation of Pocklington type. Hallens integral equation usually necessitates the
postulation of a delta-gap voltage at the feed point and also
requires the inversion of an (N 1)-order matrix. The advantages of Pocklingtons integral equation is that it is possible
to incorporate different source configurations and it requires
inversion of a matrix of order N.
For a current-carrying perfectly conducting wire, the Hallens integral equation obtained by solving the scalar wave
equation is given by (1)
Iz (z )
n=1
is given by
1
2
[N ] =
..
.
N
(38)
where f f(x) is the unknown function to be determined,
g(x) is a known function and L is a linear operator. The seven
steps (16) in obtaining the solution of Eq. (38) is the same as
the steps for the solution of Eq. (37).
The steps are as follows:
f, g =
(43a)
(37)
L( f ) = g
N
(42)
where N and gN are the column vectors given by
N
n
Wm , L( f n) =
Wm , g
n=1
where Js is the electric current on the wire, and G is the free
space Greens function.
Let us discuss the solution of inhomogeneous scalar equation given by
f =
393
exp( jkR)
dz = j / [B1 cos(kz) + B2 sin(kz)] (45)
4R
where
Iz(z ) the current flowing through an elementary length
of the wire
R distance of the observation point from the elementary length
394
LINEAR ANTENNAS
I the total length of the center-fed wire
, , 0 are the primary constants of the medium in
which the antenna is radiating and k is the derived
secondary constant, namely, the wave vector of the
medium
B1 and B2 are constants to be determined
able in Richmonds work (17) and also in Ref. 8 are reproduced in Table 1.
2. Subdomain Expansion Functions. Subdomain expansions are attractive, convenient, and less expensive in
terms of computer time. This stems from the fact that
the current is matched on part of the integration path,
whereas for the entire domain the integration is performed over the whole path and for all N terms of the
expansions and coefficients determined.
The Pocklingtons integral equation is given by
Iz (z )
2
2
+ k G(z, z ) dz = jEzi
z2
(at = a) (46)
where for thin wire the radius a , the free-space Greens
function G(z, z ) is given by
exp( jkR)
with R = a2 + (z z )2
4R
Ezi = the incident field (this is the source field for
antennas and scattered field for scatterers)
(47)
G(z, z ) = G(R) =
The availability of high-speed computers, graphics, and software packages, along with the enormous development of microcomputers, has made electromagnetic numerical computation extremely viable. An attractive feature of numerical
methods is their ability to handle arbitrarily shaped and electrically large bodies and bodies with nonuniformity and anisotropy where exact solutions can only at best provide some
physical insight. For large problems, it is possible to get a
linear system with a minimum set of equations to achieve a
certain accuracy.
An account of the area of numerical computation of thin
wire problem is well-documented in the literature (1,3,613).
As described in Ref. 8, there are many important computational issues involved in thin wire problems. These are (a)
segmentation of the structure and the correct number of segments, (b) choice of right current expansion functions, (c) the
thin wire approximation (radius a ), (d) Roundoff errror
due to matrix factorization, (e) near-field numerical anomaly,
(f) treatment of the junctions of the segments, (g) wire-grid
modeling, and (h) computer time required. Also, the errors (7)
involved are of concern. There are two types of errors encountered: (a) the physical modeling error, because in the absence
of an exact solution for a variety of semicomplex and complex
stuctures, it is the natural departure of the assumed structural details from the actual structure, and (b) the numerical
modeling error, since all numerical methods are approximate
but sufficiently accurate for the application.
Miller and Deadrick (8) provide a table containing the many
basis and weighting functions which have been tested in computer runs. This table also compares the suitability of the use
of different functions in different problems. The table is too
big to reproduce here, and it is left to the reader to look up. It
tabulates the method, integral equation type, basis function,
current conditions for interior and end segments, the basis
function in terms of unknown for unknown and end segments,
weighting function for interior and end segments, number of
unknowns, and specific comments on the applicability of the
expansion functions.
The Electric Field Integral Equation
and Its Matrix Representation
Figure 5 gives the geometry of the arbitrarily oriented
straight wire. The wire is broken into segments, or subsections. The mini numerical electromagnetic code (MININEC)
relates the current distribution on the wire due to the incident field. The integral equation relating the incident field
Ei, magnetic vector potential A, and electric scalar potential
are given by
A t t
E i t = jA
where
1. Entire Domain Functions. As the name suggests, these
functions are defined over the entire integration path
C, and the subdomain is defined over pieces on the integration path C. Some of these are Fourier, MacLaurin,
Chebyshev, Hermite, Legendre, and so on. Tables avail-
A=
0
4
1
4
S (s)k(r) ds
I (s)S
(48b)
q(s)k(r) ds
(48c)
Table 1. Entire Domain Current Expansions Using
Different Polynomials
A. The Polynomials
Fourier: I(z) I1 cos(x/2) I2 cos(3x/2) I3 cos(5x/2)
MacLaurin: I(z) I1 I2 x2 I3 x4
Chebyshev: I(z) I1T0(x) I2T2(x) I3T4(x)
Hermite: I(z) I1 H0(x) I2 H2(x) I3 H4(x)
Legendre: I(z) I1 P0(x) I2 P2(x) I3 P3(x)
where 1/2 x 2z/L 1/2
Formulation
Before we discuss the formulation of the thin-wire integral
equation, comments on the expansion functions used in this
study are in order. There are broadly two types of expansion
functions:
(48a)
B. Typical Results for Functions
L 0.5; a 0.005; i 90
In
Fourier
MacLaurin
Chebyshev
Hermite
Legendre
1
2
3
4
5
3.476
0.170
0.085
0.055
0.040
3.374
4.037
3.128
4.101
1.871
1.7589
1.5581
0.0319
0.0112
0.0146
8.2929
14.3644
4.4135
0.3453
0.0073
2.2763
2.1005
0.0655
0.0421
0.0322
LINEAR ANTENNAS
p1 ......... pn1
pn ......... pn+1
1 ......... n1
pN1
n+1 ......... N1
n+1 ......... N1
(a)
1 ......... n1
n
(b)
Figure 5. Wire segmentation with pulses for current and charges.
(a) Unweighted current pulses. (b) Unweighted charge representation. The whole length is broken into several segments. Each segment
is assigned a pulse, and the pulses represent the assumed current
distribution.
t is a unit vector tangential to the wire at any distance along
the path of integration which is the length of the wire and
k(R) is given by
k(R) =
1
2
2
=0
exp( jkR)
d
R
1 dl
j ds
(49)
(50)
The MININEC solves the integral equation using the following steps:
1. The wires are divided into small equal segments such
that the length of the segment is still large compared to
the radius of the wire (a , typically 1/100th of a
wavelength). The radius vectors m, n 0, 1, 2, . . . are
defined with reference to a global origin.
2. The unit vectors are defined as
sn+1/2 =
r n+1 r n
|rr n+1 r n |
(51)
The testing and expansion functions are pulse functions
which are defined by
1
for sn1/2 < s < sn+1/2
Pn (s) =
(52)
0
otherwise
where the points sn1/2 and sn1/2 are the segment midpoints
and are given by
sn+1/2 =
sn+1 + sn
,
2
sn1/2 =
sn sn1
2
The exact kernel treatment developed above is for observation
points on source segments. For observation points near but
not on the source, a segment has been developed by Wilton
and MININEC has incorporated it (16).
Expansion of Currents
The currents are expanded in terms of pulse functions as
shown in Fig. 5, excluding the end points where the currents
are chosen as zeros to satisfy boundary conditions at the ends.
The current expansion is given by
I(s) =
N
In Pn (s)
(56)
Using this current expansion in Eq. (48b) and after mathematical manipulations which are available in Ref. 1 and are
not detailed here, we get the linear system matrix equation
in N unknowns:
[Vm ] = [Zmn ][In ]
(57)
where m, n 1, 2, . . ., N, [Zmn] is the square impedance
matrix, and [Vm] and [In] are applied voltage and current column vectors:
1
k2 (rm+1/2 rm1/2 ) (sn+1/2m,n,n+1/2
Zm =
4 j
m+1/2,n,n+1 m+1/2,n1,n
+ sn1/2m,n1/2,n )
+
(58)
sn+1 sn
sn sn1
m1/2,n,n1
m1/2,n1,n
+
sn+1 sn
sn sn1
This matrix has elliptical integrals involved in it. These elliptical integrals can be evaluated numerically.
The above equations are valid for any radius other than
small, for which the expression for breaks down and Harrington (18) provided an approximate formula for . This is
given by
=
=
s
k
1
ln
j 0
2
s
4
exp( jkrm )
4rm
for m = n
for m = n
(59a)
(59b)
The integral is given by
(53)
In terms of global coordinates,
+ rn
r
,
rn+1/2 = n+1
2
When the pulse functions of Eq. (52) are inserted in parentheses, we obtain
s s
sm sm1
m
m+1
i
E (sm )
Sm1/2 +
Sm+1/2
2
2
sm sm1
Sm+1 sm
Sm1/2 +
Sm+1/2
= jA(sm )
2
2
(55)
n=1
The continuity equation given below determines the relationship between the charge q(s) and the rate of change of current
with distance:
q(s) =
395
m,u,v =
k0 (sm s ) ds
(60)
Inclusion of Nonradiating Structures
rn1/2
rn + rn1
=
2
(54)
The Ground Plane. When the wire structure near the
ground plane is assumed to be perfectly conducting, an image
396
LINEAR ANTENNAS
is created. The structure and the ground plane is equivalent
to the structure and the image.
The voltage and current relationship is given by
[Vm ] = [Zmn ][In ]
(61)
0.1
0.1
2m
2m
Wire 3
Wire 1
where
Wire 2
Zmn = Zmn + Zm,2Nn+1
y
0.04 m
Wire Attached to Ground. When a wire is attached to the
ground on one or both sides, there will be a residual component of current at one or both ends. In this case, a current
pulse is automatically added to the end point in the formulation.
Lumped Parameter Loading. If an additional complex load is
added to the perfectly conducting wire (Fig. 6), there will be
an additional voltage drop created at that point if the location
of the load (Zl Rl jXl) is at a point of nonzero pulse function. The impedance matrix is modified to
[Vm ] = [Zmn ][In ]
=
Zmn + Zload
for m = n
Zmn
for m = n
Figure 7. Geometry of the Tee antenna. Typical dimensions are
shown.
to meet the following requirements: (a) the segmentation density, (b) thin-wire criteria, (c) small radius calculation, (d)
step changes in wire radius, (e) spacing of wires, (f) loop antenna, and (g) monopoles and antennas above ground.
Operation of CurrentsLU Decomposition
where Z mn is the modified impedance matrix and is given by
Zmn
Validation of the MININEC Code
Extensive work has been reported on the validation of the
MININEC. Numerous validation runs have been carried out
The operation is oriented around the Menu shown below.
Here we describe the DOS version (19,20), but the Windows
version is also available (2123).
MENU
1 - COMPUTE/DISPLAY CURRENTS
2 - CHANGE EXCITATION
3 - CHANGE FREQUENCY
4 - CHANGE LOADING
5 - LOAD GEOMETRY
6 - SELECT OUTPUT DEVICE
7 - RETURN TO SYSTEM SUPERVISOR
0 - EXIT TO DOS
SELECTION (17 OR 0)?
(a)
Some Examples Using MININEC
Wire 1
Wire 2
Overlapping pulse
(b)
Overlapping pulse
Wire 3
Wire 2
Wire 1
Overlapping pulse
(c)
Figure 6. Overlap scheme used at a multiple junction of wires. (a)
Wire 1 with no end connections. (b) Wire 2 overlaps onto wire 1. (c)
Wire 3 overlaps onto wire 1.
Tee Antenna. Figure 7 shows the geometry of the Tee antenna fed from the base by a coaxial line. The impedance calculations of the Tee antenna using different computer programs including CURLU in MININEC and have been
compared with measurements (25).
Near and Far Fields. The near- and far-field programs
(FIELDS) calculate near and far fields using the current distribution on the structure obtained by integral equation formulations. The current distribution can be computed using
three programs: CURLU, CURTE, and CURRO. The current
distribution can be computed using perfect and imperfect
grounds, although the real ground corrections are applied to
the far field only. The real ground correction is included in the
form of reflection coefficients for parallel and perpendicular
polarizations. For details, the reader is referred to Chap. 8 of
Ref. 7. The menu is given below. User input(UI) means the
user is expected to respond at that point.
LINEAR ANTENNAS
104
100
103
MININEC
2
NEC
Antenna:
10.67 m monopole
0.102 m wire radius
20-Segment model
Near fields:
Magnitude of Ez and Er at z = 1.5 m
Radiated power = 8.775 mW at f = 1 MHz
101
Electric field (V/m)
Electric field (V/m)
Vertical electric field
at 1 m above perfect ground
1 kW radiated at 2.0 MHz
10
102
MININEC
NEC
103
104
8
12
16
Horizontal distance (m)
NEC
105
20
Figure 8. Monopole near fields: Ez versus horizontal distance.
MENU
1 - COMPUTE NEAR FIELDS
2 - COMPUTE FAR FIELDS
3 - SELECT/CHANGE ENVIRONMENT
4 - SELECT/CHANGE CURRENTS FILE
5 - SELECT OUTPUT DEVICE
6 - RETURN TO SYSTEM SUPERVISOR
0 - EXIT TO DOS
SELECTION (16 OR 0)? User Input
NAME OF INPUT CURRENT FILE? User Input (UI)
ELECTRIC OR MAGNETIC NEAR FIELDS (E/M)? User
Input
X-COORDINATE
Y-COORDINATE
INITIAL VALUE? UI INITIAL VALUE?
INCREMENT?
UI INCREMENT?
NO. OF PTS?
UI NO. OF PTS?
Z-COORDINATE
UI INITIAL VALUE? UI
UI INCREMENT?
UI
UI NO. OF PTS?
UI
PRESENT POWER LEVEL IS : CURRENT VALUE
CHANGE POWER LEVEL (Y/N)? UI
NEW POWER LEVEL (WATTS)? UI
Once the parameters are specified, the near- and far-field
results are printed out in words. Figures 8 through 15 show
the near-field characteristics of the monopole.
200
300
400
Radial distance (m)
500
A thick cylindrical dipole (Fig. 16) is the inexpensive way to
increase the bandwidth of linear antennas. The increase in
thickness leads to circumferential component I of the otherwise linear current. This can be handled with the integral
equation formulation. Figure 17(a) and Fig. 17(b), respectively, show the variation of input resistance and reactance of
the dipole with l/2a ratios 25 (thick), 50, and 10,000 (thin),
where 2a is the diameter of the wire.
102
Near fields:
H component at z = 1.5 m
Radiated power = 8.775 mW at f = 1 MHz
103
104
MININEC
105
NEC
Antenna:
Height = h = 75 m
Radius = a = 0.3 m
15-Segment model
THE THICK LINEAR ANTENNA
The thin linear antenna is frequency-sensitive. In practical
communication scenarios, the transponders use wideband signals to increase the channel capacity and therefore needs antennas that can handle a large band of frequencies. One way
of increasing the bandwidth is to use electrically thick wires.
100
MININEC
Er
Figure 9. Monopole near fields: Electric fields Ez and Er versus radial distance.
Magnetic field (A/m)
Ez
Antenna:
Height = h = 75 m
Radius = a = 0.3 m
15-Segment model
Measured
101
397
106
100
300
400
200
Radial distance (m)
500
Figure 10. Monopole near fields: H versus radial distance.
398
LINEAR ANTENNAS
101
MININEC compared to NEC
4
100
Near fields:
3
Vertical and radial components of E-fields
z = 1.5 m
Power radiated = 8.775 mW
f = 1 MHz
Antenna:
Height = h = 75 m
Radius = a = 0.3 m
15-Segment model
Er
Ez
50
100
MININEC
150
200 250 300 350
Radial distance (m)
400
450
500
Vertical electric field (V/m)
Percent difference in electric fields
Antenna:
Height = h = 75 m
Radius = a = 0.3 m
15-Segment model
101
102
103
Figure 11. Percentage difference in electric fields Ez and Er versus
radial distance.
NEC
104
Antenna:
z - Component at wire surface
Radiated power =
8.775 mW at
f = 1 MHz
THE SLEEVE DIPOLE
The structure that closely resembles an asymmetric dipole is
the sleeve dipole (Fig. 18), which is a base-fed monopole on a
ground plane. The outer conductor of the coax line is extended
to give wideband characteristics apart from impedance control and mechanical strength. Another way of increasing the
bandwidth will be to coat the metal wire with a low-loss dielectric or just push the metal wire into a sleeve (a thin dielectric pipe). It can be fed symmetrically, but the problem of
impedance matching and using a proper matching network
will increase the complexity and cost of the device. For further discussion on sleeve dipoles, the reader is referred to
Ref. 1.
MININEC compared to NEC
Percent difference in H
0.5
Near fields:
H component
at z = 1.5 m
Radiated power =
8.775 mW at
f = 1 MHz
1.5
50
100
150
Antenna:
Height = h = 75 m
Radius = a = 0.3 m
15-Segment model
200 250 300 350
Radial distance (m)
36
108
72
Vertical distance (m)
144
180
Figure 13. Monopole near fields: Vertical electric field versus vertical distance.
COMPUTER CODES
A Summary of Different Available Codes
The following subsections summarize the scope of the codes
currently used for linear antenna analysis.
The Electromagnetic Surface Patch Code. The electromagnetic surface patch (ESP) code Version IV (26), which uses
the Method of Moments (MOM), is written for the analysis of
radiation and scattering from three-dimensional geometries.
These geometries include the interconnection of thin wires
and perfectly conducting and thin dielectric polygonal plates.
This code works for open as well as closed surfaces since the
formulation is based on electric field integral equation. It uses
polygonal plates modeling, which can generate an object as
complex as an aircraft with only 12 polygonal plates. The ESP
code is capable of doing the following:
0.5
105
400
450
500
Figure 12. Monopole near fields: Percentage difference in H versus
radial distance.
1. It can model an arbitrary thin wire by using a number
of piecewise straight segments.
2. It can model an arbitrary perfectly conducting surface
by a combination of polygonal plates. The plates can be
of thin dielectric sheets, which have been modeled using
impedance approximation.
3. The code incorporates the wirejunction with the restriction that the junction must be at least one-tenth
wavelength away from the nearest edge; it can take
care of several plates generated from a common edge.
LINEAR ANTENNAS
399
101
101
NEC
102
MININEC
MININEC
NEC
103
101
Magnetic field H (A/m)
Radial component of electrical field (V/m)
100
Near fields:
x-Component of E field
Radiated power =
8.775 mW at
f = 1 MHz
102
103
Near fields:
At x = 0.3 m
Radiated power =
8.775 mW at
f = 1 MHz
104
105
106
Antenna:
Height = h = 75 m
Radius = a = 0.3 m
15-Segment model
104
105
107
108
36
72
108
Vertical distance (m)
Antenna:
Height = h = 75 m
Radius = a = 0.3 m
15-Segment model
144
180
Figure 14. Radial component of electric field at wire surface versus
vertical distance.
In order to use this code, the user specifies the (x, y, z) coordinates of the corners of each plate by entering the (x, y, z)
coordinates of the corners of each plate and the maximum
segment size in wavelengths for calculating the MOM modes.
The code automatically takes care of the frequency independence of the models.
In summary, the ESP code can treat a variety of geometries: (a) thin wires with finite conductivity and lumped loads,
(b) perfectly conducting or thin dielectric polygonal plates (30
wireplate junctions) at least 0.1 from the edge of a plate,
(c) plateplate junctions, including several plates of different
sizes which intersect along a common edge, and (d) excitation
by either a voltage generator or an incident plane wave.
ESP can compute the key characteristics of engineering interest of an antenna, namely, current distributions, antenna
input impedance, radiation efficiency, mutual coupling, near
or far zone radiation patterns for all polarizations, and near
or far zone back, bistatic, and forward scattering patterns
(full scattering matrix). There are still bugs in the ESP code,
and new features are being added to the codes. Nevertheless,
it should be possible to customize the code for a specific application.
The Mini Numerical Electromagnetic Code (MININEC). The
MININEC (20,21) is a computer program prepared in BASIC
language using the method of moments for the analysis of
linear antennas. It uses a Galerkin procedure (19) to solve
for wire currents using an integral equation formulation that
36
72
108
Vertical distance (m)
144
180
Figure 15. Monopole magnetic field H at surface versus vertical distance.
dz
l/2
z A
y
l/2
d = 2a
Figure 16. The center-fed thick dipole. A thick dipole has a frequency characteristic over a much wider band than a thin antenna.
The diagram shows how such a dipole will receive power from the
feeder network.
400
LINEAR ANTENNAS
1000
l/d = 25
(Moment method)
l/d = 50
(Moment method)
l/d = 104
(Sinusoidal current)
Input resistance ()
800
600
400
200
0
0.0
0.2
0.4
0.6
0.8
1.0
Dipole length (wavelengths)
1.2
1.4
1.6
1.2
1.4
1.6
(a)
400
Input reactance ()
200
l/d = 25
(Moment
method)
l/d = 50
(Moment
method)
l/d = 104
(Sinusoidal
current)
200
400
600
0.0
0.2
0.4
0.6
0.8
1.0
Dipole length (wavelengths)
(b)
Figure 17. The input impedance of wire dipoles with dipole length in wavelengths. (a) Input
resistance in ohms. (b) Input reactance in ohms. Every antenna presents an input complex impedance at its feed terminals. It is very important to know how the real and imaginary parts of
the impedance vary with parameters so that the antenna can be matched to the feeder.
relates the vector and scalar potentials to the electric field.
This formulation can easily be programmed for use in microcomputers. The code can solve for impedance and currents on
arbitrarily wires including configurations with multiple wire
junctions. The code incorporates lumped impedance loading
and near-zone electric and magnetic fields when the wire is
in free space as well as over a perfectly conducting ground.
MININEC is written in a IBM PC-DOS-compatible BASIC
language.
FEED FOR LINEAR ANTENNAS
In all practical antennas, there exists a mechanism of feeding
the antenna to launch the desired current distribution. This
requires a finite gap with a nonzero current and alters the
normally assumed current distribution. Hence, a correction
factor should be introduced to accurately model the feed point
current. Several methods of feeding half wave antennas are
shown in Fig. 19(a) through 19(d). Normally since the com-
LINEAR ANTENNAS
401
Figure 19(b) shows a delta-match or shunt-feed feeder
which provides a good impedance mismatch and low VSWR
on the feeder line. The simplest feeding arrangement is the
end-feed [Fig. 19(c)] where the vertical transmission line radiates energy. The configuration in Fig. 19(d) can be made
efficient by tapping the vertical line at an appropriate location.
To account for the nonzero current at the finite gap feed
terminals and consequent distortion in the current distribution.
The modified current is given by (28)
Iz (x , y , z )
a {I sin[k(l/2 z )] + jI0
z 0
=
az {I0 sin[k(l/2 + z )] + jI0
z=h
z=0
Figure 18. The sleeve dipole. Apart from making the wire thick, another way of increasing the bandwidth of a thin wire is to put it in a
sleeve, which is basically a dielectric coating, and terminate by a
ground plane on one side.
plex input impedance of the dipole is quite different from the
real input resistance (under high frequency approximation),
there is a need to use a matching network to match the input
impedance to the feeder and keeping the voltage standing
wave ratio (VSWR) close to unity in the frequency range of
operation. Figure 19(a) is the balanced line-type center feed.
(a)
(b)
(c)
(d)
Figure 19. Methods of exciting linear antennas: (a) center-feed; (b)
delta-match or shunt-feed; (c) end-feed; (d) impedance matched taping. Excitation mechanisms are very important in order to launch
maximum power. Not shown are matching transformers to ensure
maximum power transfer.
[cos(kz ) cos(kl/2)]}
for 0 z l/2
[cos(kz ) cos(kl/2)]}
(62)
for l/2 z 0
where the coefficient dependent upon the length of the
gap and the antenna. If the radius of the wire and the gap
are small the value of this coefficient is small.
For a half wave dipole,
I0 (x , y , z ) = az I0 (1 + j) cos(kz )
for 0 |z | /4
The shape of the current does not change for half wave dipole,
but it does for other lengths, such as a wavelength.
BANDWIDTH OF LINEAR ANTENNAS
The bandwidth of an antenna is the frequency range over
which the performance of the antennas meets a specific requirement, such as the gain does not fall below 3 dB of the
maximum, or the mid-frequency value. There is no unique
characterization of bandwidth since the different properties
such as input impedance, radiation pattern, polarization and
gain of an antenna vary in entirely different ways in a frequency range. The definition meets the requirement of specific application. Nevertheless, the bandwidth is defined in
three different ways: (1) Half power bandwidth is the frequency range within which the gain does not fall by more
than 3 dB. (2) The percentage bandwidth normally defined
for narrow band antenna is defined as the bandwidth (the
difference between upper and lower frequencies of operation)
divided by the center frequency and then multiplied by 100,
and (3) for wideband or frequency independent antennas it is
the ratio of higher and lower frequencies of operation.
The thin linear antenna that we have dealt with in this
article is based on the assumption that the radius is small
compared with the wavelength of operation. This necessarily
means that the current on the wire is linear and has no tangential component. In a practical situation an antenna has to
operate in a frequency band. This assumption is no longer
valid and the thin linear antenna becomes frequency sensitive. Therefore, something needs to be done to the thin antenna so that it develops the capability of handling a wide
frequency band.
The ways of achieving wider bandwidth will be to use electrically thick dipoles or to coat thin metallic wires with lossy
402
;;
LINEAR ANTENNAS
terial. They are given by
18
16
14
G (mmhos)
12
10
Dielectric coated
dipole
L =
2a =
2b =
2 =
6
4
8"
0.025"
0.146"
2.360
2
0
0.1
0.2
0.3 0.4 0.5 0.6
Dipole length L/
0.7
1
b
ln
r
a
b
Q = (r 1) ln
a
P=
Calculated
Measured
at O.S.U.
0.8
(a)
10
(63a)
(63b)
where a radius of the thin wire, b a thickness of the
coating, r the relative permittivity of the coating or the
medium in which the thin wire is embedded, r the relative
permeability of the coating or the medium in which the thin
wire is embedded.
It turns out that the effects of increasing the real part of
both P and Q are to increase the peak value of input impedance, to increase the electrical length which lowers the resonant frequency and to reduce the bandwidth of operation. The
effects of increasing the imaginary part of P and Q are to
decrease the peak value of input impedance, to decrease the
electrical length which increases the resonant frequency, and
to increase the bandwidth. This means a proper choice of a
lossy dielectric with maximum imaginary parts of P and Q
and minimum real parts of P and Q can achieve an optimum
bandwidth. But this will be at the cost of antenna efficiency
because of the lossy coating.
BIBLIOGRAPHY
B (mmhos)
1. C. A. Balanis, Antenna Theory: Analysis and Design, New York:
Wiley, 1997.
2. C. A. Balanis, Advanced Engineering Electromagnetics, New
York: Wiley, 1979, chap. 12, pp. 670742.
3. J. Moore and R. Pozer (eds.), Moment Methods in Electromagnetics: Techniques and Applications, Letchworth, UK: Research Studies Press, 1984; New York: Wiley, 1984.
4. W. A. Imbraile, Applications of the Method of Moments to ThinWire Elements, in R. Mittra (ed.), Computer Techniques for Electromagnetics, New York: Pergamon, 1973, chap. 1.
4
2
0
2
4
6
0
0.1
0.2
0.3 0.4 0.5 0.6
Dipole length L/
0.7
0.8
(b)
Figure 20. Comparison of measured and calculated admittance for
a coated dipole of length L 8 in.; 2a 0.025 in.; 2b 0.146 in.;
2 2.30. The figures show the bandwidth of antennas: (a) conductance and (b) susceptance of an insulated dipole vs. length in wavelengths.
5. C. A. Balanis, Antenna Theory: Analysis and Design, New York:
Wiley, 1997, chap. 4, chap. 8.
6. G. A. Thiele, Wire antenna, in R. Mittra (ed.), Computer Techniques for Electromagnetics, New York: Pergamon, 1973, chap. 2,
pp. 770.
7. J. Rockway et al., The MININEC System Microcomputer Analysis
of Wire Antenna, Norwood, MA: Artech House, 1988.
8. E. K. Miller and F. J. Deadrick, Some Computational Aspects of
Thin-Wire Modeling, in R. Mittra (ed.), Numerical and Asymptotic
Techniques in Electromagnetics, New York: Springer-Verlag,
1975, chap. 4, pp. 89127.
9. J. J. H. Wang, Electromagnetics: Formulation and Computer Solution of Integral Equation, New York: Wiley, chap. 5, 1991.
dielectrics. A thick dipole and a dielectric coated dipole with
their centrally located feeding source are shown in Fig. 20(a)
and (b), respectively. The effect of coating the thin linear antenna with a layer of electrically and magnetically lossless
and lossy material is discussed in two papers in (28) and (29)
and summarized in (1). The analytical technique used is a
moment method solution. Two parameters P and Q (28,29)
involving the electrical and magnetic parameters, inner and
outer radii, are found to be of interest in moment method solutions and help in designing antenna characteristics using
coating of electrically and magnetically lossless and lossy ma-
10. E. K. Miller, L. Medguesi-Mitchang, and E. H. Newman, part 3,
Thin Wires, Computational Electromagnetics: Frequency-Domain
Method of Moments, New York: IEEE Press, 1992.
11. R. Mittra (ed.), Numerical and Asymptotic in Electromagnetics,
Berlin: Springer-Verlag, 1975.
12. C. A. Balanis, Advanced Engineering Electromagnetics, New
York: Wiley, 1979, pp. 150.
13. K. K. Mei, On the integral equations of thin wire antennas, IEEE
Trans. Antennas Propag., AP-13: 1965.
14. C. M. Butler and D. R. Wilton, Analysis of various techniques
applied to thin-wire scatterers, IEEE Trans. Antennas Propag.,
AP-23: 534540, 1975.
LINEAR DYNAMICAL SYSTEMS, APPROXIMATION
15. G. J. Burke and A. J. Poggio, Numerical electromagnetic code
NEC-method of moments, Tech. Document 116, Naval Ocean Systems Center, San Diego, CA, September 6, 1982.
16. IEEE Standard definition for terms for antennas (IEEE Std. 1451983), IEEE Trans. Antennas Propag., AP-17: 1969, AP-22: 1974,
AP-31: part II, 1983.
17. A. K. Bhattacharyya and D. L. Sengupta, Radar cross section
analysis and control, Norwood, MA: Artech House, 1991.
18. J. H. Richmond, Proc. IRE, 53: 796, 1965.
19. R. F. Harrington, Field Computation by Moment Methods, New
York: Macmillan, 1968.
20. J. C. Logan and J. W. Rockway, The new MININEC Version 3: A
mininumerical electromagnetic code, Tech. Document 93B, Naval
Ocean Systems Center, San Diego, CA, September 1986.
21. A. J. Julian, J. C. Logan, and J. W. Rockway, MININEC: A mininumerical electromagnetics code, NOSC TD 516, September 1982.
22. Windows Version of MININEC, EM Scientific, Available E-mail:
emsi@aol.com
23. Windows versions of NEC2 and NEC4, Nitany-Scientific Available [Online]. www.nittany-scientific.com
24. A. R. Djordjevic et al., AWAS for Windows: Analysis of Wire Antennas and ScatterersSoftware and Users Manual, Norwood, MA:
Artech House, 1995.
25. S. Prasad and R. W. P. King, Experimental Study of Inverted
L-, T- and Related Transmission Line Antenna, J. Res. Natl. Bur.
Standards-D Radio Propag., 65D (5): 1961.
26. E. H. Newman, A users manual for electromagnetic surface patch
code ESP, OSUESL, Department Electr. Eng., Rep. 713602-1,
prepared under Contract DAAG-39-81-K-0020 for the Department of the Army, US Army Res. Office, Res. Triangle Park, NC,
July 1981.
27. J. H. Richmond, Radiation and scattering by thin-wire structures
in the complex frequency domain, OSU Res. Foundation Rep., Rep.
No. RF-2902-10, 1973.
28. S. A. Schelkmoff and H. T. Fris, Antennas: Theory and Practice,
New York: Wiley, 1952, pp. 229244, pp. 351353.
29. J. H. Richmond and E. H. Newman, Dielectric coated wire antennas, Radio Science, 11(1): 1320, 1976.
30. J. Y. P. Lee and K. G. Balmain, Wire antennas coated with magnetically and electrically lossy material, Radio Science, 14 (3):
437445, 1979.
ASOKE K. BHATTACHARYYA
Lincoln University
LINEAR CIRCUITS, SENSITIVITY AND ACCURACY. See SENSITIVITY ANALYSIS.
403
LOOP ANTENNAS
597
LOOP ANTENNAS
The IEEE Standard Definitions of Terms for Antennas (see
Ref. 1) defines the loop antenna as an antenna whose configuration is that of a loop, further noting that if the current
in the loop, or in the multiple parallel turns of the loop, is
essentially uniform and the loop circumference is small compared with the wavelength, the radiation pattern approximates that of a magnetic dipole. That definition and the further note imply the two basic realms of loop antennas:
electrically small and electrically large structures.
There are more than 200 million loop antennas currently
used by subscribers of personal communications devices, primarily pagers [see Ref. (2)]. More than a million a month are
currently being manufactured. Furthermore, loops have appeared as transmitting arrays, such as the massive multielement loop array at shortwave station, call sign HCJB, in
Quito, Ecuador, and as fractional wavelength-size tunable
high-frequency transmitting antennas. The loop is indeed an
important and pervasive communications antenna.
The following analysis of loop antennas reveals that the
loop, when small compared with a wavelength, exhibits a radiation resistance proportional to the square of the enclosed
area. Extremely low values of radiation resistance are encountered for such loops, and extreme care must be taken to
effect efficient antenna designs. Furthermore, when the small
loop is implemented as a transmitting resonant circuit, surprisingly high voltages can exist across the resonating capacitor even for modest applied transmitter power levels. The
wave impedance in the immediate vicinity of the loop is low
but at further distances (up to 2 wavelengths) exceeds the
intrinsic free-space impedance before approaching that value.
A loop analysis is summarized, which applies to loops of
arbitrary circular diameter and of arbitrary wire thickness.
The analysis leads to some detail regarding the current density in the cross section of the wire. Loops of shapes other
J. Webster (ed.), Wiley Encyclopedia of Electrical and Electronics Engineering. Copyright # 1999 John Wiley & Sons, Inc.
598
LOOP ANTENNAS
than circular are less easily analyzed, and are best handled
by numerical methods such as moment method described in
Ref. 3.
Loops are the antennas of choice in pager receivers and
appear as both ferrite-loaded loops and as single-turn rectangular shaped structures within the radio housing. Body-worn
loops benefit from a field enhancement due to the resonant
behavior of the human body with respect to vertically polarized waves. In the high-frequency bands, the loop is used as
a series resonant circuit fed by a secondary loop. The structure can be tuned over a very large frequency band while
maintaining a relatively-constant-feed point impedance.
Large loop arrays of one-wavelength-perimeter square loops
have been successfully implemented as high-gain transmitting structures at high-power shortwave stations.
The dipole uniform current I flowing over an elemental length
h is the dual of a magnetic current MzS Ih and the surface
area is S h/k. The fields due to the infinitesimal loop are
then found from the vector and scalar potentials.
Vector and Scalar Potentials. The wave equation, in the form
of the inhomogeneous Helmholtz equation, is used here with
most of the underlying vector arithmetic omitted; see Refs. 10
to 12 for more details. For a magnetic current element source,
the electric displacement D is always solenoidal (the field
lines do not originate or terminate on sources), that is, in the
absence of source charges the divergence is zero,
D = 0
(1)
and the electric displacement field can be represented by the
curl of an arbitrary vector F,
ANALYSIS OF LOOP ANTENNAS
Loop antennas, particularly circular loops, were among the
first radiating structures analyzed, beginning as early as
1897 with Pocklingtons analysis of a thin wire loop excited
by a plane wave (4). Later, Hallen (5) and Storer (6) studied
driven loops. All these authors used a Fourier expansion of
the loop current, and the latter two authors discovered numerical difficulties with the approach. The difficulties could
be avoided, as pointed out by Wu (7), by integrating the
Greens function over the toroidal surface of the wire. The
present author coauthored an improved theory (8,9) that specifically takes into account the finite dimension of the loop
wire and extends the validity of the solution to thicker wires
than previously considered. Additionally, the work revealed
some detail of the loop current around the loop cross section.
Arbitrarily shaped loops, such as triangular loops and square
loops, as well as loop arrays can be conveniently analyzed using numerical methods, such as by the moment method (3).
The Infinitesimal Loop Antenna
The infinitesimal single turn current loop consists of a circulating current I enclosing an infinitesimal surface area S, and
is solved by analogy to the infinitesimal dipole. The fields of
an elementary loop element of radius b can be written in
terms of the loop enclosed area, S b2, and a constant excitation current I (when I is rms, then the fields are also rms
quantities). The fields are near in the sense that the distance parameter r is far smaller than the wavelength but far
larger than the loop dimension 2b. Hence, this is not the close
near-field region. The term kIS is often called the loop moment and is analogous to the similar term Ih associated with
the dipole moment. The infinitesimally small loop is pictured
in Fig. 1(a) next to its elementary dipole analog [Fig. 1(b)].
D = 0 E = F
(2)
where F is the vector potential and obeys the vector identity
F 0. Using Amperes law in the absence of electric
sources
H = j0 E
(3)
and with the vector identity () 0, where represents an arbitrary scalar function of position, it follows that
jF
F
H =
(4)
and for a homogeneous medium, after some manipulation we
get
F + j0 0 )
2F + k2F = 0M + (
(5)
where k is the wave number and k2 200. Although Eq.
(2) defines the curl of F, the divergence of F can be independently defined and the Lorentz condition is chosen:
F
j0 0 =
(6)
We define 2 as the Laplacian operator
2 =
2
2
2
+ 2 + 2
2
x
y
z
(7)
Substituting the simplification of Eq. (6) into Eq. (5) leads to
the inhomogeneous Helmholtz equation
2F + k2F = 0 M
(8)
Similarly, by using Eqs. (6) and (4) it is seen that
2b
h
S
I
(a)
(b)
Figure 1. Small-antenna geometry showing (a) the parameters of
the infinitesimal loop moment, and (b) its elementary dipole dual.
[Source: Siwiak (2).]
2 + k2 = 0
(9)
Using Eq. (4) and the Lorentz condition of Eq. (6) we can find
the electric field solely in terms of the vector potential F. The
utility of that definition becomes apparent when we consider
a magnetic current source aligned along a single vector direction, for example, M zMz for which the vector potential is
F zFz, where z is the unit vector aligned with the z axis,
and Eq. (8) becomes a scalar equation.
LOOP ANTENNAS
Radiation from a Magnetic Current Element. The solution to
the wave equation, Eq. (8), presented here, with the details
suppressed, is a spherical wave. The results are used to derive the radiation properties of the infinitesimal current loop
as the dual of the infinitesimal current element. The infinitesimal magnetic current element M zMz located at the origin satisfies a one-dimensional, hence scalar form of Eq. (8).
At points excluding the origin where the infinitesimal current
element is located, Eq. (8) is source-free and is written as a
function of radial distance r,
1
Fz (r)
r2
r2 r
r
+ k2 Fz (r) = 0
(10)
which can be reduced to
d 2 Fz (r) 2 dFz (r)
+ k2 Fz (r) = 0
+
dr2
r dr
(11)
Since Fz is a function of only the radial coordinate, the partial
derivative in Eq. (10) was replaced with the ordinary derivative. Equation (11) has a solution
Fz = C1
ejkr
r
(12)
There is a second solution in which the exponent of the phasor quantity is positive; however, we are interested here in
outward traveling waves so we discard that solution. In the
static case the phasor quantity is unity. The constant C1 is
related to the strength of the source current and is found by
integrating Eq. (8) over the volume including the source, giving
0
kIS
4
C1 =
(13)
and the solution for the vector potential is in the z unit vector
direction,
F =
ejkr
0
kIS
z
4
r
Hr =
kIS jkr 2
e k
2
j
1
+
(kr)2
(kr)3
cos( )
j
kIS jkr 2
1
1
e k +
H =
+
4
kr
(kr)2
(kr)3
E = 0
kIS jkr 2
e k
4
1
kr
j
(kr)2
The Wave Impedance of Loop Radiation. The wave impedance can be defined as the ratio of the total electric field divided by the total magnetic field. We can study the wave impedance of the loop fields by using Eqs. (15) to (17) for the
infinitesimal loop fields, along with their dual quantities for
the ideal electric dipole. Figure 2 shows the loop field wave
impedance as a function of distance kr from the loop along
the direction of maximum far-field radiation. The wave impedance for the elementary dipole is shown for comparison.
At distances near kr 1 the wave impedance of loop radiation
exceeds 0 376.73 , the intrinsic free-space impedance,
while that of the infinitesimal loop is below 376.73 . In this
region, the electric fields of the loop dominate.
The Radiation Regions of Loops. Inspection of Eqs. (15) to
(17) for the loop reveal a very complex field structure. There
are components of the fields that vary as the inverse third
power of distance r, inverse square of r, and the inverse of r.
In the near-field or induction region of the idealized infinitesimal loop, that is, for kr 1 (however, r kS for the loop and
r h for the dipole), the magnetic fields vary as the inverse
third power of distance.
The region in which kr is nearly unity is part of the radiating near field of the Fresnel zone. The inner boundary of that
(14)
which is an outward propagating spherical wave with increasing phase delay (increasingly negative phase) and with amplitude decreasing as the inverse of distance. We may now solve
for the magnetic fields of an infinitesimal current element by
inserting Eq. (14) into Eq. (4) with Eq. (6) and then for the
electric field by using Eq. (2). The fields, after sufficient manipulation, and for r kS (see Ref. 10), are
Equations (15) and (16) for the magnetic fields Hr and H
of the infinitesimal loop have exactly the same form as the
electric fields Er and E for the infinitesimal dipole, while Eq.
(17) for the electric field of the loop E has exactly the same
form as the magnetic field H of the dipole when the term kIS
of the loop expressions is replaced with Ih for the infinitesimal ideal (uniform current element) dipole. In the case for
which the loop moment kIS is superimposed on, and equals
the dipole moment Ih, the fields in all space will be circularly polarized.
Equations (15) to (17) describe a particularly complex field
behavior for what is a very idealized selection of sources: a
simple linear magnetic current M representing a current loop
I encompassing an infinitesimal surface S b2. Equations
(15) to (17) are valid only in the region sufficiently far (r
kS) from the region of the magnetic current source M.
sin( )
sin( )
(15)
10,000
Small dipole antenna
Wave impedance,
2 Fz (r) + k2 Fz (r) =
599
1,000
100
Small loop antenna
(16)
(17)
where 0 c0 376.730313 is the intrinsic free-space impedance, c is the velocity of propagation in free space (see Ref.
13 for definitions of constants), and I is the loop current.
10
0.1
10
kr
Figure 2. Small loop antenna and dipole antenna wave impedances
compared. [Source: Siwiak (2).]
600
LOOP ANTENNAS
zone is taken by Jordan and Balmain (12) to be r2
0.38D3 / , and the outer boundary is r 2D2 / , where D is the
largest dimension of the antenna, here equal to 2b. The outer
boundary criterion is based on a maximum phase error of /
8. There is a significant radial component of the field in the
Fresnel zone.
The far field or Fraunhofer zone is the region of the field
for which the angular radiation pattern is essentially independent of distance. That region is usually defined as extending from r 2D2 / to infinity, and the field amplitudes
there are essentially proportional to the inverse of distance
from the source. The far-zone behavior is identified with the
basic free-space propagation law.
The Induction Zone of Loops. We can study the induction
zone in comparison to the far field by considering induction
zone coupling, which was investigated by Hazeltine (14), and
which was applied to low-frequency radio receiver designs of
his time. Today the problem might be applied to the design of
a miniature radio module in which inductors must be oriented
for minimum coupling. The problem Hazeltine solved was one
of finding the geometric orientation for which two loops in
parallel planes have minimum coupling in the induction zone
of their near fields and serves to illustrate that the nearfield behavior differs fundamentally and significantly from
far-field behavior. To study the problem we invoke the principle of reciprocity (see Ref. 10), which states
Z
V
E b J a H b M a ) dV =
(E
Z
V
E a J b H a M b ) dV
(E
(18)
That is, the reaction on antenna (a) of sources (b) equals the
reaction on antenna (b) of sources (a). For two loops with loop
moments parallel to the z axis we want to find the angle for
which the coupling between the loops vanishes, that is, both
sides of Eq. (18) are zero. The reference geometry is shown in
Fig. 3. In the case of the loop, there are no electric sources in
Eq. (18), so Ja Jb 0, and both Ma and Mb are aligned with
z, the unit vector parallel to the z axis. Retaining only the
inductive field components and clearing common constants
in Eqs. (15) and (17) are placed into (18). We require that
(Hrr H)z 0. Since r z sin() and z cos(), we
are left with 2 cos2() sin2() 0, for which 54.736.
When oriented as shown in Fig. 3, two loops parallel to the
xy plane whose centers are displaced by an angle of 54.736
with respect to the z axis will not couple in their near fields.
To be sure, the angle determined above is exactly correct for
Detector
Oscillator
Figure 4. A metal detector employs two loops initially oriented to
minimize coupling in their near fields.
infinitesimally small loops; however, that angle will be nominally the same for larger loops. Hazeltine (14) used this principle, placing the axes of the inductors in a common plane
each at an angle of 54.7 with respect to the normal form the
radio chassis, to minimize the coupling between the inductors.
The same principle can be exploited in the design of a
metal detector, as depicted in Fig. 4. The loop a is driven with
an audio frequency oscillator. Loop b, in a parallel plane and
displaced so that nominally 54.7, is connected to a detector that might comprise an audio amplifier that feeds a set of
headphones. Any conductive object near loop a will disrupt
the balance of the system and result in an increased coupling
between the two loops, thus indicating the presence of a conducting object near a.
The Intermediate- and Far-Field Zones of Loops. The loopcoupling problem provides us with a way to investigate the
intermediate- and far-field coupling by applying Eq. (18) with
Eqs. (15) and (16) for various loop separations kr. In the farfield region only the H term of the magnetic field survives,
and by inspection of Eq. (16), the minimum coupling occurs
for 0 or 180. Figure 5 compares the coupling (normalized
to their peak values) for loops in parallel planes whose fields
are given by Eqs. (15) to (17). Figure 5 shows the coupling as
a function of angle for an intermediate region (kr 2) and
for the far-field case (kr 1000) in comparison with the induction zone case (kr 0.001). The patterns are fundamentally and significantly different. The coupling null at 54.7
is clearly evident for the induction zone case kr 0.001 and
for which the (1/kr)3 terms dominate. Equally evident is the
far-field coupling null for parallel loops on a common axis
when the 1/kr terms dominate. The intermediate-zone cou-
0
kr = 0.001
kr = 2
90
270
kr = 1,000
b
y
a
x
Figure 3. Two small loops in parallel planes and with 54.736
will not couple in their near fields. [Source: Siwiak (2).]
180
(deg)
Figure 5. Normalized induction zone, intermediate zone, and far
zone coupling between loops in parallel planes. [Source: Siwiak (2).]
LOOP ANTENNAS
pling shows a transitional behavior in which all the terms in
kr are comparable.
The Directivity and Impedance of Small Loops. The directive
gain of the electrically small loop can be found from the farfield radially directed Poynting vector in ratio to the average
Poynting vector over the radian sphere:
D(, ) =
1
4
E H ) r|
|(E
Pd =
(19)
Qloop
1.5I 2 Rr
= H2 0 =
4r2
kS k 2
4 r
(20)
for radiated power I 2Rr, hence, we can solve for the radiation
resistance:
Rr =
the Q, the quality factor defined in (2), is inversely proportional to the third power of the loop radius, a result that is
consistent with the fundamental limit behavior for small antennas.
If we use Eq. (22) and ignore the dipole mode terms and
second-order terms in a/b, the unloaded Q of the loop antenna is
E H ) r | sin( ) d d
|(E
Only the component of H and the component of E survive
into the far field. If we use Eq. (16) for H and Eq. (17) for
E and retain only the 1/kr terms, Eq. (19) yields D 1.5
sin2() by noting that the functional form of the product of E
and H is simply sin2() and by carrying out the simple integration in the denominator of Eq. (19).
Taking into account the directive gain, the far-field power
density Pd in the peak of the pattern is
(k2 S)2
0 = 0 (kb)4
6
6
(21)
for the infinitesimal loop of loop radius b.
When fed by a gap, there is a dipole moment that adds
terms not only to the impedance of the loop but also to the
close near fields. For the geometry shown in Fig. 6, and using
the analysis of King and Harrison (15), the electrically small
loop, having a diameter 2b and wire diameter 2a, exhibits a
feed point impedance given by
6
8b
ln
2
a
=
(kb)3
(23)
which for b/a 6 becomes
Qloop =
3.6
(kb)3
(24)
which has the proper limiting behavior for small loop radius.
The Q of the small loop given by Eq. (23) is indeed larger than
the minimum possible Qmin (kb)3 predicted by Siwiak (2)
for a structure of its size. It must be emphasized that the
actual Q of such an antenna will be smaller than given by Eq.
(24) due to unavoidable dissipative losses not represented in
Eqs. (22) to (24). We can approach the minimum Q but never
go smaller, except by introducing dissipative losses.
The Gap-Fed Loop
The analysis of arbitrarily thick wire loops follows the method
in Ref. 8, shown in simplified form in Ref. 9 and summarized
here. The toroid geometry of the loop is expressed in cylindrical coordinates , , and z with the toroid located symmetrically in the z 0 plane. The relevant geometry is shown in
Fig. 6.
Loop Surface Current Density. The current density on the
surface of the toroidal surface of the loop is given by
XX
J =
Zloop = 0 (kb)4 [1 + 8(kb)2] 1 2
6
b
8b
2
2 + (kb)2 [1 + 2(kb)2]
+ j0 kb ln
a
3
2
601
Bn, p e jn Fp
(25)
n= p=
(22)
including dipole mode terms valid for kb 0.1. The leading
term of Eq. (22) is the same as derived in Eq. (21) for the
infinitesimal loop. Expression (22) adds the detail of terms
considering the dipole moment of the gap fed loop as well as
refinements for loop wire radius a. The small loop antenna is
characterized by a radiation resistance that is proportional to
the fourth power of the loop radius b. The reactance is inductive, hence, is proportional to the loop radius. It follows that
where the functions Fp are symmetrical about the z axis and
are simple functions of cos(n), where is in the cross section
of the wire as shown in Fig. 6 and is related to the cylindrical
coordinate by z a sin(). These functions are orthonormalized over the conductor surface using the GramSchmidt
method described in Ref. 16, yielding
F0 =
and
F1 = F0
2b
2 ab
a
2
cos( )
2
2b
1 (a/2b)
(26)
(27)
The higher-order functions are lengthy but simple functions
of sin(p) and cos(p).
2a
Figure 6. Parameters of the thick-wire loop. [Source: Siwiak (2).]
Scalar and Vector Potentials. The electric field is obtained
from the vector and scalar potentials
jA
A
E =
(28)
LOOP ANTENNAS
The boundary conditions require that E , E, and E are zero
on the surface of the loop everywhere except at the feed gap
. Because this analysis will be limited to wire diameters
significantly smaller than a wavelength, the boundary conditions on E and E will not be enforced. In the gap E
V0 /2, where V0 is the gap excitation voltage.
The components of the vector potential are simply
A =
1
4
Z Z
J cos( ) dS
(29)
1000
Radiation resistance,
602
and
A =
1
4
Z Z
J sin( ) dS
(30)
= 10
= 15
800
600
400
200
0.5
1.5
2.5
kb
and the vector potential is
=
j0
4k
Z Z
S
Figure 7. Loop radiation resistance.
1 J
G dS
(31)
given by
where the value of dS [b a sin()]a d. The Greens function G is expressed in terms of cylindrical waves to match the
rotational symmetry of the loop,
1
G=
2j
jm( )
m=
Jm (1
(2)
v)Hm
(2
v)e
j (zz )
d
Hz =
(32)
where
v=
k2 + 2
1 = a cos( )
2 = + a cos( )
and where Jm() and H(2)
m () are the Bessel and Hankel functions.
Matching the Boundary Conditions. Expression (32) is now
inserted into Eqs. (29) to (31) and the electric field is then
found from Eq. (28) and the boundary condition is enforced.
For constant on the wire
E e j p d =
V0 sin(p)
p
H =
A
z
(34)
H =
A
z
(35)
A
A
1 A
+
(36)
The loop current across a section of the wire is found by integrating the function J in Eq. (25) around the wire cross section. The loop radiation impedance is then the applied voltage V0 in the gap divided by the current in the gap. Figure 7
shows the loop feed radiation resistance, and Fig. 8 shows the
corresponding loop reactance, as a function of loop radius kr
for a thin wire, 15, and a thick wire, 10, where
2 ln(2b/a) is Storers parameter (6). The thin-wire loop has
very sharp resonant behavior compared with the thick-wire
loop, especially for a half-wavelength diameter (kb 0.5)
structure. The higher resonances are less pronounced for both
loops. Thick-wire loops exhibit an interesting behavior in that
over a diameter of about a half wavelength, the reactance is
1500
(33)
= 10
= 15
This condition is enforced on the wire as many times as there
are harmonics in . Truncating the index p as described in
Ref. 9 to a small finite number P, we force E 0 except in
the feeding gap along the lines of constant on the surface of
the toroid. If we truncate to P, the number of harmonics Fp in
, and to M the number of harmonics in , we find the radiation current by solving M systems of P by P algebraic equations in Bm, p. In Ref. 9, P 2 and M in the several hundreds
was found to be a reasonable computational task that led to
useful solutions.
Reactance,
1000
500
0
500
1000
1000
Loop Fields and Impedance. With the harmonic amplitudes
Bm, p known, the current density is found from Eq.(1). The electric field is found next from Eq. (2) and the magnetic field is
0.5
1.5
kb
Figure 8. Loop reactance.
2.5
LOOP ANTENNAS
Table 1. Parameter Y for Various Loop Thicknesses and
b 0.01 Wavelengths
19.899
17.491
15.294
12.886
10.689
8.2809
a/
0.000003
0.00001
0.00003
0.0001
0.0003
0.001
0.0039
0.0090
0.020
0.048
0.098
0.179
603
numerical codes, such as the numerical electromagnetic code
(NEC) described in Ref. 3, and often used in the numerical
analysis of wire antenna structures.
When the small loop is used as an untuned and unshielded
field probe, the current induced in the loop will have a component due to the magnetic field normal to the loop plane as
well as a component due to the electric field in the plane of
the loop. A measure of E field to H field sensitivity is apparent
from Eq. (40). The electric field to magnetic field sensitivity
ratio of a simple small-loop probe is proportional to the loop
diameter. The small gap-fed loop, then, has a dipole moment,
which complicates its use as a purely magnetic field probe.
essentially always capacitive and the total impedance remains well behaved.
LOOP APPLICATIONS
Small Gap-Fed Loops. The detailed analysis of the thick,
gap-fed wire loop, as shown in Refs. 8 and 9, reveals that the
current density around the circumference of the wire, angle
in Fig. 6, is not constant. An approximation to the current
density along the wire circumference for a small diameter
loop is
J =
I
[1 2 cos()(kb)2][1 + Y cos( )]
2a
(37)
where I is the loop current, which has cosine variation along
the loop circumference, and where the variation around the
wire circumference is shown as a function of the angle . Y is
the ratio of the first- to the zero-order mode in and is not a
simple function of loop dimensions a and b, but can be found
numerically [Siwiak (2)] and from the analysis of the preceding section. For the small loop Y is negative and of order
a/b so Eq. (37) predicts that there is current bunching along
the inner contour ( 180) of the wire loop. Table 1 gives
representative values for Y as a function of a/b.
This increased current density results in a corresponding
increase in dissipative losses in the small loop. We can infer
that the cross-sectional shape of the conductor formed into a
loop antenna will impact the loss performance in a small loop.
The small loop fed with a voltage gap has a charge accumulation at the gap and will exhibit a close near electric field.
For a small loop of radius b and in the xy plane, the fields
at (x, y) (0, 0) are derived in Ref. 9 and given here as
E = j
0 kI
2
(38)
where I is the loop current and
Hz =
I
2b
The Ferrite-Loaded Loop AntennaA Magnetic Dipole
Let us examine a small ferrite-loaded loop antenna with dimensions 2h 2.4 cm, 2a 0.4 cm, and at a wavelength of
about 8.6 m as pictured in Fig. 9. When the permeability
of the ferrite is sufficiently high, this antenna behaves like a
magnetic dipole. The magnetic fields are strongly confined to
the magnetic medium, especially near the midsection of the
ferrite rod, and behave as the dual of the electric dipole excited by a triangular current distribution. We can therefore
analyze its behavior using a small dipole analysis shown by
Siwiak (2). The dipole current is replaced by the equivalent
magnetic current along the ferrite rod length 2h.
The impedance at the midpoint of a short dipole having a
current uniformly decreasing from the feed point across its
length 2h is
2h
0
ln
1
2
a
Zdipole = 0 (kh)2 j
6
kh
(41)
(39)
Expression (39) is recognized as the classic expression for the
static magnetic field within a single-turn solenoid. Note that
the electric field given by Eq. (38) does not depend on any
loop dimensions, but was derived for an electrically small
loop. The wave impedance Zw at the origin is the ratio of E
to Hz and from Eqs. (38) and (39) is
Zw = j0 kb
Loop antennas appear in pager receivers as both ferriteloaded loops and as single-turn rectangular shaped structures
within the radio housing. When worn on a belt the loop benefits from coupling to the vertically resonant human body. In
the high-frequency bands, the loop has been implemented as
a series resonant circuit fed by a secondary loop. The structure can be tuned over a very large frequency band while
maintaining a relatively-constant-feed point impedance. Onewavelength-perimeter square loops have been successfully
implemented as high-gain transmitting structures.
2a
2h
(40)
In addition to providing insight into the behavior of loop
probes, Eqs. (38) to (40) are useful in testing the results of
>> 1
Figure 9. A ferrite-loaded loop antenna. [Source: Siwiak (2).]
604
LOOP ANTENNAS
The corresponding unloaded Q of the dipole antenna is
3 ln
Qdipole =
2h
1
a
(kh)3
15
12
9
Equation (42) has the expected behavior of the inverse third
power with size for small antennas, and for h/a 6
Qdipole =
(43)
0
10 log(QA /Q) = 40 dB
(44)
and the actual resultant 3 dB bandwidth is about 700 kHz.
Such an antenna is typical of the type that would be used in
a body-mounted paging receiver application. As detailed by
Siwiak (2), the body exhibits an average magnetic field enhancement of about 6 dB at this frequency, so the average
belt-mounted antenna gain is 34 dBi. This is typical of a
front position body-mounted paging or personal communication receiver performance in this frequency range.
Body Enhancement in Body-Worn Loop Antennas
Loops are often implemented as internal antennas in pager
receiver applications spanning the frequency bands from 30
MHz to 960 MHz. Pagers are often worn at belt level and
Table 2. Paging Receiver Performance Using Loops
30 to 50
85
160
280 to 300
460
800 to 960
Source: After Siwiak (2).
6
3
4.5
(kh)3
Comparing the Q for a small dipole given by Eq. (43) with the
Q of a small loop of Eq. (24) we see that the loop Q is small
even though the same ratio of antenna dimension to wire radius was used. We conclude that the small loop utilizes the
smallest sphere that encloses it more efficiently than does the
small dipole. Indeed, the thin dipole, here masquerading as
the analog of a long thin ferrite loaded loop, is essentially a
one-dimensional structure, while the small loop is essentially
a two-dimensional structure.
We can use Eqs. (41) and (42) for the elementary dipole to
examine the ferrite-loaded loop antenna since it resembles a
magnetic dipole. The minimum ideal Q of this antenna is
given by Eq. (42), 1.01 106. The corresponding bandwidth
of such an antenna having no dissipative losses would be 2
35 MHz f/Q 70 MHz/1.01 106 69 Hz. A practical ferrite
antenna at this frequency has an actual unloaded QA of
nearer to 100, as can be inferred from the performance of beltmounted radios shown in Table 2. Hence, an estimate of the
actual antenna efficiency is
Frequency Band
(MHz)
Vertical
Horizontal
(42)
dB
Paging Receiver,
at Belt,
Av. Gain
(dBi)
Field Strength
Sensitivity
(dB V/m)
32 to 37
26
19 to 23
16
12
9
12 to 17
13
10 to 14
10
12
18 to 28
3
10
100
1,000
Frequency (MHz)
10,000
Figure 10. Gain-averaged body-enhanced loop response. [Source:
Siwiak (2).]
benefit from the body enhancement effect. The standing adult
human body resembles a lossy wire antenna that resonates
in the range of 40 MHz to 80 MHz. The frequency response,
as seen in Fig. 10, is broad, and for belt-mounted loop antennas polarized in the body axis direction, enhances the loopantenna azimuth-averaged gain at frequencies below about
500 MHz.
The far-field radiation pattern of a body-worn receiver is
nearly omnidirectional at very low frequency. As frequency is
increased, the pattern behind the body develops a shadow
that is manifest as a deepening null with increasing frequency. In the high-frequency limit, there is only a forward
lobe with the back half-space essentially completely blocked
by the body. For horizontal incident polarization there is no
longitudinal body resonance and there is only slight enhancement above 100 MHz.
The Small Resonated High-Frequency Loop Antenna
The simple loop may be resonated with a series capacitor having a magnitude of reactance equal to the loop reactance, and
indeed is effectively implemented that way for use in the
high-frequency (HF) bands as discovered by Dunlavy (17).
When fed by a second untuned loop, this antenna will exhibit
a nearly constant-feed-point impedance over a 3 : 1 or 4 : 1
bandwidth by simply adjusting the capacitor to the desired
resonant frequency. The reactive part of the loop impedance
is inductive, where the inductance is given by ImZL L,
so ignoring the higher-order terms
8b
0 kb ln
2
a
L=
(45)
which with the substitution 0k/ 0 becomes
8b
L = 0 b ln
2
a
(46)
The capacitance required to resonate this small loop at frequency f is
C = 1/(2f )2 L
(47)
LOOP ANTENNAS
The loop may be coupled to a radio circuit in many different
ways, including methods given in Refs. 17 and 18. When used
in transmitter applications, the small loop antenna is capable
of impressing a substantial voltage across the resonating capacitor. For a power P delivered to a small loop with an unloaded Q of Eq. (23) and with resonating the reactance XC
given by the reactive part of Eq. (22), it is easy to show that
the peak voltage across the resonating capacitor is
/4
0.14 to 0.25
/4
XC QP
(48)
2 Irms XC
(49)
Vp =
605
by recognizing that
Vp =
where Irms is the total rms loop current
Irms =
Reflector
element
Feed
Driven
element
Stub
tuner
Figure 11. Two-element loop array.
P
Re{Zloop}
(50)
along with Q at the resonant frequency in Eq. (23).
Transmitter power levels as low as 1 W delivered to a moderately efficient small-diameter ( /100) loop can result in
peak values of several hundred volts across the resonating
capacitor. This is not intuitively expected: the small loop is
often viewed as a high current circuit, which is often described as a short-circuited ring. However, because it is usually implemented as a resonant circuit with a resonating capacitor, it can also be an extremely-high-voltage circuit as will
be shown later. Care must be exercised in selecting the voltage rating of the resonating capacitor even for modest transmitting power levels, just as care must be taken to keep resistive losses low in the loop structure.
As an example, consider the Q and bandwidth of a small
loop antenna, 2b 10 cm in diameter, resonated by a series
capacitor and operating at 30 MHz. The example loop is constructed of 2a 1 cm diameter copper rod with conductivity
5.7 107 S/m. The resistance per unit length of round
wire of diameter 2a with conductivity is
1
1
=
Rs =
2as
2a
element tips severe enough to damage the antenna when operated at high power levels (10 kW) in a high-altitude (10,000
ft) shortwave broadcasting application in the 25 m band.
Moore sought an antenna design with no tips that would
support extremely high electric field strengths that caused
the destructive arcing. His solution was a one-wavelength-perimeter square loop, later with a loop director element as
shown in Fig. 11. The configuration exhibited no arcing tendencies, and a new shortwave antenna configuration was born.
As pictured in Fig. 11, the driven element is approximately
one-quarter wavelength on an edge. Actually, resonance occurs when the antenna perimeter is about 3% greater than a
wavelength. The reflector element perimeter is approximately
6% larger than a wavelength, and may be implemented with
a stub tuning arrangement. Typical element spacing is between 0.14 and 0.25 wavelengths. The directivity of a quad
loop is approximately 2 dB greater than that of Yagi antennas
with the same element spacing.
BIBLIOGRAPHY
(51)
1. IEEE Standard Definitions of Terms for Antennas, IEEE Standard
145-1993, SH16279, 1993.
where s is the skin depth for good conductors, is the radian
frequency, and 0 4 107 H/m is the permeability of free
space, so Rs 0.046 . From Eq. (22) the loop impedance is
Z 0.00792 71.41j. Hence the loop efficiency can be found
by comparing the loop radiation resistance with loss resistance. The loop efficiency is Rs /(Rs ReZ) 0.147 or 14.7%.
From Eqs. (46) and (47) we find the resonating capacitance
C 74.3 F. From Eqs. (48) to (50) we see that if 1 W is
supplied to the loop, the peak voltage across the resonating
capacitor is 308 V and the loop current is 4.3 A. The resonated
loop is by no means the low-impedance structure that we
normally imagine it to be.
2. K. Siwiak, Radiowave Propagation and Antennas for Personal
Communications, 2nd ed., Norwood, MA: Artech House, 1998.
The Quad Loop Antenna
The quad loop antenna, sometimes called the cubical quad,
was developed by Clarence C. Moore in the 1940s as a replacement for a four-element parasitic dipole array (Yagi
Uda array). The dipole array exhibited corona arcing at the
3. G. J. Burke and A. J. Poggio, Numerical electromagnetics code
(NEC)method of moments, Lawrence Livermore Laboratory,
NOSC Tech. Document 116 (TD 116), Vol. 1, 2, January 1981.
4. H. C. Pocklington, Electrical oscillations in wires, Proc. Cambridge Phys. Soc. London, 9: 324333, 1897.
5. E. Hallen, Theoretical investigation into transmitting and receiving qualities of antennae, Nova Acta Regiae Soc. Ser. Upps., Vol.
II, Nov. 4: 144, 1938.
6. J. E. Storer, Impedance of thin-wire loop antennas, Trans. AIEE,
75: 609619, 1965.
7. T. T. Wu, Theory of the thin circular antenna, J. Math. Phys., 3:
13011304, 1962.
8. Q. Balzano and K. Siwiak, The near field of annular antennas,
IEEE Trans. Veh. Technol., VT36: 173183, 1987.
9. Q. Balzano and K. Siwiak, Radiation of annular antennas, Correlations, Motorola Engineering Bulletin, Schaumburg, IL: Motorola Inc., Vol. VI, No. 2, Winter 1987.
606
LOSS-ANGLE MEASUREMENT
10. C. A. Balanis, Advanced Engineering Elecromagnetics, New York:
Wiley, 1989.
11. R. E. Collin, Antennas and Radiowave Propagation, New York:
McGraw-Hill, 1985.
12. E. C. Jordan and K. G. Balmain, Electromagnetic Waves and Radiating Systems, 2nd ed., Englewood Cliffs, NJ: Prentice-Hall,
1968.
13. R. Cohen and B. N. Taylor, The 1986 CODATA recommended
values of the fundamental physical constants, J. Res. Nat. Bur.
Stand., 92 (2), 1987.
14. L. A. Hazeltine, Means for eliminating magnetic coupling between coils, US Patent No. 1,577,421, 1926.
15. R. W. P. King and C. W. Harrison, Jr., Antennas and Waves: A
Modern Approach, Cambridge, MA: MIT Press, 1969.
16. R. Courant and D. Hibert, Methods of Mathematical Physics, New
York: Interscience Publishers, 1953.
17. J. H. Dunlavy, Jr., Wide range tunable transmitting loop, US
Patent No. 3,588,905, 28 June 1971.
18. T. Hart, Small, high-efficiency loop antennas, QST, 70 (6): 33
36, 1986.
KAZIMIERZ SIWIAK
Motorola, Inc.
LOOP PARALLELISM. See PROGRAM CONTROL STRUCTURES.
LORAN NAVIGATION. See RADIO NAVIGATION.
592
METEOR BURST COMMUNICATION
METEOR BURST COMMUNICATION
Billions of small meteors enter the earths atmosphere daily.
Upon entering the atmosphere, these meteors quickly vaporize, leaving behind a trail of ionized particles. Ionized particles reflect, or actually reradiate, radio signals; so if properly
oriented, a meteor trail can be used to establish a long-range
communication link between a pair of radios. Meteor trails
diffuse quickly, however, resulting in a rapid decay in signal
strength. In many systems, usable trail lifetimes are on the
order of 1 s. Communication that takes place via meteor trails
is known as meteor-burst communications.
The utility of meteor-burst communications is evidenced
by the snowpack telemetry system (SNOTEL) (1), which has
been in operation for over 15 years, as well as by the commercial success of companies such as Meteor Communications
Corporation and StarCom, which design meteor-burst communication systems. SNOTEL is a system that collects snowfall and weather data for 12 states in the western United
States. In addition to remote telemetry, applications of meteor-burst communications include vehicle tracking and twoway messaging (1). The U.S. military was one of the first proponents of meteor-burst systems because it was determined
that such systems would be among the first beyond-line-ofsight media to resume operability after a nuclear war.
The two primary alternatives for wireless communication
at ranges beyond 100 km are satellite communications and
terrestrial communication in the high-frequency (HF) band of
3 MHz to 30 MHz. For certain commercial applications, however, satellite communication is prohibitively expensive and
HF communication is too unreliable. With regard to military
applications, satellite and HF systems have additional disadvantages. For example, satellite and HF communication signals can be received over a large area and are therefore easy
to intercept. In addition, satellite and HF communication
links can be relatively easy to disrupt. These problems can be
overcome by meteor-burst systems, which provide relatively
low-cost, reliable, survivable, long-range communication.
PROPERTIES OF METEOR-BURST COMMUNICATION SYSTEMS
A typical meteor-burst protocol and network topology is as
follows. The meteor-burst terminals are arranged in a network with one master station and many remote stations. The
master station transmits a probe signal continuously using 1
kW to 5 kW of power. Once a meteor of appropriate size and
trajectory enters the atmosphere, the probe signal is reflected
from the meteor trail down to a remote station, which remains idle until the probe signal is detected. Once the probe
signal is detected, the remote station transmits a burst of digital data to the master station via the meteor trail. Once the
trail has diffused, the master station begins transmitting the
probe signal again, and the process repeats. This basic protocol is illustrated in Fig. 1.
Meteor trails are generally categorized as either underdense or overdense based on their electron line density (1).
For underdense trails, the received signal power decreases
J. Webster (ed.), Wiley Encyclopedia of Electrical and Electronics Engineering. Copyright # 1999 John Wiley & Sons, Inc.
METEOR BURST COMMUNICATION
Probe
signal
Master
station
Remote
station
Step 1: Probe signal is transmitted until
response from remote station is received.
Meteor
trail
593
nal decreases and the trail decay rate increases, so the operating frequency should be as small as possible according to
this argument. Important practical considerations favor
higher frequencies, however. For example, the desired signal
may be reflected by the ionosphere at frequencies below approximately 30 MHz, and these lower-frequency signals are
subject to higher atmospheric noise and are unreliable. Also,
antenna size and cost increase as the operating frequency is
decreased. As a result of the opposing criteria, the operating
frequency of most meteor-burst systems lies between 40 MHz
and 60 MHz. Several key characteristics of meteor-burst systems and channels are presented in Table 1.
THE METEOR-BURST CHANNEL
Master
station
Remote
station
Step 2: Probe signal is detected at remote station.
Master
station
Remote
station
Step 3: Confirming message is sent from remote
station to master station.
Figure 1. Basic meteor-burst communication protocol.
with time and is generally modeled as an exponential decay.
Underdense trails occur much more frequently than overdense trails, and the underdense trail model is used almost
exclusively in the analysis of system performance.
Underdense trails are formed by meteors whose mass is
between 105 g and 103 g. The number of meteors of a given
size that enter the atmosphere per day is inversely proportional to mass. For example, the number of earth-bound meteors per day that have a mass of 105 g is 1010, and the number
of earth-bound meteors per day that have a mass of 103 g is
108 (2), where all numbers are approximate.
The average data rate of a meteor-burst system is a function of the frequency of occurrence of useful meteor trails,
which has a daily and a seasonal variation. The maximum to
minimum daily variation is approximately 4 : 1, and the maximum to minimum seasonal variation is between 2 : 1 and 4 : 1
(1). A meteor-burst system is usually designed based on pessimistic assumptions about the frequency of useful meteor
trails.
The selection of the operating frequency for a meteor-burst
system is a compromise between opposing criteria. As the operating frequency increases, the strength of the received sig-
As a meteor enters and descends into the earths atmosphere,
it collides with increasing numbers of atmospheric particles.
The collisions convert kinetic energy into a combination of
heat, light, and ionization. As collisions continue, the outer
layers evaporate and the meteor loses velocity and breaks up.
Although portions of the very largest meteors retain solid
form all the way to the earths surface, the vast majority of
meteors completely evaporate long before this point. Because
the outer layers of the meteor are the first to lose velocity, the
result is a tail of ionized particles that forms behind the
meteor. It is the ionized particles that provide a communication medium, because free electrons in the meteor tail are capable of re-radiating a radio-frequency signal.
The geometry between the meteor tail and the transmit
and receive antennas plays a crucial role in the degree to
which the transmitted signal is received, if at all. Maximum
signal power is received if two things occur: (a) The tail forms
a tangent line to an ellipsoid for which the foci coincide with
the two antennas, and (b) substantial ionization occurs at and
around the point of tangency (3). Ionized particles all along
the tail contribute to the signal to some extent, but most important are the particles close to the point of tangency. These
particles add constructively to the received signal provided
that the distance from transmitter to meteor particle to receiver does not vary more than a half-wavelength from the
corresponding distance for the point of tangency.
The length of the region on the tail for which this constructive contribution occurs is proportional to the square root of
the wavelength of the signal. Because wavelength is equal to
the speed of light divided by the carrier frequency, it follows
that the length of the region decreases as the carrier frequency increases. A reduction in the length of the region corresponds to a reduction in the number of ionized particles
that contribute to the signal strength.
Table 1. Typical Meteor-Burst System and
Channel Parameters
Carrier frequency:
Transmitter power:
Communication range:
Trail height:
Typical application:
Trail lifetime:
Throughput:
4060 MHz
1005000 W
Up to 2000 km
80120 km
Vehicle tracking, environmental monitoring, two-way messaging
0.21.0 s
1000 bits/min
594
METEOR BURST COMMUNICATION
For those ionized particles for which the distance from
transmitter to particle to receiver is between a half-wavelength and a full wavelength different from the corresponding
distance for the point of tangency, destructive interference results. Differences in distance of additional multiples of a halfwavelength alternatively add constructively and destructively
to the signal.
When the meteor first enters the atmosphere, very little
ionization is present, and the received signal level is zero.
Fairly quickly, ionization begins, and while the tail is relatively short, most of the ionized particles add constructively.
The optimal time for a meteors usefulness typically occurs
when the meteor is at an altitude between 120 km and 80 km
(4). As the meteor continues to descend, the tail elongates and
a higher percentage of particles add destructively until the
numbers tend to cancel out and the received signal level again
drops to zero. The period of time for which a usable signal is
present is called a meteor burst.
The length of a meteor burst ranges from milliseconds to
seconds. The time duration is a function of many factors including the ionization density along the trail. If the density is
sufficiently low, the trail is termed underdense. Theoretical
analysis and experimental data show that underdense trails
provide a received signal level that very rapidly rises to a
peak value and then decays exponentially. Most trails are underdense. If the density is not low enough to result in the
exponentially shaped signal level, the trail is termed overdense. The received signal level of overdense trails typically
rises and decays more slowly, although eventually the trail
usually exhibits an exponential decay.
For longer bursts, multipath fading may occur. In terms of
the received signal, the result is multiple short dropouts of
the signal occurring before the burst ends. Multipath can be
caused by upper-atmosphere winds that bend the meteor trail
in such a way that multiple communication paths between
transmitter and receiver result. Because of their shorter duration, meteor bursts due to underdense trails typically are
less affected by multipath than overdense trails.
Between meteor bursts, communication is not possible, unless other beyond-line-of-sight propagation mechanisms occur. Additional propagation mechanisms seen to occur on meteor links include sporadic-E propagation, ionospheric scatter,
auroral scatter, diffraction, and troposcatter (3). Some of
these effects, when present, can provide much higher data
rates than can be obtained when propagation occurs solely by
the meteor-burst mechanism.
PERFORMANCE MODELING AND CHARACTERIZATION
One of the factors that affect the performance of a meteorburst communication system is the time waveform for the received signal power. For underdense trails, the received signal power is given by
P(t) = P0 exp
t0
where P0 is the peak received signal power and is a time
constant related to the decay rate. Both P0 and depend on
the trail characteristics and vary from trail to trail. Since
most trails are underdense, this model is often used in the
design and analysis of meteor-burst communication systems.
For overdense trails, the received signal power is more diffi-
cult to model. One model that does reasonably well for overdense trails has the form (5)
P(t) = P0 exp
1 at + b
2
ln
c
,
at + b
t0
where a, b, and c are constants that vary from trail to trail.
Unfortunately, a significant number of overdense trails are
not well-fit to this model. The general shape of P(t) for an
overdense trail is an initial rapid rise followed by a slower
rise, which is in turn followed by a slow decay and, finally, an
exponential decay. Although meteor bursts due to overdense
trails usually have a longer duration than those due to underdense trails, they may or may not have larger peak power.
With regard to underdense trails, because P0 varies from
trail to trail, it can be modeled as a random variable (6). Experimental studies indicate that P0 is reasonably well-modeled as a Gaussian random variable (3). The parameter can
also be modeled as a random variable, and experimental studies indicate that either a Rayleigh or a log-normal distribution provides a reasonably good fit (7). An apparently untested assumption in the literature is that P0 and are
statistically independent.
The number of meteor bursts in a given time window is
well-modeled as a Poisson random variable; that is, the probability that i bursts occur in T seconds is (aT)i exp(aT)/i!,
where a is a parameter that depends on the time of day, the
time of year, and the region of the sky to which the antennas
point. This model corresponds to a Poisson process (6), and it
follows that the time between meteor bursts is exponentially
distributed; that is, the probability that the time between
bursts is at least T seconds is exp(aT).
An additional important characterization of the performance of meteor-burst systems relates to the geographical region for which communication is possible using a particular
meteor trail. The footprint of a meteor trail can be defined as
the region on earth for which, at a particular time and a particular location of the transmitter, the received signal power
exceeds some specified threshold. This footprint might more
properly be termed the instantaneous footprint, because the
size and location of the footprint changes over the course of
the burst. As might be expected, the size grows and then
shrinks, and the location drifts as a result of the fact that the
meteor trail drifts with the winds in the upper atmosphere.
The trail footprint can be defined as the total accumulated
region in which the instantaneous footprint ever extends over
the course of the burst. The size of the trail footprint gives an
indication of the degree that the communication is secure. A
small footprint implies that the communication link has low
probability of detection (LPD).
The size of a footprint is significantly affected by the location of the trail relative to the locations of the transmitter
and receiver. Meteor trails close to the transmitter tend to
have larger footprints than trails close to the receiver (3).
This fact implies that the footprints for two radios transmitting back and forth to each other may be of different sizes.
CODING AND MODULATION FOR METEOR-BURST SYSTEMS
The nature of the meteor-burst channel implies that for all
trails, the received signal power is time-varying and the trails
are short-lived. Therefore, the capability to adapt transmis-
METEOR BURST COMMUNICATION
sion parameters, such as the ratio of data to redundancy in
an error-control coding system or the rate at which channel
symbols are transmitted, is desirable provided that the following two conditions are satisfied: (1) The channel conditions
can be estimated quickly and accurately, and (2) the cost of
adaptivity is not prohibitive.
Error-control coding, also referred to as forward error correction coding, must be used in many communication systems
to provide acceptable levels of performance. It is not clear at
the outset, however, that error-control coding will improve
the performance of a meteor-burst system. This is because the
use of error-control coding requires the inclusion of redundant
symbols in the encoded packet, and this in turn requires a
larger transmission time than an uncoded packet if the symbol transmission rate is fixed. The received signal strength
may be large during early portions of a trail, but the power
decays rapidly so there is a penalty for longer transmission
times. The following question therefore arises: Do the benefits
of error-control coding outweigh the penalty that results from
a longer transmission time? As shown in several research articles, the answer is yes. Several approaches to the use of
error-control coding in meteor-burst systems along with selected results are discussed below.
There are at least three approaches to the use of errorcontrol coding for meteor-burst communication systems. In
the first approach we consider, referred to as standard coding,
a fixed-rate code is used. In the second approach, which we
refer to as singly adaptive-rate coding, a fixed-rate code is
used for each trail but the rate of the code is allowed to vary
from trail to trail. In the third approach, which we refer to as
doubly adaptive-rate coding, the ratio of data to redundancy
is allowed to vary throughout a trail lifetime as well as from
trail to trail. In the singly and doubly adaptive-rate approaches, the channel characteristics must be measured in order to adapt the code rate correctly, but these approaches
have the potential to offer much higher throughput than the
standard coding approach.
In Refs. 8 and 9 the authors study the performance of systems that use standard coding with block codes and confirm
that systems that use error-control coding outperform systems that do not use coding. In Ref. 10 a study of doubly
adaptive-rate coding is presented for systems that use Reed
Solomon codes, and in Ref. 11 a doubly adaptive-rate system
that uses adaptive trellis-coded modulation is presented and
analyzed. The latter two articles demonstrate that a doubly
adaptive-rate coding can have a much larger throughput than
uncoded systems and standard coding systems. In Ref. 12 an
implementation of an adaptive-TCM system is presented, and
a standard coding system has been fielded (13) that uses a
rate convolutional code with Viterbi decoding.
Some form of automatic-repeat-request (ARQ) may be required in some applications, and ARQ may be used with any
of the three error-control-coding approaches described above.
Additional ways of using ARQ with error-control coding are
presented in Refs. 14 and 15. In Ref. 14 a system is investigated in which only part of the redundancy of the error-control code is transmitted initially along with the data. If errors
are detected, the transmitter is alerted and additional redundancy is transmitted thereby increasing the error-correcting
capability of the code that was sent previously. In Ref. 15 two
practical methods for implementing doubly adaptive-rate coding are investigated which use feedback from ARQ to determine when to change the code rate. Other work in ARQ (with-
595
out error-control coding) for meteor-burst communications
includes Ref. 16, in which three different ARQ protocols are
compared. The protocols are designed to study the relative
advantages of a simple stop-and-wait ARQ scheme as well as
the ability to detect the presence of a meteor channel. Symbol
interleaving can also be used effectively with error-control
coding (17). The idea is that interleaving can provide additional burst error correction capability. This is especially important near the end of the packet where the rapidly decreasing received signal power results in a large number of errors.
Adaptive-rate coding is one approach to adapting the
transmission parameters to the channel conditions. Another
approach is to adapt the rate at which channel symbols are
transmitted, and this approach is known as variable-rate signaling. The goal is to vary the signaling rate in direct proportion to the received signal power so that the received symbol
energy is maintained within a desired range. In Ref. 18 an
analysis of a system that uses variable-rate signaling is presented. Furthermore, at least two systems (19,20) that use
variable-rate signaling have been implemented. In the approach used for these systems, a feedback channel is maintained for each meteor trail that allows the receiver to communicate information about the new signaling rate to the
transmitter. (Note that the systems are designed to change
the signaling rate during lifetimes of each usable trail.) Demodulator outputs are used to estimate the current signal-tonoise ratio, and the signal-to-noise ratio estimate is used in
turn to determine the new signaling rate.
Finally, note that it is generally believed that channel disturbances, such as multipath propagation, are not severe
enough to make tracking of the carrier phase impossible. The
existence of several commercial systems (1) that use a carrierphase tracking system (e.g., a phase-locked loop) for the purposes of coherent demodulation supports this point. Therefore, the evidence suggests that coherent demodulation
should be used if the additional cost of a phase tracking device is not prohibitive and if the phase can be acquired in a
small period of time.
BIBLIOGRAPHY
1. J. Z. Schanker, Meteor Burst Communications, Norwood, MA: Artech House, 1990.
2. G. R. Sugar, Radio propagation by reflection from meteor trails,
Proc. IEEE, 1964, pp. 116136.
3. J. A. Weitzen, Meteor Scatter Communication: A New Understanding, in D. L. Schilling (ed.), Meteor Burst Communications:
Theory and Practice, New York: Wiley, 1993, pp. 958.
4. R. A. Desourdis, Jr., Modeling and Analysis of Meteor Burst
Communications, in D. L. Schilling (ed.), Meteor Burst Communications: Theory and Practice, New York: Wiley, 1993, pp. 59342.
5. C. O. Hines and P. A. Forsythe, The forward scattering of radio
waves from overdense meteor trails, Can. J. Phys., 35: 1033
1041, 1957.
6. A. Papoulis, Probability, Random Variables, and Stochastic Processes, 3rd ed. New York: McGraw-Hill, 1991.
7. J. A. Weitzen and W. T. Ralston, Meteor scatter: An overview,
IEEE Trans. Antennas Propag., AP-36: 18131819, 1988.
8. K. Brayer and S. Natarajan, An investigation of ARQ and hybrid
FEC-ARQ on an experimental high latitude meteor burst channel, IEEE Trans Commun., COM-37: 12391242, 1989.
596
METEOROLOGICAL RADAR
9. S. L. Miller and L. B. Milstein, Error correction coding for a meteor burst channel, IEEE Trans. Commun., COM-38: 1520
1529, 1990.
10. M. B. Pursley and S. D. Sandberg, Variable-rate coding for meteor-burst communications, IEEE Trans. Commun., COM-37:
11051112, 1989.
11. J. M. Jacobsmeyer, Adaptive trellis-coded modulation for bandlimited meteor-burst channels, IEEE J. Select. Areas Commun.,
10: 550561, 1992.
12. J. M. Jacobsmeyer, Adaptive data rate modem, U.S. Patent No.
5,541,955, 1996.
13. E. J. Morgan, Meteor burst communications: An update, Signal,
42: 5561, 1988.
14. M. B. Pursley and S. D. Sandberg, Incremental-redundancy
transmission for meteor-burst communications, IEEE Trans.
Commun., COM-39: 689702, 1991.
15. M. B. Pursley and S. D. Sandberg, Variable-rate hybrid ARQ for
meteor-burst communications, IEEE Trans. Commun., COM-40:
6073, 1992.
16. S. L. Miller and L. B. Milstein, A comparison of protocols for a
meteor-burst channel based on a time-varying channel model,
IEEE Trans. Commun., COM-37: 1830, 1989.
17. C. W. Baum and C. S. Wilkins, Erasure generation and interleaving for meteor-burst communications with fixed-rate and variable-rate coding, IEEE Trans. Commun., COM-45: 625628,
1997.
18. S. Davidovici and E. G. Kanterakis, Performance of Meteor Burst
Communication Using Variable Data Rates, in D. L. Schilling
(ed.), Meteor Burst Communications: Theory and Practice, New
York: Wiley, 1993, pp. 383410.
19. D. L. Schilling et al., The FAVR Meteor Burst Communication
Experiment, in D. L. Schilling (ed.), Meteor Burst Communications: Theory and Practice, New York: Wiley, 1993, pp. 367381.
20. D. K. Smith and T. G. Donich, Maximizing throughput under
changing channel conditions, Signal, 43 (10): 173178, 1989.
CARL W. BAUM
CLINT S. WILKINS
Clemson University
J. Webster (ed.), Wiley Encyclopedia of Electrical and Electronics Engineering
c 1999 John Wiley & Sons, Inc.
Copyright
MICROSTRIP ANTENNAS
Microstrip circuitry consists of a metal strip or patch on a dielectric substrate backed by a metal ground plane.
Microstrip antennas are becoming increasingly popular owing to their advantages in size, cost, conformity to
the supporting structure, low profile, and ease of fabrication. By using simple etching techniques, it is possible
to fabricate a wide variety of microstrip circuits, including arrays of antennas, feeding networks, and active
devices such that preamps or distributed transmitters can be conveniently placed next to the antenna elements.
In addition, diode phase shifter circuits can also be etched on the substrate to form single-board phased arrays.
This article describes the microwave properties of patch antennas.
The dimensions of microstrip patch antennas are large in comparison with the width of conventional
microstrip lines. The ma purpose of a patch antenna is to confine microwave energy to a small finite region.
Such confinement may be achieved with the resonant behavior of a finite guided structure supporting standing
modes. Electromagnetic energy radiates from that part of the patch that is open to free space, similar to a
slot cut along the side of a waveguide. To continuously radiate microwave energy away, the patch needs to be
electrically connected to a so-called feeder line. When the characteristic impedance of the feeder line matches
the impedance of radiation waves, power is directly dumped into the confined space of the patch without
causing much reflection. This means that all of the input power to the patch is radiated away into free space. A
conventional transmission line radiates little power, because the fringing fields alternate in sign over a short
distance, resulting in cancellation of their radiated field. In microstrip circuitry, power also radiates from open
circuits and from discontinuities, such as corners. The radiated power is small, however, because the radiation
impedances there are usually much higher than the characteristic impedance of the microstrip transmission
line encompassing these discontinuities.
This article introduces two calculation methods capable of describing antenna performance quantitatively.
The first method is less rigorous, but it has the advantage of being more analytical and hence can be applied
with ease. Called the resonant cavity model, it assumes perfect metal boundaries at the metal patch and
ground plane and magnetic-wall boundary conditions at the periphery, or sides of the patch antenna. It is
well known, that there are fringing fields at the periphery of the patch antenna. The influence of such fields
may be included in this approximate analysis by extending the linear dimension of the patch antenna by
a small amount consistent with the depth of the fringing fields. Radiation and material imperfections are
then considered as perturbations to the lossless cavity. As such, the resonant frequency, far-field pattern,
input impedance, radiation linewidth, and efficiency are all calculable. The only shortcoming of this simple
cavity model is that it is not able to address the possibility of the propagation of surface waves. Surface wave
generation becomes significant in comparison with patch radiation for a thick substrate with high dielectric
constant.
The second calculation method concerns Greens function analysis. In general, a Greens function is defined
as the solution of a differential or integral equation resulting from a (delta-function) point source satisfying the
required (hogeneous) boundary conditions. Thus, under arbitrary source excitation, the solutn of the equation
can be obtained by a superposition ofreens functions, still satisfying the boundary conditions. In the presence
of a point (dipole) current source in the background of a stratified structure consisting of dielectric or magnetic
1
MICROSTRIP ANTENNAS
layers, the excited electromagnetic field is termed a dyadic Greens function from the vectorial nature of the
source field and the observer field. Although the construction of the dyadic Greens function is straightforward,
this article pays special attention to its physical interpretations as it appeared in the original work by 1.
Actually, Sommerfeld solved for the first time the Greens functions associated with a horizontal point
dipole and a vertical point dipole in the presence of a semiinfinite conductor half-space: the Earth (1). Radio
waves are radiated into free space either directly from the point dipole source or indirectly from the image
dipole source induced by the (imperfect) ground plane, the Earths surface. These are called spatial waves,
because they exhibit a 1/r spatial dependence, where r denotes the distance from the original source point (or
image point) to the observer point. However, Sommerfeld also showed a second kind of radiation field, which
is tied to the airearth interface exhibiting a 1/ dependence, where denotes the 2-D distance between the
source point projected on the Earths surface to the observer point located also on the Earths surface (1). They
are called surface waves, because they decay exponentially when departing away from the airearth interface.
This article explains how the spatial wave solutions and the surface wave solutions are constructed from a
(horizontal) point dipole source in the presence of a layered dielectric or magnetic structure.
By using the dyadic Greens function, the general field solution in a stratified structure with external
current excitation can then be formulated in terms of an integral equation, which is subsequently solved by applying the numerical Galerkin method. This article outlines procedures to calculate engineering parameters of
general microstrip antennas and gives calculation results for circular microstrip antennas, including radiation
frequencies, bandwidths, far-field patterns, input impedances, effects of feeder-line position, and interference
between antennas. Recent developments on printed-circuit antennas are briefly reviewed, including the important broadband technique employing stacked parasitic elements to achieve high gain and low cross-polarization
levels. More sophisticated treatments on microstrip antennas may be found in Ref. 2.
Cavity Model
A microstrip patch antenna is a narrow-band device, and typically, the bandwidth covers approximately 5
percent of the radiation frequency. When the bandwidth, f , can be related to the Q of a resonator
it implies that power dissipation is not significant and the circuit of a microstrip patch antenna can be
approximated as a low-loss cavity resonator. This suggests that the performance of a microstrip patch antenna
can be analyzed using a perturbation method. That is, the zeroth order solution of the antenna is described
in terms of the eigen modes, or normal modes, of a lossless cavity. Losses are then added to the analysis as
first-order perturbations, including conductor loss, dielectric loss, and radiation loss. Magnetic loss can also be
included in the total loss if ferrites are used as the substrate material.
Analogous to radiation loss, surface wave loss may also be treated as a perturbation in the cavity model.
Radiation and surface wave losses may be represented as leaky waves from the antenna, the former directly
from air and the latter from the guided structure of the substrate. However, the analysis of surface-wave loss
may be difficult, because it involves Sommerfeld-type integrals where simple poles and branch cuts appear in
the complex k plane (1). Usually, surface waves are included only in a full-wave analysis in which Maxwell
equations are solved numerically in the microstrip geometry.
In contrast, the cavity model is easier to apply, allowing engineering parameters of a microstrip patch
antenna to be calculated analytically, including bandwidth, input impedance, radiation efficiency, and nearfield and far-field patterns. Furthermore, the physical meaning of normal modes is evident in the cavity
MICROSTRIP ANTENNAS
Fig. 1. Feeding the rectangular patch antenna by a microstrip line. The resonant cavity locates directly under the patch
bound by a magnetic wall, shown shaded in the figure.
model. For example, the left- or right-hand polarized radiation from a microstrip ferrite patch antenna can be
readily calculated by using the cavity model. The cavity model provides satisfactory quantitative answers if
surface wave losses are minimal. Surface wave losses are minimal for a thin substrate exhibiting low dielectric
constant, but not for a thick substrate with high dielectric constant. Also, for microstrip patches in proximity
to each other a full-wave analysis is required, since the cavity model is not sufficient to describe these complex
situations. The full-wave analysis involving the use of dyadic Greens functions is discussed later.
Resonant Frequency. Let a metal patch be deposited on top of a dielectric substrate backed by a ground
plane. The patch considered here is either of rectangular or circular geometry. Only dielectric substrates are
considered. Analysis of patch antennas with ferrite substrates may be found in Ref. 3. In the cavity model one
assumes a lossless substrate in which the cavity is bounded by an electric and a magnetic wall. The metal
surfaces of the microstrip patch and the ground plane may be approximated as perfect conductors or electric
walls, but the peripheral surface surrounding the antenna cavity directly under the patch is assumed to be a
magnetic wall. This is illustrated in Fig. 1, in which a microstrip rectangular patch antenna is represented by
a cavity with its periphery shown shaded as a magnetic wall. An electric or magnetic wall is defined such that
the tangential component of the electric or magnetic field vanishes at the wall boundary. Thus, a normal metal
boundary condition approaches that of an electric wall if the value of conductivity goes to infinity. A magnetic
wall imposes imaginary boundary conditions that insulate the inside of the cavity from the outside, allowing
no electromagnetic energy to propagate across it.
Under quasi-static assumptions a magnetic wall can be located at the periphery of a patch antenna,
provided that it encloses the total volume of the antenna plus its fringing fields. That is, the effective volume
of the cavity is slightly larger than the physical one, so that it accounts for both the electromagnetic energy
stored directly under the metal patch and its fringe. This is shown in Fig. 2, for the case of a circular microstrip
antenna, where the effective radius of the cavity is larger than that of the metal patch. Thus, the resonant
length at the side of a rectangular patch antenna, denoted as L = L + 2, is larger than the physical length,
MICROSTRIP ANTENNAS
Fig. 2. Feeding the circular patch antenna by a coax line. Due to the fringing-field effect the effective radius of the
resonator is larger than its physical value.
L, and the effective increment in length due to the fringing-field effect (4) is 2 with
where d is the thickness of the substrate, eff denotes the effective dielectric constant of the patch cavity (5)
given by
and r is the dielectric constant of the substrate material. The rationale behind Eq. (3) is that, depending on
the ratio of d/L, electric field lines can still propagate away from the cavity region, and so eff < r , and eff
r if d/L 1. The resonant frequency of the cavity resonator is therefore
where c denotes the speed of light in vacuum and n the order of the resonant mode (n = 1 for the fundamental
mode).
Based on the same fringing-field consideration, we expect that for a circular patch antenna the effective
resonant radius of the cavity, denoted as R = R + , is larger than the physical radius of the metal patch (6),
denoted as R, and
as shown in Fig. 2. A similar expression for eff as Eq. (3) is expected for a circular microstrip cavity. However,
such an expression is lacking in the present literature. We therefore use the value of r for eff , bearing in mind
MICROSTRIP ANTENNAS
that this approximation is true only when R d. The resonant frequency of the cavity is, therefore
where X mn is the nth zero of the derivative of the Bessel function J m (x) of order m.
We note that the above expressions for resonant frequencies for rectangular and circular microstrip patch
antennas [ Eqs. (4) and (6)] are derived based on the cavity model according to which standing modes are
excited within the cavity. At high frequencies the quasi-static assumption is no longer valid, and surface waves
of high orders are generated and propagating away from the cavity region guided by the dielectric substrate.
In this case, magnetic-wall boundary conditions are not appropriate at the periphery of the antenna cavity.
For these situations we must resort to a rigorous full-wave analysis to adequately address the leaky feature
of surface wavesfor example, by using the dyadic Greens functions to be discussed later. In the following we
use r for eff and L and R for L and R , respectively, with the difference understood.
Normal Modes and Feeder-Line Excitations. The normal modes excited in a microstrip patch antenna are usually transverse electric (TE) waves exhibiting no nodal points along the z axis, the direction
perpendicular to the substrate plane. Thus, the normal mode solutions show no z dependence, and the electromagnetic components can be uniquely solved from Ez , satisfying the following Helmholtz equation
where t denotes the transverse part of the del operator (with respect to the z axis), kmn is given by
mn is the angular frequency, and {m, n} refers to the index of the normal mode. The boundary condition
imposed on Ez is that the derivative of Ez along the normal direction of the boundary vanishes at the magnetic
wall. Once Ez is solved from Eq. (7), the corresponding H field is
where ez denotes the unit vector along the z axis.
Let the rectangular patch be located at 0 x L, and 0 y W, [see Fig. 1]. The normal modes therefore
are
with the corresponding modal wavenumber
For a circular cavity of radius R the normal modes are, [see Fig. 2]
MICROSTRIP ANTENNAS
and the modal wavenumber is determined by the zeroing condition of the derivative of Bessel functions of order
m
and (, , z) denotes the cylindrical coordinate.
Let us consider the patch antenna excited by either a microstrip line at the edge of the patch, Fig. 1.
or a coax feeder directly under the patch, Fig. 2. The angular frequency of the external driving field is =
2f . We expect that cavity to be driven with maximum intensity when the frequency of excitation approaches
the normal mode frequencies, Eq. (8). This phenomenon is generally known as forced oscillation, and it occurs
commonly in many branches of physics. Thus, at resonance, the cavity is driven by the external field undergoing
forced oscillation giving rise to maximum radiation efficiency. We first consider the excitation of the cavity by
a microstrip feeder line. The boundary conditions for the cavity are that electric walls are located at metal
boundaries and magnetic walls at the periphery which is not adjacent to the microstrip feeder line. At the
input junction between the cavity and the feeder line, we define a window where we assume that the excitation
currents are uniformly distributed, producing a uniform magnetic field there
Here n denotes the unit vector pointing outward along the normal direction of the window, and h0 is a
constant specifying the excitation current. In Fig. 1 the excitation-current window is located on the shad area
directly under the microstrip feeder line with n being coincident with ey . The excitation field governed by
Eqs. (7) and (9) can be uniquely solved subject to the boundary condition of Eq. (14).
For the rectangular patch depicted in Fig. 1, let the microstrip feeder connection to the patch at the line
segment a x b, and y = 0. The excitation field therefore is
where
MICROSTRIP ANTENNAS
and
Note that at = mn , Am = , indicating that when the frequency of the external driving field equals one
of the normal mode frequencies, only that normal mode will be excited in the cavity responsible for radiation.
However, when losses are included, other modes with much smaller amplitudes can also be excited at resonance.
For the circular patch we consider the microstrip feeder line to join the metal patch at the arc
, and = R. The excitation field can be then written as follows
where
and and k are given in Eq. (19). Again, from Eqs. (25) and (26) it is seen that the coefficient of Cn goes to
infinity if normal mode frequencies are approached [Eq. (13)].
Let us consider the excitation of the patch antenna by a coax line directly under the patch. We may assume
the coax line to possess an inner filament of zero diameter ending in a point charge at the junction with the
patch (see Fig. 2). The excitation current density therefore is
MICROSTRIP ANTENNAS
In the presence of a driving current it appears as an inhomogeneous term at the right-hand side of Eq.
(7):
The boundary conditions are the same as those leading to normal mode solutionsthat is, electric walls
at metal boundaries and magnetic walls at the periphery of the patch cavity. Thus, Eq. (29) can be solved in
terms of normal mode solutions:
Here Eqs. (30) and (31) are for a rectangular patch and a circular patch, respectively, and the normal
modes, Ez mn z (x, y) and Emn z (, ), are given in Eqs. (10) and (12). Equations (30) and (31) imply that when k
approaches kmn , the normal mode {m, n} acquires an infinite amplitude. The corresponding magnetic field can
be derived using Eq. (9).
Thus, for the rectangular patch, we have
MICROSTRIP ANTENNAS
For the circular patch we have
where ij is the Kronecker delta function and
In Eqs. (32) to (34) and (35) to (37), kmn is given by Eqs. (11) and (13), respectively, and and k are defined
in Eq. (19). Once the excitation fields are known, Eqs. (16) to (18), (22) to (24), (32) to (37), losses of various
kinds, and hence the quality factor, Q, of the patch cavity can be calculated, as discussed later.
Input Impedance. Having solved the electromagnetic fields inside a lossless patch cavity, we can
relax the assumption of perfect electric and magnetic walls and allow electromagnetic energy to propagate
across these boundaries. This results in ohmic loss and radiation (surface wave) loss. Dielectric and magnetic
losses occur in the interior of the patch cavity, assuming a lossy medium exhibiting complex permittivity and
permeability, respectively. We consider first the radiation loss.
The KirchhoffHuygens principle, which is a vector analog of Greens theorem, states that the electromagnetic field inside a closed volume, V, can be derived by the volume charge and current distributions inside
V and the surface charges and currents distributed on the enclosing surface of V, denoted as S. The effective
10
MICROSTRIP ANTENNAS
electric surface current density, K e , magnetic surface current density, K m , and electric surface charge density,
, are (7), respectively,
where n denotes the unit outward vector normal to S, and is the permittivity. Thus, if we consider the outside
volume of the patch cavity as the volume V, we conclude that the effective magnetic current density on the
magnetic wall is
where n denotes the unit outward vector normal to the magnetic wall, with sign opposite that is Eq. (38), and
the factor of two accounts for the presence of the ground plane. K e and
do not appear on a magnetic wall.
The radiation field arising from the magnetic current density K m can be derived by using the same
formula describing the electric current density K e converted from the duality rule (7). The duality rule states
that the electromagnetic theory remains valid if all of the elic quantities are changed into the corresponding
magnetic quantities and the magnetic quantities are changed into the negative of the corresponding electric
quantities. Thus, the vector potential associated with K m can be written as (8)
where d denotes the thickness of the substrate, Cm the contour of the magnetic wall with counterclockwise
tangential unit vector t, e r the unit vector along r, a vector connecting the coordinate origin toward the
observation point, and
is the wavenumber in air. In Eq. (40) primed quantities originate at the source and assume |r r | d.
The fields in the far-zone are
where
MICROSTRIP ANTENNAS
11
is the wave impedance in air (= 377 ). We therefore have
where e and e denote unit vectors along the and directions at the observation point r. By using Eq.
(45) the far-field radiation associated with a circular patch antenna fed by a microstrip line, Eq. (22), is, for
example,
where it is understood that the ratio of sin m to m is 1 when m = 0. The far-field radiation pattern can be
calculated by using the following equation:
and the total radiated power is then
Conductor loss and dielectric loss can be derived with ease (7). The conductor loss is given by
where Sc denotes the metal surface with Rs the surface resistance
and is the conductivity of metal. The dielectric loss is given by
where V c denotes the volume of the cavity and tan is the dielectric loss tangent.
12
MICROSTRIP ANTENNAS
The radiation or antenna efficiency is the ratio of radiated power to input power, or
The total stored electric energy is then
Since, by definition, at resonance the stored magnetic energy, W m , equals the stored electric energy, W e ,
we have for the total stored electromagnetic energy
From the definition of Eq. (1), the total Q factor of the cavity is
upon which the voltage standing wave ratio (VSWR) bandwidth can be defined. We define Smax to be the
maximum value of VSWR that can be tolerated. We then have (9)
Typically, Smax = 2.
The input susceptance B, of the antenna can be calculated from the lossless patch cavity, which is a
zeroth-order quantity. However, the input conductance, G, must be calculated from the total loss of the cavity,
which is a first-order quantity. Thus
where I and V are the averaged input current and voltage experienced at the input feeder position. For a coax
feed with excitation current expressed by Eqs. (27) and (28), assuming a thin inner filament as shown in Fig.
2, I is known and V can be calculated from Ez evaluated at the input position multiplied by the thickness of
the substrate d. We therefore have
MICROSTRIP ANTENNAS
13
for rectangular and circular patch antennas, respectively. For a microstrip feed, the input voltage is obtained
according to Faradays law by averaging Ez over the feeder-line window joining the cavity multiplied by the
thickness of the substrate, d. The input current can be derived by applying Amperes law at the feeder-line
window, which is ho times the transverse length of the window, that is, ho (b a) for the rectangular antenna
and ho (2R) for the circular antenna. We therefore have
for rectangular and circular patch antennas, respectively. Here kmn in Eq. (60) is given by Eq. (13) and m in
Eq. (61) is given by Eq. (18).
From Eqs. (59), (60) and (62), we note that at the normal mode frequencies the input susceptance B =
0, and hence B1 = . These points are called antiresonance points as plotted in the Smith chart (see Fig. 3,
example for). The resonant points are defined to be purely resistive when the input resistance of the antenna
matches the feeder-line impedance, resulting in zero reflection. Therefore, B 1 = 0 at resonance. The resonance
point occurs slightly above the antiresonance frequency, which requires participation of normal modes of all
orders, although the normal mode responsible for antiresonance is excited with the greatest intensity. Thus,
by definition, at resonance the capacitive part of the stored energy equals the inductive part, rendering the
overall input reactance zero. In calculating the susceptance, we should also include the capacitance associated
with near-field excitation. However, its contribution has been accounted for as the increment in the effective
resonance length of the antenna, [See Eqs. (2) and (5)]. The input impedance is
We note that, in reality, at antiresonance, B is finite due to the presence of losses occurring at the antenna
patch cavity.
One severe drawback of the cavity model is that it is not able to address the surface-wave loss. 10 estimated
that surface-wave excitation is not important if d/0 < 0.09 for r = 2.3 and d/0 < 0.03 for r = 10, where
0 is the free-space wavelength. The criterion given by 11 is more quantitative: d/0 < 0.07 for r = 2.3, and
d/0 < 0.0023 for r = 10, if the antenna is to emit no more than 25 percent of the total radiated power as surface
waves. Recent work by Fonseca et al. showed that the size of the patch is also a parameter (12), as discussed in
more detail in the next section. When ferrite material is used as the substrate, magnetic loss can be estimated
14
MICROSTRIP ANTENNAS
Fig. 3. Integration contours for Sommerfeld integrals illustrating contribution from spatial wave and surface wave
excitations.
using a formula similar to Eq. (51):
where denotes the imaginary part of the permeability. In concluding this section we note that the cavity model
depicts a semiempirical picture where the parameters have been adjusted to fit experiments, for example, the
effective dielectric constant eff , Eq. (3), and the increment in the resonant length of the patch, , Eqs. (2) and
(5). Nevertheless, the calculated radiation pattern and input impedance compare very well with measurements
(13).
Dyadic Greens Function
As we have mentioned in the previous section, the open structure of a microstrip patch antenna can be
rigorously accounted for only in a full-wave analysis resorting to numerical solutions. Maxwells equations
can be explicitly solved numerically in the frequency domain or in the time domain using the generic 3-D
finite-element and finite-difference methods (14). However, it is more informative to use the dyadic Greens
functions, since the electromagnetic fields generated by a point dipole-current source has already been solved
analytically in the same microstrip geometry, which is termed the dyadic Greens function. The electromagnetic
fields excited by a patch antenna can then be composed as superpositions of the point-dipole solutions in the
MICROSTRIP ANTENNAS
15
context of a conventional Greens function method. The numerical technique applies only when the Galerkin
method is used to solve the resultant integral equations relating the unknown current variables to local electric
fields distributed across the metal patch boundaries.
The physical meaning of the Greens function is clear, and surface waves have been used with equal
importance as spatial waves in the Greens function formalism. Furthermore, the Greens function solution
usually requires 2-D calculations, in contrast to the generic 3-D computational methods. This is true when the
metal thickness is much smaller than the thickness of the substrate, as is usually the case. Material losses can
be readily included in the Greens function if complex permittivity and permeability values are used. It turns
out that efficient CAD tools can be constructed using the dyadic Greens function solutions, which calculate
engineering parameters of a microstrip antenna, including radiation frequency, far-field pattern, efficiency,
input impedance, and so on and analyze the cross-talk problem inherent in common microstrip circuitries.
To illustrate the physics in the application of a Greens function, we have decided not to regenerate
many mathematical formulas in this section. Instead, we concentrate on Sommerfelds approach to the Greens
function analysis (1), because it lends itself to a more physical understanding of the problem. We wish to
introduce the methodology leading to the formulation of the dyadic Greens function for a general stratified
structure, consisting of a finite number of dielectric and magnetic layers as constituents. We assume this layered
structure is infinite in both horizontal and vertical directions, although it is possible to include finite substrate
and radiation space by employing periodic boundary conditions, for example. Also, we assume the current
distribution is two-dimensional, resulting in 2-D analysis on the Galerkin elements. The finite conductivity of
the ground plane can be accounted for by invoking complex permittivity of the conductor layer (1). Results of
calculations are cited mainly from Ref. 15. Background material can be found in Ref. 16.
Point-Dipole Solutions. We start by introducing the vector potential, A, and scalar potential, V, in
electrodynamics subject to a Lorentz gauge (17), which has been implicitly used by 1
This results in uncoupled equations for A and V as follows:
where J and are the current and charge densities, respectively, and normally = 0 (oscillating charges can
not easily be realized physically). We note that Eq. (68) does not specify A uniquely for a finite volume under
consideration (17). However, this gauge freedom is almost fixed for a system of infinite volume, because the only
source-free radiation for the entire space are the incoming waves from infinity, which can be readily checked
out and excluded from the solution of A by performing proper gauge transformation. From Eq. (67), the scalar
potential V is obtained from the divergence of A, and hence only the vector potential A needs to be solved.
We are now solving the vector potential, A(x, y, z), Eq. (68), induced by a point dipole current source, J 0 (x,
y, z), in the background of a dielectric or magnetic layered structure. A point dipole is also called a Hertzian
16
MICROSTRIP ANTENNAS
dipole, which can be approximated by a dipole antenna whose arms are much smaller than the wavelength of
radiation (1). Assuming the point dipole to be located at ( 0, 0, z0 ), the current density associated with the point
dipole is
where I 0 dl (= finite) denotes the strength of the dipole. Without loss of generality we assume I 0 is along the
x direction. From Eq. (68) we know that Ax is nonzero, corresponding to the radiation field from a horizontal
dipole. This is the only field that would be induced in empty space by the point dipole of Eq. (70). In the
presence of the layered structure, however, Az is also nonzero, because of the oblique reflection of Ax from
the layer interfaces, corresponding to the radiation field from a vertical dipole in the absence of the stratified
structure. Finally, Ay is identically zero as implied by the symmetry of the problem.
We require the tangential components of E and H to be continuous across the layer interfaces, which can
be directly written down from Eqs. (65) and (66). According to 1, these boundary conditions can be integrated
with respect to the transverse coordinates, x and y, and the constants of integration can be justified as zeros
by letting x and y go to infinity. For example, suppose one boundary condition requires the derivative of the
function f (x, y, z) with respect to x to be continuous across an interface. If f (x, y, z) vanishes as x goes to infinity,
either decreasing exponentially to zero as for a decaying wave or averaging to zero as for an oscillating wave,
we can integrate this boundary condition with respect to x and conclude that the function f (x, y, z) itself must
be continuous across the interface. As such, the boundary conditions imposed on the vector potential A = Ax e
x + Az e z as derived by Sommerfeld are the continuity conditions on the following four quantities (1)
Therefore there are four boundary conditions at each layer interface.
Recognizing the fact that the layered structure is homogeneous in the transverse directions, say, x and y,
it implies that the Helmholtz equation, Eq. (68), can be conveniently solved in the transverse Fourier-spectral
domain. For a given transverse spectral vector, kt = [kx , ky ], we denote the corresponding spectral-domain
vector potential components as x (kx , ky , z) and x (kx , ky , z) which relate to Ax (x, y, z) and Ax (x, y, z) via double
Fourier integrals, respectively. For each layer that does not contain the dipole source, we solve the Helmholtz
equation, Eq. (68), to obtain the following solutions:
where a, b, c, d are four unknowns to be determined by the boundary conditions and is given by
where and are the permittivity and permeability of the layer under consideration. For the uppermost layer
a = 0 = c and for the lower most layer b = 0 = d, as required by the boundary conditions at z = . For the
layer that contains the dipole source, we integrate Eq. (68) from z = z0 z = z0 + and note that the double
Fourier transform of the function (x) (y) is 1/2, and we derive the following discontinuity requirement on
MICROSTRIP ANTENNAS
x (kx ,
17
ky , z)/z on both sides of the plane z = z0
We therefore insert a fictitious interface at z = z0 assuming different values of a, b, c, d of for the two
subregions above and below this interface [Eqs. (72) and (73)], but with the same value of [Eq. (74)]. The
boundary conditions imposed on this fictitious interface are the same as before, except that the requirement
on the continuity of the quantity (Ax /z) is now being replaced by Eq. (75). If the z0 plane occurs at a single
layer interface of the stratified structure, no virtual interface needs to be created; we need only replace the
continuity requirement on (Ax /z) by Eq. (75).
Thus, we arrive at 4N unknowns with 4N boundary conditions for N interfaces, including the fictitious
one, if there is one. In the boundary conditions we replace [Eq. (71)] the operator /x by jkx , and /z by ,
whichever is applicable according to Eqs. (72) and (73). The 4N boundary conditions now become algebraic
ones and hence the 4N unknowns can be solved. The amplitudes of these unknowns are all proportional to the
dipole strength I0 d l, which can be conveniently chosen to be 1, as required by the Greens function. The vector
potentials Ax (x, y, z) and Az (x, y, z) can then be solved from x (kx , ky , z) and x (kx , ky , z) by applying the inverse
two-dimensional Fourier transforms. The electric and magnetic fields can finally be obtained by using Eqs. (65)
to (67).
Material imperfection results in losses of various kinds, upon which the permittivity and permeability
values become complex numbers. The dielectric loss is described by a loss tangent, tan, and the permittivity
takes the form of 0 r (1 jtan), where r denotes the dielectric constant. For a demagnetized magnetic
substrate, the permeability is 0 ( +j ) and and are the real and imaginary parts of the relative
permeability (3). For a metal conductor the permittivity contains both the displacement current and the
conduction current and hence the permittivity is modified as 0 j/, where denotes the conductivity. This is
the permittivity that was explicitly considered by 1, solving the electromagnetic fields generated by a Hertzian
dipole in response to the earths surface.
If a perfect metal is used as the ground plane, the electromagnetic field will not penetrate into it. Therefore,
the boundary condition on the metal surface is that the tangential components of the electric field vanish, and
it implies
Thus, we have 4N 2 unknowns and 4N 2 boundary conditions for a layered structure possessing a
perfect metal ground plane.
Actually, the interface boundary conditions [Eq. (71)] specify the flection and transmission of electromagnetic waves from one layer to another, and x is proportional to the magnetic field component and z is
proportional to the electric field component. Oblique-angle reflection and transmission can be readily written
down in terms of Snells law (18) and hence the coefficients of x and z , [Eqs. (72) and (73)] are determined
with relation to the preceding and succeeding layers. Thus, without the need of solving the boundary conditions
explicitly, [Eq. (71)], all of the a, b, c, d coefficients for the layers are correlated with each other and only four
unknowns remain, corresponding to those at the outermost layers (top and bottom), which can now be solved
by using the boundary conditions at the fictitious interface imposed by the point dipole [Eq. (75)], and so on.
For example, if the last layer is a perfect metal, the a, b, c, d coefficients of the layer adjacent to it satisfy the
following relationships
which are recognized as the total reflection condition. Equation (77) can be shown to be identical to Eq. (76).
18
MICROSTRIP ANTENNAS
For completeness, we list the transverse spectral-domain vector potential induced by a horizontal point
dipole located on top of a dielectric substrate backed by a perfect metal ground plane
and DTE and DTM are defined by
Here the dipole is located at the plane z = 0, the dielectric substrate is of a thickness d and permittivity ,
and 0 and are given by Eq. (74) with subscript 0 referring to air. The zeros of Eqs. (80) and (81) correspond to
surface-wave TE and TM modes, respectively. While there exists at least one TM surface mode, it is not always
the case that TE surface modes will be excited. The threshold for TE mode excitation is
Sommerfeld Integrals. Within each layer the dipole field solutions are obtained by inverse Fourier
transforms of x (kx , ky , z) and z (kx , ky , z) over the two-dimensional kx ky plane, or, equivalently, k k
plane. Here (k , k ) denotes the polar coordinate and (kx , ky ) the Cartesian coordinate. In general
and
as can be checked from the boundary condition of Eq. (71). In other words, x is an even function of kx and z
is an odd function of kx , as implied by the symmetry of the problem. Therefore, after integration over k , we
MICROSTRIP ANTENNAS
19
obtain
Although, Eqs. (85) and (86) can be numerically integrated, as discussed later, it is instructive to discuss
a procedure outlined by 1. Actually, Eqs. (85) and (86) are termed Sommerfeld integrals. The most significant
discovery of Sommerfeld was that the seemingly real integration starting at the fixed point k = 0 can be
converted into a complex k integration over a path, W, that closes at infinity (See Fig. 4
where H 0 (2) are Hankel functions of the second kind of order 0 and 1, respectively. In Fig. 4, the contour W is
detoured slightly in the complex k plane in order not to run into branch cuts and poles such that the integrands
remain finite and single valued. Only two-layer substances are illustrated in Fig. 4, where the air is denoted by
the subscript 0 and a dielectric layer is denoted by the subscript 1. For example, the respective permittivities
for air and the dielectric layer are 0 and 1 (the permeability of the dielectric layer is assumed to be 0 , the
same as air). In Fig. 4, k0 and k1 are the branch points given by
and is the smallest zero assumed by the function DTM , see Eq. (81). It is understood that more branch points,
and hence more branch cuts, will appear in Fig. 4 if more layers are presented in the layered structure. Also,
more simple poles will appear if DTE and DTM of Eqs. (80) and (81) admit more zeros. In Fig. 4, the branch
point at k = 0 is associated with the Hankel functions H 0 (2) ; For Eq. (88), this branch point, and hence its
associated branch cut, is replaced by a simple pole due to the different singular behavior of H 1 (2) at the origin.
According to Sommerfield, the couour W shown in Fig. 4 can be analytically deformed into three or more
components surrounding the respective branch cuts and simple pole, denoted as Q0 , Q1 , and P. 1 showed that
contour integrals of Q0 and Q1 give rise to spatial wave radiation, and for a large (spherical) distance r, they
exhibit the following asymptotic dependence
Actually, Q0 is due to the free-space dipole radiation, and Q1 is the radiation wave diffracted by the
dielectric layer (or from the image dipole induced by that layer) (1). Assume the air-dielectric interface is
20
MICROSTRIP ANTENNAS
Fig. 4. Input impedance loci of the cited antenna. The solid line is from calculation, and small circles represent measurements made in Ref. 13.
Fig. 5. Radiation profile of the cited antenna (13) in the = 0 plane.
located at z = 0. For z > 0, contribution form contour P has the following asymptotic forms
MICROSTRIP ANTENNAS
21
Fig. 6. Radiation profile of the cited antenna (13) in the = 90 plane.
Fig. 7. Calculated radiation frequency as a function of substrate thickness (fed by microstrip line).
They are surface waves tied to the interface and decrease at a rate proportional to 1/ in the lateral
directions, in contrast to the spatial wave radiation exhibiting a 1/r dependence, [Eq. (90)].
In the presence of material losses, the branch points and simple poles associated with spatial and surface
waves, respectively, acquire imaginary components, which are then pushed off from the real axis toward the
lower half of the complex k plane. As such, numerical integration of Eqs. (65) and (66) can be properly carried
22
MICROSTRIP ANTENNAS
Fig. 8. Calculated radiation resistance as a function of substrate thickness (fed by microstrip line).
out. However, care needs to be taken to avoid large truncation errors. Integration of k along the real positive
axis can be distinguished in three regions
Region I 0 < k < k0
Region II k0 < k < kc
Region III kc <k <
Here, kc denotes a cutoff wavenumber to be discussed later. In Region I the integrands are well behaved.
However, at the resonant frequencies of a metal patch dictated by the cavity model, both the numerator and
denominator of the integrand vanish, although their ratio remains finite. We call these geometric resonant
points quasisingularities (15). Near these quasisingular points the numerator and denominator need to be
expanded in Taylor series upon which their common zeros cancel out.
All of the surface poles are contained in Region II, and, conventionally, the upper bound of Region II, kc ,
is defined to be 10 times the real part of the largest surface pole occurring at the integrand. In Region II the
integrand behaves wildly in the vicinity of a surface pole. When coming across a surface pole, the integrand
transits from positive infinity to negative infinity, resulting in sharp cancellation during numerical integration.
To circumvent this difficulty we expand the integrand in Laurent series in the vicinity of a surface pole kp = p,
and the quasisingular terms (with negative exponents) are then evaluated analytically (remember p is now a
complex number). Equivalently, the singular part of integration is obtained via residual calculations. After the
singular part is subtracted from the integrand, the integrand becomes regular and can then be numerically
integrated in Region II.
In Region III we are involved with integration of Bessel functions at infinity, which oscillate indefinitely
as infinity is approached without exhibiting a strict period. This renders the conventional extrapolation scheme
MICROSTRIP ANTENNAS
23
Fig. 9. Calculated radiation linewidth as a function of substrate thickness (fed by microstrip line).
inaccurate. To perform integration in this region we consider asymptotic expansion of Bessel functions
where denotes the gamma functions. As such, integrands are written in series containing terms of the
following form
which can be readily evaluated by exploiting sine and cosine integrals and their derivatives if p is a positive
integer, or error functions and their derivatives if p is a positive half-integer larger than 1 (15).
When gyromagnetic layers appear in the layered structure, instead of using the vector potential A, it
is more convenient to express the electric field E and magnetic field H directly in the formulation. This is
because the dispersion relation for a gyromagnetic medium together with the associated E and H fields are
already known (3), which can be readily used to compose the boundary conditions at the layer interface and the
metal patch position. We are still working in the transverse Fourier spectra domain but do not use the point
dipole solutions. Thus, as before, by starting from the outermost layer interfaces, we postulate the unknown
24
MICROSTRIP ANTENNAS
Fig. 10. Input impedance loci of the antenna for several coax feeder positions.
coefficients for Et and H t and translate the field components layer after layer until reaching the interface
containing the metal patch. For dielectric layers, the field solutions can be expressed in the form of Eqs. (72)
and (73) relating Et and H t on both sides of the layer. For a gyromagnetic layer, a similar expression is used,
except that the propagation constant in Eqs. (72) and (73) become nondegenerate; together with the four a,
b, c, d coefficients, these propagation constants need to be solved from a quartic dispersion relation (19). This
nondegeneracy for wave propagation with respect to different propagation direction and polarization comes
from the tensor behavior of the permeability, called Polder tensor (3). In principle, there is no fundamental
difference in handling dielectric and magnetic layers, except that more algebraic steps are involved in solving
the dispersion relation imposed by the Polder tensor. At the metal patch boundary the discontinuity in H t gives
rise to surface current densities th relates to Et through Ohms law and are expressed in the form of an integral
equation discussed later.
Numerical Solutions. The dyadic Greens function, G(r, r ), is defined as the electric field at location
r produced by a unit point dipole located at r . We can solve the current distribution over a microstrip metal
patch of negligible thickness deposited on a layered structure backed by a ground plane. Therefore, r and r are
both located at the metal patch, and essentially we have a two-dimensional problem with the third dimension,
the z-direction, being absorbed into the Greens function. Thus, we are required to solve the following integral
MICROSTRIP ANTENNAS
25
equation
where S denotes the metal patch with surface impedance Rs = (1 + j)
. The first term in Eq. (95) is
the electric field returned by the background layered structure the second term relates to the Ohms current
(induction and conduction), and Ee is the electric field generated by an external current. Of special concern, in
Eq. (95), te results from integration along the z-axis, and hence it defines the effective thickness of the metal
patch to be either the skin-depth thickness, = (2/0 )1/2 , or the physical thickness of the metal patch, t,
whichever is smaller. For a perfect metal patch, te = 0. Thus, te represents the singular behavior of the current
distribution exhibiting a delta-function profile in the thickness direction. However, we expect that te will not
appear in the final expressions evaluating engineering parameters of the antenna, as is the case in Eqs. (96),
(98), (99), and (103). In Eq. (95) we have used and as the two-dimensional position vectors on the metal
patch for which the z coordinate is clear.
We denote by {J mn ()} a complete orthonormal vector basis for currents on the metal patch. We note that
{J mn ()} is the regular part of the surface current density, and the singular part is factored out in Eq. (97) as
te 1 , characterizing, again, the delta-function-like distribution of surface current along the z-direction. Thus,
the dimension of J mn is amperes permeter. We apply the Galerkins method to convert the integral equation
[Eq. (95)], into the following matrix form
where [ amn ] are the unknown coefficients to be solved expressed in terms of the current basis {J mn ()},
and [ Bmnm n ] the matrix elements derived from Eq. (95)
where I denotes the identity dyad. The inhomogeneous term of Eq. (96), [ bmn ], is associated with current
driving given by
26
MICROSTRIP ANTENNAS
where E mn is the electric field generated by J mn , J e is the driving current density, and integration is over the
whole volume, V. In Eq. (99) the reciprocity theorem has been used, which states that a response of a system
to a source is unchanged when the source and observer are interchanged (20). The unit drive-current density
may be specified as
Equations (100) and (101) are for coax and microstrip feeders, respectively, In Eqs. (100) and (101), S(x)
denotes the step function and
Therefore, assuming unit excitation current, the input impedance is
The radiation field associated with a point (Hertzian) dipole located at the air dielectric interface of a
microstrip structure is (16)
where
MICROSTRIP ANTENNAS
27
and
The radiation pattern for a given patch current distribution J(), Eq. (107), is therefore
where 0 = (0 /0 )1/2 and ko = 2/0 = (0 0 )1/2 are, respectively, the impedance and wavenumber in air.
A convenient current basis can be derived from the intrinsic functions associated with the metal patch
geometry (21). We define a set of current potential {mn ()} satisfying the two-dimensional Helmholtz equation
from which the current basis {J mn ()} is defined as
The boundary condition imposed on {J mn ()} is that the current is not allowed to flow across the metal
patch boundary, or, at the patch boundary
where n denotes the unit outward normal at the metal patch boundary. That is
Equation (112) determines the eigenvalues kmn in Eq. (109). The current potential can thus be determined,
and, for a rectangular patch located at 0 x L, 0 y W, we have
For a circular patch located at 0 R, we have
28
MICROSTRIP ANTENNAS
Fig. 11. Calculated radiation frequency as a function of coax feeder position.
and mn denotes the root of the derivative of the Bessel function, J m (x)
The advantages of using current potentials are that not only the vector Galerkin equations are converted
into a scalar ones, but also most of the integrations can be carried out analytically, including those required
to perform inverse transverse Fourier transforms, rendering only one-fold Sommerfeld-type integrals to be
evaluated numerically. This is true even in the presence of multiple patches exhibiting close coupling. We note
that the current basis introduced in Eq. (110) corresponds to the normal-mode currents appearing in the cavity
model of the antenna resonator. Because the basis current is defined as the gradient of a scalar function [the
current potential, Eq. (110)], we call it an irrotational current (21).
In the following we present the results of some calculations for circular microstrip patch antennas (15).
For rectangular patch antennas, see Ref. 22. The first calculation applies to the published data of a circular
antenna characterized by the following parameters (13): R = 6.75 cm, d = 0.1588 cm, r = 2.62, and tan
= 0.001. The calculated resonance frequency of the fundamental mode is 0.7936 GHz (15), which compares
exactly with the measured value of 0.794 GHz (13). This is contrasted with the calculation for a cavity model
presented in the first section, which predicts a resonant frequency of 0.805 GHz (13). The calculated input
impedance of the antenna is shown in Ref. 15, which compares nicely with measurements shown as small
circles in Fig. 3 (13). Figures 5 and 6 show the calculated radiation pattern of the antenna in the = 0 and
= 90 planes, respectively. We note that only copolarized radiations are generated from the fundamental
mode excitation; the cross-polarized field cancels out for the two m = 1 and m = 1 modes at the fundamental
resonant frequency.
The second patch antenna geometry considered is a microstrip disk of radius R = 1 cm, which is fed by
either a coax line or a microstrip line (15). The substrate has of dielectric constant r = 2.2, loss tangent tan
= 0.001, whose thickness d that varies Figures 7 to 9, assuming = 0.2 rad, show the calculated resonant
MICROSTRIP ANTENNAS
29
Fig. 12. Calculated radiation resistance as a function of coax feeder position.
frequency, input impedance, and radiation linewidth of the fundamental mode as a function of the substrate
thickness, respectively. In Fig. 7 the calculated resonant frequency of the antenna decreases monotonically
with the substrate thickness d, indicating that the effective dimension of the patch resonator increases with d,
as expected for a leaky cavity. Figure 8 shows that the input impedance of the antenna is relatively a constant,
unless d becomes very small, say, smaller than 0.05 cm. In Fig. 9, we see that the radiation bandwidth increases
with d, d hence the Q factor of the antenna decreases with d. This finding is consistent with Fig. 7 which exhibit
the leaky feature of the antenna patch cavity.
Figure 10 shows the calculated input impedance loci of the above circular antenna of substrate thickness
d = 0.1 cm fed by a coax line located at (0 , 0). In Fig. 10, the parameter re is defined as re = 0 /R, and re
= 1.0, 0.8, 0.6, and 0.416. For each feeder location, the two resonance frequencies shown in Fig. 10 are the
resonance frequencies of the probe-inductance in parallel with the detuned patch-resonator, forming a parallelresonant circuit, and the resonance frequency of the probe-inductance in series with the detuned patch-parallel
resonator, forming a series-resonant circuit. The field distribution of the patch at both frequencies shows very
little difference, and the patch is operated at the same resonance mode yet more or less detuned from its
resonance frequency, which occurs between these two frequencies. It can be seen from Fig. 10 that these
resonant frequencies appear only if 0.416 re 1. For re 0.416, no patch resonance can possibly be excited.
Figure 11 shows the calculated radiation frequency as a function of re , or 0 . It is seen in Fig. 11 that radiation
frequency remains roughly constant for 0.6 re 1, which increases rapidly when re is further reduced. Figure
12 shows the calculated input resistance as a function of re , or 0 . From Fig. 12, we see that the input impedance
decreases with re or 0 , and hence it is possible to design 50 input resistance of the antenna by feeding the
antenna at re = 0.442 (or re = 0.488 for 75 input impedance).
Finally, we consider the interaction between two identical circular microstrip patch antennas excited by
microstrip feeders of equal amplitude and phase. Let the antennas be deployed in parallel exhibiting the same
parameters considered before: R = 1 cm, d = 0.1 cm, r = 2.2, and the feeder lines having 50 resistance
(width 0.312 cm). The separation between the antennas is designated R 12 measured between their respective
30
MICROSTRIP ANTENNAS
Fig. 13. Calculated and measured input impedance loci of two coupled patch antennas.
centers. Figure 13 shows the calculated and measured input impedance for the coupled case, R 12 = 2 cm (the
patches are touching each other), and the uncoupled case, R 12 = 5 cm (the patches are separated by 3 cm at
their edges). Measurements were performed with respect to patch antennas fabricated using RT/Duroid 5880
material (Rogers Co., Chandler, AZ). The measured resonant frequencies were 5.514 and 5.561 GHz for the
coupled and uncoupled cases, respectively, which compare almost exactly with their calculated values of 5.5137
and 5.5642 GHz. The resonant frequencies of the patches and their input impedance have also been calculated
as a function of the patch separation, R 12 . These are shown in Figs. 14 and 15, respectively, with measurements
shown as solid squares. From Fig. 14, it is seen that the resonant frequency changes most rapidly when R12
is small, say, when 2 R 12 3 cm. Further increase in R12 does not change the resonance frequency much.
However, the input impedance does show long-range interference between two patch antennas. As shown in
Fig. 15 the input impedance is still increasing when the two antennas are separated by 5 cm, although the rate
of increase has slowed compared with its initial rate at R12 = 2 cm.
We conclude that the leaky feature of an antenna cavity can be well characterized by using a full-wave
analysis outlined previously. Among many numerical methods, the dyadic Greens function approach might
prove to be the simplest one to apply, not only because the analysis is two dimensional, but also because
the evaluation of the scalar Galerkin elements involves only onefold Sommerfeld-type integrals. Surface-wave
generation is significant for a thick dielectric substrate with large dielectric constant. The coupling between
MICROSTRIP ANTENNAS
31
Fig. 14. Calculated and measured resonant frequency of two coupled patch antennas as a function of their separation
distance.
microstrip elements is of long-range nature, although the radiation frequency of a patch is less influenced by
its neighboring antenna elements.
Recent Developments
Wideband Techniques. A microstrip patch antenna is inherently narrowband, stemming from the
cavity model that an electromagnetic resonator confining energy locally in space with little dissipation is
necessarily a high-Q, and hence narrowband, device. For a high-Q resonator the generation of surface waves
is insignificant. For a thin substrate, this condition amounts to d/0 < 0.07 for r = 2.3 and d/0 < 0.023 for
r = 10, as stated previously (10,11). For a microstrip antenna fabricated on a thin substrate the impedance
bandwidth is typically 1% to 3%. This is in contrast to the bandwidth of 16% of a half-wave dipole with a radius
to length ratio equal to 0.01, and 70% of a medium-length helix operating in the axial mode.
A number of techniques have been proposed to increase the bandwidth of a microstrip patch antenna.
These techniques are generally classified into three categories. The first category involves a straightforward
approach based on the use of a thick substrate whose dielectric constant is small. The second approach is
to design a matching network to enhance the bandwidth (23). The third method uses parasitically coupled
elements in a variety of ways to produce closely spaced multiple resonances of the antenna (24,25).
By using a thick substrate whose dielectric constant is considerably different from that of air, surface
waves will be generated and inevitably reduce the radiation efficiency and introduce interference between
array elements. To resolve this problem, air cavities or holes may be introduced with the substrate to effectively
reduce the dielectric constant of the patch. For example, 26 machined closely spaced holes in a Duroid substrate
underneath a microstrip patch to lower the effective dielectric constant of the antenna. Using a micromachining
technique, Zheng et al. measured a 12.8% impedance bandwidth on cavity-backed microstrip patch antennas
fabricated on silicon wafers (27).
32
MICROSTRIP ANTENNAS
Fig. 15. Calculated and measured input impedance at resonance of two coupled patch antennas as a function of their
separation distance.
The bandwidth of the antenna is primarily determined by the rate of transition that the imaginary
part of the impedance changes sign at resonance, as discussed previously. Thus, it is possible to introduce
a cancellation mechanism on the input inductance so as to smooth the impedance variation. For example,
the inductance associated with the long wire-lead of the coax probe will limit the bandwidth to <10% for a
thick substrate. By etching a small circular slot around the probe on the patch, capacitance is introduced that
cancels the probe inductance to produce a bandwidth of 16% (28). Recently, with the use of a U-shaped slot,
a substantial increase in bandwidth (32%) has been demonstrated (29). Alternatively, an L-shaped probewire
has been shown to result a bandwidth of 28 percent (30).
The third method to achieving broadband operation is to couple the microstrip patch antenna parasitically
with other dielectric resonators characterized by approximately the same resonant frequency of the patch. For
an isolated patch the resonant frequency is determined by its lateral dimension. In the presence of fringe fields
at the patch periphery, however, the boundary of the antenna is neither sharply nor rigidly defined, leading to
a slightly larger effective dimension of the patch, as described by Eqs. (2) and (5) for rectangular and circular
microstrip patches, respectively. Thus, in contrast to a metal cavity, patch antennas have soft boundaries, which
in turn give rise to finite bandwidth in radiation: the softer the boundary, the wider the bandwidth. For this
reason, a patch antenna fabricated on a thick substrate will show a wide bandwidth. Similarly, when coupled
together, many soft-boundary microstrip resonators or patches, the overall radiation bandwidth is consequently
enlarged.
The disadvantage of using a thick substrate is that, besides its cost, surface waves may be generated
in the substrate so as to reduce the antenna feeding efficiency. By using multiple electromagnetically coupled
patches this disadvantage can be overcome. Electromagnetically coupled patches can be deployed either side by
side (laterally coupled geometry) or layer by layer (vertically coupled geometry). For laterally coupled patches,
the antenna size will increase considerably, ultimately restricting its usage large-array applications (31).
MICROSTRIP ANTENNAS
33
The two-layer electromagnetically coupled patch antenna consisting of a driven (feeder) patch in the
bottom and a parasitic (radiating) patch on the top has been investigated by several authors. For circular
(32), equitriangular (33), and rectangular (34) patches, experimental results have shown an enhanced gain
and impedance bandwidth with low cross-polarization levels as compared with the conventional single-layer
microstrip antenna. Actually, like a Yogi antenna, the gain of a stacked antenna can be increased to above 20
dBi at any scan angle if the thickness of the substrate and multiple superstrate layers is chosen pperly (35).
Therefore, using stacked parasitic elements in microstrip arrays could improve the overall array performance,
offering a higher gain and broad bandwidth, meeting with the same array design criteria of a conventional
single-layer microstrip array but with fewer array elements.
Size Reduction. Microstrip antennas have a number of advantages over conventional antennas,
namely, small size, light weight, low production cost, and natural conformity. For many applications, however, for example, hand-held mobile communications systems, half-wave microstrip antennas etched on a
low-cost dielectric substrate are still too large to be accommodated on a portable phone. A well-known method
of reducing the size of a half-wave patch to a quarter wave is to introduce an electric-shorting wall at one of the
radiating edges. Hiraswa and Haneishi (36) have shown that the length of the patch can be made sufficiently
shorter than a quarter wave by replacing the shorting wall with a shorting pin at the corner of the patch.
Recently, Wong and Lin have shown that by replacing the shorting pin with a chip resistor of low resistance,
antenna size can be further reduced, with an additional increase in bandwidth (37).
The other approach to reduce antenna size is to use a meandered geometry of the patch antenna (36).
By meandering the patch, the effective electrical length is larger than the physical length. Consequently, the
resonant frequency of the meandered antenna can be much lower than that of a conventional design with the
same physical length (38).
Finally, we note that not only is it possible to achieve high gain for printed circuit antennas, it is also
feasible to shape the radiation pattern in a prescribed manner. Some very interesting phenomena, such as
radiation into the horizon, radiation pattern monodirectionality, and azimuthal-dependent radiation have been
found possible (39). Furthermore, laterally coupled patches can serve as adaptive array antennas, because the
phase of radiation from a parasitic element can be adjusted via a varactor diode inserted at some feeding
position (40). The important feature of an adaptive array antenna is that it can provide beam steering as
required by many communications and traffic control systems (41).
BIBLIOGRAPHY
1. A. Sommerfeld, Partial Differential Equations, New York: Academic Press, 1962.
2. J. R. James P. S. Hall (eds.), Handbook of Microstrip Antennas, Vols. 1 and 2, London: Peter Peregrinus, 1989.
3. H. How, T.-M. Fang, C. Vittoria, Intrinsic modes of radiation in ferrite patch antennas, IEEE Trans. Microw. Theory
Tech., MTT-42: 988, 1994.
4. E. O. Hammerstad, Equations for microstrip circuit design, Proc. 5th Eur. Microstrip Conf., Hamburg, September, 1975,
pp. 268272.
5. M. V. Schneider, Microstrip lines for microwave integrated circuits, Bell System Tech. J., 48: 14211444, 1969.
34
MICROSTRIP ANTENNAS
6. L. C. Shen, Resonant frequency of circular disc, printed circuit antenna, IEEE Trans. Antennas Propagat., AP-25: 596,
1977.
7. J. A. Stratton, Electromagnetic Theory, New York: McGraw-Hill, 1941.
8. R. S. Elliott, Antenna Theory and Design, Englewood Cliffs, NJ: Prentice-Hall, 1981.
9. A. G. Derneryd, Microsip array antenna, Proc. 6th Eur. Microw. Conf., 1976, pp. 339343.
10. J. R. James A. Henderson, High-frequency behavior of microstrip open-circuit terminations, IEE J. Microw. Opt. Acoust.
3: 205218, 1979.
11. C. Wood, Analysis of microstrip circular patch antennas, IEE Proc., 128 (H): 6976, 1981.
12. S. B. A. Fonseca A. J. Giarola, microstrip disk antennas, part I: Efficiency of space wave launching, IEEE Trans.
Antennas Propagat., AP-32: 561567, 1984.
13. Y. T. Lo, D. Solomon, W. F. Richards, Theory and experiment on microstrip antennas, IEEE Trans. Antennas Propagat.,
AP-27: 137145, 1979.
14. R. C. Booton, Jr., Computational Method for Electromagnetics and Microwaves, New York: Wiley, 1992.
15. H. How C. Vittoria, Directions for the next generation of MMIC devices and systems, in N. K. Das and H. L. Bertoni,
(eds.), Computer Aided Design Tools for Microstrip Circuitries, New York: Plenum, 1997, pp. 399406.
16. J. R. Mosig, R. C. Hall, F. E. Gardiol, Numerical analysis of microstrip patch antennas, in Handbook of Microstrip
Antennas, J. R. James and P. S. Hall, (eds.), London: Peter Peregrinus, 1989, pp. 393453.
17. J. D. Jackson, Classical Electrodynamics, New York: Wiley, 1975.
18. S. Y. Liao, Microwave Devices and Circuits, Englewood Cliffs, NJ: Prentice-Hall, 1985.
19. C. Vittoria, Microwave Properties of Magnetic Films, Singapore: World Scientific, 1993.
20. R. F. Harrington, Time Harmonic Electromagnetic Fields, New York: McGraw-Hill, 1961.
21. H. How C. Vittoria, Radiation modes in dielectric circular patch antennas, IEEE Trans. Microw. Theory Tech., MTT-42:
1939, 1994.
22. D. M. Pozar, Input impedance and mutual coupling of rectangular microstrip antennas, IEEE Trans. Antennas Propag.,
AP-30: 1191, 1982.
23. H. An, K. J. C. Nauwelaers, A. R. Van de Capelle, Broadband microstrip antenna design with the simplified real
frequency technique, IEEE Trans. Antennas Propag., AP-42: 129136, 1994.
24. R. Q. Lee, K. F. Lee, J. Bobinchak, Characteristics of a two layer electromagnetically coupled rectangular patch antenna,
Electron. Lett., 23: 10701072, 1987.
25. C. K. Aanandan, P. Mohanan, K. G. Nair, Broadband Gap-Coupled Microstrip Antenna, IEEE Trans. Antennas Propag.
AP-38: 15811586, 1990.
26. G. P. Gauthier, A. Courtay, G. M. Rebeiz, Microstrip antenna on synthesized low-dielectric constant substrates, IEEE
Trans. Antennas Propag., AP-45: 13101314, 1997.
27. M. Zheng, et al., Broadband microstrip patch antennas on micromachined silicon substrates, Electron. Lett., 34 (1):
34, 1998.
28. P. S. Hall, Probe compensation in thick microstrip patches, Electron. Lett., 23: 606607, 1987.
29. K. F. Lee, et al., Experimental and simulation studies of coaxially fed U-slot rectangular patch antenna, IEE Proc.
Microw. Antennas Propag., 144: 354358, 1997.
30. K. M. Luk, et al., Broadband microstrip patch antenna, Electronics Lett., 34 (15): 14421443, 1998.
31. S. Dey R. Mittra, A compact broadband microstrip antenna, Microw. Opt. Technol. Lett., 11 (6): 295297, 1996.
32. A. Sabban, A new broadband stacked two-layer microstrip antenna, IEEE AP-S Int. Symp. Dig., 6366, 1983.
33. P. S. Bhatnagar, et al., Experimental study on stacked triangular microstrip antennas, Electron. Lett., 22: 864865,
1986.
34. R. Q. Lee K. F. Lee, Experimental study of the two-layer electromagnetically coupled rectangular patch antenna, IEEE
Trans. Antennas Propag., AP-38: 12981302, 1990.
35. H. Y. Yang G. N. Alexopoulos, Gain enhancement methods for printed circuit antennas through multiple superstrates,
IEEE Trans. Antennas Propag., AP-35: 860863, 1987.
36. K. Hiraswa M. Haneishi, Analysis, Design, and Measurement of Small and Low-Profile Antennas, London: Artech
House, 1992, Ch. 5.
37. K. Wang Y. Lin, Small broadband rectangular microstrip antenna with chip-resistor loading, Electron. Lett., 33 (19):
15931594, 1997.
38. S. Dey R. Mitttra, Compact microstrip patch antenna, Microw. Opti. Technol. Lett., 13 (1): 1214, 1996.
MICROSTRIP ANTENNAS
35
39. H. Y. Yang N. G. Alexopoulos, Generation of nearly hemispherical and high gain azimuthal symmetric patterns with
printed circuit antennas, IEEE Trans. Antennas Propag., 35: 972977, 1987.
40. D. Cailleu, N. Haese, P. A. Rolland, Microstrip adaptice array antenna, Electron. Lett. 32 (14): 12461247, 1996.
41. R. J. Dinger, A planar version of a 4.0 GHz reactively steered adaptive array, IEEE Trans. Antennas Propag., 34:
427431, 1986.
HOTON HOW
ElectroMagnetic Applications
CARMINE VITTORIA
Northeastern University
MONOPOLE ANTENNAS
513
MONOPOLE ANTENNAS
This article describes the concepts underlying monopole antennas and their many applications, such as for broadcasting,
car radios, and more recently for cellular telephones. In its
simplest form, the monopole antenna above an infinite ground
plane shown in Fig. 1(a) can be considered as one-half of a
corresponding double-length center-fed linear dipole shown
for comparison in Fig. 1(b). The current distribution for a vertical monopole antenna of height h is assumed to be a standing wave of the following form, in terms of the feed point current I(0):
I(z ) =
I(0)
sin k(h z )
sin(kh)
(1)
for 0 z h. This occurs because the perfect ground plane
creates an image of the monopole with a current distribution
identical to that for the lower arm of the dipole. Together with
this image, the monopole antenna appears to be a center-fed
dipole for the upper half-space. There is negligible penetration of fields into a high conductivity ground for a monopole
antenna, and all that radiation is directed into the upper halfspace creating a power density for any angle (see Fig. 1)
that is twice as high as that for a dipole radiating the same
amount of power. This gives a directivity or gain of the monopole antenna that is twice that for the double-length dipole.
Many of the other properties of a monopole antenna above
;;;
z
z = h
I(z )
Ground
z = 0
(a)
z = h
I(z )
z = 0
z = h
(b)
Figure 1. A vertical monopole antenna above ground (a) and the corresponding center-fed dipole (b). Shown also are the current variations I(z) over the lengths of both the monopole and the corresponding dipole.
J. Webster (ed.), Wiley Encyclopedia of Electrical and Electronics Engineering. Copyright # 1999 John Wiley & Sons, Inc.
514
MONOPOLE ANTENNAS
Table 1. Relationships Between Monopole and Dipole Antennas
Monopole Above Ground
Length h
Radiation pattern
Corresponding Dipole of
Twice Length L 2h
Feed-point resistance Ra
Same as that for the dipole but only
for angle 0 90
Ra Ra,d (2h)
Feed-point reactance Xa
Xa
Directivity Da
Da
monopole
Xa,d (2h)
2Da,d (2h)
monopole
monopole
ground are also related to those for the corresponding doublelength dipole in a fairly simple manner.
We give in Table 1 the relationships between a monopole
antenna of length h above ground and the corresponding dipole antenna of length L 2h. A graph of the variation of the
feed-point resistance and reactance of the monopole antenna
above ground is given in Fig. 2 as a function of length h/ ,
where is the free-space wavelength at the radiation frequency (1). Note that the reactance Xa depends on the conductor radius a, whereas the feed-point resistance Ra is relatively
independent of conductor radius a for thin antennas (a/
1). The radiation patterns of a few typical monopole antennas
are given in Fig. 3. Because a monopole antenna is symmetric
with respect to angle in the azimuthal plane (xy plane),
these patterns are independent of angle and depend only
on angle (see Fig. 1). The following relationship is given for
power density S():
Ra,d : function of length
L 2h (see Fig. 2)
Xa,d : function of length
L 2h (see Fig. 2)
Da,d : function of length
L 2h
where W is the radiated power, r is the distance to the field
point, k 2/ is the propagation constant, and P2() is the
pattern factor given by
P2 ( ) =
cos(kh cos ) cos(kh)
sin
2
;;;
;;;
;;;
;;;
;;;
(3)
Because monopole antennas help reduce the length of the
needed dipole by a factor of two and result in directivities that
(a)
S( ) =
30 W P ( )
r2 Ra sin2 (kh)
2
100
(2)
70
(b)
100
200
a
= 0.0035
60
0.0014
50
0.00035
0.00014
300
40
30
0.000035
400
20
500
10
600
0
0
0.10 0.15 0.20 0.25 0.30
Monopole length h/ (wavelengths)
Resistance Ra ()
Reactance Xa ()
Figure 2. Variation of feed-point resistance Ra and reactance Xa for
an end-fed monopole above ground as a function of height h/ [source:
E. C. Jordan and K. G. Balmain (1)].
(c)
(d)
Figure 3. Radiation patterns for some typical end-fed vertical monopoles above infinite ground: (a) h/ l 0.05 (short monopole), (b)
h/ 0.25 (quarter-wave monopole), (c) h/ 0.5 (half-wave monopole), and (d) h/ 0.75 ( monopole) [adapted from E. C. Jordan
and K. G. Balmain (1)].
MONOPOLE ANTENNAS
are twice as large, vertical monopole antennas above ground
are extensively used for amplitude modulated (AM) broadcasting in the frequency range 5351605 kHz. For this application, the wavelengths are long on the order of 200600 m,
and the monopole antennas are immensely helpful in reducing the required height. It is necessary, of course, to create a
good conductivity ground, which for dry or rocky soil conditions is often obtained by burying a conducting screen made
of radially spread metal wires with angular separations of
23 that extend to a radius at least equal to the height of the
antenna, but preferably 20 to 50% larger than this minimal
requirement. This screen, called a counterpoise, is often buried a few inches below the surface of the natural ground but
may also be left slightly above ground for rocky or otherwise
difficult terrain.
EFFECT OF FINITE CONDUCTIVITY GROUND
AND CURVATURE OF THE EARTH
Ra
Ra + Ro
Insulator
;;
;;;
;;;;;
Support
tower
Vertical
radiator
Feed wire
Support
tower
Insulator
Ground
Figure 4. A top-loaded short-monopole antenna typical of VLF/LF
communication systems.
Equation (2) gives the power density radiated from a monopole antenna for a ground of infinite conductivity and for flat
earth. For finite conductivity ground, the power density at the
surface of the earth diminishes more rapidly than the square
of the distance r from the antenna (2). Furthermore, the rate
at which the power density at the surface of the ground diminishes with distance is steeper at higher frequencies. The
rate of reduction of the radiated power density as compared
to Eq. (2) is also higher for lower conductivity soil such as
sandy or rocky lands, cities, etc., rather than propagation over
seawater, marshy, or pastoral lands, etc.
For distances in excess of about 1000 miles, there is a precipitous drop in the surface power density at all frequencies
due to the increased effect of the curvature of the earth (2).
Monopole antennas are also the antennas of choice for very
low frequency (VLF) (3 to 30 kHz) and low frequency (LF) (30
to 300 kHz) communication systems. For these applications,
the height h of the antenna is generally a small fraction of
the wavelength. From Fig. 2, for small values of h/ , the feedpoint resistance Ra is very small, on the order of a few ohms,
and the ohmic losses resulting from the surface resistance of
the antenna can be significant by comparison. This results in
reduced antenna radiation efficiency r given by the following
equation:
r =
Flat top
Insulator
515
(4)
where Ro is the effective ohmic resistance of the antenna.
Top loading of the monopole antenna illustrated in Fig. 4
is often used to improve the feed-point or antenna-equivalent
resistance Ra. Because the top set of wires acts as a capacitance to ground rather than the open end of the antenna, the
current at the top of the antenna (z h) no longer needs to
go to zero as it does for an open-ended monopole [from Eq.
(1), I 0 at z h]. The upper region of the antenna can,
therefore, support a substantial radio frequency (RF) current,
resulting in improved radiation efficiency. The antenna-equivalent resistance Ra 395 h2 / 2 for open-ended short monopoles may be improved by a factor of 2 to 3 by use of top
loading, resulting in higher radiation efficiency.
It should be mentioned that the various concepts discussed
here are also usable at higher frequencies. Open-ended monopole antennas are, in fact, used up to ultrahigh frequency
(UHF) and microwave frequencies where metallic conducting
sheets, either solid or in the form of a screen, may be used in
lieu of the ground to create the image antenna. A variety of
top-loading hats illustrated in Fig. 5 may also be used at
these higher frequencies.
Because a monopole antenna above ground or similar reflector acts as a dipole antenna, several of the concepts that
are useful for dipole antennas are also used for the monopole
antenna. Some of these concepts that are often used in practice are discussed next.
Folded Monopoles. A folded monopole antenna is illustrated in Fig. 6(a) where the feed point is connected to arm 1,
and arm 2 is grounded at the bottom end. For dipoles, this
;;
;;
;;
;;
Ground plane
or reflector
(a)
(b)
(c)
Figure 5. Some typical geometries used for top-loaded monopole antennas: (a) plate, (b) umbrella of wires, and (c) inverted L monopole.
516
;;
;;
;;
MONOPOLE ANTENNAS
Ground plane
or reflector
(a)
Z0
(b)
Reflector
Driven antenna
Directors
(c)
Figure 6. Some special adaptations of monopole antennas: (a) folded
monopole antenna, (b) shunt-fed grounded monopole, and (c) four-element Yagi-Uda array of monopoles.
provides a method of increasing the feed-point impedance
given in the following relationship (3):
Zi = (1 + c)2 (Ra + jXa )
(5)
where Ra and Xa plotted in Fig. 2 are the resistance and reactance of the single arm (not-folded) monopole antenna, and
the factor c is given approximately by
c
ln(d/a1 )
ln(d/a2 )
(6)
In Eq. (6), a1 and a2 are radii of wires or rods used for arms 1
and 2, respectively, and d is the center-to-center spacing between the two arms of the antenna, generally much smaller
than the height of each of the arms. For a folded monopole
antenna where equal radii arms are used, the feed-point impedance is, therefore, four times higher than that for a monopole antenna that is not folded. This can be truly advantageous, because the feed-point resistance may now be
comparable to the characteristic impedance Z0 of the transmission or feeder line, and the reactance of the antenna may
easily be compensated by using a lumped element with a reactance that is negative of the reactance (1 c)2 Xa at the
terminals of the folded monopole antenna.
Shunt-Fed Monopoles. A typical arrangement of a shunt-fed
grounded monopole is shown in Fig. 6(b). This corresponds to
one-half of a delta match that is often used for dipoles. Because the base of the monopole is grounded for this arrangement, the antenna is fed typically 15 to 20% farther up on the
antenna; the exact location is determined by matching the
feeder line of characteristic impedance Z0, which is typically
200 to 600 . Shunt-fed monopole antennas were often used
in the early days of AM broadcasting when high-quality insulators were not generally available and it was convenient to
ground the tall steel/aluminum masts being used for monopole antennas. With the advent of high compression strength
insulators, the masts are now more typically mounted on
these insulators, and the monopole towers are fed at the base
of these antennas.
Parasitic Monopoles. Much like the Yagi-Uda arrays of dipoles (3), parasitic monopoles of grounded elements are used
to help direct the beams away from the reflector monopoles in
the direction of director monopoles. An illustration of a fourelement Yagi-Uda array of monopoles is shown as Fig. 6(c).
Forward gains on the order of 10 to 12 dB with front-to-back
ratios of 5 to 8 dB may be obtained using such arrangements.
One or more of the parasitic monopoles have also been occasionally used for beam shaping by AM broadcast stations.
Two-element antenna arrays consisting of a parasitic reflector
monopole and a driven monopole antenna have also been proposed for use in hand-held wireless telephones (see, for example, Ref. 4). If they are carefully designed, such antennas may
allow electromagnetic fields to be partially directed away from
the human head.
Arrays of Monopole Antennas. Like arrays of dipole elements, monopole arrays of equal or arbitrarily spaced and
phased elements may be used for directing principal lobe(s) of
radiation in desired directions. Arrays of monopole antennas
are used for VLF/LF communications, for AM broadcasting,
and, in general, for all frequencies up to microwave frequencies. AM broadcasters for non-clear-channel stations have, for
example, used this technique to alter the day and night radiation patterns by varying the phases and the magnitudes of
excitations in the various elements.
Conical Monopole. A conical monopole, realized as a cone
above a perfect ground plane, is commonly used because of its
broadband properties. The cone may be either solid or built
from a series of wires above a ground plane. Similar to conventional monopoles, the ground may also be built from a series of radial wires as shown in Fig. 7.
The analysis of this antenna may be done either using rigorous analytical solutions (5,6) or an approximate method (6).
Support wires and radiators
Support poles
Insulators
Guys
o
To
transmitter
Ground counterpoise
Figure 7. A conical monopole.
Guys
MONOPOLE ANTENNAS
The rigorous approach is based on a series expansion of transverse magnetic (TM) modes and is computationally complex.
Numerical methods, based on the assumption that the antenna is built from a series of wires, have been commonly
used for specific monopole arrangements.
An example of a conical monopole constructed of 60 evenly
spaced wires above a ground plane is an antenna that has
been in use by the U.S. Navy (7). For use in the HF band 2
to 14 MHz, the dimensions of the monopole are l 30.6 m,
a 0.001 m, 0 45 (see Fig. 7). The radius of the ground
screen is approximately 37.5 m, constructed of 120 radial
wires over average soil (r 15 and 0.012 S/m), and the
radius of the screen wire is 0.002 m. This antenna performs
over a 7 : 1 frequency band (2 to 14 MHz) with an input resistance varying from 44 to 64 (6). Because of the fairly narrow range of variation of input resistance, matching of this
antenna to a 50 line is possible over at least a 7 : 1 frequency range.
Monopole Antennas for Car Radio and Mobile Communications. Monopole antennas used for car radio and mobile communications commonly rely on the body of the vehicle to provide the ground plane. Variations in the shape of the vehicle
and placement of the antenna affect the radiation pattern and
performance of these antennas (8). For an antenna mounted
on the automobile, it would no longer radiate isotropically in
the horizontal plane but may radiate with a slightly higher
gain in directions where the automobile body extends farther,
thus simulating a larger width ground plane in such directions.
For communication antennas in general, it is often desirable to have the shortest possible antenna. Although a quarter-wave monopole is easier to match, a shorter antenna has
an associated capacitive reactance. This can, however, be canceled out by adding a loading coil (inductance) to the base of
the antenna.
Another arrangement often used is to place an inductive
coil at the center of the antenna (9). Because of the altered
current distribution along the length of the antenna (similar
to a top-loaded monopole), this increases the feed-point impedance of the end-fed antenna to roughly twice the value,
making it easier to match to such an antenna. The value of
the inductance needed at the center is approximately double
that which would be needed at the ends. This consequently
increases the coil resistance.
In order to reduce the inductance of the loading coil required to match the antenna at either the center or top, a
capacitive load (a small metal ball) may be added to the tip
of the antenna.
In all this discussion of monopoles, we have assumed that
the wave propagation from the antenna is over a perfectly
conducting or infinite conductivity ground plane. This is
hardly ever the case. For ground wave propagation ( 90)
over a finite conductivity ground plane, the fields drop off
more rapidly than 1/r2 given by Eq. (2). This effect is more
pronounced for higher frequencies and is generally small only
at very low frequencies less than a few tens of kilohertz. Also,
as expected, the departure from the idealized Eq. (2) is more
pronounced the lower the conductivity of the soil. Rocky and
jagged hilly ground, for example, would result in a more rapid
drop off of the power density S along the ground plane than
propagation over sea water.
517
Yet another factor that has an impact on surface wave
propagation is the curvature of the earth. For distances larger
than about 100 miles (62 km), the power density at the
ground level diminishes very rapidly particularly for frequencies in excess of about 1 MHz.
EFFECT OF FINITE-SIZE GROUND PLANE ON
IMPEDANCE AND RADIATION PATTERN
Whereas a ground plane of diameter 10 wavelength or larger
has a fairly small effect on the feedpoint impedance of a
monopole antenna as compared to the values given in Fig. 2,
finite-size ground plane has a fairly significant effect both on
the feedpoint impedance and radiation pattern (10).
An example of the effect on the radiation pattern is illustrated in Fig. 8(a) for the radiation pattern of a monopole antenna of a cellular telephone in free space. Since this monopole antenna is mounted on a plastic-covered metal box of
finite dimensions, only a finite-size ground plane of box shape
is used for this radiator. One effect is to alter the radiation
pattern [see Fig. 8(a)] from a perfect circle in the azimuthal
plane to one that is crooked with a lower gain in the sector
where the extent of the ground plane is minimal, i.e., on the
side where the antenna shown by a circle in the insert of Fig.
8(a) is mounted. Also the gain for this quarter wave monopole
antenna of about 1 dBi is considerably lower than 5.16 dBi (a
factor of 3.28) for the same antenna mounted on a flat, infinite
ground plane. The actual reason for this considerably reduced
gain is that part of the energy is also radiated in the lower
half space for this vertical monopole antenna above the box
while none would be radiated for this antenna mounted above
an infinite ground plane.
SOME RECENT APPLICATIONS OF MONOPOLE
ANTENNAS FOR PERSONAL WIRELESS DEVICES
A monopole antenna with or without a helix is the antenna of
choice for todays personal wireless devices, including cellular
telephones (11). Antennas of different lengths, typically from
a quarter wavelength to a half wavelength at the irradiation
frequency, are generally used for hand-held devices that operate at 800 to 900 MHz for cellular telephones and 1800 to
1900 MHz for PCS (personal communication system) devices.
Often, these monopole antennas are in the form of a whip
antenna, which can be pulled out to its full length during a
telephone conversation. At times, a small length helical antenna is permanently mounted on the top of the handset
through which the monopole antenna can be retracted and to
which this monopole connects when it is completely pulled
out, giving a monopole loaded with a helix at the bottom as
the radiating antenna during telephone usage. Of necessity,
these antennas use a finite-size reflecting plane, which is generally the metallic shielding box used either for the top RF
section, or at times for the entire handset. Several authors
have calculated and/or measured the radiation patterns of
such monopole antennas mounted on a handset held in air or
against the human head (1214).
Given in Fig. 8(a, b) are the computed and measured radiation patterns in the azimuthal plane for a typical commercial
telephone at the center-band frequency of 835 MHz in free
space and in the presence of the human head, respectively
518
MONOPOLE ANTENNAS
3
3.6
10.2
stantially lower due to absorption in the tissues of the head.
Radiation efficiencies for monopole antennas used for personal wireless devices are typically on the order of 30 to 50%,
with 30 to 50% of the power being absorbed in the head and
another 5 to 15% of the power being absorbed in the hand
(13,15).
16.8
Medical Applications
23.4
End-fed monopole antennas excited by a coaxial line have
been used for a number of medical applications such as hyperthermia for cancer therapy (16) and cardiac ablation catheters
(17). For these applications, both straight wire and small-diameter helix monopoles have been used.
BIBLIOGRAPHY
1. E. C. Jordan and K. G. Balmain, Electromagnetic Waves and Radiating Systems, 2nd ed., Englewood Cliffs, NJ: Prentice-Hall,
1968, p. 548.
2. ITT Handbook, Reference Data for Radio Engineers, 5th ed., Indianapolis, IN: Howard W. Sams & Co., Inc., Chap. 26, 1973.
(a)
3
3. W. L. Stutzman and G. A. Thiele, Antenna Theory and Design,
New York: Wiley, 1981.
3.6
4. O. P. Gandhi and J. Y. Chen, Electromagnetic absorption in the
human head from experimental 6-GHz hand-held transceivers,
IEEE Trans. Electromagn. Compat., 37: 547558, 1995.
10.2
5. C. T. Tai, On the theory of biconical antennas, J. Appl. Phys., 19:
11551160, 1948.
16.8
6. R. W. P. King, Theory of Linear Antennas, Cambridge: Harvard
University Press, 1956.
23.4
7. M. T. Ma and K. P. Spies, A Simplified Formulation Computation
for Conical Monopole Antennas, U.S. Dept. of Commerce Report
OT 74-53, 1974.
Head model
8. D. E. Hicks, CB Radio Antennas, Indianapolis, IN: Howard W.
Sams & Co., 1967.
9. The A.R.R.L. Antenna Book, West Hartford, CT: The American
Radio Relay League, 1956.
10. R. C. Johnson and H. Jasik, Antenna Engineering Handbook, 2nd
ed., New York: McGraw-Hill, Chap. 4, 1984.
11. R. Luebbers et al., FDTD calculation of radiation patterns, impedance and gain for a monopole antenna on a conducting box,
IEEE Trans. Antennas Propag., 40: 15771583, 1992.
(b)
Figure 8. The computed and measured radiation patterns (in dBi)
in the azimuthal plane for a commercial cellular telephone at 835
MHz (a) in free space and (b) held against the human head.
(14). This telephone uses a short helical antenna of diameter
4.2 mm, pitch 1.4 mm, and total length 16 mm. Dimensions
of the handset are similar to those for many telephones and
are approximately 2.4 5.2 14.7 cm along the three axes,
respectively. Fairly similar radiation patterns are also obtained for antennas using either monopoles or monopole-helix
combinations. In the presence of the human head, gains on
the order of 1.0 to 3.0 dBi are obtained away from the human
head, with gains in directions through the head that are sub-
12. M. A. Jensen and Y. Rahmat-Samii, EM interaction of handset
antennas and a human in personal communications, Proc. IEEE,
83: 717, 1995.
13. M. Okoniewski and M. A. Stuchly, A study of the handset antenna and human body interaction, IEEE Trans. Microw. Theory
Tech., 44: 18551864, 1996.
14. G. Lazzi et al., Comparison of FDTD-computed and measured radiation patterns of commercial mobile telephones in presence of
the human head, IEEE Trans. Antennas Propag., 46, 1998.
15. O. P. Gandhi, G. Lazzi, and C. M. Furse, Electromagnetic absorption in the human head and neck for mobile telephones at 835
and 1900 MHz, IEEE Trans. Microw. Theory Tech., 44: 1884
1897, 1996.
16. J.-C. Camart et al., Coaxial antenna array for 915 MHz interstitial hyperthermia: Design and modelizationpower deposition
and heating patternphased array, IEEE Trans. Microw. Theory
Tech., 40, 22432250, 1992.
MONTE CARLO ANALYSIS
17. S. Labonte et al., Monopole antennas for microwave catheter ablation, IEEE Trans. Microw. Theory Tech., 44: 18321840, 1996.
O. P. GANDHI
G. LAZZI
C. M. FURSE
University of Utah
519
MULTIBEAM ANTENNAS
553
MULTIBEAM ANTENNAS
Multiple beam, or multibeam, antennas are systems designed
to produce several antenna beams from a single or multiple
apertures. Traditional antennas use a single focusing device,
J. Webster (ed.), Wiley Encyclopedia of Electrical and Electronics Engineering. Copyright # 1999 John Wiley & Sons, Inc.
554
MULTIBEAM ANTENNAS
such as a lens or paraboloid reflector, illuminated by a single
feed or illuminator to produce a single antenna beam. When
two or more antenna beams are required, it is necessary to
use two or more traditional antennas. Therefore, the reduced
volume required by a multibeam antenna is the main purpose
for its use.
A typical communication link consists of two terminals,
each of which has an antenna that produces a single antenna
beam. When the beams of these antennas point toward the
cooperating terminal, a radio link is formed, or closed. A satellite-borne relay terminal permits Earth terminals to communicate with one another when it is not possible to form a
line-of-sight (LOS) path between these terminals. For economic reasons, the satellite relay terminal provides access to
several Earth terminals simultaneously. Multiple access requires multiple antenna beams, each pointing to a different
coverage area. Usually the location of these coverage areas is
either not known prior to placing the satellite in orbit, or it
varies during normal operation of the communication system.
Suitable sites for several antenna apertures on-board the
space craft, or the cost of designing and manufacturing several single-beam antennas, prohibit the use of several singlebeam antennas to provide a relay between multiple Earth terminals simultaneously. In order to relay signals between two
Earth terminals a transponder consisting of a receiver, an
amplifier, and a transmitter, is required. Usually these terminals require a two-way, or duplex, communication link.
Therefore each antenna beam will have a transponder connected to it and the input and output terminals of the transponder may be connected to different beams. In this case one
beam points toward one Earth terminal and another beam
points toward the other Earth terminal.
A multiple beam antenna (MBA) system generally consists
of four subsystems:
1. A converging device, such as a lens, paraboloid reflector,
or Cassegrain reflector system
2. A feed horn array
3. A beam switching network (BSN), or a beamforming
network (BFN)
4. A control subsystem for configuring the BSN or BFN
Design and performance of a reflector, lens, or other converging device that is used in an MBA system is essentially the
same as when the device is used with a single-beam antenna
system. However, the converging device (lens or reflector) of
an MBA must be able to accommodate illumination by feeds
that are not located at the focal point of the device. This accommodation implies that the concomitant degradation in
performance is tolerable. This degradation in performance is
commonly referred to as beam scanning loss.
Array antennas can also be used as a converging device.
In this case it is customary to excite the array via a butler
beamforming matrix (BBFM). A butler beamforming matrix
has the same number of input and output ports. (Discussion
in this section assumes that the antenna is transmitting signals. The same principles and comments apply if the antenna
is receiving signals.) The array antenna system is formed by
connecting the output ports of a BBFM to the radiators of an
array antenna. Exciting an input port of the BBFM produces
an antenna beam pointing in a given direction. Exciting a dif-
ferent input port produces an antenna beam pointing in a different direction. The collection of input ports will produce
beams that span the antennas field of view. This field of view
is usually a rectangular sector of space. This type of MBA, as
with the reflector and lens antennas, has a scanning loss that
must be tolerated.
Feed horns are pyramidal, or conical, extensions of the
waveguide used to excite them. Their apertures are usually
packed together to form an equilateral triangular grid. The
centers of their apertures are spaced to produce the desired
angular spacing between adjacent beams. There is one horn
for each beam and each horn has at least one input port.
These input ports are connected to the output ports of a BSN
or BFN. In some cases the antenna system is dual-polarized
and each feed horn has two input ports, one for each polarization. In this case a BSN or BFN is connected to each port, of
like polarization, of the feed horn array.
A BSN consists of a corporate tree arrangement of singlepole double-throw switches at each junction of the corporate
tree. There is a single-input port and usually 2m (where m is
an even integer) output ports.
This circuit permits the input port to be connected to any
one of the output ports. A BFN consists of a corporate tree
arrangement of variable power dividers and phase shifters.
There is a single-input port and usually 2m output ports. This
circuit permits the input signal to be divided among the output ports in accordance with any desired amplitude and
phase distribution. Thus a BFN permits one to produce any
beam shape that the converging device is capable of producing. This includes a pencil beam pointing anywhere in the
antennas field of view, or a single beam that covers the entire
field of view.
A control subsystem consists of electronic equipment capable of interpreting input signals and converting them into
control signals that operate the switches of a BSN, or the
power dividers and phase shifters of a BFN. The input signals
may be in the form of voltages and the addresses to which
they must be supplied, or in the form of beam pointing angles.
The actual form is dictated principally by the system within
which the MBA must operate. It is possible to include auxiliary equipment, in the control subsystem, that will detect the
presence of undesirable signals and reduce their amplitude
by shaping the antenna pattern.
GENERAL THEORY
The general theory and design procedures are similar if either
a reflector or a lens is used as the focusing device. The following discussion uses a lens system as a model and assumes
that it is transmitting. A general configuration is shown in
Fig. 1. Exciting the feed horn labeled #1 produces an antenna
beam pointing along the focal axis of the lens. Exciting the
feed horn labeled #2 produces a beam pointing in the direction , measured from the focal axis. Exciting the feed horn
labeled #3 produces a beam pointing in the direction , also
measured from the focal axis. Exciting additional feed horns,
located in the focal region of the lens, will produce additional
antenna beams. Usually the focal length of the lens is approximately equal to its diameter and the angle is approximately equal to . Usually the feed horns are arranged on
an equilateral triangular grid; consequently the centers of the
MULTIBEAM ANTENNAS
of required field of view. The approximate number of beams
Nb is given by
Lens
d
555
Nb = int[( / )2 ]
(2)
2
1
3
1
2
Focal axis
Note that function int(x) returns the largest integer value of
x. Gain at the edge of a coverage area can be expressed in
dBi, that is,
G 20log(D/) 8.2
(3)
Equation (3) assumes the antenna will be 40% efficient and
the gain at the edge of coverage is 4.2 dB less than the gain
at the peak of the beam. Equations (1) and (3) can be combined to give the approximate relation
Typical MBA
Figure 1. Typical MBA.
G 20log(70/ ) 8.2 dBi
antenna beams are located on an equilateral triangular grid.
It is customary to represent this array of beams as shown in
Fig. 2. Each circle represents a beam. The area within the
circle represents the coverage area of that beam. The lens diameter D determines the angular diameter d of the coverage
area, in accordance with the approximate relationship
d = 70/D
(1)
where is the operating wavelength. The diameter d, of the
feed horn aperture, determines the angular separation between adjacent beams, the intensity at the edge of the lens
aperture, and that amount of the feed horn energy that is not
intercepted by the lens. The latter loss is referred to as spillover loss, or simply spillover. Appropriate design of an MBA
determines that lens diameter that results in an acceptable
balance between spillover and antenna gain at the edge of a
coverage area. Minimum antenna gain usually occurs at the
edge of the coverage area.
Number of Beams and Minimum Gain
An MBA can have a few, tens, or hundreds of beams, depending on the required antenna gain G, the angular diameter of the required coverage area, and the angular diameter
Coverage area
Field of view
Beam coverage areas
Figure 2. Beam coverage areas.
(4)
Recall that is expressed in degrees.
Exciting a Cluster of Feed Horns
In deriving Eq. (3) it is assumed that the antenna efficiency
is 40%. This is low when compared to a typical single-beam
antenna. It is low because of the spillover that results in a
tolerable edge of coverage gain. Exciting a cluster of feed
horns, instead of a single feed horn, increases the antenna
efficiency substantially. Instead of exciting a single feed horn,
the six adjacent feed horns surrounding it are also excited but
with about 10 dB lower intensity. This increases the halfpower beamwidth of the beam and decreases the spillover by
about 2 dB. Loss in peak gain due to the increase in beamwidth is more than compensated by the reduction in spillover.
Gain at the edge of coverage is increased because the peak
gain and the beamwidth of the antenna beam are increased
and because the spillover is decreased. Exciting a cluster of
seven feeds increases the antenna efficiency to 60%, lowers
the sidelobes adjacent to the main beam about 10 dB, and
modifies Eq. (3) to
G7 20log(D/) 6.2 dBi
(5)
When a cluster of feed horns is excited to produce a single
beam, the number of beams Nb given by Eq. (2) must be increased. That is, a ring of beam must be added to surround
those required to cover the field of view when a single beam
is excited. Thus Eq. (2) becomes
Nb7 int[( / )2 + /]
(6)
Exciting three or more feed horns to produce a single beam
will increase the antenna gain and edge of a coverage area.
Experience has indicated that exciting a cluster of seven feed
horns is a best compromise between the increased complexity
of the BSN, or BFN, and the increase in edge of coverage gain.
When exciting a cluster of three or more feed horns, Eq. (6)
gives the number of beams.
When more than one feed is excited to produce a single
beam, the term beam becomes ambiguous. Consequently it is
common to refer to the beam produced when a single feed
horn is excited as a beam port. In the foregoing discussion,
exciting a cluster of seven beam ports produces a single beam;
and the number of beam ports required to cover the field of
556
MULTICAST ALGORITHMS
view is given by Eq. (5). Equation (2) gives the number of
beams required to cover the field of view.
W. G. Scott, H. J. Luh, and W. E. Matthews, Design tradeoffs for
multibeam antennas in communication satellites, presented at
1976 Int. Conf. Commun., Philadelphia, PA, 1976.
Multiple-Aperture MBA
L. J. Ricardi, Satellite antennas, in R. C. Johnson and H. Jasik (eds.),
Antenna Engineering Handbook, 2nd ed., New York: McGraw-Hill,
1984, Chap. 3.
Using three or more lenses is another way to reduce spillover
loss and increase edge of coverage gain. For example, using
four lenses instead of one permits about one quarter of the
beams to be produced by each lens. In addition any one of the
four lenses lens does not produce adjacent beams in the antenna field of view. This permits the feed horn aperture to
be significantly larger and reduces the spillover by 2.2 dB,
resulting in a 2.2 dB increase in gain at the edge of coverage. The resulting antenna aperture is four times that of a
single-aperture MBA, and pattern shaping by exciting more
than one beam port is very difficult if impossible. This is because the distribution or the lens apertures introduces a
grating lobe effect, which may not be tolerable.
Use of three lenses in an MBA will reduce spillover 1.2
dB. In this case, one third of the beam ports are located in
each lens. As with a four-lens MBA, a single-beam port is
excited to produce an antenna beam. An MBA using eight
lenses was designed and used to improve the spatial discrimination by producing a null in the direction of an interfering
signal source, while providing adequate gain in the direction
of a nearby desired signal source. This type antenna is more
appropriately referred to as a thinned phased array, using an
MBA as each element.
Unfocused Aperture
Using a lens with diameter much larger than the value implied by Eq. (1) could reduce spillover loss. The lens is then
designed to produce a beam that is broader than a diffractionlimited beam. This might be called a flat-nosed beam. The
increased size of the lens reduces the spillover to a negligible
level and defocusing the lens broadens its beam. Performance of this MBA configuration is comparable to that of an
MBA that uses four lenses.
BIBLIOGRAPHY
A. R. Dion and L. J. Ricardi, A variable-coverage satellite antenna
system, Proc. IEEE, 59: 252262, 1971.
J. W. Duncan, S. J. Hamada, and P. G. Ingerson, Dual polarization
multiple beam antenna for frequency reuse satellites, AIAA/CASI
6th Commun. Satellite Systems Conf., Montreal, Canada, 1976.
IEEE Definitions of Terms for Antennas, IEEE Standards 45, 1973.
J. T. Mayhan and L. J. Ricardi, Physical limitations of interference
reduction by antenna pattern shaping, IEEE Trans. Antennas Propag., AP-23, 1975.
J. T. Mayhan, Nulling limitations for a multiple-beam antenna, IEEE
Trans. Antennas Propag., AP-24, 1976.
B. M. Potts, Radiation pattern calculations for a waveguide lens multiple-beam antenna operating in the AJ mode, MIT, Lincoln Laboratory, Technical Note 1975-25, 1976.
L. J. Ricardi et al., Some characteristics of a communication satellite
multiple-beam antenna, Defense Tech. Inf. Center AD/A-006405,
1975.
L. J. Ricardi, Methodology of assessing antenna performance, Lincoln
Laboratory Technical Note 1978-24, 1978.
L. J. Ricardi, Multiple beam antennas, in A. W. Rudge et al. (eds.),
The Handbook of Antenna Design, Stevenage, UK: Peter Peregrinus, 1982, Vol. 1, Chap. 6.
LEON J. RICARDI
Creative Engineering
J. Webster (ed.), Wiley Encyclopedia of Electrical and Electronics Engineering
c 1999 John Wiley & Sons, Inc.
Copyright
RADAR ANTENNAS
Although the principle of radar has been known and advocated by many scientists since the early 1900s, it
did not come into its own until its widespread development and application in World War II when the word
RADAR was coined. Since then, the use of radar has expanded prodigiously. Important applications are found
not only in numerous military problems but also in many public and commercial areas (e.g., civil air-traffic
control, aircraft navigation, ship safety, spacecraft, ground detection, and remote sensing of the environment).
It is even applied to law enforcement when police use radar to check the speed of cars.
The major types of radar are pulse radar, high-range-resolution radar, continuous waveform (CW) radar,
moving target indication (MTI) radar, airborne radar, tracking radar, pulse Doppler radar, ground penetrating
radar, synthetic aperture radar (SAR), and inverse SAR. Each of these radar types employs a different characteristic waveform, which is determined by the corresponding radar antenna and signal-processing method.
The radar antenna is an important component of radar. For different radar types, the requirement for
antenna performances and functions is varied. Because the subject of radar antennas is very extensive, only
selected types and performances of radar antennas are discussed in this article. The reader should consult the
bibliography for further details (1,2,3,4,5).
Radar antennas serve as transduces to couple electromagnetic (EM) energy into the atmosphere. A simple
description of the antenna in a radar system is shown in Fig. 1. The functions of the antenna when transmitting
are to concentrate the radiated energy from the transmitter into a shaped directive beam in a predetermined
direction. When receiving, the antenna again forms a beam in a particular direction and collects the energy
contained in the reflected target echo signals. Received energy is sent to the receiver via transmission lines.
Therefore, the primary purpose of the antenna is to determine accurately the angular direction of the target
in both transmit and receive modes. For this purpose, a highly directive beamwidth is required, not only to
achieve accurate angular targets but also to resolve closely located targets. This important feature of radar
antennas can be quantitatively expressed in terms of the beamwidth, gain, and sidelobes.
Generally, the functions of radar antennas can be classified into beam scanning and target tracking. Beam
scanning fulfills the search or surveillance function of a radar. It requires that the narrow directive beam be
scanned rapidly and repeatedly over the region where the detected target may appear. Target tracking, however,
requires that the narrow beam follow the target once it has been detected. In some radar systems, particularly
airborne radars, the antenna is designed to perform both searching and tracking functions.
The preceding functional description of radar antennas implies that a single antenna is used for both
transmitting and receiving via the duplexer, a device that permits both transmission and reception of EM
waves with the single antenna. This is true for most radar systems, but there are exceptions (e.g., the bistatic
radars in which separate transmit and receive antennas are used). This article considers only the single
antenna.
Radar antennas are mainly classified into two types: quasi-optical antennas, which can be analyzed by
optics methods like geometrical optics and physical optics, and array antennas. In some cases, small antennas
are used (e.g., in ground-penetrating radars). Quasi-optical antennas include reflector antennas and lens
antennas. The former are still widely used for radar, whereas lens antennas, although still used in some
1
RADAR ANTENNAS
Fig. 1. Basic elements in a radar system in which the antenna is an important part. It can be served as transmitting and
receiving.
communication and military applications, are no longer used in modern radar systems. Therefore, because of
limited space, we discuss only reflector antennas and array antennas.
Basic Parameters and Requirements
Consider a radar antenna located at the origin of a spherical coordinate system as shown in Fig. 2. The
observation point is on a sphere having a very large radius R, which is located in the far field of the antenna.
Assuming that the antenna is reciprocal, so that the transmit and receive patterns are identical. When the
antenna is transmitting, its radiation efficiency is defined as
where P0 is the total power consumed by the antenna, and Pr is the power radiated by the antenna. The later
can be expressed by the radiation intensity (, )
in which is the azimuth angle and is the elevation angle, as shown in Fig. 2. From Eq. (2), the average
radiation intensity is easily obtained
From Eqs. (1)(3), we can define some basic parameters of an antenna. Directivity is a measure of the
ability of the antenna to concentrate the radiated power in a particular direction. Thus it can be defined as the
RADAR ANTENNAS
Fig. 2. Spherical coordinate system. The azimuth angle and elevation angle are adopted by most antennas.
ratio of the achieved radiation intensity in the direction to the average:
In practice, one is usually interested in the maximum directivity of the main lobe.
Gain is another important parameter of an antenna. It represents the ability to concentrate the power
accepted by the antenna in a particular direction:
Thus antenna gain is always less than directivity except for a lossless and reflectionless antenna (= 1). Again,
the peak value of the gain, G0 , is of more interest in practice.
The aperture of an antenna is its physical area projected on a plane perpendicular to the desired direction.
If the antenna is lossless and the aperture of area A is uniformly illuminated with equiphase, the directivity is
given by
which is the maximum available gain from an aperture A. Practical antennas are not uniformly illuminated,
but they have a maximum in the center of the aperture and are less tapered toward the edges in order to reduce
the sidelobes of the pattern. In this case, an effective aperture is defined from the directivity
The concept of effective aperture is very useful when considering the antenna in its receiving mode, which
measures the effective absorption area of the antenna to an incident plane wave. Let Ae be the peak value of
RADAR ANTENNAS
Ae (, ). Clearly, the effective aperture Ae is always less than the physical aperture A by a factor a
which is usually called the aperture efficiency. From the definitions in Eqs. (6)(8), we can see that the aperture
efficiency measures only how effectively a given aperture is used, but does not involve the EM energy loss.
Therefore, the efficiency of the antenna should be the multiplication of a and antenna losses L .
When the power radiation intensity (, ), the directivity D(, ), the gain G(, ), and the effective
aperture Ae (, ) are normalized to their peak values, they will be identical and are called the antenna radiation
pattern, which in fact represents the EM-energy distribution in three-dimensional (3-D) angular space. There
are several ways to plot the radiation pattern: rectangular or polar, voltage intensity or power density, absolute
value or dB value, or power per unit solid angle. However, the 3-D plots of radiation pattern require extensive
data. Thus two-dimensional (2-D) plots are usually adopted as a result of the ease in measurement and plotting.
Generally, the two cuts of 3-D plots in the principal elevation and azimuth planes are sufficient to describe the
pattern performance of an antenna, which is much less costly.
Instead of azimuth and elevation, E- and H-plane patterns are usually used in actual radar antennas
because the terms azimuth and elevation imply the earth-based reference coordinates, which are not applicable
to some space-based systems like airborne and satellite.
The main lobe of the radiation pattern is in the direction of maximum gain; all other lobes are called
sidelobes. In addition to the peak gain of the main lobe, two other important features of an antenna pattern are
the beamwidth of the main lobe, which is usually specified at the half-power level (3 dB), and the maximum
sidelobe level. The half-power beamwidth (HPBW) is usually a measure of the resolution of an antenna.
Therefore, if two identical targets at the same range can be separated by the HPBW, they are said to be
resolved in angle.
The beamwidth of an antenna is determined by the size of the antenna aperture as well as the amplitude
and phase distributions across the aperture. For a given distribution, the half-power beamwidth in a particular
plane is inversely proportional to the size of the aperture in that plane
where L is the aperture dimension and K is a constant for the given distribution, which is known as the
beamwidth factor. Accurate estimate of the beamwidth factor must take into account the aperture illumination function. However, K = 70 can give a rough estimate in most cases. For reflector-type antennas, a good
estimation to K is given by (6)
in which I is the absolute value of edge illumination in decibels.
From the half-power beamwidths in two orthogonal principle planes, one can obtain a practical formula
for predicting the gain of a relatively lossless antenna
where C is a unitless constant and 1 and 2 are HPBWs in degrees. The accurate value of C depends on the
antenna efficiency, but the rough estimate can be taken from 26,000 to 35,000 (1,7).
RADAR ANTENNAS
Fig. 3. Common reflector antenna types. (a) Paraboloid; (b) parabolic cylinder; (c) shaped reflector; (d) multiple beam; (e)
monopulse; (f) cassegrain.
Ideally, an antenna radiation pattern would consist of a single main lobe and no sidelobes. However, for all
practical radar antennas, it contains numerous sidelobes. These sidelobes can be a source of problems for the
radar system. For example, a radar for detecting low-flying aircraft targets can receive strong ground or ocean
echoes, which is called the clutter, through the sidelobes. This signal will interfere with the desired echoes
coming from the low radar cross-section targets through the main lobe. Another fatal problem of the sidelobes
is coming from jamming, which threatens most military radars. Therefore, it is often (but not always) desirable
to design radar antennas with sidelobes as low as possible to minimize such problems. However, low sidelobes
and high gain are competing requirements. Thus, the trade-off between sidelobes level, gain, and beamwidth
is an important consideration for designing radar antennas.
Reector Radar Antennas
Reflector antennas are one of the most important radar antennas and are widely used in practical radar
systems. They provide an economical way to distribute energy over a large aperture area and produce shaped
or pencil beams with high gain. In accordance with different demands, there are many types of reflector
antennas with different shapes and feed systems.
Types of Reectors. A wide variety of reflector antennas have been used in radar system. Figure 3
shows the most common reflector antennas.
Paraboloidal Reector Antennas. The paraboloidal reflector antenna shown in Fig. 3(a) is a classical
one with a round aperture. The theory and design of this kind have been well developed and discussed in the
RADAR ANTENNAS
literature (1,5). In this antenna, the feed is placed at the focus of the paraboloid. Then the spherical wave
emerging from the feed strikes on the reflector and is transformed into a plane wave after reflection to form
a pencil beam. Usually, the paraboloidal antenna provides a high gain and a minimum beamwidth with the
simplest and smallest feed.
An important parameter for this paraboloid is the ratio of the focal length f to the aperture diameter D.
For fixed aperture, when f increases, the reflector becomes flatter and introduces less distortion for polarization
and off-axis beam. However, a narrow primary beam that leads to a larger feed is also required. For example,
the size of the horn to feed a reflector of f /D = 1.0 is approximately four times that of f /D = 0.25. In general,
most reflectors are chosen to have a ratio between 0.25 and 0.5.
Besides the round aperture, a variety of reflector outlines are also used in practice (2,5). For example, an
oblong shape is chosen when the azimuth and elevation beamwidths have different requirements.
Parabolic-Cylinder Antennas. It is very common that only one of the azimuth and elevation beams must
be steerable or shaped while the other is not. In this case, a parabolic-cylinder reflector fed by a line source
shown in Fig. 3(b) can meet the requirement at a modest cost. The line source feed may be a parallel-plate lens,
a slotted waveguide, a phased array, etc. (1). Of course, the parabolic-cylinder antenna is not limited to these
applications. It can even be applied when both patterns are fixed in shape (e.g., the AN/TPS-63).
Parabolic cylinders suffer from large blockage if they are symmetrical, thus they are often built offset. Properly designed, however, a cylinder fed by an offset multiple-element line source can have excellent
performance (8).
Shaped-Reector Antennas. A good method of shaping the beam in one plane is using shaped reflectors
(9), as shown in Fig. 3(c), in which the surface itself is no longer a paraboloid. From Fig. 3(c), each portion of the
reflector is aimed in a different direction, making the analysis complicated. However, with modern computers,
arbitrary beam shapes can be approximated accurately by direct integration of the reflected primary pattern.
A disadvantage of the shaped reflector is that its effective aperture is very small because a large fraction
of its energy is radiated to a wide angle. However, it can reduce the blockage from the feed. In most shaped
reflectors, the feed is usually placed outside the secondary beam. Then the blockage can be virtually eliminated
even though the feed appears in front of the reflector, as shown in Fig. 3(c).
Comparing shaped-reflector antennas with parabolic-cylinder antennas, the shaped reflector is cheap and
simple to construct. But it has less control over the beam shape because only the phase of the wave across the
aperture is changed, whereas both phase and amplitude can be adjusted in the linear array for the parabolic
cylinder.
Multiple-Beam Reector Antennas. Very often the radar designer needs multiple beams to provide
extended coverage. This requirement can be achieved by using a reflector with multiple feeds (10), as shown
in Fig. 3(d). As we know, a feed at the focal point of a paraboloid forms a beam at the direction parallel to the
focal axis. The feeds displaced from the focal point, however, will form additional beams at distinct angles from
the axis with nearly full gain. Therefore, the coverage range of the radar can be increased. However, when the
feed is off the focus, a distortion occurs on the antenna pattern. As the angular displacement increases, the
beamwidth becomes larger, and additional sidelobes arise. The other limitation of this kind of antennas is that
the extended feeds increase the blockage of the aperture.
Monopulse Reector Antennas. An especially common multiple-beam radar antenna is the monopulse
reflector illustrated in Fig. 3(e). In fact, the illustration shows an amplitude comparison system, which is far
more prevalent than the phase comparison system. In the amplitude system, the sum of the even-numbered
source outputs fed by a single pulse forms a high-gain and low-sidelobe beam, and the difference forms a
precise deep null at the peak-value direction of sum pattern. The sum beam is used on bolt transmit and
receive to detect the target, whereas the difference beams provide angle determination in both azimuth and
elevation angles. Therefore, the monopulse reflector is normally used in tracking systems in which the antenna
is removable and keeps the target near the null in difference pattern.
RADAR ANTENNAS
Fig. 4. Offset dual reflector. The blockage of the subreflector can be virtually eliminated.
A trade-off to design the monopulse radar antenna is the choice of feed size. The conflict arises between
the goals of high sum-beam gain and large difference-beam slopes. For example, for a four-horn feed, the former
requires a small overall horn size to reduce the blockage, whereas the latter requires large individual horns.
Also, high difference sidelobes are required for tracking purpose in the design. To overcome these problems,
many methods have been presented and several suggested configurations of the feed are given in Ref. (11).
Multiple-Reector Antennas. Multiple-reflector systems can offer more degrees of flexibility by shaping
the primary beam and then overcoming some of the shortcomings of single reflector. Figure 3(f) shows the most
common type of multiple reflector, the Cassegrain antenna (12), in which the feed system is conveniently located
behind the main reflector. In the Cassegrain system, the subreflector is a hyperboloid. The feed is placed at the
left focus of the hyperboloid, and the paraboloid focus coincides with the right focus. From geometry optics, the
spherical wave radiated by the feed and then reflected to the main reflector by the hyperboloid is equivalent to
the one emerging directly from the right focus. Therefore, the wave reflected from the main reflector is also a
plane wave that forms a pencil beam. A similar antenna to Cassegrain is the Gregorian antenna, which uses
an ellipsoidal subreflector rather than a hyperboloidal one.
Aperture blockage can be very large for symmetrical Cassegrain antennas, as illustrated in Fig. 3(f). There
are two ways to reduce blockage. The first method is to make the diameter of the subreflector equal to that of
the feed, but the reduction is limited. The second method, which is generally used, is achieved by offsetting
both the feed and subreflector. For example, Fig. 4 shows an offset dual reflector with blockage and supporting
struts virtually eliminated. This antenna usually has very low sidelobes.
The geometry of the Cassegrain antenna is especially attractive for monopulse tracking radar because
the radiofrequency (RF) plumbing can be placed behind the reflector to avoid blocking of the aperture. Also, the
long runs of transmission line out to the feed at the focus of a conventional monopulse reflector are avoided.
Therefore, the Cassegrain antennas have usually a low noise temperature.
Feeds for Reector Antennas. In the previously discussed antennas, a spherical wave from the focus
of the paraboloidal reflectors will be transferred into a plane wave to form a pencil beam at the transmit mode,
while the incoming plane wave will be converted into spherical phase fronts centered at the focus on their
receive mode. Thus the feeds for reflector antennas must be point-source radiators so that they can radiate
spherical waves. Other requirements for reflector feeds include the proper illumination of the reflector with a
prescribed amplitude distribution, a minimum spillover, and the correct polarization with no or minimum cross
polarization. These are the basic factors chosen to design a feed. For a different application of the antennas,
other considerations must be included.
RADAR ANTENNAS
Fig. 5. Various types of feeds used in radar antennas.
The most common feeds for reflector antennas are the flared waveguide horns, which are very suitable
for microwave-frequency radar systems (L band and higher). At lower frequencies, dipole feeds, Yagi-Uda or
log-periodic feeds are sometimes used. For example, a linear array of dipoles can feed a parabolic-cylinder
reflector. Rectangular waveguide horns propagating the dominant TE01 mode are widely adopted because they
meet the high power and other requirements, although in some cases circular waveguide horns propagating
TE11 mode have also been used. These single-mode horns provide just one linear polarization. When such
additional requirments as polarization diversity, multiple beams, high beam efficiency, or ultralow sidelobes
are needed, the design for feeds becomes more complicated. In these cases, segmented horns, finned horns,
multimode horns, and corrugated horns must be used. Figure 5 illustrates some of the types, which have been
well studied (1,13).
In addition, the sitting of the feeds has also a great influence to the performance of the reflector antennas.
Figure 6 shows several placement styles of the feeds.
Reector Antennas Design and Analysis. The analysis and design of conventional paraboloidal
reflectors are well developed (1,5) and not stated here. However, most of the rotating search radars require
that the antenna patterns have a narrow azimuth beamwidth for angular resolution and a shaped elevation
beam for multiple requirements, for example, the ARSR-3 radar for air-traffic control and the AN/TPN-19 for
ground control. Thus, a shaped reflector is nearly always the practical choice. In this section, we will consider
the analysis and design of such kinds of antennas.
Requirements. When a search radar works, small nearby targets like birds and insects may clutter the
output of the radar because of the inverse fourth-power variation of signal strength with range. For example,
the radar that can detect a 1 m2 target at 100 nmi can detect a 10 4 m2 target at 10 nmi. [Nautical mile (nmi)
where 1 nmi = 1852 m is usually used as the range unit.] To avoid the clutter, the receiver gain should be
reduced at short range and increased at distant range so that the received signal from a target of constant size
remains unchanged with range. The programmed control of the receiver gain to keep a constant echo signal
strength is called sensitive time control (STC), which is an effective method of eliminating the radar echoes
from unwanted objects. Therefore, STC is the usual requirement for designing the search radar.
A typical range coverage requirement for the rotating search radars is shown in Fig. 7, which is an
idealized simplification of the STC requirement (14). It gives a relationship between the radar range, height,
RADAR ANTENNAS
Fig. 6. Various placements of the feeds in radar antennas.
and elevation angle. From Fig. 7, at low elevation angles (0.5 to 3.5 ), the maximum range is the critical
requirement. When the elevation angle increases (3.5 to 9.5 ), however, the height becomes the governing
requirement. In this period, the corresponding pattern is a cosecant square. At high elevation angles (9.5 to
30 ), the STC is used.
The two-way coverage requirement shown in Fig. 7 is inconvenient for antenna design. It can be converted
into a pattern requirement if we plot only the range against angle because the range is directly proportional
to the pattern amplitude, as shown in Fig. 8 (the solid line).
Analysis. Antenna analysis is very important not only in analyzing a given antenna but also in antenna
design. Because the accurate design is usually an iterative procedure, in each iterative cycle, an analysis is
needed.
There are two methods to analyze a given antenna: current-distribution method and aperture-field method
(5). For shaped antennas, however, the first one is more suitable. From the magnetic field H radiated by the
feed, one can get the induced current on the reflector surface J = 2I H. Then the radiated fields by the
reflector can be obtained by integration.
This type of accurate analysis is quite time consuming. In practical design, the antenna pattern and
gain are usually approximated in the first several iterative cycles. For rotating search radars, the antenna
pattern can be obtained directly from the coverage requirement; however, the designer must realize that the
range-angle plot represents the minimum range requirement. If some ripple is anticipated (normally about 1
dB), the design curve should be raised 1 dB in the height limit and STC coverage regions. For the range limit
region, a reasonable situation is achieved by assuming that a section or several sections of the reflector are
shaped so that the 3 dB points of the resulting pattern will coincide with the two corners of the range limit
portion, as shown in Fig. 8 (dashed line), which is just the realizable elevation pattern of the antenna.
For the gain of shaped beam considered here, we assume that the azimuth and elevation power patterns
are separable. Then using the definition in Eq. (5), the maximum gain (in main beam direction) can be expressed
10
RADAR ANTENNAS
Fig. 7. Typical two-way coverage requirement of the search radar, in which a relationship between the radar range,
height, and elevation angle is given.
as
where f az () and f el () are azimuth and elevation power patterns. For simplicity, the azimuth power pattern
can be approximated by a Gaussian function: f az () = exp[2.7726(/BWAZ)2 ]. Then
in which BWAZ is the azimuth half-power beamwidth. For the other integral in Fig. 8.
Design. For shaped-beam antennas, the shaping is generally in one plane (elevation), with a narrow
pattern of conventional design in the orthogonal plane (azimuth). The parabolic-cylinder antenna fed from a
line source is convenient for obtaining independent control of the patterns. However, it is much more expensive
and heavier than a reflector fed from a point source. In order to obtain a shaped beam, the antenna with a
point-source feed requires a reflector surface with double curvature. An easy but efficient way to do this is to
use an offset paraboloidal antenna by adding several shaped sections. Each successive section points slightly
higher than the preceding section, as shown in Fig. 9. Clearly, the projections of added sections are virtually
missing the feed. So there is no feed blockage. The design of the added sections can be done iteratively by using
a computer. However, the antennas width and height can be directly approximated from the antenna patterns
(14)
RADAR ANTENNAS
11
Fig. 8. Coverage requirement (solid line) and realizable pattern (dashed line).
in which BWEL is the elevation half-power beamwidth, and is the integral in the range-limit region.
Practical Consideration. In the analysis and design of radar antennas, some practical considerations
must be taken into account because they can affect the performances of the antennas. There are a lot of
practical details for different radar antennas, but here we consider only four common effects: feed-support
blockage, surface leakage, surface tolerance, and radomes.
Feed-Support Blockage. No matter whether the feed is placed in the front or rear, there is always
blockage from the feed supports. In the case of Cassegrain or Gregorian, the blockage comes from the support
of the subreflector. Like the blockage of feeds, the supports have two effects on the antennas performance: they
put a shadow in the desired distribution, and they scatter the intercepted power. Both effects can reduce the
gain and increase the sidelobe level.
The effective size of the blockage may be larger than its projected area depending upon the material and
polarization. For conducting supports, the effective size equals the actual dimension in the E plane, whereas
(where W and d are the width and thickness of a strut) in the H
the effective width becomes W +
plane (15). Therefore, the blockage of a strut of 0.125 square in cross section will be three times larger when
it is parallel to the E field than when it is parallel to the H field. In addition, various ways of sitting the
supports have different effects. Figure 10 shows three common ways of supporting the feeds. Note that the
tripod arrangement produces sidelobe ridges that are broader but 6 dB lower than those of the quadrapods
(15). The feed-support blockage can be reduced by using a dielectric strut.
Surface Leakage. Many reflectors are designed with a tubular grid or a wire-mesh instead of a solidmetal surface in order to reduce the wind resistance and the antennas weight. To accomplish this, the opening
12
RADAR ANTENNAS
Fig. 9. An example of shaped-beam antenna design.
Fig. 10. Some common feed supports used in radar antennas.
area in the structure should be as large as possible. However, the openings lead to surface leakage, from which
backlobe is produced. The larger the area of the opening is, the higher the backlobe is. Therefore, a suitable
design of the surface structure is important to the antennas performance. Figure 11 illustrates some of the
common reflector surfaces, in which the gap S between conductors must be much smaller than half wavelength
in the H plane so that the passage of EM energy will be below cutoff (14).
Surface Tolerance. Mechanical consideration to design reflector antennas is very important in a radar
system. In most cases, it is even more difficult and complicated than the electrical design, especially for a
high-performance radar. The surface tolerance is one important part of the mechanical design. The reflector
surface must be smooth enough so that it remains within close tolerances of the ideal surface. Typically, the
surface error must be less than /28.
The roughness of reflector surface can cause small phase error on the aperture. From the antenna tolerance
theory, the loss of gain due to phase error is approximated by Eq. (16):
RADAR ANTENNAS
13
Fig. 11. Common types of reflector surfaces that can reduce the wind resistance.
where G1 and G0 are gains with and without phase error, and 2 is the mean square phase error. From Eq. (16),
one clearly sees that for 1 dB loss of gain the phase error
must be less than 0.45 rad, which means the
reflector surface error must be less than /28.
Radomes. Antennas for ground-based radars are often subjected to high winds, icing, and temperature
extremes so they must be sheltered if they are to survive and perform under adverse weather conditions. The
shelter to protect antennas is called a radome. Antennas mounted on aircraft must also be housed within a
radome to offer protection from large aerodynamic loads. An ideal radome should be perfectly transparent to
the rf radiation from (or to) the antenna but mechanically strong if they are to provide the necessary protection.
These two requirements are usually competing, thus the design of radomes must be compromised (17).
The design of radomes for antennas can be divided into two separate classes, depending upon whether the
antenna is for airborne or ground-based (or ocean-based) applications. The airbone radome is characterized by
smaller size than ground-based radomes because the antennas that can be carried in an aircraft are generally
smaller. The airborne radome must be strong enough to form a part of the aircraft structure and usually it is
designed to conform to the aerodynamic shape of the aircraft, missile, or space vehicle in which it is to operate.
If the antenna scans inside the streamlined radome, the incident angle may be varied around the normal
direction. This will lead to great degradation of antenna performance.
A properly designed radome should distort the antenna pattern as little as possible. The presence of a
radome can affect the gain, beamwidth, sidelobe level, beam direction, the voltage standing-wave ratio (VSWR),
and antenna noise temperature. Sometimes in tracking radars, the change rate of the beam direction can be
very important.
14
RADAR ANTENNAS
Phased-Array Radar Antennas
Besides reflectors, phased-array antennas are also important in radar systems. An attractive advantage of
the phased-array antenna for radar applications is that it can steer a beam without having to move a large
mechanical structure. For example, when a single radar simultaneously carries out the functions of surveillance
and tracking, it must pause for an instant to confirm or disprove a possible alert as a surveillance radar, however,
it is required to follow several targets by pointing at each of them successively without any loss of time as a
tracking radar. In this case, a phased-array antenna is the best choice because it can steer the beams rapidly
by means of electronic control.
The first large steerable directive phase-array antenna for the reception of transatlantic shortwave communication was developed and installed by the Bell Telephone Laboratories in the late 1930s. However, it was
limited because the beam was scanned by mechanically actuated phase shifters. A major advance in phasedarray technology was made in the early 1950s with the replacement of the mechanically actuated phase shifters
by electronic phase shifters, from which the scanning in one angular coordinate can be electronically controlled.
The introduction of digitally switched phase shifters employing either ferrites or diodes in the early 1960s made
a significant improvement in the practice use, in which a phased array could be electronically steered in two
orthogonal angular coordinates. With the development of modern computer and solid-state microwave devices,
multifunction phased-array antennas have been developed.
Array antennas can take many different forms for various applications. However, they always include
some common components. Figure 12 shows a basic schematic of the phased-array antennas. Clearly, a phasedarray system contains an element array composed of many similar radiating elements, a power splitter, the
phase shifters, the transmitter and receiver, and a central computer. The radiating elements might be dipoles,
open-ended waveguides (or small horns), slots cut in waveguide, printed-circuit patches, or any other type of
antennas. On its transmission mode, the power splitter divides the energy from the transmitter to the various
elements of the array via the phase shifters. Then the phases and amplitudes of the array can be controlled
by the computer. Generally, the amplitude characteristic does not vary during scanning except for adaptive
and optimum arrays. Therefore, one can adjust the phase and amplitude taper on array by using the phase
distribution and the amplitude distribution to meet different uses.
Basic Characteristics. See Antenna arrays for a detailed analysis of phased-array antennas. Here we
give only some general characteristics important to the analysis and design of radar antennas.
For a general planar array, when the elements are spaced by half wavelength to avoid the generation
of grating lobes, the number of the radiating elements N for a pencil beam can be approximated from the
broadside beamwidth by (18)
where HPBW is the 3 dB beamwidth (in degree) of the antenna pattern. When the beam points in the broadside
direction to the aperture, the corresponding antenna gain is
where is the antenna efficiency, which accounts for antenna losses (L ) and reduction in gain caused by
weighting the elements with a nonuniform amplitude distribution (a ).
RADAR ANTENNAS
15
Fig. 12. General structure of a phased array in which the phase shifter driver provides different phases to the array.
When the antenna is scanning to an angle 0 , the scanned beamwidth is increased from the broadside
beamwidth
whereas the gain of the planar array is reduced to that of the projected aperture
These characteristics can provide an easy way to estimate quickly when analyzing and designing array
antennas.
Types of Arrays. The arrays used in radar systems can be classified by such different means as
geometry and functions. This subsection will discuss several typical arrays.
Linear Arrays. The linear array generates a fan beam when the phase relationships are such that the
radiation is perpendicular to the array. If the radiation is at some other angles, the antenna pattern is a
conical-shaped beam. The broadside linear-array antenna may be used where broad coverage in one plane and
a narrow beamwidth in the orthogonal plane are required. An active application of linear array is that it can
act as the feed for a parabolic-cylinder antenna, as discussed earlier.
Planar Arrays. The two-dimensional planar array is capable of steering the beams in two angular
coordinates. It is probably the array of most interest in radar applications because of its versatility. Generally,
a rectangular aperture can produce a fan-shaped beam, whereas a square or a circular aperture can generate a
pencil beam. The array may also be made to produce simultaneously many search and or tracking beams with
the same aperture. Most of arrays discussed later will be the planar arrays.
Conformal Arrays. Sometimes, a radar designer is required to place array elements on an arbitrary
surface to achieve a directive beam with good sidelobe level and efficiency, which can be easily scanned
16
RADAR ANTENNAS
electronically. For example, on an aircraft, arrays arranged along the nose, on the wings, or on the fuselage
would be attractive options. In this case, a conformal array, which conforms to the geometry of a nonplanar
surface, should be used (1).
In principle, the array on any surface can be made to radiate a beam in some given direction by applying
the proper phase, amplitude, and polarization at each element. In practice, however, it is very difficult to control
the beam shape and obtain low sidelobes from an arbitrary surface when the beam is electronically scanned.
Furthermore, the mechanisms for feeding the elements and generation of the phase-shifter command are also
more complicated than those of a planar array. Therefore, the application of conformal array is only possible
for some simple shapes like cylinder and cones.
Most of the work on conformal arrays has been with the cylinder because it has a geometry suitable for
antennas that scan 360 in azimuth. Although it is of a relatively simple shape compared with others, the
properties of the cylindrical array are not as suitable as those of planar array. In the cylindrical array, the
radiation pattern cannot be separated into an element factor and an array factor as they can in the planar
array. Considering the additional difficulties in the practical control, the conformal arrays are not widely used
in the radar systems.
Thinned Arrays. Most array antennas have equal spacings between adjacent elements. In order to
obtain a given beamwidth with considerably fewer elements, a thinned array has sometimes been considered.
Generally, the shape of the main beam has little distortion after thinning. However, the average sidelobes are
degraded in proportion to the number of elements removed. Figure 13 shows an example of thinned arrays, in
which 77.5% of the elements are randomly removed from a regular grid (19). Then the gain, which is caused by
the actual element number, will drop by 6.5 dB, while almost 77.5% of the power is delivered to the sidelobes
because the main beam is nearly unchanged.
Therefore, the design of a thinned array consists of selecting the size of the aperture to give the desired
beamwidth, selecting the number of actual elements to give the desired gain, and arranging the element
distribution to obtain some required properties of the sidelobes.
The thinned arrays have seen only limited applications in radar system because of the reduction in gain
and the increases in sidelobes. Instead of thinning elements, the number of phase shifters can also be thinned,
in which some of the phase shifters in the array can be used to adjust the phase of more than one element. A
50% saving of the phase shifters might be possible (20).
Phase-Scan Arrays. The beam of an antenna points to the direction perpendicular to the phase front.
In phase-scan arrays, the phase front is adjusted by controlling the phase of each radiating element to steer
the beam, as shown in Fig. 14(a). The phase shifters are electronically actuated to perform rapid scanning and
are adjusted in phase to a value between 0 and 2 rad. If the spacing between two adjacent elements is S, the
incremental phase shift will be = (2/)S sin 0 , in which 0 is the scan angle. Because the phase shifters
have phase shift that is virtually independent of frequency, the scan angle 0 is frequency-dependent.
Time-Delay-Scan Arrays. Phase scanning was seen to be frequency-sensitive. Time-delay scanning
provides a way that is independent of frequency. Instead of the phase shifters, delay lines are used in the
time-delay-scan arrays, as illustrated in Fig. 14(b). For the spacing distance between two adjacent elements S,
the incremental time delay from element to element is t = (S/c) sin 0 , in which c is the velocity of propagation
of wave and 0 is again the scan angle. The time-delay circuits accompanied by all radiating elements are
normally too cumbersome. A reasonable compromise solution is sharing one time-delay network by a group of
elements in which each element has its own phase shifter. However, this will cause a grating lobe problem.
Frequency-Scan Arrays. A change in frequency of an EM signal propagating along a transmission line
produces a change in phase. This provides a simple way for obtaining the electronic phase shift, as shown in
Fig. 14(c). In this case, the incremental phase shift is
RADAR ANTENNAS
17
Fig. 13. (a) Thinned array with a 4000-element grid containing 900 elements; (b) typical pattern for a thinned array in
which SA is the average sidelobe level. [From Willey (15, Figs. 5 and 6) courtesy of Bendix Radio Div., Bendix Corporation.]
in which l is the length of line connecting adjacent elements. When the beam points broadside (0 = 0), Eq. (21)
yields m = l/0 , where 0 is the wavelength corresponding to the beam position at broadside. Then the direction
of beam pointing can be obtained from Eq. (21)
Comparing with other ways, the frequency-scanning system is relatively inexpensive and easy to implement.
Frequency-scan arrays have been developed and used to provide elevation-angle scanning. Combined with the
mechanical horizontal rotation, it served as 3-D radars in the past.
18
RADAR ANTENNAS
Fig. 14. Scanning of arrays. (a) Phase scan; (b) time-delay scan; (c) frequency scan.
Multiple-Beam-Forming Arrays. One of the properties of the phased array is the ability to generate
multiple independent beams simultaneously from a single aperture. In principle, an N-element array can
generate N independent beams. Multiple beams allow parallel operation and can obtain a higher data rate
than a single beam. The multiple beams may be fixed in space, steered independently, or steered as a group. In
some applications, multiple beams are generated on receive and connected to separate receivers, whereas only
one wide radiation pattern is needed on transmit so that it can cover all the received beams. Such multibeam
systems have found application with mechanical rotation for 3-D coverage.
Low-Sidelobe Arrays. A low sidelobes level is one of the most important requirements for the radar
antennas, especially for military radars that may be threatened by the jamming. Therefore, the low sidelobe
level has been and will be of great interests to radar designers. By modern technology, an ultralow sidelobe has
been achieved in the Airborne Warning and Control System (AWACS) radar, which now supports sidelobe levels
of more than 50 dB below the main-beam peak (21). However, as mentioned earlier, the low sidelobe and high
gain are competing requirements. Thus the cost to achieve low sidelobe is a reduction in gain and an increase in
beamwidth. Considering the practical implementation, the low sidelobes also require a high-tolerance control
and a good environment free from obstructions that could increase the sidelobes so low-sidelobe antennas are
usually expensive. In spite of these drawbacks, the trend to low-sidelobe arrays has accelerated because low
sidelobes provide an excellent deterrent to electronic countermeasures (ECM).
From the aperture theory, antenna sidelobes are related to the aperture amplitude distribution. For
phased arrays, the amplitude taper is generally fixed and does not change. However, the amplitude of each
radiating element can be controlled individually by the computer, which makes it possible to achieve very low
sidelobes.
RADAR ANTENNAS
19
The relation between the antenna pattern and the aperture illumination has been studied extensively
in many literatures. It has been shown that the far-field pattern is just the Fourier transform of the aperture
distribution. Some typical illumination functions and their corresponding gains, sidelobes, and beamwidths
can be found in Refs. 5 and 18. For low-sidelobe arrays, the Taylor illumination for the sum patterns and
the Bayliss illumination, a derivative form of the Taylor illumination, for the difference patterns are usually
chosen because they can provide low sidelobes at a minimum loss in gain. Note that the sidelobes predicted by
the aperture illuminations are suitable for perfect phase and amplitude distributions (error free). In practice,
aperture illuminations must be chosen to provide peak sidelobes below the requirements to allow for errors.
The effect of errors on the design will be mentioned next.
Practical Considerations. The array theory gives only the theoretical analysis and design of the
phased arrays. For practical use in radar systems, a lot of details must be considered. In this subsection, some
important considerations are addressed.
Phase Shifters. As already stated, there are three basic techniques to steer the beam electronically:
frequency scanning, time-delay scanning, and phase scanning with phase shifters. In practice, the use of phase
shifters is the most popular technique, but many other techniques have been developed for a variety of phase
shifters (22). Generally, the phase shifters can be classified into two categories: reciprocal and nonreciprocal.
For reciprocal phase shifters, the phase change does not depend on the direction of propagation. Therefore, it
is not needed to switch the phase states between transmit and receive if the reciprocal phase shifters are used.
For a nonreciprocal phase shifter, however, it is necessary to change the phase states between transmit and
receive. Typically, it takes a few microseconds to switch the nonreciprocal phase shifters. During this period,
the radar cannot detect targets.
A perfect phase shifter could change its phase rapidly; handle high power; require control signals of little
power; be of low loss, light weight, small size, and reasonable cost; and have a long life. But in practice, no
one device is universal enough to meet all the requirements. Various phase shifters possess these properties
in varying degrees. Presently, three types of phase shifters are used in the phased-array antennas: the diode
phasers which are all reciprocal, the nonreciprocal ferrite phasers, and the reciprocal ferrite phasers. Each
of the three types has its own advantage. The choice of phase shifters is highly dependent on the radar
requirements.
Array Elements and Matching. Almost any type of radiating antenna element can be considered for an
array antenna. In practice, the dipole, open-ended waveguide (or small horn) and slotted waveguide have found
wide applications. Detailed descriptions of these antennas have been investigated in the standard textbooks
(1,5). However, note that the properties of a radiating element in an array is significantly different from its
properties when in free space. For example, the radiation resistance of a half-wavelength dipole in free space
is 73
, but when it is in an infinite array with half-wavelength spacing and a back screen of quarter-wave
separation, it will be 153
when the beam is broadside. The impedance also varies with scan angles.
The change of impedance with scan angle makes the matching of an array antenna difficult. There are
conditions where an antenna that is well matched at broadside may have some angles at which most of the
power is reflected. Unlike a conventional antenna, the mismatch of array antennas will affect both the level of
the radiated power and the shape of the antenna pattern.
Mutual Coupling. If two radiating elements are widely separated, the energy coupled between them is
small, and the influence of one to the other on the current excitation and pattern can be negligible. However,
when the elements are placed closer, their coupling will increase. Generally, the strength of the coupling in a
phased array is related to the distance between elements, the pattern of the elements, and the structure of the
array. For example, the radiation pattern of a dipole has a null at = 90 and a peak value at = 0 . Thus, the
dipoles in a straight line are loosely coupled, whereas the dipoles parallel to each other have a strong coupling.
For a phased array, the effect of coupling between elements in mainly focused on the pattern and impedance
of the element, which are usually called the active element pattern and the active element impedance. The
exact analysis of coupling in practical arrays is rather complicated. A convenient way is to assume that the
20
RADAR ANTENNAS
array is infinite in extent and has a uniform amplitude distribution and a linear phase taper from element
to element. Under such an assumption, every element in the array has the same environment; thereby a
significant simplification in the calculations can be made. It has been shown that the infinite-array model has
a good prediction to the actual arrays. Even for a modest array that has fewer than 100 elements, the predicted
results can also give reasonable agreement (23).
Errors in Arrays. As indicated in the section entitled Low-Sidelobe Arrays, the errors in the amplitude
and phase of the current at individual elements of the array may cause distortion of the radiation pattern.
Additional factors of the error include the missing or inoperative elements, rotation or translation of an element
from its correct position, and variations in the individual element patterns. These errors can result in a decrease
in gain, increase in sidelobes, and shift in the direction of the main beam.
When errors occur in the phase and amplitude of the aperture, the energy will be removed from the main
beam and distributed to the sidelobes. If the errors are purely random, they will produce random effects on the
main beam and all sidelobes. When the errors are correlated, the sidelobe energy will be lumped at discrete
locations in the far field. Usually, the correlated errors provide higher sidelobes than the random errors, but
they are located at some certain directions only. Both of the correlated and random errors should be considered
in the practical design of phased arrays (18,24).
Applications. In many cases, the phased-array antennas have been of considerable interest in the radar
systems because they have different properties from those of reflector antennas. Hence, a number of phasedarray radar systems have been built for different use (18,25). Examples include the AN/SPS-33 radar for the
purpose as aircraft surveillance from on-board ship; the AN/FPS-85 for satellite surveillance; the PAR and
MSR for ballistic missile defense; the AN/SPY-1 and PATRIOT for air defense; the EAR (Electronically Agile
Radar) for airborne bomber; the AN/TPN-19 and AN/TPS-32 used in aircraft landing systems; the AWACS and
AN/TPS-70 for airborne warning and control systems; the MESAR as a multifunction electronically scanned
adaptive radar; the AN/TPQ-37 Firefinder radar; the PAVE PAWS for providing early warning of ballistic
missiles and performing satellite tracking; the COBRA DANE for tracking of ballistic missile; and the COBRA
JUDY for collecting data on foreign ballistic missile tests (18,25).
Although the array antennas have many unique characteristics that make them candidates for use in
radar systems, they are expensive. As technology advances, the costs can be reduced, particularly in the areas
of phase shifters and drivers. In the meantime, the high demand for better performance with lower sidelobes
and wider bandwidth makes the costs high. In the future, the great potential for cost reduction may be the
application of solid-state systems with a transmit/receive module at each element.
BIBLIOGRAPHY
1.
2.
3.
4.
5.
6.
7.
8.
R. C. Johnson, H. Jasik (eds.), Antennas Engineering Handbook, 2nd ed., New York: McGraw-Hill, 1984.
M. I. Skolnik (ed.), Radar Handbook, 2nd ed., New York: McGraw-Hill, 1990.
L. J. Cantafio (ed.), Space-Based Radar Handbook, Norwood, MA: Artech House, 1989.
M. I. Skolnik, Introduction to Radar Systems, 2nd ed., New York: McGraw-Hill, 1980.
C. A. Balanis, Antenna Theory Analysis and Design, 2nd ed., New York: Wiley, 1997.
M. J. Komen, Use simple equations to calculate beam width, Microwaves, 20 (13): 6163, 1981.
W. L. Stutzman, G. A. Thiele, Antenna Theory and Design, New York: Wiley, 1981.
R. L. Fante et al., A parabolic cylinder antenna with very low sidelobes, IEEE Trans. Antennas Propag., AP-28: 5359,
1980.
9. K. Aoki et al., Design method for an offset dual-shaped reflector antenna with high efficiency and an elliptical beam,
IEEE Proc. H. Microw. Antennas Propag., 140 (2): 121128, 1993.
10. Y. T. Lo, On the beam deviation factor of a parabolic reflector, IRE Trans. Antennas Propag., AP-8: 347349, 1960.
11. P. W. Hannan, P. A. Loth, A monopulse antenna having independent optimization of the sum and difference modes,
IRE Int. Conv. Rev., Part 1, 1961, pp. 5760.
RADAR ANTENNAS
21
12. C. Granet, Designing axially symmetric Cassegrain or Gregorian dual-reflector antennas from combinations of prescribed geometric parameters, IEEE Antennas Propag. Mag., 40 (2): 7682, 1998.
13. A. W. Love, Electromagnetic Horn Antenna, New York: IEEE Press, 1986.
14. H. E. Schrank, G. E. Evans, D. Davis, Reflector antennas, in M. I. Skolnik (ed.), Radar Handbook, 2nd ed., New York:
McGraw-Hill, 1990.
15. C. L. Gray, Estimating the effect of feed support member blocking on aperture gain and sidelobe level, Microwave J.,
(3), 8891, 1964.
16. J. Ruze, Antenna tolerance theoryA review, Proc. IEEE, 54: 633640, 1966.
17. J. D. Walton, Jr. (ed.), Radome Engineering Handbook, New York: Dekker, 1970.
18. T. C. Cheston, J. Frank, Phased array radar antennas, in M. I. Skolnik (ed.), Radar Handbook, 2nd ed., New York:
McGraw-Hill, 1990.
19. R. E. Willey, Space tapering of linear and planar arrays, IRE Trans. Antennas Propag., AP-10: 369377, 1962.
20. N. Goto, D. K. Cheng, Phase shifter thinning and sidelobe reduction for large phased array, IEEE Trans. Antennas
Propag., AP-24: 139143, 1976.
21. G. E. Evans, H. E. Schrank, Low sidelobe radar antennas, Microwave J., 26 (7): 109117, 1983.
22. M. R. Iravani, D. Maratukulam, Review of semiconductor-controlled (static) phase shifters for power systems applications, IEEE Trans. Power Syst., 9: 18331839, 1994.
23. J. Frank, Phased array antenna development, Appl. Phys. Lab. Rep., Johns Hopkins Univ., TG 882, March 1967.
24. J. L. Allen, The theory of array antennas, MIT Lincoln Lab. Rep. 323, July 1963.
25. P. E. Rawlinson, H. R. Ward, Radar antennas, in R. C. Johnson and H. Jasik (eds.), Antenna Engineering Handbook,
2nd ed., New York: McGraw-Hill, 1984.
TIE JUN CUI
WENG CHO CHEW
FU-CHIARNG CHEN
University of Illinois at UrbanaChampaign
J. Webster (ed.), Wiley Encyclopedia of Electrical and Electronics Engineering
c 1999 John Wiley & Sons, Inc.
Copyright
RADAR CROSS-SECTION
Radar cross-section (RCS) is a measure of the magnitude of the scattered electromagnetic wave from a body
relative to the magnitude of the wave incident on the body. To many people RCS is synonymous with stealth.
Stealth technology has become an integral part of all military systems since it was first employed in the 1970s.
However, RCS is just one aspect of stealth, the design philosophy that seeks to make a platform (such as an
aircraft, ship, or ground vehicle) undetectable to a wide range of sensors. Such a platform is referred to as
low-observable (LO). In addition to radar, the infrared, acoustic, and visible signatures must be controlled.
In the past, the RCS has received the most attention because the detection range of radar is generally much
greater than that of other sensors. Undetectability does not necessarily imply a very low signature, only that
the target must be indiscernible against the background. In the visible region (wavelengths visible to the eye)
stealth has been used since World War II in the form of camouflage.
A typical radar and target arrangement is shown in Fig. 1. The transmitter illuminates the target from
a distance Rt . If the body is sufficiently far from the transmitting antenna, the incident spherical wave front
will be approximately planar across the target. The incident wave induces currents on the surface of the body,
and even within its volume if the material is not a perfect electric conductor. The induced currents give rise to
a scattered field, which is detected by the receiver at a distance Rr .
The arrangement shown in Fig. 1 is referred to as bistatic; the directions of the transmitter and receiver
are different as viewed from the target. The monostatic case is more common; that is, the directions of the
transmitter and receiver are the same as viewed from the target. For a truly monostatic radar the transmit
and receive channels should use the same antenna. Frequently the radar will use separate antennas for
transmit and receive, but they will be closely spaced on the same platform. This configuration is referred to as
quasimonostatic, and can be considered as monostatic for the purpose of specifying the RCS of a distant target.
The majority of radars currently in use for both military and civilian applications are monostatic.
Most radars are designed to provide information on the targets location and velocity. Location is given in
terms of the range (radial distance) from the radar along with direction. The range to the target is obtained by a
time delay measurement. An electromagnetic wave is transmitted in the direction of the target, and the round
trip travel time converted to distance using the known velocity of propagation of the wave. The direction of the
wave is obtained from antenna pointing. Velocity can be measured using the frequency shift due to the Doppler
effect or two range measurements at closely spaced times to obtain the targets range rate. Any measurement
of the target parameters assumes that the reflected wave is strong enough to be detected and processed by the
radar. Search radars are specifically designed for this purpose. They are used to scan large volumes of space
for targets at great distances (hundreds or even thousands of kilometers). Once a target has been detected it is
handed over to a tracking radar. Modern multifunction radars are designed to perform both search and track
functions.
Although both the search and track operations are more difficult for targets with low RCS, the primary
objective of RCS reduction is to defeat search radars (i.e., avoid detection). The role of the RCS in the detection
of a target is demonstrated by the basic form of the radar equation derived later. It shows that when all other
1
RADAR CROSS-SECTION
Fig. 1. Bistatic RCS refers to an arrangement where the transmitting and receiving antennas are at different locations
as viewed from the target. For the monostatic case the transmitting and receiving antennas are co-located.
radar system parameters are fixed, the maximum detection range of a target varies as the fourth root of RCS.
Thus by reducing the target RCS, the range at which it is first seen by the radar is also reduced.
The RCS of modern aircraft and ships has been significantly reduced, but in response to the lower cross
sections, more sensitive radars have evolved. The RCS designer is at a disadvantage in this respect because
a truly LO platform must be designed from the start. Stealth cannot simply be added on after all other
design tradeoffs have been made. The entire design and test cycle of an aircraft can be a decade, and its
operational life several decades. On the other hand, the radar designer may be able to significantly increase
the capability of a radar by simply replacing a system component such as an antenna or signal processor.
These are relatively minor changes and can be integrated into deployed radars in a couple of years, thereby
decreasing the effectiveness of the LO platform for the rest of its operational lifetime.
The scattering properties of the target are only one of several considerations that enter into the selection of
the operating frequency of a radar (1,2). Others are size constraints, antenna gain and beam width, transmitter
power, ambient noise, Doppler shift, and atmospheric attenuation. The frequency spectrum is divided into bands
with letter designations as shown in Table 1. A few long-range ballistic missile defense radars operate in the
300 MHz region (UHF), but most others use frequencies greater than 1 GHz (L band and above). Devices at
low frequencies are larger than similar devices at a higher frequency, and therefore the systems components
require more volume. Likewise the antenna must be larger at a low frequency than at a high frequency to
achieve the same beam width and gain. Low frequencies are capable of handling more power because the
applied voltages can be higher without causing breakdown. Finally, ambient noise is lowest in the 1 to 10 GHz
range, and low-altitude atmospheric attenuation favors frequencies below 18 GHz.
RADAR CROSS-SECTION
The function of the radar also influences the choice of frequency. For instance, an airborne imaging radar
must be compact and lightweight, but also have a narrow antenna beam for resolving scatters. These all favor
a high frequency. On the other hand, a ground-based search radar must radiate high power to achieve a large
detection range. In this case the larger and heavier equipment that is necessary for high-power operation is
not a problem.
Denition of Radar Cross Section
Radar cross section is defined as (3)
In general it is dependent on many parameters. They include
(1) The size, shape, and material composition of the body
(2) The frequency and polarization of the incident electromagnetic wave
(3) The direction from which the wave illuminates the body, as well as that at which the observer is located
(the angle at which the target is viewed is referred to as the aspect angle)
These factors are taken into consideration when a radar is designed to detect a specific target. On the
other hand, if a vehicle is to face a radar with known specifications, the target can be designed with the radars
performance in mind. For example, if an aircraft will be flying directly at a radar in most mission scenarios,
then it will be wise to put extra effort into reducing the nose-on RCS, perhaps even at the expense of raising it
at broadside aspect angles. Furthermore, if the frequency of the radar is known, then the RCS reduction effort
need only concentrate on that frequency.
The conventional symbol for RCS is and it has units of square meters. As evident in Table 2, typical
values of the RCS range from 0.0005 m2 for insects to 100,000 m2 for a large ship. The decibel unit is convenient
when working with quantities that extend over several orders of magnitude. The RCS in decibels relative to a
square meter (dBsm) is defined as
Implicit in the definition of RCS is the assumption that the frequency is fixed.
Mathematically the RCS is given by
where
Ei = plane wave electric field intensity incident from the direction (i , i )
Es = scattered spherical wave electric field intensity observed from the direction (, )
Rr = distance from the target to the observer as shown in Fig. 1
RADAR CROSS-SECTION
At first glance is may appear that RCS is a function of range because of the R2 r in Eq. (1). However, the
scattered field is a spherical wave whose electric field has a 1/Rr dependence, which, when squared, cancels the
factor in the denominator. Hence RCS is a range-independent quantity.
Electromagnetic scattering and radiation problems can be formulated in either the time domain (TD) or
the frequency domain (FD) (4,5). In fact, the scattering body can be represented as a linear system with impulse
and frequency responses. The RCS as a function of time is related to the RCS as a function of frequency by
the Fourier transformation. There are advantages and disadvantages to working in each domain. Early radar
systems operated over narrow frequency ranges, and the need to compare calculated and measured data
dictated that calculations also be done in the frequency domain. Modern radars operate over wide frequency
ranges, and in order to characterize their performance adequately, the target RCS must also be determined
over the same wide range of frequencies. For a large, complex target, the calculation of the RCS at each
frequency is potentially a time-consuming task. It may be more efficient to compute the targets response to a
waveform in the time domain and then perform a Fourier transformation to obtain the RCS at many frequencies
simultaneously. Thus the measurement or calculation of RCS can be performed by either of two procedures:
(1) Frequency Domain The received power is obtained as the target is rotated through a range of angles while
a continuous wave (CW) fixed-frequency source is radiating. The frequency can be changed and another plot
of RCS versus angle obtained, and so on. At any fixed angle the RCS from a large number of frequencies
can be Fourier transformed to obtain the time-domain response of the target at that particular angle.
(2) Time Domain The received power as a function of time is obtained at each aspect angle of interest. The
time histories are inverse Fourier transformed to yield the frequency response of the target at each angle.
If the Fourier transform of the waveform and the impulse response of the radar system are known, then
the targets frequency response can be converted to RCS.
There are several aspects of the RCS that enter into the design of an LO platform, and therefore the
discussion of the RCS logically falls along similar lines. They include the topics:
(1)
(2)
(3)
(4)
Electromagnetic (EM) scattering mechanisms and the physical aspects of scattering
Analytical and computational methods
Reduction methods
Measurement techniques
A brief overview of each of these four topics is given, followed by mathematically based discussions of RCS
prediction and reduction.
RADAR CROSS-SECTION
Basic Scattering Characteristics of Bodies
Frequency Regions. The scattering characteristics of a target are strongly dependent on the frequency
of the incident wave. There are three frequency regions in which the RCS of a target is distinctly different.
They are referred to as the (1) low, (2) resonance, and (3) high frequency regimes (6). The labels are somewhat
misleading in that low and high are defined relative to the size of the target when normalized to the incident
wavelength, rather than to its physical size. If the target is smooth and it can be roughly described by a
characteristic length L, then the three frequency regimes can be defined in terms of the quantity L, where
and is the wavelength.
Low Frequency Region (L 1) At these frequencies the phase variation of the incident plane wave across
the extent of the target is small. Thus the induced current on the body is approximately constant in
amplitude and phase. The particular shape of the body is not a factor in the RCS pattern shape. For example, both a small sphere and a small cube have essentially isotropic (direction-independent) scattering
patterns. In general, versus L is smooth and varies as 4 . This region is also called the Rayleigh
region.
Resonance Region (L 1) When L is on the order of 1, the phase variation of the incident field across the
body is significant and all parts contribute to the scattering pattern. A plot of versus L is oscillatory.
This region is also referred to as the Mie region.
High Frequency Region (L 1) There are many cycles in the phase variation of the current across the
body, and consequently the scattered field will be very angle-dependent. The peak scattering levels are
primarily due to isolated points. For example, the peak scattering from large flat plates originates from
specular points on the surface. (They are the mirrorlike reflection points for which the angle of reflection
equals the angle of incidence.) A plot of versus L is smooth in this region and may be independent of
. This is also called the optical region.
The RCS of a sphere, which is plotted in Fig. 2, clearly illustrates the three frequency regions. The
appropriate characteristic dimension of the sphere is its radius; therefore, let L = a. (One could argue that
based on the previous discussion that L = 2a should be used. However, the boundaries between the frequency
regions are defined by orders of magnitude; they are not sharp. Hence either choice for L is acceptable. For a
sphere quantities are generally normalized to the radius, and therefore L = a is the most convenient choice.)
For a < 1, the curve is almost linear, but above 1 it begins to oscillate. This is the resonance region. The
oscillations die out at higher values, and above a 10 the curve approaches a constant equal to a2 .
Polarization and the Scattering Matrix. In addition to frequency, polarization plays a major factor
in determining the scattering characteristics of a body. Polarization refers to the orientation of the electric
field vector. The incident wave polarization is determined by the orientation of the transmit antenna, while the
receive antenna determines the receive polarization. If the antennas are linearly polarized and their electric
field vectors are aligned, then the copolarized scattered field is measured. On the other hand, if linear transmit
and receive antennas are orthogonal (electric field vectors crossed at 90 ), then the cross-polarized component
of the scattered field is sensed.
The copolarized scattered field is generally used, for several reasons. First, the copolarized scattered
field component is usually much higher than the cross-polarized component. Secondly, for monostatic radars
it is more efficient to use the same antenna for transmit and receive, and hence the transmit and receive
RADAR CROSS-SECTION
Fig. 2. The RCS of a sphere illustrates the three frequency regions: (1) Rayleigh (linear region at low frequencies), (2)
resonance (oscillatory region), and (3) optical (flat high frequency region).
polarizations are identical. However, measuring both components provides the most information available
from the scattered field and may assist in target identification.
A polarization-specific RCS can be expressed as
where p denotes the polarization of the incident wave and q the polarization of the receiver. Common polarization designations are defined with regard to a reference such as the ground or surface. They include: horizontal
and vertical (H and V); parallel and perpendicular ( and ); rectangular (x and y). A scattering matrix relates
the scattered field components to the incident field components:
The elements of the scattering matrix can be determined by computation or measurement.
The RCS of a cylinder illustrates the effect of polarization on scattering level and pattern shape. As shown
in Fig. 3, a relatively strong copolarized RCS exists even for thin wires, whereas the cross-polarized RCS tends
to zero. At high frequencies the cross sections for both polarizations asymptotically approach the same value.
Monstatic versus Bistatic RCS. The vast majority of radars currently in use are monostatic, because
most systems must operate from a single platform. However, the monostatic RCS (backscatter) of electrically
large smooth flat targets is very small unless the radar is positioned normal to the surface. Besides the normal,
large cross sections are observed in the specular ( = i ) and forward ( = i + ) scattering directions.
Bistatic radars have the transmit and receive antennas in two separate locations. Thus a bistatic radar
with a receiver at either of these two locations would have the advantage of larger target RCS. However
there are also disadvantages such as reduced operational flexibility because of the restricted transmitreceive
relationships, and maintaining a coherent reference for the transmitter and receiver.
RADAR CROSS-SECTION
Fig. 3. A cylinder illustrates how a targets RCS can depend on the incident wave polarization. At low frequencies the
cylinder becomes a thin wire that is transparent to a perpendicularly polarized wave.
Fig. 4. Simple geometric shapes such as plates, cylinders, and spheres can be used to build more complicated targets. The
interaction between the individual parts is important in determining low-level RCS.
Scattering Mechanisms
The RCSs of targets encountered by most radars are complicated geometrical shapes. There are a few exceptions
such as weather balloons and buoys. However, simple shapes such as plates, spheres, cylinders and wires are
useful in studying the phenomenology of RCSs. Furthermore, complex shapes can be decomposed into basic
geometrical building blocks (primitives) that can be assembled to form a more complex shape. Figure 4 shows
an aircraft represented by cylinders, plates, cones, and hemispheres. The RCS of the model in Fig. 4 is shown in
Fig. 5. A collection of basic shapes gives an acceptable RCS estimate that can be used during the initial design
stages of a platform. The locations and levels of the largest RCS lobes are of most concern at this stage of the
design process. The accuracy of the RCS estimate at other angles depends on how the interactions between
the various shapes are handled. These are more difficult to include, as will be seen later. Even when they
are included, the agreement with the measured RCS may not be particularly good if the interactions are not
modeled accurately. For an LO target more sophisticated numerical techniques must be used.
As evident from the pattern in Fig. 5, the RCS of a large, complex target is a rapidly varying function of
angle. However, as complicated as it looks, isolated features of the pattern shape and level can be associated
with the behavior of the currents induced in and on the scattering body. For example, the high narrow spikes
RADAR CROSS-SECTION
Fig. 5. An azimuth plot of the RCS of the aircraft in Figure 4 is shown. The two large lobes are due to the fuselage, which
is modeled as a cylinder.
Fig. 6. The total RCS of a target can often be decomposed into a sum of contributions from distinct scattering mechanisms.
Some important mechanisms for electrically large targets are depicted in this figure.
result from mirrorlike specular reflections from large flat surfaces. Other lobes can be associated with surface
waves, multiple reflections, diffraction, and so on. These so-called scattering mechanisms are depicted in Fig.
6. The lines in the figure are ray representations for propagating waves. They denote the direction of power
flow and are normal to the planes of constant phase.
For any particular aspect angle, frequency, and polarization, if one scattering mechanism dominates over
all others, then it may be possible to express the total scattered field as a combination of scattered fields Es1 ,
Es2 , Es3 , . . ., each due to a separate scattering mechanism:
and if one of these terms is much greater than all others,
RADAR CROSS-SECTION
It follows from Eq. (1) that the total RCS can be written as a sum of RCSs, each attributed to a separate
scattering mechanism:
Note that Eq. (5) is a convenient way of decomposing the scattered field for the purpose of assisting the
RCS modeling and reduction efforts. First, if sufficiently accurate models exist for the individual scattering
mechanisms, then Eq. (3) suggests that they can be combined under the appropriate circumstances to model
their collective effects for a complex target. Secondly, since the largest terms in Eq. (3) correspond to the source
of the most intense scattering, it may be possible to determine which scattering mechanism dominates, and
thereby choose an appropriate reduction technique.
Reection. This is the mechanism that yields the highest RCS peaks, but they are limited in number
because Snells law must be satisfied. When multiple surfaces are present, multiple reflections are possible.
For instance, the incident plane wave could reflect off of the fuselage of an aircraft, hit a fin, and then return
to the radar.
Diffraction. Diffracted fields are those scattered from discontinuities such as edges, corners, and vertices. The waves that are diffracted from these shapes are less intense than reflected waves, but they can
emerge over a wide range of angles. At aspects requiring low RCS, the diffracted waves can be significant.
Surface Waves. The term surface wave refers to the current traveling along a body and includes several
types of waves (7). In general, the target acts as a transmission line, guiding the wave along its surface. If
the surface is a smooth closed shape such as a sphere, the wave can circulate around the body many times.
Such waves are called creeping waves because they appear to creep around the back of a curved body. On
curved bodies the surface wave continuously radiates. Radiating surface waves on flat bodies are usually called
leaky waves. Traveling waves appear on slender bodies and along edges, and suffer little attenuation as they
propagate (8). If the surface is terminated with a discontinuity such as an edge, the traveling wave will be
reflected back along the surface and radiate towards its origin.
Ducting. Also called waveguide modes, ducting occurs when a wave is trapped in a partially closed
structure. An example is an air inlet cavity on a jet. Once the wave enters the cavity, many bounces can occur
before a ray emerges. There are many paths that the ray can take, and therefore rays emerge at almost all
angles. The result is a large, broad monostatic RCS lobe. An optical analogy of this is the glowing of a cats eye
when it is illuminated by a light.
Interactions. It is possible that these mechanisms will interact with each other. For example, a wave
reflected from a flat surface can subsequently be diffracted from an edge or enter a cavity. For a complex target
the interactions are not always obvious and the decomposition of the scattered field no longer makes sense.
Prediction Methods
The analytical methods used to calculate RCSs are similar to those used in antenna analysis. A radar target is
essentially acting as an antenna. The incident wave induces a current on the target, and the induced current
radiates a field just as an antenna would. However, in the radar case it is called a scattered field as opposed to
a radiated field. A good example is a reflector antenna or satellite dish. The incident field from a feed horn
excites currents on the reflector, which in turn give rise to a radiated field. In the radar analogy, the feed horn
represents the radars transmit antenna, while the reflector is the target. If the distribution of current is known
for a target, then the currents can be used in the radiation integrals (StrattonChu integrals) to determine the
scattered field. Once the scattered field is known, the RCS can be computed.
10
RADAR CROSS-SECTION
Of course, the problem is that the induced current on the target is unknown and generally very difficult
to determine. An analytical solution is possible in only a few simple cases. In other cases it is feasible to solve
Maxwells equations or integral equations for the current numerically. These solutions are rigorous in that there
are no assumptions or restrictions imposed. The only error in the solution is due to the numerical evaluation
of the integral or differential equations. The alternative to a numerical solution is to take an intuitive guess at
the current, in which case the resulting RCS is only approximate.
Several methods of RCS prediction are described below. The classical solution techniques are not discussed
in this list, because most are limited to one- or two-dimensional structures or simple three dimensional ones.
The methods of interest here are those that can be applied to complex three-dimensional targets. The methods
most commonly encountered are physical optics, microwave optics (ray tracing), the method of moments, and
finite-difference methods.
Physical Optics. The physical optics (PO) approximation (9) provides a means of estimating the surface
current induced on an arbitrary body. On the portions of the body that are directly illuminated by the incident
field, the induced current is simply proportional to the incident magnetic field intensity. On the shadowed
portion of the target the current is assumed to be zero. The current is then used in the radiation integrals to
compute the scattered field far from the target.
PO is a high-frequency approximation that gives best results for electrically large bodies (L 10). It is
most accurate in the specular direction. Since PO assumes that the current abruptly decays to zero at a shadow
boundary, the computed field values at angles far from the specular directions and in the shadow regions are
inaccurate. Furthermore, surface waves are not included. PO can be used in either the time or the frequency
domain. It can be supplemented by the physical theory of diffraction (PTD), which provides corrections to PO
approximation that improve the accuracy of the current distribution near edges.
Microwave Optics. Microwave optics refers to a collection of ray-tracing methods that can be used
to analyze electrically large targets of arbitrary shape. The rules for ray tracing in a simple medium (linear,
homogeneous, and isotropic) are similar to reflection and refraction in optics. Geometrical optics (GO) (10) is
the classical theory of ray tracing used since the days of Newton. It provides a formula for computing the
reflected and refracted fields. In addition, the geometrical theory of diffraction (GTD) is used in conjunction
with GO to include diffraction. Diffracted rays originate from the scattering of the incident wave at edges,
corners, and vertices. The formulas are derived on the basis of infinite frequency ( ), which implies an
electrically large target. Ray optics is frequently used in situations that severely violate this restriction and
still yields surprisingly good results. The major disadvantage of ray tracing is the bookkeeping required for a
complex target. It is used primarily in the frequency domain.
Method of Moments. The method of moments (MM) is a technique used to solve an integral equation
(11). Integral equations are so named because the unknown quantity appears in an integrand. In electromagnetics they are derived from Maxwells equations and the boundary conditions. The unknown quantity can be a
combination of volume or surface current (either electric or magnetic). The MM reduces the integral equations
to a set of simultaneous linear equations that can be solved using standard matrix algebra. The size of the
matrix involved depends on the size of the body; current computer capabilities allow bodies on the order of 10
or 20 wavelengths to be modeled.
Most MM formulations require a discretization (segmentation) of the body. Therefore, they are compatible
with finite-element methods used in structural engineering, and the two are frequently used in tandem during
the design of a platform. The MM can be used to solve both time- and frequency-domain integral equations.
Finite-Difference Methods. Finite differences are used to approximate the differential operators in
Maxwells equations in either the time or the frequency domain (5,6). As with the MM technique, the target
and surrounding space must be discretized. Maxwells equations and the boundary conditions are enforced on
the surface of the target and at the boundaries of the discretization cells. This method has found extensive use
in computing the transient response of targets to various waveforms. Finite-difference methods do not require
the large matrices that the MM does, because the solution is stepped in time throughout the scattering body.
RADAR CROSS-SECTION
11
Reduction of Radar Cross Sections
RCS reduction techniques are traditionally classified as geometrical shaping, materials selection, and passive
and active cancellation. The boundaries between these categories are not always distinct, and a reduction
approach may draw on all of these simultaneously.
Shaping. Target shaping is the first step in RCS reduction. The first rule of shaping is to avoid presenting large flat surfaces to the radar. For large surfaces specular reflection is the dominant scattering mechanism.
The direction of strongest reflection for a smooth surface is given by the law of reflection. Therefore, by examining the relationship between the direction of incidence and the surface normal, the direction of maximum
scattering can be predicted. If a target is composed of several surfaces and one could visualize a plot of the
normals of the surfaces, then it is expected that there would be few normals pointed in directions of low RCS.
It will be shown that the maximum monostatic RCS of a large flat surface occurs when the incident wave
and observer are located normal to the surface. The maximum value is
where A is the area of the surface. Thus the RCS can be minimized by minimizing the area presented to the
radar. If it is not possible to tilt a large surface sufficiently, then it can be broken into a number of smaller
surfaces, each tilted in a slightly different direction.
These approaches have led to rather unconventional aircraft shapes such as that of the F-117 Nighthawk
(commonly referred to as the Stealth Fighter) (12). To date, shaping has not been applied extensively to ships
and ground vehicles, although that is beginning to be done. Modern ship designs are making some concessions
to RCS control, such as canting deckhouse walls and enclosing mast structures.
Another aspect of shaping is edge alignment. Once the specular reflections have been controlled, the
contributions from edges (such as traveling waves and diffraction) may become important. Aligning edges is
one way of maintaining low-RCS spatial sectors. For example, in plan form it is seen that the edges of the B2
bomber fall along only a couple of fixed angles.
The second rule of shaping is to avoid or hide retroreflectors. Retroreflectors are simple geometrical
structures that tend to redirect the incident radar wave back in the direction from which it arrives. Surfaces
that meet at a 90 angle are strong retroreflectors, and therefore the angles between the tail rudder and
elevator surfaces of a stealthy aircraft are always canted. New designs such as the tailless McDonnell Douglas
X-36 have no vertical surfaces at all.
Cavities are another type of retroreflector. They have a very broad scattering pattern. Thus jet intakes are
often located above the wing or else employ an electromagnetic screen to prevent penetration by the incident
wave.
Obviously there is a limit to the degree of shaping that can be incorporated while simultaneously meeting
other operational requirements. For example, extreme shaping applied to an aircraft degrades its aerodynamic
performance and control characteristics. Another disadvantage of low-RCS shapes is that they have slim
profiles, which reduces the useful enclosed volume relative to a conventional shape.
Materials and Coatings. Obviously the material composition of a body affects its RCS. For example,
a metal plate has a higher RCS than a glass plate. Similarly, a plastic aircraft body has a lower RCS than one
with a metal skin. The electrical characteristics of a material are defined by its constitutive parameters (13):
(1) Permittivity, = 0 r = j (F/m)
(2) Permeability, = 0 r = j (H/m)
(3) Conductivity, c (/m)
12
RADAR CROSS-SECTION
where 0 and 0 are the permittivity and permeability of free space. The imaginary parts are associated
with dissipation (attenuation) of the fields in the medium.
The permittivity determines the fundamental relationship between the electric field intensity E and flux
density D. For the vast majority of materials that occur in nature one can write
where the boldface denotes a matrix. Expanding Eq. (7) and explicitly including the spatial and frequency
dependences gives
Likewise, the permeability determines the fundamental relationship between the magnetic field intensity H
and magnetic flux density B, which for most materials can be written as
or
The following special cases occur frequently:
(1) Isotropic
(2) The diagonal components of the permittivity and permeability matrices are equal, and the off-diagonal
elements are zero. That is, pp = r 0 and pp = r 0 ; pq = 0 and pq = 0 if p = q (p, q = x, y, or z).
(3) Nondispersive
(4) The constitutive parameters are independent of frequency. Most materials exhibit some dispersion (frequency dependence) if the frequency change is great enough. However, we define a nondispersive material
as one whose r and r are sufficiently constant within the frequency band of the radar illuminating the
material.
(5) Homogeneous
(6) The constitutive parameters are independent of location in the material. Almost all materials are inhomogeneous on a microscopic scale. For our purpose a homogeneous material is one that has a sufficiently small
variation in r and r throughout its volume.
To examine the dependence of the RCS on the constitutive parameters, consider an infinite interface
between free space and a half-space of material with r and r . The Fresnel reflection coefficient for a plane
RADAR CROSS-SECTION
13
wave normally incident on the interface is
where the intrinsic impedances are defined by
and
From Eq. 1 the RCS is proportional to the square of the scattered electric field intensity. Since the scattered
field intensity (in this case the reflected field) is proportional to , it follows that | |2 . Thus the RCS will
be reduced in direct proportion to the reduction in reflection coefficient squared. This is an ideal limit, because
in practice the material thickness is finite, and there will likely be some reflection from its back face and
scattering from its edges.
Unfortunately, in most applications the target materials cannot be selected solely on the basis of the
electromagnetic properties r and r . Mechanical properties such as strength, weight, and thermal conductivity
have a higher priority. In the drive to reduce weight and increase strength, composite materials have found
widespread military and civilian application (14).
Composites are formed from multiple constitutive materials in a manner that yields a final product having
desirable properties from each material. Graphite is an example of a composite that is used extensively in the
aircraft industry. Although the primary motivation for the use of composites has been structural, the reflection
from most composite materials is lower than that for conductors. However, the majority of the incident field
that is not reflected from the composites surface is transmitted through the material, and the transmitted field
can potentially be scattered from internal metallic objects. The use of nonconductors also complicates other
electrical design issues such as the grounding of electrical systems and lightning protection.
An absorbing layer is another example of the application of materials for RCS reduction. A layer of
absorbing material is deposited on the surface of the target. As the incident wave propagates through the
material, energy is extracted from the field and its intensity diminishes (15). Thus a weaker field is reflected
at the target surface, and the reflected wave is attenuated further as it travels back through the material. The
circuit analog of this type of material is a resistor.
A material that performs the function just described is referred to as a radar absorbing material (RAM).
High absorption is desired, yet the material thickness and weight must be minimized. Unfortunately, many
materials with high loss also have high reflection coefficients. Thus there is significant reflection of the incident
wave at the absorbing-layer face, and little energy enters the material where attenuation takes place.
There are some materials that possess more complicated or exotic electromagnetic properties that can
potentially be exploited for RCS reduction. For example, since the pq affect the velocity of propagation along
the coordinate axes of the material, it may be possible to construct for a material so that the polarization
of the scattered field is different than that of the incident field. Thus the polarization of a wave rotates as it
propagates through the medium, and scattered energy is converted to a field component that the radar is not
capable of sensing.
14
RADAR CROSS-SECTION
Another example is a spatially dispersive medium. The electric flux density depends on both the electric
field intensity and the magnetic flux density. A chiral material (16) has the constitutive relationship
This type of behavior is associated with materials that have a helical molecular structure. It is possible to
produce this type of behavior artificially by dispersing small helical coils throughout a bulk material.
In the past, the reduction of RCSs has focused on treating each scattering source, or hot spot, individually.
For instance, an absorbing layer might be installed to reduce scattering from a wing edge. The layer will reduce
the edge scattering, but will also introduce new scattering sources at the perimeter. For a useful reduction
method, the new scattering sources must be much weaker than those of the original edge. But as the scattering
intensities of the sources become smaller, one quickly reaches the point where the new scattering sources
introduced by the treatments are comparable to those being treated. At this level the interaction between
scatterers becomes important, and this complicates the design and application of RCS treatments.
Cancellation Techniques. Cancellation refers to RCS reduction methods that are based on the introduction of a secondary scatterer whose scattered field is adjusted to cancel the reflection of the primary target
(17,18). The total field at the radar receiver is the vector sum of the fields scattered by the target and the
cancellation element. Therefore, if the cancellation field can be made equal and opposite to the field scattered
from the body at the angle, polarization, and frequency of interest, then destructive interference occurs. The
secondary scatterer is referred to as the cancellation element, and its scattered field the cancellation field.
The cancellation element can be a detached body located near the original target, or an attached protrusion. Both are referred to as parasitic elements. The combination of the target with the scattering element is
frequently called the loaded body. If the cancellation field is fixed, then the system is a passive one. On the
other hand, if the system can respond to changes in the incident field, then it is an active one. A fair amount
of information about the radar and the target is required for an active system: the threat radar frequency, its
direction and polarization, and the targets RCS in the direction of the radar.
Passive Cancellation. Since the cancellation of the two fields is essentially a tuning method, this technique is only effective over a narrow frequency band and usually limited to a small spatial sector. If large
parasitic elements are to be avoided, then the magnitude of RCS that can be canceled is small. Thus passive
cancellation is usually used to supplement shaping and absorbers.
The most common cancellation elements are slots and wires. Two orthogonal elements are required to
cancel an incident wave of arbitrary polarization. The elements can be increased in size or number so as to
provide sufficient gain to cancel the RCS in the direction of interest.
Two examples of passive cancellation are the Salisbury screen and Dallenbach layer (19). A Salisbury
screen consists of a resistive film located a distance d in front of the target surface. The film resistivity is chosen
so that the reflection from its surface is exactly canceled by the field that passes through the film, is reflected
from the target, and then is transmitted back through the film.
The Dallenbach layer shown in Fig. 7 is similar to the RAM coating discussed earlier, except that the
layer thickness is chosen to provide destructive interference between the fields scattered from the top of the
layer and the boundary at the back of the layer, rather than absorption within it. Thus the top of the layer
serves the same purpose as the resistive film does for the Salisbury screen.
Active Cancellation. Active cancellation (also known as adaptive cancellation) is the extension of passive
cancellation to handle dynamic threat scenarios. In the context used here, it does not include deception, which
comprises techniques of modifying and retransmitting a signal. For the present discussion, the source of RF
energy is the radars signal. Two levels of sophistication are considered:
RADAR CROSS-SECTION
15
Fig. 7. A Dallenbach layer is used to reduce the RCS of a conductor. The material electrical characteristics and thickness
are chosen to provide a reflection coefficient of zero.
Fig. 8. Traveling waves along a surface are treated using a transmission line matching method: the quarter wave
transformer.
(1) Fully active The cancellation network receives, amplifies, and retransmits the threat signal in such a way
that it is out of phase with the static (skin) RCS of the target. The transmitted signal amplitude, phase,
frequency, and polarization can be adjusted to compensate for changing threat parameters.
(2) Semiactive No amplification of the incident signal energy is provided by the cancellation network, but
adjustable devices in the network allow the reradiated signal to compensate for limited changes in the
radar signal parameters.
The demands for a fully active system are almost always so severe as to make it difficult to implement.
It requires amplifiers and antennas that cover the anticipated threat angles, frequencies, incident power
densities, and polarizations. A knowledge of the radars direction is required, as well as the targets own RCS.
A semiactive system is not as complicated with regard to hardware, but the use of adjustable devices still
requires bias lines, controller units, and a computer with the appropriate databases.
Because of the growing importance of low-level scattering sources (in particular, manufacturing-induced
errors that differ from platform to platform), adaptive cancellation techniques are receiving increased attention. Adaptive cancelers have the ability to modify the amplitude and phase of the new scattering source to
compensate for changes in the original scattering source that may occur with frequency or observation angle. In
principle, adaptive techniques are capable of adjusting to changing physical and electrical conditions brought
on by exposure to the environment. However, cancellation is most effective when there are a small number of
coherent scattering sources rather than a large number of random sources.
Treatments for Surface Waves. Surface waves, and in particular traveling waves, present a special
challenge as far as RCS reduction is concerned (6). The target acts as a combination of antenna and transmission
line, collecting incident energy and guiding it along the surface. As depicted in Fig. 8, when a discontinuity
(such as a termination) is encountered, the surface wave is reflected. As it propagates in the reverse direction,
it loses energy by radiation. A dielectric-coated ground plane can support a surface wave for incident plane
waves having a component of E normal to the surface; a magnetic-coated ground plane can support a surface
wave for incident plane waves having a component of H normal to the surface.
The most common technique for treating surface-wave reflections that arise from surface discontinuities
is one borrowed from transmission line theory: the use of transformer sections. Typically a quarter-wave
matching section is appended to the termination of the surface, and the surface characteristics of the matching
16
RADAR CROSS-SECTION
section are chosen to allow the wave to continue propagating in the forward direction, thereby eliminating the
reflected field.
Special Stealth Issues
To achieve and maintain low observability requires careful design, manufacturing and maintenance procedures
to a level that may seem trivial to the nonspecialist. Instances of bird droppings, loose screws, and pebbles
kicked up from the takeoff roll having resulted in increased RCS have been reported (12). Some special stealthrelated issues are discussed below.
Integration of Sensors and Sensor RCS. Active sensors that transmit energy, such as microwave
and laser radars, need to have a low probability of intercept (LPI). One aspect of LPI involves controlling
the spatial distribution of radiation by using narrow antenna beams and low sidelobes. Another is to control
the power distribution in the frequency spectrum, which is directly related to the transmitted waveform. For
example, wide-bandwidth spread-spectrum pseudorandom waveforms distribute their power almost uniformly
over a wide frequency band and hence appear like noise, making detection extremely difficult.
The RCSs of sensors installed on an LO platform, especially antennas in the radar frequency bands,
are an area of concern. Unlike cavities, antennas cannot always be hidden from the radars field of view
without degrading the performance of the accompanying system. Ideally, if the incident wave frequency is in
the operating band of the antenna (where the antenna is efficient), then its RCS is low because most of the
incident wave is collected and appears at the antenna terminals.
The antenna scattering occurs when the frequency of the radar wave is outside the antennas operating
band. Large reflection sources may occur inside the antenna at frequencies out of its operating band, resulting
in scattering characteristics similar to those of a cavity. Furthermore, if the antenna is composed of widely
spaced (relative to wavelength) periodic elements, Bragg diffraction can occur. Bragg lobes are high-RCS spikes
that result at multiple angles when the round trip path length differences between adjacent elements are an
integer multiple of 2. (This is similar to array grating lobes in the radiation case.)
The obvious antenna RCS reduction methods, such as adding lossy materials to attenuate out-of-band
reflections, also decrease the antennas gain in its operating band. The preferred approach is to incorporate a
frequency-selective surface (FSS) into the design. An ideal FSS permits radiation in the antennas frequency
band to pass through unattenuated while all other frequencies are reflected or absorbed. FSS shielding has
been applied to antennas on ship masts (20) and as missile radomes (21).
In the drive to reduce the number of low-level scattering sources, continuity of the platform external
surface should be maintained. The number of apertures (openings) must be kept to a minimum. Apertures
are needed for radar and communication antennas, which operate at microwave frequencies, and infrared and
optical sensors. By integrating the infrared and optical sensors into the antennas, two or three apertures can
be replaced by a single one.
Computational Advances. The need to predict the mutual interactions between many low-level scattering sources has fueled research in the area of computational electromagnetics (CEM). Numerical solutions
of the rigorous equations of scattering from complex bodies have existed since the early 1960s. They involve
reducing the electromagnetic equations to a set of simultaneous linear equations, which are solved by matrix
methods. The dimension of the matrix increases as the size of the target measured in wavelengths increases
(that is, all target dimensions are scaled by the wavelength of the incident wave). It has only been in recent
years that the dramatic increase in computer memory and speed has made this approach practicable for aircraft and ships. Even so, rigorous methods are still limited to low frequencies for targets that are physically
large.
Approximate solutions have been developed based on the assumption that scattering is localized, so that
interactions between distant parts of the structure can be neglected. The approximations have constantly been
RADAR CROSS-SECTION
17
improved by adding higher-order corrections based on ever more detailed electromagnetic models as well as
empirical data. Both the rigorous and the approximate computer models have incorporated geometry builders
that can utilize computer-aided drawing (CAD) files generated by commercially available drawing programs.
This capability allows for concurrent engineeringthe ability to make a structural change and immediately
assess its effect on the RCS.
Effect of Imperfections and Flaws. An important limitation in achieving low RCS is the imperfections
introduced during the manufacture and assembly of the platform and the installation of its systems. Their
effect on RCS performance depends on whether the imperfections are periodic or random in nature. Most error
sources are nearly random, and their net effect on RCSs is to shift energy from the scattering mechanisms
discussed previously to diffuse scattering. Diffuse surfaces reflect uniformly in all directions, independent of the
angle of incidence; they are electrically rough surfaces. The magnitude of the diffuse scattering increases with
the extent of the surface roughness (imperfection). Since diffuse scattering is not angle-dependent, shaping is
not a practical way of reducing the RCS for it, and therefore very close tolerances must be maintained to keep
the diffuse scattering to acceptable levels. Maintaining close tolerances is one of the primary cost drivers in
the manufacture of LO platforms. Eventually a point is reached where the question must be asked: How much
money is an additional RCS reduction of a few percent worth?
Maintenance and Testing. A second important cost driver is the maintenance of an LO platform.
The natural elements, along with the harsh operating environment, stress and degrade the electrical and
mechanical properties of materials. Special testing methods are required to determine if and when the RCS
has been affected by exposure to the environment. So-called gono-go tests compare a set of measured data from
the platform under test with a reference data set obtained from a known clean one of identical configuration.
A significant deviation in the data sets would signify a correspondingly significant change in RCS. A test of this
type does not require a high level of precision, and therefore can be conducted using portable test equipment.
Data can be taken by probes located close to the platform surfaces, that is, in the near-field region of the
scatterers.
If a platform fails the gono-go check in the field, then it will be sent to a test and repair facility
for further evaluation. Specialized facilities and repair techniques are required because of the uncommon
materials involved and the exacting tolerances that must be satisfied. The integrity of localized repairs can be
checked using hand-held reflection devices that verify local surface reflection characteristics.
Tradeoff between Stealth and Electronic Warfare. Low observability demands the most of technology and forces concessions in many other aspects of platform design and performance. At some point the
burden of increasing stealth may become unacceptable. Electronic warfare (EW) techniques (now referred to
as electronic attack and electronic protect in the United States) are a complement to low RCS. Common EW
techniques currently in use around the world include jamming, decoys, and chaff (22). The major disadvantage
of EW is that the radar is aware of the targets presence. However, the radar is denied information on the
targets range and velocity. Even though the radar is not completely disabled, a combination of EW and low
RCS can cause the radar to loose lock, forcing it to reacquire a target repeatedly. This is a time-consuming
process that ties up valuable radar resources. With all other factors constant, chaff, deception, and jamming
are more effective for low-RCS targets than for conventional ones.
Some older platforms currently in the field are being retro-fitted or upgraded to reduce the RCS and other
signatures. All new aircraft, ships, and ground vehicles are incorporating some degree of stealth design and
technology. However, it appears that the complete dominance of stealth in all aspects of design, as seen in the
F-117, has been abandoned. Programs like the F-22, the Comanche helicopter, and the Euro-fighter are opting
for a balance between stealth and other performance measures.
18
RADAR CROSS-SECTION
The Radar Equation
Consider the situation shown in Fig. 1 with the radar operating at a frequency f . The corresponding wavelength
is = c/f , where c 2.998 108 m/s is the speed of light in free space. Define the following quantities: where
Pt = transmitter power delivered to the transmit antenna
Gt = transmit antenna gain, a measure of how efficiently the antenna concentrates power in the
direction of the target
Gr = receive antenna gain
Aer = effective area of the receive antenna, which can approach the physical area of the antenna
aperture for a very efficient antenna
From antenna theory, antenna gain and effective area are related by
The power scattered by the target that is delivered to the receive antenna terminals is given by
The first factor on the right-hand side is the power density incident on the target. Multiplying this factor by
gives the fraction of incident power scattered in the direction of the receiver. The product of the first three
factors represents the scattered power density at the receive antenna. Finally, multiplying by Aer gives the
scattered power collected by the antenna.
Specializing to the monostatic case (Gt = Gr G and (Rt = Rr R) and using the relationship between
effective area and gain give the result
This form of the radar equation is too simple to be of practical use in predicting a radars performance. However,
two important features are evident. The first is the R4 in the denominator, which is a disadvantage to the radar
system designer, whose goal is to increase the detection range of the radar. The second is that the received
power is linearly related to , which is a disadvantage to the RCS engineer.
If the minimum scattered power that the radar receiver can detect is Pm , then the corresponding maximum
detection range is obtained by solving the radar equation for R:
Therefore, to halve the maximum detection range requires that the RCS be reduced by a factor of 24 (=16).
Finally, note that the radar equation was derived for a single-frequency wave. Typical radar waveforms,
such as pulse trains, occupy a band of frequencies. The radar equation may have to be evaluated at several
frequencies in the radar band if the quantities in the equation are not constant.
RADAR CROSS-SECTION
19
Calculation of Radar Cross Section
RCS prediction methods can be broadly categorized as either rigorous or approximate. Rigorous methods are
those that have been shown by mathematical proof to converge to the exact result under the appropriate
conditions. However, in practice, there is always some small error present in the computed data due to the
numerical evaluation of integrals, computer roundoff, and so on. Examples of rigorous methods are the finitedifference time domain (FDTD) technique and the MM.
Approximate formulations are based on assumptions and therefore are subject to limitations in their
application and accuracy. Unlike the rigorous methods, even if there is no computational error, the approximate
result will not necessarily converge to the true result. If the approximation is a good one, then the error will not
be significant for the problem under consideration. Examples of approximate methods are PO and microwave
optics.
Phasors. Phasor notation is commonly used when working with time-harmonic fields. Time-harmonic
fields are those that have a sinusoidal time variation of the form cos t, where = 2f . Since the sources and
fields have the same time dependence, it can be factored out of all electromagnetic quantities and suppressed
(i.e., dropped). When the calculations are completed, the time variation is reintroduced. Thus a time-varying
quantity E (x, y, z, t) can be obtained from the phasor E (x, y, z) by multiplying the phasor by ejt and taking
the real part
All quantities appearing throughout this article are phasors unless otherwise indicated.
Spherical and Plane Waves. Ideal spherical waves (i.e., waves for which the planes of constant phase
are spheres) originate from point sources, which do not actually exist. However, for practical purposes, spherical
wave behavior results if the observer is located far enough from an extended source as depicted in Fig. 1.
Spherical waves have the phasor form
where r is the distance from the source. The vector constant E0 (which may have complex components) is
determined by the amplitude, phase, and polarization of the extended source.
At great distances from the source (r ) the target only intercepts a small section of the spherical wave
front with a very large radius of curvature. Thus it appears that planes of constant phase exist over the extent
of the target, and the electric field can be approximated by
where r is now the distance from the phase reference and E0 is determined from the intensity of the spherical
wave at the target. If the source is at the origin of a spherical coordinate system, then E0 can only have and
components in the far zone. (This would be the case for a radar antenna located at the origin.)
Both spherical and plane waves have the characteristics of a transverse electromagnetic (TEM) wave:
(1) The electric and magnetic field vectors are perpendicular to each other and the direction of propagation.
(2) The magnitudes of the electric and magnetic fields are related by the intrinsic impedance of free space.
20
RADAR CROSS-SECTION
Fig. 9. Current flows on the surface of a PEC. The total scattered field at the observation point is obtained by an integration
of the current over the entire surface.
Thus, if is a unit vector in the direction of propagation,
A plane wave of arbitrary polarization is given by the formula
where r is a position vector to the point x, y, z at which the field is to be evaluated:
The corresponding magnetic field intensity for a plane wave propagating inward toward the origin ( = r) is
The component of the electric field is called transverse magnetic to z (TMz ) because the magnetic field vector
associated with this component [E0 term in Eq. (16)] lies entirely in a plane transverse to the z axis. Similarly,
component of the electric field is referred to as transverse electric to z (TEz ) because this component of the
the
electric field [E0 term in Eq. (15)] lies entirely in a plane transverse to the z axis. Thus, for a TMz -polarized
plane wave E0 = 0; for a TEz -polarized plane wave E0 = 0. Any arbitrary polarization can be expressed as a
linear combination of the two. In a similar manner one could define TEx , TMy , and so on.
Radiation Integrals. Scattered fields are set up by a combination of electric and (in general) magnetic
currents. Currents are induced such that the total fields satisfy the boundary conditions and Maxwells equations. The radiation integrals provide a means of calculating the fields due to a prescribed set of currents.
The radiation integrals, or StrattonChu integrals, are integral solutions to Maxwells equations. They can be
derived directly by taking the curl of Maxwells first two equations, using the vector Greens theorem, and then
integrating (23).
For RCS calculations, according to Eq. (1), only the scattered field far from the target is of interest,
because R . Consider a scattering body of volume V that is enclosed by a surface S as shown in Fig.
9. The standard convention is to denote observation point quantities by unprimed symbols, and source point
quantities by primed symbols.
RADAR CROSS-SECTION
21
Fig. 10. If a target is not a PEC, then volume currents may exist inside of the target. The volume currents can be replaced
by surface currents using the surface equivalence principle.
In the far zone, the following approximations are employed:
(1) The vector from the origin to the observation point and the vector from any source point on S to the
observation point are approximately parallel (R parallel to r).
(2) For large distances the difference in amplitude between a spherical wave emanating from the origin and
one emanating from any point on the surface S is negligible (1/R 1/r).
With these assumptions, the radiation integral for the electric field components tangential to a sphere of
radius r centered at the origin can be expressed as
where J (A/m2 ) and M (V/m2 ) are the volume electric and magnetic currents, respectively. (Magnetic current is
a fictitious quantity because isolated magnetic charge does not exist, but under some circumstances it is used
as a matter of mathematical convenience.)
The use of volume current is advantageous for computing the scattered fields from highly inhomogeneous
bodies. However, for most targets, it is more efficient to use a surface current formulation. As shown in Fig.
10, current flows on all interfaces between materials, as well as the external surface of the target. For surface
currents, the far-field form of the radiation integral becomes
where S includes all of the surfaces over which nonzero current exists. If the current is known, then the
integrals in Eqs. (17) or (18) can be evaluated numerically.
22
RADAR CROSS-SECTION
If the scattering body is a perfect electrical conductor (PEC), then only electric current exists, and the
surface of integration corresponds to a real physical surface. However, S in Eq. (18) need not exist; it can be
a fictitious surface defined for convenience. The currents on the fictitious surface can be determined from the
actual tangential electric and magnetic fields at every point on S. Thus
where n
is the unit vector normal to the surface. This is a consequence of the equivalence principle of electromagnetics (4).
Approximate Prediction Methods. Approximate prediction methods are used for a variety of reasons,
but primarily because they are computationally less demanding, thereby allowing data to be generated more
rapidly. This is an advantage in the early stages of design, when major tradeoff studies are being performed.
Coarse approximations can be applied in the beginning, and then refined as the design progresses. Eventually
it may be necessary to verify the performance of a specific design by a rigorous calculation.
At microwave frequencies, useful approximations should include scattering contributions from large surface areas, edges, surface waves, and interactions between and among surfaces and edges. There are two
approaches to computing the RCS. One is based on estimating the induced currents and is referred to as a
current-based approach: the PO approximation. The second is based on postulating relationships between the
incident and scattered waves, or rays: GO. This is a ray-based approach.
PO and GO are used to calculate the scattered field from surfaces; the effects of the edges of surfaces are
not properly modeled or not included. Supplemental theories have been developed to allow for edge scattering.
The PTD is the current-based method used in conjunction with PO. The GTD is a ray-based method for
calculating diffracted fields. Both the current and ray approaches can be shown to yield identical results under
the same conditions. Table 3 summarizes the two methods.
Physical Optics. A rigorous calculation of the currents J s and M s typically requires a great amount
computational effort. The PO approximation is an estimate of the electric current that is accurate for a large,
RADAR CROSS-SECTION
23
smooth PEC. The current is approximated by (9)
where H i is the incident magnetic field intensity.
Equation (20) is explained by approximating the target surface in the vicinity of the point (x, y, z) by
an infinite, flat surface. Assume that the incident wave is due to a source at a distance R and the electric
field vector is parallel to the surface. By the method of images, the target can be removed and replaced by an
equivalent scattering source at a distance R below the surface. To satisfy the boundary conditions (i.e., the
tangential component of the total electric field must be zero) the electric field of the equivalent scattering source
must be equal and opposite to that of the incident field. Because the two waves are propagating in opposite
directions toward the surface, from Eq. (14) the magnetic field vectors will be oriented in the same direction.
Thus the factor of 2 in Eq. (20).
Perhaps the simplest application of the PO approximation is for a flat plate (6). Let the plate be centered
at the origin in the z = 0 plane and have dimensions of a in the x direction and b in the y direction. A plane
wave of arbitrary polarization is incident at an angle an angle i . The PO approximation for the current is
where
and
The subscript i denotes incident wave quantities. The x, y, and direction cosines are defined by
Using Equation (21) in 18a and the fact that z = 0 gives
24
RADAR CROSS-SECTION
Fig. 11. A comparison of the RCS of a plate with dimensions a = 5 and b = 2.5 in the plane = 0 . The approximate
(PO) and rigorous (MM) methods are in close agreement in the region of the main lobe but disagree at wide angles.
For the monostatic RCS ui = u and vi = v; therefore the exponents in the integrand can be combined. Furthermore, the x and y integrals can be separated, and both are of the form
Now assume that a TMz -polarized plane wave is incident (E0 = 0) and the radar receive antenna is sensitive
to the same polarization. Thus we use the notation where the first subscript denotes the polarization of the
receiver and the second the polarization of the incident wave. Using Eqs. (23) and (24) in the definition of RCS
yields
where A = ab is the area of the plate.
The monostatic RCS of a square plate with a = 5 and b = 2.5 is shown in Fig. 11 for = 0 . The xz and
yz planes are referred to as principal planes, because they are transverse to the plate edges. For comparison,
the RCS obtained by a rigorous calculation of the current (MM) is also shown. (The MM result is converged
and can therefore be considered exact.) The maxima of the sinc() functions in Eq. (25) occur at = = 0 and
have values of 1. Thus the maximum value of the RCS is given by Eq. (6). As the dimensions of the plate in
wavelengths increase, the width of the central lobe narrows. Note that the major difference between the two
methods occurs at wide angles (near 90 ). In this region the RCS is primarily determined by the current
distribution near the plate edges, which is where the PO approximation fails.
Physical Theory of Diffraction. The PTD, also known as the method of edge waves (MEW), is a currentbased approach to obtaining the edge scattered field (24). The total current on a conductor with an edge can be
decomposed into a PO component and a nonuniform (edge or fringe) component. If the fringe current is known,
RADAR CROSS-SECTION
25
then the edge-scattered field can be computed by using this current in the radiation integral. The total field at
an observation point is the sum of the PO and edge-scattered fields.
For a surface that is very large relative to the wavelength of the incident wave, the current distribution at
a point near the edge is primarily determined by the local features of the geometry. If the edge is very long and
the point under consideration is not close to the end of the edge, then the current is approximately the same as
that for an infinitely long edge. Thus edge scattering can be considered a local phenomenon, and the current
approximated by the current of the corresponding canonical scattering problem. (A canonical problem is one
that has fundamentally the same geometry as the one under consideration, but whose solution is known.) If
the total current can be determined for a canonical problem, in this case an infinite knife edge, then the PO
component can be subtracted out and the result used as an estimate for the fringe current on a finite edge. The
fringe currents are then used in the radiation integrals to obtain an edge-scattered field.
This approach has the advantage that the edge-scattered fields remain finite at shadow and reflection
boundaries. However, the fringe currents extend a significant distance from the edge (a minimum of about
0.5), so that two-dimensional integrals must be evaluated. Michaeli has reduced the PTD fringe currents to
filaments that flow along the body edge, thus reducing the integration to one dimension (25).
The electric and magnetic current filaments for a knife edge are given by
where
The PTD is simple in principle, but can be inconvenient in practice. The currents in Eqs. (26) and (27)
are a function of incidence angle, so for monostatic calculations the integrations must be performed again after
each angle change. Another disadvantage is that modeling the fringe current interaction is difficult. On the
plus side is that the edge-scattered field is simply added to the PO field. This is the major attraction of the
PTD, since PO is so convenient and widely used.
Geometrical Optics. GO is the theory of ray tracing that has been used for centuries to design systems of
lenses such as telescopes. It works well at optical frequencies because the lens and mirror dimensions are much
larger than a wavelength and thus the interaction of the wave with the lens becomes a localized phenomenon.
The reflected ray appears to originate from a single point on the surface, called a specular or reflection point.
GO is the most basic theory that describes wave behavior upon reflection or refraction at an interface
between two materials (10). GO is based on seven assumptions or postulates:
(1) Waves are everywhere locally plane and TEM.
26
RADAR CROSS-SECTION
(2) The wave direction is specified by the normal to the equiphase surfaces, that is, a ray. (Equiphase surfaces
are also referred to as eikonal surfaces.)
(3) Rays travel in straight lines in a homogeneous medium (Fermats principle).
(4) Polarization is constant along a ray.
(5) At the reflection point, the reflected fields are linearly related to the incident fields by reflection coefficients.
(6) Reflected rays satisfy the law of reflection.
(7) Power in a tube of rays is conserved. The field strength is inversely proportional to the square root of the
cross-sectional area of the flux tube.
Ray tubes are also called ray bundles or flux tubes. For a plane wave, all of the rays are parallel and
therefore a bundle of rays has a constant cross section. However, for a spherical wave, a bundle of rays expands
as the distance from the source increases. The power flowing through a slice of the bundle must be constant if
the medium is lossless and power is conserved.
For the general scattering problem the incident wave front and scattering surface may have some curvature, as illustrated in Fig. 12. The local curvature at a point on a doubly curved surface, whether it is a wave
front or target surface, can be expressed in terms of two principal radii of curvature. Let the incident wave
front be characterized by the principal radii of curvature Ri 1 and Ri 2 . Similarly, let the scattering surface be
characterized by the principal radii of curvature Rs 1 and Rs 2 . If these two sets of radii are known, then the
radii of curvature of the reflected wave front can be computed using the following equation:
where f 1,2 are the focal lengths in the two principal planes. (The focal length is another measure of curvature.)
The focal lengths can be computed by (4)
where the angles are given by
The vectors I1 and t2 are tangent to the surface at the reflection point and lie in principal planes 1 and 2,
respectively.
RADAR CROSS-SECTION
27
Fig. 12. GO can be used to predict the reflected field from a doubly curved surface. The curvatures of the incident wave
front and surface affect the curvature of the reflected wave front.
The postulates allow the reflected field to be computed from the incident field at the reflection point by
multiplying by a reflection coefficient. In matrix form,
where
and the reflection coefficient matrix is
The subscripts and refer to a polarization where the electric field is parallel or perpendicular to the plane
of incidence (the plane defined by the vectors I and i ).
28
RADAR CROSS-SECTION
The divergence and spreading factor for the reflected wave is
The quantity s is the distance from the source point S to the reflection point Q, and s is the distance along the
reflected ray from the reflection point to the observation point P.
The last factor in Eq. (31) allows for phase transitions through caustics, which are points or lines at which
the cross section of a ray tube is zero (and thus the power density is infinite); we have
A well-known formula for the RCS of a curved surface can be derived using the above formulas. The
fact that for a plane wave Ri 1 = Ri 2 = and the observer is far from the target (Rs 1 , Rs 2 s) yields the
approximation
Equation (37) has been widely applied to basic geometric shapes such as spheres and ellipsoids. Note that GO
predicts zero field in shadow regions. That is, if the path between the source and observation points is blocked,
then there is no received field.
Geometrical Theory of Diffraction. GO provides a means of calculating the field reflected from a surface.
The associated edge effects require a mathematical model for edge diffraction. The GTD is such a ray-based edge
diffraction theory (26). For the GTD, rays are hypothesized that obey diffraction laws similar to reflection laws.
A diffraction coefficient is defined, which depends on the edge geometry and polarization of the incident wave.
Upon diffraction, the scattered field is given by the incident field times the diffraction coefficient. Furthermore,
the diffracted wave follows prescribed straight line paths in free space. The total field at an observation point
is the vector sum of all the reflected and diffracted fields arriving at that point.
As first developed, the GTD had several shortcomings with regard to singularities in the field. Several
modified versions of the GTD, such as the uniform theory of diffraction (UTD) (27) and the asymptotic theory
of diffraction (ATD) (28), have eliminated the problems in many cases by introducing correction factors. In the
following discussions GTD will refer to the collection of these diffraction theories.
The postulates of the GTD are similar to those of GO; only numbers 5 and 6 change:
(1) At the diffraction point, the diffracted fields are linearly related to the incident fields by diffraction coefficients.
(2) Diffracted rays emerge radially from the edge.
In addition, rays traversing a caustic encounter a phase advance of 90 ; an edge is always a caustic. At a
focus there is a phase advance of 180 .
The behavior of diffracted rays from an infinite conducting wedge is illustrated in Fig. 13. The diffracted
rays lie on the surface of a cone (referred to as the Keller cone) with its tip located at the diffraction point and
RADAR CROSS-SECTION
29
Fig. 13. The GTD is used to predict edge diffracted fields. As long as the diffraction point is not near the end of the edge,
the diffracted field is similar to that of an infinitely long edge.
axis along the edge. The cone half angle is equal to the angle that the incident ray makes with the edge,
(a consequence of Fermats principle). The coordinate variables (s, , ) and (s , , ) uniquely define the
directions of diffraction (observation) and incidence (source), respectively.
In the light of postulate 5 the diffracted field from an edge can be written as
The vectors Ed and Ei have the same form as in Eq. (32) and Eq. (33), except that parallel and perpendicular
are defined relative to the edge rather than the plane of incidence. In Eq. (38) D is a two-dimensional matrix
of diffraction coefficients,
and the divergence and spreading factor is
where c is the distance from the caustic.
The diffraction coefficients are derived from the appropriate canonical problem. The GO field is subtracted
from the total scattered field. The difference is the diffracted field, which can be used to determine the elements
of D, given the known form of Eq. (36). For a knife edge the diffraction coefficients are
30
RADAR CROSS-SECTION
where the transition function is
with
In special cases where Ed is tangent to the surface of a PEC, the surface diffracted ray is zero. For a
rectangular plate, there will not be an interaction (higher-order diffraction) between the two parallel edges. In
fact, there is an interaction between two edges of a parallel plate even when the electric field is parallel to the
edge, and it is included by adding a slope diffraction term to Eq. (38)
where D is a matrix of slope diffraction coefficients.
Rigorous Methods.
Method of Moments. The radiation integrals provide a means of calculating the scattered field from
a target once the induced current on the surface is known. Except for the most simple shapes, the current
distribution on the target is not known accurately. For electrically large bodies that are smooth and sufficiently
flat, the PO approximation provides a good estimate at points far from discontinuities and shadow boundaries.
By a discontinuity is meant any abrupt change in the surface contour or composition. Examples are edges,
steps, cracks, and material joints.
In order to compute the current rigorously, an integral equation (IE) for the current can be formulated
from Maxwells equations and the boundary conditions. IEs have the unknown quantity in the integrand and
possibly in other terms outside of the integral. In most problems of practical interest, the IE must be solved
numerically using a computer. The most popular solution technique, called the method of moments (MM),
consists in expanding the current into a series with unknown coefficients (11). The expansion is inserted into
the integral equation and a weighting process applied. The properties of the expansion functions are used to
reduce the integral equation to a set of simultaneous linear equations that can be solved for the coefficients.
The E-field integral equation (EFIE) is the most commonly used. It can be derived by using the general
form of the radiation integrals and placing the observation point on the surface of the body. [The general form
of the radiation integrals does not make the assumption of a far-field observation point as in Eqs. (17) and
(18).] If the target is a PEC, then (Ei + Es )tan = 0 on the surface. The incident field is known, and the radiation
integrals give the scattered field in terms of the unknown surface current. Thus the following form of the EFIE
RADAR CROSS-SECTION
31
is obtained:
where, as usual, primes denote source point quantities associated with the current and unprimed denotes
observation point quantities. The free space Greens function is defined by
with
Equation (45) is reduced to a matrix equation using the MM. First the current is expanded into a series of basis
functions {J n } with unknown current coefficients {In }:
The basis functions are chosen to be mathematically convenient and adequately represent the current over the
surface. They are of two general types: (1) subdomain or (2) entire domain. In the subdomain case, each function
is nonzero over only a part of the surface. On the other hand, each entire-domain basis function is nonzero over
the entire surface. Using a function of time as an analogy, it can be expressed as a series of sinusoids (Fourier
series), which are an example of entire-domain basis functions. The time function could also be expressed as a
series of steps (a stair step approximation), which is an example of subdomain basis functions.
Once the basis functions are chosen, Eq. (47) is inserted into Eq. (45) and a test is performed that involves
multiplying by a set of weighting functions W n and integrating. Generally the complex conjugates of the basis
functions are selected as the weighting functions (Galerkins method): W n = J n . This leads to N equations
which can be cast into the matrix form
32
RADAR CROSS-SECTION
Fig. 14. Any arbitrary shaped surface can be represented by triangular facets. Triangular (rooftop) basis functions span
each edge. The total current is a superposition of all of the basis functions at a point.
where
which is an impedance, and
which is a voltage. Thus Eq. (48) can be solved for I, then used in Eq. (47), which in turn is used in the radiation
integrals to obtain the scattered field.
A commonly encountered basis function is the triangular subdomain (29). The surface of a target is
represented by a collection of triangular facets. This type of subdomain has the flexibility to model curved
surfaces with arbitrary edge contours. Each basis function is associated with an edge as shown in Fig. 14:
The current at a point in a subdomain is the vector sum of the currents crossing all edges, weighted by the
distance from the edge.
RADAR CROSS-SECTION
33
Using Eq. (51) in Eq. (49) gives
where
The excitation vector elements are
where the electric field at the centroid of the facet is
These basis functions result in a linear approximation to the current over each facet. The rigorous result for
the rectangular plate shown in Fig. 11 was computed using the method of moments.
There are a few instances when other basis functions may be more efficient than triangular subdomains.
For example, consider a body of revolution (BOR), which is generated by rotating a plane curve around an
axis. Included in this class are cylinders, cones, spheres, and ellipsoids. A hybrid basis function, that is, a
combination of entire-domain and subdomain basis functions, is an efficient alternative to triangular facets
(30). In these cases overlapping triangular basis functions can be used to represent the current along the
generating curve axis, while entire-domain complex exponentials are used in the azimuthal direction (around
the circumference of the body). The result is a linear approximation to the current in the axial direction (from
back to front) and a Fourier series representation for the current around the body. The major advantage of
this approach is that each Fourier mode, which is actually a matrix whose size is determined by the number of
subdomains along the axis, is independent of all other modes. Thus a large-matrix problem is decomposed into
a series of smaller matrices, thereby permitting larger bodies to be modeled for a given computer resource.
Hybrid Methods. Hybrid computation methods combine several prediction techniques in a single solution (31,32). A hybrid solution decomposes a problem into subproblems, which are solved using different
techniques. Generally the mutual effects (interaction between the subproblems) are included to some degree.
For example, PO is known to give excellent results for large smooth surfaces away from edges and discontinuities. A hybrid approach might use PO on regions of the surface away from discontinuities and the MM near
discontinuities. Larger problems can modeled than with a pure MM approach, and more accuracy is achieved
than using PO separately. The disadvantage is that each hybrid combination is tailored for a specific type of
problem and therefore may not yield accurate results under all conditions.
One hybrid method that is used extensively in RCS calculation for large targets at high frequencies is
the shooting and bouncing ray (SBR) technique (33). It combines GO and PO, and can be supplemented by
34
RADAR CROSS-SECTION
Fig. 15. A complex multilayered target can be replaced by an equivalent target having the same surface contour but an
equivalent surface impedance.
the PTD. Current distributions are determined from the trajectories of rays in a dense bundle. The current
distributions can be used to:
(1) Initiate secondary rays to evaluate multiple reflections
(2) Compute PTD fringe currents for edge diffraction contributions
(3) Compute the far scattered fields
Currently this technique is the most accurate and flexible for computing the RCSs of large complex targets
such as aircraft with engine inlets and cockpits.
Surface Impedance Approximation. The impedance boundary is an approximate boundary condition
that can be used to replace a surface composed of multiple materials and layers by a single surface. It was first
proposed by Leontovich (34) and is frequently referred to as the Leontovich boundary condition. The concept
is illustrated in Fig. 15, which shows the use of a surface current approach to calculate the scattered field from
a target composed of several materials.
The possibility of surface currents flowing on all of the interfaces dramatically increases the mathematical
and computational complexity of the problem over that of a single surface. In many cases it may be sufficient
to replace the collection of surfaces with a single layer that has approximately the same external scattering
properties as the combination of original surfaces. (For the RCS the fields inside the body are not of interest.)
The same scattering properties for the original and equivalent problems are ensured if the boundary conditions
on the tangential field components are the same. The required relationship between the electric and magnetic
fields is referred to as a surface impedance approximation.
Mathematically it relates the tangential components of the electric and magnetic fields at every point on
the surface:
RADAR CROSS-SECTION
35
where s is the surface impedance in ohms. In vector form,
There is a dual to this equation:
Comparing Eqs. (53) and (54) reveals that the surface impedance relates the electric and magnetic currents at
every point on the surface:
Note that s = 0 corresponds to a PEC, and the resulting reflection coefficient for a wave normally incident on
the PEC is
The surface impedance approximation is most accurate when:
(1) The surface is electrically large and the incident wave is near normal incidence to the surface (i 0 ).
1).
(2) The relative dielectric constant of the material is high (
(3) The local radius of curvature of the surface is large compared to a wavelength.
Impedance boundary conditions can be used to derive integral equations (35) or to simplify an approximate
solution. They are widely used in spite of the fact that they usually cannot be justified rigorously.
An estimate of the surface impedance can be obtained from the permittivities and permeabilities of the
constituent materials. For layers, a transmission line model can be used to compute the reflection coefficient
at the outer surface, and then Eq. (56) solved for s . Using this approach, s is a function of the angle of
incidence because the Fresnel reflection coefficients are angle-dependent. Generally an average value near
normal incidence is used, because it is near the local normal that each region of the surface makes its greatest
contribution to the RCS.
A second set of approximate boundary conditions are used to represent thin films (36). They are the
resistive sheet and conductive sheet boundary conditions, which provide a relationship between the currents
on the surface and the tangential field components. A resistive sheet is an infinitely thin imperfect electric
conductor (finite conductivity) and it does not support a magnetic current:
where Rs is the surface resistivity in ohms per square. The limiting cases are Rs = 0 for a PEC and Rs = for
a perfectly transparent sheet. The reflection coefficient of the sheet is
36
RADAR CROSS-SECTION
Fig. 16. The RCS of a metal skin aircraft is compared to that of a composite aircraft. The monostatic RCS is lower for
the composite aircraft because more of the incident field is scattered in the forward direction. The surface impedance
approximation has been used.
Resistive films are an effective means of controlling the reflection coefficient of a surface. They consist of
thin metallic deposits on low-dielectric films such as Mylar. A common example of a commercially available
resistive film is window tinting of the type used for automobiles.
The dual of the resistive sheet is the conductive sheet, with the boundary condition
where Gs is the surface conductance. The conductive sheet does not support an electric current. A combination
of resistive and conductive sheets can be viewed as an impedance sheet.
Figure 16 compares the RCSs of PEC and composite aircraft wings. The RCS of the composite wing is
significantly reduced because more of the wave is transmitted and less reflected.
Rough Surfaces. Imperfections and errors in the assembly and manufacturing processes have a significant effect on the low-level RCS. The sources of errors are variations in the physical dimensions and electrical
characteristics of the target materials. Most of the errors behave in a random fashion. Random deviations
in the characteristics of a target result in an increase in diffuse scattering and a corresponding decrease in
specular reflection.
The Rayleigh condition is the traditional criterion for a rough surface:
RADAR CROSS-SECTION
37
where h is the average height of the irregularities and the grazing angle. The surface impedance boundary
condition can also be used to simulate surface irregularities (37).
The calculation of the RCS for targets with random characteristics can be performed using a Monte
Carlo simulation (method of repeated trials) or by applying probability theory to the governing equations
of electromagnetics. Both approaches start by writing the RCS in terms of the random variables of interest
(for example, target dimensions). In the Monte Carlo method, the equations are computer-programmed and
repeatedly executed using a different random seed each time. The average RCS and its standard deviation can
be computed from the output data.
Another approach is to take the expected value of the RCS expression. If the probability density functions
of the random variables are known, then the expectation operation results in an expression for the average
RCS. Using this approach, the RCS of a lumpy surface with a correlation interval c (the distance beyond
which on average the surface deviations become independent) is given approximately by (6)
where
is the variance of the surface deviations (assumed small). P0 is the error-free power pattern
normalized to A2 . (Using a flat plate as an example, P0 = [sinc(ua) sinc(vb) cos ]2 .) The second term in the
brackets of Eq. (60) represents a random noise that increases with the level of the error. A small correlation
interval yields lower, more uniformly dispersed scattering than a large correlation interval.
Measurement of Radar Cross Section
Measurements are a crucial step in the design and verification of an LO platform. In general, measured RCS
data can serve three purposes:
(1) To establish a targets RCS for design verification
(2) For comparison with calculated RCS to evaluate numerical methods or the validity of approximations
(3) As a diagnostic tool
Measurements are performed throughout the design of a stealthy platform. During the conceptual stage,
scale models of candidate designs may be built and tested to verify that the RCS levels are close to those
expected. At various stages in the development of the full-scale components (such as wings and antennas),
measurements can be used to verify assembly tolerances, RCS treatments, and other manufacturing processes.
Problem areas can be reworked and then checked again. Finally, if the target is not too large, a measurement
can be made on the completed product. This step may serve as part of the official fulfillment of a contract.
Measurement facilities can be categorized by a variety of attributes that include physical characteristics,
instrumentation capabilities, and data analysis and presentation modes. Table 4 summarizes some commonly
used descriptors.
Most systems measure monostatic RCSs by using a fixed transmit/receive antenna and rotating the target.
The traditional continuous wave (CW) measurement involves two steps. First, a calibration target of known
RCS (typically a sphere) is placed in the chamber, and the received power is recorded for reference. Next,
the calibration target is replaced by the object with unknown RCS, and the power measured as the target is
rotated. By comparing the received and reference powers, the RCS of the unknown target is obtained. In an
analog system, the power difference drives the vertical (RCS) axis of a plotter, and the horizontal (angle) axis
38
RADAR CROSS-SECTION
is slewed to the target mount. The final result is a plot of RCS versus aspect angle. The data obtained using
the classical far-field CW RCS measurement of a rectangular plate are illustrated in Fig. 17.
For a true far-field measurement the target must be illuminated by a plane wave. This condition is
never rigorously satisfied in practice, because the need for a directive transmit antenna usually results in
a nonuniform-amplitude illumination of the target. Furthermore, the wave incident on the target is always
spherical, not planar. Finally, because the system components are near the floor, ceiling, and walls, reflections
from these surfaces affect the measurement.
Figure 17 illustrates the types of inaccuracies that can occur, such as sidelobe asymmetry and the existence
of a noise floor. The noise floor sets a lower limit on the RCS that can be reliably measured. Measures of chamber
performance are:
(1) Pattern symmetry in angle
(2) Noise floor (noise level)
(3) Quiet zone (linear extent of the region where the amplitude and phase of the incident wave are approximately plane)
(4) Repeatability
(5) Allowable target size and weight for the mount
(6) Allowable range of target motion
(7) Instrumentation capabilities
(8) Data-processing capabilities
(9) Data-presentation capabilities
The state-of-the-art instrumentation used in RCS measurement is essentially a programmable multimode
radar. Data can be collected in either the time or the frequency domain and transformed back and forth between
the two. In the time domain, gating can be employed to discard signals that do not correspond to the time delay
of a scattered signal from the targets extent. Processing such as windowing can be performed on the raw
RADAR CROSS-SECTION
39
Fig. 17. The measured RCS of a 16 inch square plate for a 2 GHz CW source. Errors in the measurement were more
pronounced in older measurement systems. Background subtraction techniques can be used to reduce measurement errors.
data to enhance various characteristics and then see how they behave as a function of frequency or aspect.
Data-processing techniques can even generate high-resolution images that highlight scattering hot spots on
the target.
Chamber Congurations. The majority of all RCS measurement facilities are completely enclosed
(indoor ranges). This provides isolation from the external environment, which is desirable for both security
and technical reasons. In an indoor chamber the interference from other RF sources is reduced or eliminated,
and the temperature is controlled for increased instrumentation stability. The result is a more accurate and
repeatable measurement than possible in an outdoor facility. The obvious limitation on an indoor chamber is
the size of the building required for large targets.
Figure 18 shows two types of indoor chambers. The first is the standard far-field configuration. The
separation of the antenna and target must be large enough so that the difference between an ideal plane wave
and the actual spherical wave is negligible. The required distance is a function of target size and the frequency.
If the distance is R and the target length L, as indicated in Fig. 19, the round trip phase difference between
rays to the edge and center is
40
RADAR CROSS-SECTION
Fig. 18. Common indoor chamber configurations include the conventional far-field range, the tapered chamber, and the
compact range. The compact range is capable of handling physically large targets over a wide frequency range.
where the right-hand side holds for R L. Based on the convention used for the definition of the far field of a
radiating antenna, one may limit the round trip phase error to /8, so that the minimum range becomes
This limit turns out to be overly restrictive and is usually not adhered to in practice. Equation (61) assumes
that all of the target surface contributes coherently to the total RCS, as will happen for example, with a smooth
plate. On the other hand, if the target is composed of many independent scatterers, the far-field distance is
primarily determined by the distribution of dominant scatterers in space. Thus when the dominant scatterers
are clustered, a much smaller range is acceptable (38).
In addition to a plane phase front at the target, the amplitude of the incident wave across the target must
be constant. The amplitude distribution is controlled by the antenna beamwidth and the ratio L/R. An isotropic
antenna provides the flattest phase front, but also results in stronger illumination of the chamber walls and
RADAR CROSS-SECTION
41
Fig. 19. The incident spherical wave from the transmitter deviates from a plane wave. The deviation leads to an error in
the measured RCS. To reduce the error the transmitter to target distance must be increased.
Fig. 20. A compact range provides a uniform field over the target with shorter antenna-to-target distances than a
conventional far-field range. Some sources of measurement error are illustrated.
other obstacles. The transmitted power must also be higher if the antenna gain is low. Thus there is a tradeoff
involved in determining the optimum antenna configuration.
The demands on chamber size required for accurate low-level RCS measurements have resulted in the
development of compact ranges. The layout of a typical compact range is shown in Fig. 20. The parent reflector
configuration is a Cassegrain, and the reflector surfaces have been arranged so that no blockage occurs. The
antenna is located at the focus, and a plane wave field is excited a short distance in front of the paraboloid. The
total volume of a compact range is much less than that of a far-field configuration.
The compact range is not without its own problems, though. To obtain uniform amplitude over the target
requires a shaped feed pattern or some subreflector shaping. Cross polarization and edge diffraction can be
significant. Since the target is close to the reflector edges, the effects of edge scattering are especially important
at low RCS levels. Much effort has gone into the design of rolled or serrated edges that control the level and
direction of edge diffraction.
Sources of Measurement Error. The received signal in a measurement is not entirely due to scattering from the target. It also includes noise introduced by the receiving system and extraneous scattered signals
42
RADAR CROSS-SECTION
Fig. 21. Radar images can be generated by plotting the intensity of the scattered field as a function of the target surface
coordinates. The downrange size of the resolution cell is determined by the radar pulsewidth; the crossrange resolution is
determined by the Doppler measurement capability of the radar.
originating from the chamber walls and other obstacles. Furthermore, some of the transmit signal couples
directly to the receive antenna. This leakage is strongest when a common antenna is used for both transmit
and receive.
Chambers that are used to measure low-level RCSs must be carefully designed, maintained, and calibrated. Great care is taken to reduce the error signals in the vicinity of the target. The cross-sectional dimension
of the chamber that is essentially free of extraneous signals and maintains the plane wave phase and amplitude
is called the quiet zone. It is usually specified in feet or meters. The quiet zone, frequency range of operation,
and noise floor effectively characterize the chamber performance.
Measurement errors are mitigated using background subtraction. First a measurement is taken with the
target absent. Ideally the received signal should be zero if the chamber walls are perfect absorbers and there
is no leakage. In practice a nonzero signal is present due to the sources indicated in Fig. 20, and this value is
stored as a reference. Next, the measurement is repeated with the target present. The measured value includes
the targets RCS as well as the error signals. If the interaction between the target and chamber is negligible,
vector subtraction of the background signal should yield a close approximation to the isolated targets RCS.
The subtraction method requires that the characteristics of the measurement equipment remain stable
during the calibration and measurement runs. If the power level or frequency drifts, the background reference
is no longer valid and the difference is attributed to the target. Even small changes in phase caused by cable
movement can affect low RCS measurements.
Target Mounts. A potential source of large measurement error is the target mount. If at all possible,
the pedestal should be made of low-density material that does not scatter. Frequently the target is suspended
and held in tension using thin polyester filaments similar to fishing line. For large, heavy targets that must
be tilted and rotated there is no alternative but to use metal mounts. In this case special shaped fixtures are
used to shield the pedestal. Absorbing material can also be used in conjunction with shaping to reduce mount
scattering.
Resolved RCS and Target Imaging. Returns from individual scattering elements on a complex target
can be isolated or resolved. The strength of the return is plotted as a function of downrange (y) and crossrange
(x) in Fig. 21. If the cell size becomes small enough, a high-resolution two-dimensional image of the target can
be constructed. As shown in Fig. 22, imaging clearly identifies the strongest scattering sources on a target.
Resolving the individual scattering sources on a target requires that accurate angle and range information be available as the returns are received. Range resolution is achieved by using a short pulse that only
illuminates a small slice of the target. Knowledge of the time elapsed after the leading edge of the pulse hits
the target is equivalent to knowledge of the downrange coordinate y.
One method of obtaining the crossrange value is by accurately pointing a narrow antenna beam. The
beam is scanned over the entire solid angle of the target, thereby providing angle data that can be processed
RADAR CROSS-SECTION
43
Fig. 22. The radar image of the target can be used to determine where RCS treatments should be applied, as well as
evaluate the effectiveness of RCS reduction methods.
Fig. 23. A point rotating on the surface of the target has a Doppler frequency shift. The frequency shift can be used to
determine the crossrange coordinate.
along with the range data to construct an image. This approach is not efficient at microwave frequencies,
however, because of the large antenna size needed for fine angle resolution. It is more practical at millimeter
wavelengths and smaller, and is highly successful at laser wavelengths. A more practical method for obtaining
crossrange exploits the Doppler shift (39). The target is rotated at an angular velocity r as shown in Fig. 23.
Target surface points rotate with a linear velocity that depends on their distance from the center of rotation.
The radial component of relative velocity (component along R) gives rise to a Doppler shift.
Referring to the coordinate system shown in Fig. 23, let R0 be the range to the targets center of rotation
along the y coordinate, and r the distance to a scattering point at a distance d from the center. Using the law
of cosines,
44
RADAR CROSS-SECTION
which simplifies to
when R0 d. The first term under the square root is associated with the round trip time delay to and from the
target center, and the second is the Doppler shift due to rotation. Since = r t (where r is the rotation rate of
the target), the Doppler frequency in hertz is
This method of rotating the target to obtain crossrange information is the same principle used in inverse
synthetic aperture radar (ISAR) method.
BIBLIOGRAPHY
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
22.
23.
24.
25.
26.
27.
M. Skolnik, Introduction to Radar Systems, 2nd ed., New York: McGraw-Hill, 1980.
B. Eddie, Radars, Engelwood Cliffs, NJ: Prentice-Hall, 1993.
G. Ruck et al., Radar Cross Section Handbook, New York: Plenum, 1970.
C. Balanis, Advanced Engineering Electromagnetics, San Francisco: Harper & Row, 1982.
K. Kunz, R. Luebbers, The Finite-Difference Time Domain Method for Electromagnetics, Boca Raton, FL: CRC Press,
1993.
D. Jenn, Radar and Laser Cross Section Engineering, Washington, DC: AIAA Education Series, 1995.
A. Ishimaru, Electromagnetic Wave Propagation and Scattering, Englewood Cliffs, NJ: Prentice-Hall, 1991.
L. Peters, End-fire echo area of long, thin bodies, IRE Trans. Antennas Propag., 133, January 1958.
R. Harrington, Time-Harmonic Electromagnetic Fields, New York: McGraw-Hill, 1961.
J. Keller, Determination of reflected and transmitted fields by geometrical optics, J. Opt. Soc. Amer., 40 (1): 48, 1950.
R. Harrington, Field Computation by Moment Methods, New York: Macmillan, 1961.
W. Streetman, J. Goodall, Lockheed F-117A, Osceola, WI: Motorbooks International, 1990.
D. Cheng, Field and Wave Electromagnetics, 2nd ed., New York: McGraw-Hill, 1989.
M. Schwartz, Composite Materials Handbook, 2nd ed., New York: McGraw-Hill, 1992.
V. Weston, Theory of absorbers in scattering, IEEE Trans. Antennas Propag., AP-11: 578, 1963.
N. Engheta, D. Jaggard, Electromagnetic chirality and its applications, IEEE Antennas Propag. Soc. Newslett., 6,
October 1968.
R. Harrington, Theory of loaded scatterers, Proc. IEE Electron., 111 (4): 617623, 1964.
R. Harrington, The control of electromagnetic scattering by impedance loading, Proc. IEEE, 53: 9931004, 1965.
E. Knott, J. Schaeffer, M. Tuley, Radar Cross Section, Norwood, MA: Artech House, 1989.
Navy Times, Jan. 12, 1998, p. 4.
E. Pelton, B. Munk, A streamlined metallic radome, IEEE Trans. Antennas Propag., AP-22: 799, 1974.
D. C. Schleher, Introduction to Electronic Warfare, Norwood, MA: Artech House, 1986.
S. Silver, Microwave Antenna Theory and Design, New York: McGraw-Hill, 1949.
P. Ufimtsev, Method of Edge Waves in the Physical Theory of Diffraction, translated from Russian by the USAF Systems
Command, Foreign Technology Division, Sept. 7, 1971.
A. Michaeli, Equivalent edge currents for arbitrary aspects of observation, IEEE Trans. Antennas Propag., AP-32:
252258, 1984; corrections, AP-33 (2), 1985.
J. Keller, Geometrical theory of diffraction, J. Opt. Soc. Amer., 52 (2): 116, 1962.
R. Kouyoumjian, P. Pathak, A uniform geometrical theory of diffraction for an edge in a perfectly conducting surface,
Proc. IEEE, 62: 1448, 1974.
RADAR CROSS-SECTION
45
28. J. Boersma, Y. Rahmat-Samii, Comparison of two leading theories of edge diffraction with the exact uniform asymptotic
solution, Radio Sci., 15 (6): 1179, 1980.
29. S. Rao, D. Wilton, A. Glisson, Electromagnetic scattering by surfaces of arbitrary shape, IEEE Trans. Antennas Propag.,
AP-30: 409, 1982.
30. L. Medgyesi-Mitschang, J. Putnam, Hybrid solutions for scattering from large bodies of revolution with material
discontinuities and coatings, IEEE Trans. Antennas Propag. AP-32: 717, 1984.
31. E. Ekelman, G. Thiele, A hybrid technique for combining the moment method treatment of wire antennas with the
GTD for curved surfaces, IEEE Trans. Antennas Propag., AP-28: 831, 1980.
32. T. J. Kim, G. Thiele, A hybrid diffraction techniquegeneral theory and applications, IEEE Trans. Antennas Propag.,
AP-30: 888898, 1982.
33. H. Ling, R. Chou, S. Lee, Shooting and bouncing rays: Calculating the RCS of an arbitrarily shaped cavity, IEEE Trans.
Antennas Propag., AP-37: 194, 1989.
34. T. Senior, Impedance boundary conditions for imperfectly conducting surfaces, Appl. Sci. Res., 8(B): 418, 1960.
35. L. Medgyesi-Mitschang, J. Putnam, Integral equation formulations for imperfectly conducting scatterers, IEEE Trans.
Antennas Propag., AP-33: 206, 1985.
36. T. Senior, Combined resistive and conductive sheets, IEEE Trans. Antennas Propag., AP-33: 577, 1985.
37. T. Senior, Impedance boundary conditions for statistically rough surfaces, Appl. Sci. Res., 8(B): 437, 1960.
38. B. Welsh, A minimum range criterion for RCS measurements of a target dominated by point scatterers, 1984 APS/URSI Int. Symp., Vol. APS-17-3, p. 666, 1984.
39. D. Mensa, High Resolution Radar Cross Section Imaging, Norwood, MA: Artech House, 1981.
40. RCS Measurement Capabilities at the Pacific Missile Test Center, Pt. Mugu, CA: Radar Signature Branch.
DAVID C. JENN
Naval Postgraduate School
RADIO SYSTEM PERFORMANCE
RADIO LINK DESIGN
Since the early introduction of radio, the signicance and
number of applications of radio services have constantly
been growing. From sophisticated satellite and microwave
links which provide world-wide communication infrastructure, mobile radio and personal communication services
that are providing seamless communication capabilities,
all the way to consumer products like cordless phones, wireless local area networks, wireless toy communicators or
garage door openers, radio services have changed everyday life in a manner that could have not been predicted
decades ago. The growing sophistication of systems and
usages is closely related to the technical understanding
necessary for putting radio systems into operation. Challenges in radio system design have grown tremendously
due to the overcrowding of radio spectrum. The need for
co-ordinating radio services has been identied early and
International Telecommunication Union (ITU) has established necessary regulations. The effective spectrum management is essential to maximizing the benets that can
be obtained from the radio spectrum (1). After restructuring of the ITU in 1993, these problems have been the focus
of the ITU Radiocommunication Sector.
Radio systems operate in different radio frequency
bands as shown in Table 1. (2).
This inuence the characteristics of the radio channel.
Signal propagation path may vary from clear line-of-sight
to severely obstructed by buildings, mountains and other
environmental effects. Since radio channels are random in
nature, understanding of the radio propagation phenomena and consequently their modeling play a major role in
the prediction of the radio reception quality which in turn
enables efcient design of wireless systems. Large-scale
propagation models characterize signal strength over long
distances, while small-signal models describe the rapid
uctuation of the received signal over very short distances
or short time duration (3).
Radio reception is inuenced by different factors, summarized in Table 2. Radio propagation inuences radio link
design, with path loss models and statistics of large-scale
fading being the major elements of power budget. In addition to power budget, radio link design includes geographical positioning of transmitter and receiver, frequency planning, radio link control scheme and handover schemes. Requirements for the type of radio service, quality of service,
coverage and availability of the radio link provide an additional set of parameters which inuence radio link design. On the other hand, fading affects the radio receiver
design, selection of modulation and coding scheme, techniques for fading compensation, synchronization circuits,
design of radio front-end, etc. Selection of the appropriate
receiver techniques determines the radio sensitivity which
is an integral part of the power budget. Results of radio
link and radio receiver design are a part of wireless networks design. Networks architecture, network interfaces
and network control are closely related to radio reception
which denes the physical layer of the network (4).
Power Budget
Power budget (link budget) is the starting point for a radio communication link design. Major factors inuencing
the power budget are shown in Fig. 1. These factors include transmitting and receiving antenna gains GT and
GR , path loss Lp (which depends on operating frequency,
environment and type of radio service), fading margin MF ,
interference margin MI , and radio receiver sensitivity RS.
Transmitting antenna gain (usually specied as the
gain relative to isotropic radiator) and receiving antenna
gain are dependant on the carrier frequency and physical size of the antenna. For a given radio system, selection
of the equipment determines antenna gains. Additional
sources of signal loss at the antenna may include noise due
to protective cover, pointing loss, feeder loss and antenna
efciency described as the ratio of the effective aperture to
the physical aperture of the antenna (5).
Link budget analysis of the radio system denes the
maximum acceptable path loss
L p [dB] = PT [dBm] + GT [dBi] + GR [dBi] MF [dB]
MI [dB] RS[dBm]
which can be used to estimate the coverage area of the
system by determining the maximum transmitter-receiver
distance.
Relationship between path loss and coverage area is
crucial for the design of radio system. Other sources of signal loss in radio propagation are caused by environmental
effects such as rain, clouds, fog, etc.
Fading margin determines the outage probability of the
system due to large-scale and/or small-scale fading. In a
similar way, interference margin can be built into the link
budget to maintain desired performance in the presence
of the interference, regardless of the interference mechanism: adjacent channel interference (ACI), cochannel interference (CCI) in cellular systems, intermodulation distortion (IMD) products created by the large interfering signals, etc.. The amount of margin depends on the required
quality of service. It is expressed in terms of signal to noise
ratio (SNR) for analog systems or bit error rate (BER) for
digital systems, and availability of the link which is the
measure of long-term link utility stated on an average basis, usually on the annual basis.
Radio receiver sensitivity describes the ability of the radio receiver to detect weak signals, and is derived from the
performance requirements of the given radio service. Sensitivity specication includes the impact of techniques that
are integral part of radio reception at the detection level,
such as diversity, coding, interleaving, as well as the impact
of radio receiver such as noise gure and noise bandwidth.
Frequency Planning
Frequency planning is a part of link control and is essential
part of radio link design. Since the radio spectrum is the
limited resource, coexistence of different services will be
practically impossible without the planning procedure. It is
effective way to optimize spectrum usage, enhance channel
J. Webster (ed.), Wiley Encyclopedia of Electrical and Electronics Engineering. Copyright 2007 John Wiley & Sons, Inc.
Radio System Performance
Table 1. Summary of frequency bands for operating radio systems
Frequency band
Frequency range
Very low frequency
(VLF)
330 KHz
Low frequency (LF)
Medium frequency
(MF)
30300 KHz 300 KHz - 3 MHz
groundwave
ionospheric
High
(HF)
330 MHz
ionospheric
30300 MHz 300 MHz-3 GHz
330 GHz
direct waves,
ground
reected
waves
direct waves
30300 GHz
direct wave
frequency
Very
high
frequency (VHF) Ultra high frequency
(VHF)
Super high frequency (SHF)
Extra high frequency (EHF)
Mode of radio propagation
Ionospheric
Radio services
Major characteristics
navigation,
long
distance
radio
telegraphy
navigation,
AM
broadcasting, aeronautical, maritime
comm.
amateur
radio,
broadcasting, aeronautical maritime
comm.
radio and television
broadcasting, land
mobile radio, paging, GPS
satellite,
radars,
short-range communications,
microwave terrestrial
links
very short range
comm, satellite to
satellite links
small bandwidth,
large antennas,
quite large antennas, high power
transmitters
high variability of
channel conditions
relatively
small
antennas, considerable bandwidth
high losses, directional antennas
enormous
bandwidths, rain absorption
Table 2. Radio system design elements relevant for radio reception
Radio propagation
Large-scale fading
-path loss
-shadowing
Small-scale fading
(multipath fading)
-delay spread
-coherence time
Radio link design
Requirements
-radio services
-quality of service
-coverage/range
Design
-power budget
-TX/ RX
allocation
-link control
Radio receiver design
-modulation
-multiple-access
-coding
-anti-fading techniques
-interfaces
suppression
-synchronization
-receiver hw design
Radio network design
-architecture
-interference
-network control
Table 3. Parameters for regression coefcients calculation
Frequency[GHz]
2
6
10
15
kH
0.000154
0.00175
0.0101
0.0367
kV
0.000138
0.00155
0.00887
0.0335
H
0.963
1.308
1.276
1.154
V
0.923
1.265
1.264
1.128
capacity and reduce different types of interference.
Some of the benets of frequency planning will be illustrated on the example of the mobile radio services. Frequency planning for mobile radio includes channel numbering, channel grouping into subsets, cell planning and
channel assignment. It controls ACI by channel separation which provides adequate isolation, and CCI by selection of frequency reuse pattern. Frequency planning can be
efciently combined with cell sectorization to improve the
system capacity. Different trade-offs are involved in frequency planning based on targeted channel capacity and
signal to interference ratio (SIR). The most popular OMNI
frequency plan in mobile radio is N=7 plan, which allows
for ACI. This plan is outperformed by N=9 plan with respect to SIR, but the drawback is the reduced channel capacity. Detailed analysis of various frequency plans can be
found in (21). Besides the impact on technical parameters
of the system, frequency planning greatly inuences radio
system growth and economics.
Large-Scale Variations: Path Loss
The prediction of the path loss is the most important step in
the radio system planning process since it determines radio
coverage. Prediction models could include only nite number of parameters that can inuence radio propagation,
Radio System Performance
Table 4. Tapped-delay fading channel models for GSM mobile radio (19)
a) typical case for rural area RA x, x-speed of mobile in km/h
Tap number
1
2
3
4
5
6
Relative time [microsec]
0.0
0.1
0.2
0.3
0.4
0.5
Average relative power [dB]
0
4
8
12
16
20
Doppler spectrum
Rice
Class
Class
Class
Class
Class
b) typical case for hilly terrain HT x
Tap number
1
2
3
4
5
6
Relative time [microsec]
0.0
0.1
0.3
0.5
15.0
17.2
Average relative power [dB]
0
1.5
4.5
7.5
8.0
17.7
Doppler spectrum
Class
Class
Class
Class
Class
Class
c) typical case for urban area TU x
Tap number
1
2
3
4
5
6
Relative time [microsec]
0.0
0.2
0.5
1.6
2.3
5.0
Average relative power [dB]
3.0
0
2.0
6.0
8.0
10.0
Doppler spectrum
Class
Class
Class
Class
Class
Class
Table 5. Tapped-delay fading channel models for WCDMA mobile radio: classical Doppler spectrum for all taps (20)
Case 1,speed 3km/h
Relative
Delay
[ns]
0
976
Relative
mean
Power
[dB]
0
10
Relative
Delay
[ns]
0
976
20000
Case 2,speed 3 km/h
Relative
mean
Power
[dB]
0
0
0
Relative
Delay
[ns]
0
260
521
781
Case 3,speed 120 km/h
Relative
mean
Power
[dB]
0
3
6
9
Relative
Delay
[ns]
0
976
Case 4,speed 3 km/h
Relative
mean
Power
[dB]
0
0
Relative
Delay
[ns]
0
976
Case 5,speed 50 km/h*
Relative
mean
Power
[dB]
0
10
Relative
Delay
[ns]
0
260
521
781
Case 6,speed 250 km/h
Relative
mean
Power
[dB]
0
3
6
9
Table 6. Parameters for calculation of the impulse noise median
Environmental category
Business
Residential
Rural
Quiet rural
c
76.8
72.5
67.2
53.6
d
27.7
27.7
27.7
28.6
Table 7. SNR improvement in dB for BER of 1% relative to to single channel reception, comparison of various diversity combining schemes (20)
Diversity branches
2
4
6
Selection
10.0
16.0
18.0
Maximal ratio
11.5
19.0
22.0
Equal gain
10.8
18.0
21.5
Radio System Performance
Figure 1. Power budget for radio link design.
and often factors such as environment (effects of buildings,
man made obstacles, vegetation) are treated separately as
well as multipath fading.
Three major mechanisms of radio propagation include
reection, diffraction and scattering. Rather than focusing
on the details of radiowave propagation which are covered
in (6), the models for large-scale path loss prediction will
be summarized indicating critical parameters that are inuencing radio reception.
Free space propagation model is applicable when there
is unobstructed line-of-sight path between the transmitter and the receiver. This scenario is typical for satellite
systems and microwave radio links. The path loss is given
by
Lfree space [dB] = 10 log(GT [dB]) 10 log(GR [dB])
+ 20 log( f [Hz]) + 20 log(d[m]) 147.6
where GT and GR are transmitter and receiver antenna
gains, f is the carrier frequency and d is the distance between the transmitter and receiver. Free-space path loss
has inverse square law dependence with the distance so
received power is reduced by 6 dB when the range is doubled (20 dB/decade decay with distance). In the same manner path loss increases with the increase of carrier frequency. While increasing the antenna gains may compensate for the loss due to high operating frequency in point-topoint links, in mobile radio links this is not possible due to
required omnidirectional coverage. The equation for freespace path loss is only valid when distance d is in the fareld of the transmitting antenna.
In practical radio channels free-space model conditions
do not apply and further corrections to the path loss have to
be accounted. A simple but practical scenario includes the
direct path and ground reected path between transmitter and receiver. This model has been found to be accurate
for path loss prediction over distances of several kilometres for mobile radio systems and for the line-of-sight microwave links. Assuming that transmitter and receiver antenna heights hT , hR are much smaller than the distance,
path loss is given by
Lground re f l. [dB] = 40 log(d[m]) 10 log(GT [dB])
10 log(GR [dB]) 20 log(hT [m])
20 log(hR [m])
In this case path loss is independent from carrier frequency; however, the inverse fourth power law decay is observed when increasing distance. For distances that are
larger than few tens of kilometres the Earth curvature
should be taken into account and the values of the reection coefcient should be modied (2). Also of importance
is dependence of the reection coefcient on polarization
and materials of the reected surface (2, 3).
Diffraction mechanism allows the radio link to be maintained even when receiver is in the obstructed area, shadowed by the object. Huygens principle could be used to explain the diffraction (2). For practical applications Fresnel
zones should be considered (4). Simple knife-edge diffraction model is often used to calculate signal attenuation.
Objects within Fresnel zones may cause diffraction loss
which is the function of the dimensionless Fresnel-Kirchoff
diffraction parameter
=h
2(d1 + d2 )
d1 d2
The geometry of knife-edge diffraction is depicted in Fig.
2. Diffraction loss can be translated into path loss relative
to free-space loss, and it is given by nomograms or by ap-
Radio System Performance
proximations which are expressed as
L()(dB) =
20 log(0.5 0.62)
20 log[0.5exp(0.95)]
0.8 < < 0
0<<1
20 log[0.4
0.1184 (0.38 0.1) ] 1 < < 2.4
0.225
> 2.4
20 log(
)
In the cases when there are more than one objects affecting
the radio propagation multiple knife-edge model should be
introduced. Various approximate solutions have been presented, with different level of accuracy. Their comparison
is given in (2).
where f is the carrier frequency, hte is the effective transmitter antenna height, hre is the effective receiver antenna
height, d is the distance and a(hre ) is the correction factor
for effective receiver antenna height which is a function
of the cell size. The correction factor for mobile antenna
height for small to medium size city (urban area) is computed as
a(hre )[dB] = (1.1 log( f [MHz]) 0.7)hre [m]
(1.56 log( f [MHz]) 0.8)
and for a large city (dense urban area) as
Path Loss Prediction Models. Prediction models have
evolved extensively in UHF/VHF bands addressing the
growing importance of mobile radio services. Most models
provide the prediction of the median path loss, the loss that
is not exceeded at 50% of locations and for 50% of the time
(2). Knowledge of the radio signal statistics further allows
estimation of the signal variability and prediction of the
area where specied signal strength is achieved for a given
percentage of location. Path loss models differ in applicability to different terrain proles and cover different level
of details, from general to very specic scenarios. Most of
the models are empirical, based on the interpretation of the
measured data in the particular area by tting curves or
analytical expressions. The advantage of this approach is
that it takes into account the variety of propagation factors
through actual eld measurements. However, the models
are strictly applicable only to the environments characterized in measurements and additional measurements are
necessary in different frequency bands and on different locations to provide correction factors. Some of the commonly
used models for outdoor and indoor environments will be
discussed in subsequent sections. For more detailed comparison reader is referred to (2, 3).
Outdoor Prediction Models. The signicance of two models has been established in practice in VHF/UHF band:
the Hata model and the Walsch-Ikegami model (8). Both
models are empirical, derived from experimental data, and
are extensively used in commercial computer-aided prediction tools. In general usage and accuracy of models depends on the propagation environment: Hatas model provides good accuracy in urban and suburban environments,
Walsch-Ikegami model is widely used for dense urban environments and micro cells. Extensions to both models are
provided by The European Co-operative for Scientic and
Technical research (COST) to cover PCS band (9).
The Hata model, based on measurements of Okumura,
has established empirical mathematical formulas for the
path loss and considerably enhanced practical value of the
Okumura method (2). It is restricted for the following range
of parameters: frequency f 150-1000 MHz, height of the
base station antenna hte 30-200 m, height of the mobile
unit antenna hre 1-10 m and distance d 1-20 km. Standard
formula for basic transmission loss in urban area is given
by
Lurban [dB] = 69.55 + 26.16 log( f [MHz]) 13.82 log(hte [m])
a(hre ) + (44.9 6.55 log(hte [m]))log(d[km])
a(hre )[dB] =
8.29(log(1.54hre [m])2 1.1, f c 300MHz
3.2(log(11.75hre [m]))2 4.97, f c 300MHz
Suburban area path loss can be calculated using the modied equation
Lsuburban [dB] = Lurban [dB] 2(log( f [MHz]/28))2 5.4
and path loss in open rural areas is given by
Lrural [dB] = Lurban [dB] 4.78(log( f [MHz]))2
18.33 log( f [MHz]) 40.98
The Hata model is well suited for large cell mobile systems, but not for the PCS cells with the radius on the order of 1 km. The COST analysis of several measurements
conducted in European cities for PCS band have resulted
in extended range of parameters for Hatas model to include 1800 MHz frequency band (9). The COST-231 proposed model for path loss is given by
Lurban [dB] = 46.3 + 33.9 log( f [MHz]) 13.82 log(hte [m]) a(hre )
+(44.9 6.55 log(hte [m]))log(d[km]) + CM
where all parameters are dened for original Hatas formula and
CM =
0dB, f or medium sized
3dB, f or dense urban area
city
and
suburban
areas
COST-231-Hatas model is valid for scenarios with base
station heights larger than roof tops in the vicinity of the
base station. This is the case in large cells and small cells
(maximum range 13 km), and path loss is determined
largely by diffraction and scattering at roof tops in the
vicinity of mobile unit. Main rays propagate in this case
above the roof tops. In practical situation it is not recommended to extend the range of validity on base station
heights below the roof tops, like in the micro cell scenario.
Wave propagation in micro cells (maximum range 0.51
km) is determined by diffraction and scattering around
buildings, and main rays propagate in street canyons. This
phenomena are further addressed by the Walsh-Ikegami
model.
The approach of Walsh-Ikegami model is restricted to
radio paths that are obstructed by buildings and is not applicable if a line-of-sight path exist between the transmitter and receiver antennas within a street canyon. Few parameters are introduced to describe the character of the
dense urban environment as depicted in Figs. 3 and 4:
heights of buildings, widths of roads, building separation
and road orientation with respect to the direct radio path.
Radio System Performance
Figure 2. Knife-edge diffraction geometry.
However, the model still considers only characteristic
values and there is no consideration of the topographical
database of the buildings. The models can be applied even
when no data on the urban parameters are available by
using the default values, which increases the prediction
error. For a line-of-sight case in a street canyon the path
loss is equal to
controling the dependence of the multi-screen diffraction
loss versus distance and
k f [dB] = 4 +
0.7( f [MHz]/925 1) f or
1.5( f [MHz]/925 1) f or
urban and suburban
dense urban area
controling the dependence of the multi-screen diffraction
loss versus radio frequency.
Lb [dB] = 42.6 + 26 log(d[km]) + 20 log( f [MHz]), f or d 20m
COST-231-Walsh-Ikegami model is restricted for the
following range of parameters frequency f 8002000 MHz,
where the rst constant in formula is determined so Lb is
height of the base station antenna hbase 450 m, height of
equal to free-space loss for 20 m. Otherwise the model is
the mobile unit antenna hmobile 13 m and distance d 0.025
composed by three terms and restricted by the free space
km If the data on the structure of buildings and roads are
loss
unknown the default values may be used: for b 2050 m,
w = b/2, hroof 3 m times the number of oors plus the roof
LO + Lrts + Lmsd
Lb [dB] =
height,
roof height is 3 m for pitched, 0 m for at and is
LO f or Lrts + Lmsd 0
90 degrees.
The free space loss is given by
LO [dB] = 32.4 + 20 log(d[km]) + 20 log( f [MHz])
the roof-top-to-street diffraction and scatter loss is
Lrts [dB] = 16.9 10 log(w[m]) + 10 log( f [MHz]) + 20 log(hmobile [m]) + Lori
where
Lori [dB] =
10 + 0.354[deg] f or 0 < 35o
2.5 + 0.075([deg] 35) f or 35 < 55o
4.0 0.114([deg] 55) f or 55 90o
and
hmobile = hroof hmobile
hbase = hbase hroof
The multi-screen diffraction loss
Lmsd [dB] = Lbsh + ka + kd log(d[km]) + k f log( f [MHz]) 9 log(b[m])
consists of number of terms:
Lbsh [dB] =
with
ka [dB] =
18 log(1 + hbase [m]) hbase > hroo f
0
hbase hroo f
hbase > hroo f
54
54 0.8 hbase [m]
d 0.5 km and hbase hroo f
54 0.8 hbase [m] d[km] /0.5 d 0.5 km and hbase hroo f
representing the increase of path loss for base station antennas below the roof tops of the adjacent buildings,
kd [dB] =
18
hbase
18 15
hroof
hbase > hroof
hbase hroof
areas
Radio System Performance
The COST-Walsh-Ikegami model has been veried for
frequencies in the 900 and 1800 MHz band and radio path
lengths from about 100 m to 3 km (9). The path loss has
very steep decay versus the height of the base station antenna when the later one is close to the height of adjacent
buildings and this case generally results in large prediction
errors. Prediction errors are larger when hbase hroo f . The
performance of the model is poor for hbase < < hroo f because
it does not consider wave guiding in the street canyons
and diffraction at corners. For good performance of small
cell area coverage the base station antenna should be installed several meters (4 m or more) above the maximum
roof tops of adjacent buildings within a radius of few hundred meters (e.g. 150 m). The prediction error may be quite
large for micro cells which requires detailed knowledge on
streets and buildings (9).
Indoor Prediction Models
Progress of personal communication services has established the need for accurate models that can support path
loss prediction inside and into the buildings. Major characteristics of indoor radio channel are much smaller distances than in outdoor cases and higher variability of the
environment. Building oor plan, construction materials
and the building type greatly inuence propagation characteristics. Partition losses at the same oor and partition
losses between oors have been extensively measured and
categorized (3).
For radio transmission being originated outdoors path
loss due to signal penetration into building should be considered. Number of factors have been found to inuence
penetration loss including signal frequency, antenna pattern and antenna position. Typical values for penetration
loss measured in 900 MHz band are on the order of 12 dB,
with 6 dB lower loss when building front had the windows
(3).
Losses within buildings and penetration losses can be
combined in the model characterizing propagation into the
building (8)
L[dB] = Lmean [dB] + 10 n log(d[m]) + kF [dB] + pWI [dB] + WE [dB]
where Lmean is the mean path loss from transmitter to the
building, n is power exponent of the distance dependence,
d is the distance into the building, k number of oors between transmitter and receiver, F oor loss factor, p number of interior walls between transmitter and receiver, WI
the internal wall loss factor and WE the external wall loss.
For inbuilding path loss Lmean becomes path loss at 1m distance from the transmitter antenna and WE is not used.
The wall loss factor depends on the construction materials
and varies from 0.4 to 29 dB in the 900 MHz band.
Analysis of the experiments carried out in UK in the 900
MHz band in a typical ofce building with several oors
resulted in path loss model
L[dB] = L0 [dB] + 10 n log(d[m]) + kF
where L0 is path loss at 1m distance from the transmitter
antenna, n is distance power law coefcient, d is vertical
range in the building, k number of oors taken into consideration and F attenuation per oor in dB. At 900 MHz
band n was approximately 4, L0 = 30dB, and F is 5.4 dB
per oor (8).
Recent advanced methods for coverage prediction
within buildings rely on site specic propagation models
and databases which support ray tracing method for deterministically modeling the propagation environment. As
the three dimensional building models and databases become widespread, deterministic methods may prevail for
determining the path loss in a wide range of operating conditions (3).
Coverage Prediction Using Path Loss Models
Propagation loss models can establish median path loss as
a function of the distance between the transmitter and the
receiver. Both theoretical and empirical models indicate
that received signal power decreases logarithmically with
distance with a slope of 10n dB per decade. The value of
the n ranges from 2 for a free-space propagation to larger
values for obstructed paths (e.g. between 35 for urban
mobile radio, 46 for obstructed indoor propagation). However, this model does not consider different clutters that
may results in different signals levels for the same distance. The value of the path loss at a particular location is
random variable with log-normal distribution (Gaussian
when measured in dB units) about the mean value which
is distance dependant. Log-normal statistics describes the
shadowing effect at different locations. In practice values
of the exponent n and standard deviation of the log-normal
fading are determined from measured data. Shadowing is
often referred to as a long-term fading.
The random effects of shadowing cause the signal level
at certain locations to be lower than the specied level.
Therefore it is useful to compute the coverage area with a
certain radius, in particular the percentage of the area with
a received signal equal or greater than the specied level.
Trade-off between the percentage of coverage (for a given
exponent loss n and shadowing variance) and the amount
of fading margin is important element of the power budget.
Details on the coverage computation procedure could be
found in (3, 4).
EFFECTS OF ENVIRONMENT
In addition to large-scale path loss different attenuation
factors due to environment may inuence the power budget. Additional signal attenuation may be result of natural
phenomena or man created environmental conditions.
Weather Effects
For microwave radio with line of sight paths different
weather related effects may contribute to fading and path
loss. Examples include absorption and scattering by snow,
hail, fog and rain. Dry snow does not have signicant effect
on the frequencies below 30 GHz, however wet snow may
cause larger attenuation. In general this attenuation is not
of the concern. However, degradation of antenna characteristics due to snow and ice build-up on the surface may
inuence antenna directivity and precautions are taken by
using protective radomes. For different types of antennas
Radio System Performance
attenuation loss due to accumulated ice is between 27 dB.
Another effect of snow is that it may affect the reection coefcient, and in most cases, depending on the type of snow
reection coefcients is close to 1 (7).
where
L0 = 35
The estimate of the path attenuation exceeded for the
0.01% of the time is given by
Molecular Absorption
The absorption from oxygen and water vapour in the atmosphere is the additional factor that has to be considered
at certain frequencies. The atmosphere extends to an altitude of approximately 20 km, yet it represents a path loss
source that cannot be neglected.
Specic attenuation (given in dB/km) is given in a form
of gure in (8). Local maxima of attenuation occur at frequencies around 22 GHz for the water vapour, and 60 and
120 GHz for the oxygen. The respective values are around
0.2 dB/km for water vapour, and around 15 dB/km and 2
dB/km for the oxygen. However, system comparison indicate that water-vapour attenuation is far less important
than the rain attenuation even in the frequency range of
around 20 GHz.
Rain Attenuation
Two components are important when considering the rain
attenuation. First one is polarization related loss due
to non-spherical shape of rain drops. The horizontally
polarized waves experience larger attenuation than the
vertically polarized ones. This can also result in crosspolarization effect that may be harmful for microwave systems using channel planning based on different polarization. The frequency related attenuation depends on the
drop-size distribution, and is not considered signicant for
the signals below 11 GHz for locations with rain climate
similar to northern hemisphere. For tropical areas critical
frequency is as low as 5 GHz.
Rain attenuation statistics prediction is based on rainrate data, which depends on rainfall microstructure. The
ITU-R model is based on rain rate R0.01% exceeded for 0.01%
of the time with an integration time of 1 min. This information can be obtained from local measurements, and in
the case that measurements are not available from ITU-R
reports. The specic attenuation is calculated as
A0.01 [dB] = R r L
Attenuation Due to Clouds and Fog
In Earth-space radio systems for frequencies higher than
10 GHz the attenuation due to clouds may be important
factor affecting the system performance. The mechanism of
attenuation is well understood since clouds and fog consist
of small droplets and it is possible to express attenuation
in terms of total water content per unit volume (12) as
[dB/km] = Kl [dB/km
kH + kV + (kH kV )cos2 ()cos(2)
2
kH H + kV V + (kH H kV V )cos2 ()cos(2)
=
2k
where is the path elevation angle and is the polarization tilt angle relative to horizontal, = 45 for circular
polarization, and other parameters are given in a Table 3.
The effective path length in this case is calculated by multiplying the actual path length L with a reduction factor
k
r=
1
(1 + L/L0 )
g/m3 ] M[g/m3 ]
where is specic attenuation within the cloud, Kl specic
attenuation coefcient and M liquid water content of the
cloud or the fog. The typical water content is about 0.05
g/m3 for medium fog with visibility of the order of 300 m
and 0.5 g/m3 for thick fog with visibility of the order of 50
m. The specic attenuation coefcient for the frequencies
up to 1000 GHz is given by
Kl =
0.189 f [GHz]
(1 + 2 )
[dB/km/g/m3 ]
where
=
2 +
and complex dielectric permitivity of water is
f (0 1 )
f (1 2 )
( f ) =
+
f p [1 + ( f / f p )2 ]
f s [1 + ( f / f s )2 ]
0 1
1 2
( f ) =
+
+2
[1 + ( f / f p )2 ] [1 + ( f / f s )2 ]
where
0 = 77.6 + 103.3( 1),
1 = 5.48,
2 =
3.51 and = 300/T,
T being temperature in Kelvin.
The principal and secondary relaxation frequencies are
given by
f p [GHz] = 20.09 142( 1) + 294( 1)2
R = k R
where R is the rain rate in mm/hour, and k, are regression coefcients which are the function of frequency and
polarization (11)
exp(0.015 R0.01 %)
f s [GHz] = 590 1500( 1)
Attenuation due to fog becomes signicant at frequencies
around 100 GHz with specic attenuation of 0.4 dB/km for
medium fog, and 4 dB/km for thick fog.
Foliage
The leaves of maple, oak, hickory and similar trees can
cause additional signal loss in the frequency range above
400 MHz. In the winter leaves fall and received signal is
generally stronger than in the summer time. Various factors contribute to this type of loss: operating frequency, type
of leaves, height of the trees, thickness of the foliage. In general, design practice is to add about 10 dB of allowance in
the forest area in addition to predicted path loss (7).
In tropical rain forest area leaves do not fall and their
shape is different producing different attenuation. Major
Radio System Performance
studies carried out in the 50 to 800 MHz band indicate
that loss increases linearly in log scale from 35 dB/decade
at 50 MHz up to 40 dB/decade at 800 MHz. The foliage loss
is approximately proportional to the operating frequency
raised to fourth power which is a good approximation for
horizontal polarization. Vertical polarization experienced
larger loss in general by 8 to 25 dB at 50 MHz and 1 to 2 dB
at 800 MHz. For a transmitting antenna over the tree tops
and receiving antenna located in the trees delay spread of
0.2s has been reported due to foliage (7).
Constructions
Although path loss models tend to capture difference between build up and rural areas, few specic conditions
should be emphasized. Street orientation has the channel
effect on the received signal strength. For the mobile unit
closer to the base station in the range of 1-2 miles the difference between signal strengths in the case when the street
is in line with base station and when it is perpendicular to
the base station is about 10 dB. This phenomenon diminishes for distances over 5 miles.
Mobile receiver moving into tunnel experiences a signal
loss which is dependant on the frequency and the transmitter/receiver positioning. Experiments carried out for the
transmitter at the entrance of the tunnel and the mobile
receiver moving through the tunnel show that at 300 m inside the tunnel about 4 dB loss is observed at 1 GHz. Attenuation at lower frequencies was much higher approaching
20 dB at 400 MHz (7).
Results from study in New York have established inverse law dependence of the path loss versus the distance
with a inverse law from 4 at 900 MHz reducing to 2 at 2400
MHz. Above this frequency loss is less than in free space
indicating some sort of guiding mechanism.
The underpass effect results in signal drop between 5 to
10 dB when a mobile receiver drives through. The period of
the attenuation depends on the vehicle speed. In cellular
radio this type of effect usually does not affect the voice
channel.
SMALL-SCALE VARIATIONS: MULTIPATH FADING
Small-scale fading (also referred as fast-fading, shortterm fading or simply fading) describes the uctuations
of the received radio signal over a short period of time or a
short travel distance. Substantial variations of the signal
amplitude are mainly caused by local multipath propagation. Multiple replicas of the signal combine at the receiver
antenna in different ways depending on the time of arrival
and the phase of individual components, resulting in uctuations that are fast compared to longer-term variation in
mean signal level (shadowing). These uctuations are seen
in the received signal as large changes of amplitude over a
small interval of time, random frequency modulation due
to Doppler shifts on different multipath arrivals and possible time dispersion due to different propagation delays of
multipath components. Although fading is basically spatial
phenomenon, it is experienced as temporal phenomenon
by a radio receiver moving through the multipath eld.
Motion of surrounding objects (reectors, scaterrers) also
impacts fading characteristics both in case of static transmitter/receiver pair, and or when it is comparable to the
speed of the mobile. Fast fading is observed over distance
of about half a wavelength, with fades of depth about 20 dB
being frequent and larger fades of more than 30 dB being
less frequent, but not uncommon (4).
In practice there is no clear border between slow and
fast fading. It is commonly assumed that in the areas where
multipath occurs radio signal consists of a local mean value
which is constant over a small area, but varies slowly as the
receiver moves, and a fast fading component superimposed
on the slowly varying signal.
In mobile radio communications multipath fading occurs in urban areas where the height of the mobile antenna
is much lower than that of the surrounding buildings and
there is no line-of-sight path between the transmitter and
the receiver. Propagation is mainly determined by scattering and diffraction around the buildings. In xed radio
links where the line of sight propagation exists, multipath
propagation is caused by the reection from the ground or
surrounding objects.
Except for the multipath propagation, other factors may
inuence the fading characteristics. Relative motion between the transmitter and the receiver in mobile and loworbit satellite communications results in a phase change
due to the difference in path lengths, which can be observed
as a Doppler frequency shift in each propagation path as
illustrated in Fig. 5. The phase change is given by
=
2 l
2 vt
=
cos
and the corresponding frequency shift is
fd =
1
2
v
= cos
t
where v is the speed of the mobile. If mobile is moving
towards the transmitter, positive Doppler shift results in
increase of the received signal frequency, and for the mobile moving away from the transmitter negative Doppler
shift results in decrease of the received signal frequency.
The angle between the incoming wave and the direction
of motion determines the maximum rate of phase change
which occurs when the waves are coming directly behind
or ahead the mobile.
While characteristics of a fading channel are consequences of physical propagation phenomena, their impact
on the received radio signal depends on the transmitted
signal characteristics as well, primarily on the bandwidth
of the transmitted signal. This relationship will be addressed in more details in the following section.
Statistical Characterization of Multipath Fading Channels
Two major characteristics of a multipath fading channel
can be observed by transmitting a short pulse (ideally impulse). The rst one is time spread (dispersion) of the channel resulting in more than one received pulse. They differ in
time of arrival, amplitude and phase. Second characteristics is the time-variation; the structure of multipath varies
with time, and the same sounding experiment will produce
different number of echoes with different arrival times, amplitudes and phases if repeated at a different time.
10
Radio System Performance
Figure 3. Parameters used in Walsh-Ikegami model.
Figure 4. Denition of the street orientation in Walsh-Ikegami model.
Figure 5. Geometry for illustration of Doppler effect.
Consequently a multipath fading channel is characterized in statistical terms. The impulse response of the timevarying multipath channel is represented by time-varying
linear lter h(, t) where variable t represents time of transmission and represents the channel multipath delay, for a
xed value of t. Thus h(, t) represents the response of the
channel at time t due to the impulse applied at time t-.
When received signal contains discrete multipath components, the equivalent complex baseband impulse response
model is given by
h(, t) =
n (t) exp( j2 f c n (t)) [ n (t)]
where n (t) is the attenuation factor for the n-th path, n (t)
is the propagation delay for the n-th path and f c is the
signal carrier frequency.
Statistical characterization of the fading channel is detailed in (15). In summary, channel impulse response h(, t)
can be represented as a zero-mean complex-valued Gaussian process, so that its envelope at any time instant is
Rayleigh distributed and the phase is uniform (16). In the
Radio System Performance
case when there exist a strong signal component (xed
scatterer, strong reection) h(, t) does not have a zeromean and the statistics of the channel is Rician (17). Regardless of the probability distribution, a wide sense stationary channel is characterized by two functions.
Multipath intensity prole or power delay prole of the
channel gives the average output power of the channel as
a function of the time delay . The multipath spread Tm of
the channel represents a range of values of over which
the power delay prole is essentially nonzero.
In practice power delay proles are found by averaging
instantaneous power delay prole measurements over a local area to determine an average small-scale power delay
prole. A typical spatial separation for sampling the channel is quarter of a wavelength, over receiver movements no
greater than 6 m in outdoor channels and no greater than
2 m in indoor channels (1). Several statistical parameters
quantify the delay characteristics of the channel. The mean
excess delay is the rst moment of the power delay prole
2
n n n
mean =
2
n
and rms delay spread is the square root of the second central moment of the power delay prole
=
where
Time variations of the channel can be described by the
Doppler power spectrum of the channel which gives the
signal intensity as a function of a Doppler frequency. The
range of Doppler frequency over which the Doppler power
spectrum is essentially nonzero is called Doppler spread Bd
of the channel. It measures the spectral broadening caused
by the time variations. The inverse of the Doppler spread
(t)c
2
2 mean
All delays are measured relative to the rst detectable signal arriving at the receiver at 0 = 0. Also relative amplitudes of multipath components are used in computations.
Typical values of rms delay spread are on the order of microseconds for outdoor mobile radio channels and nanoseconds for indoor radio channels (3).
The maximum excess delay is dened as X 0 , where
X is the maximum delay at which the multipath component is X dB lower than the strongest multipath arrival,
which is not necessarily at 0 . It is important to note that
all parameters highly depend on the noise threshold in the
power delay prole which is used to differentiate multipath
components and thermal noise.
The Fourier transform of the power delay prole determines frequency-coherence properties of the channel. The
inverse of the multipath spread
( f )c =
1
Tm
is a measure of the coherence bandwidth of the channel.
Effectively, two signal components separated in frequency
by more than a coherence bandwidth will by affected differently by a channel i.e. they will fade independently. Coherence bandwidth is practically quantied by a level of
frequency correlation function and the rms delay spread.
For example, for a correlation function above 0.9 the coherence bandwidth is approximately (3)
( f )c
1
.
50
1
Bd
is the coherence time of the channel, and denes the time
interval over which the characteristics of the channel can
be regarded as constant. If the signal components are separated in time more than a duration of a coherence time,
they can be regarded as independent.
However, Bd is not uniquely dened. On the other hand,
if a pure sinusoidal tone of frequency fc is transmitted over
a channel, the received signal is within [ f 0 f d , f o +
f d ], where fd is the Doppler shift, showing that spectral
broadening depends on the velocity of the mobile as well. A
commonly used relationship describing the coherence time
is given by (3)
(t)c =
2 2
2
n n n
=
2
11
0.423
fd
Scattering function of the fading channel, representing
power density as a two dimensional function of delay and
Doppler frequency, and different correlation functions of
the channel are related via different Fourier transform
pairs (15).
In a case of Rayleigh fading channel, two important statistical parameters may clarify design choices of diversity
techniques and error control codes. The level crossing rate
is dened as the expected value of the rate at which the
envelope of the Rayleigh fading process crosses a specied
level in a positive going direction (3). The number of level
crossings per second is given by
NR =
2
2 f d e
where fd is the maximum Doppler frequency and = R/Rm
is the normalized value of the specied level R by the local
rms amplitude of the fading envelope. For example in 900
MHz band with vehicle speed of 48kmph, the maximum
value is 39 crossings/sec.
The average fade duration denes the average period of
time for which the received signal is below a given level R.
For a Rayleigh fading, it is given by
2
e 1
f d 2
showing the dependency on the mobile speed. For a given
fading margin, determined in a power budget, it is important to evaluate the rate at which the received signal
falls below the certain level and the average time during
which it remains below that level. These parameters directly translate into a SNR reduction during the fade which
can be used for the prediction of the radio performance.
12
Radio System Performance
Channel Classication
The impact of the fading channel on the received signal
depends on the signal characteristics such as signal bandwidth and signal duration. The relationship between signal
characteristics and channel parameters will serve as the
criterion for channel classication. Let us consider digital
communication system with signalling interval T.
For a signalling interval T much larger than the multipath spread of the channel, T > > Tm , channel introduces
negligible amount of intersymbol interference. On the
other hand for a bandwidth of a signalling pulse W 1/T ,
previous relationship imply that the signal bandwidth is
much smaller than the coherence bandwidth of the channel, resulting in frequency-nonselective or frequency-at
fading. This implies that amplitude attenuation and phase
shift introduced by channel are the same for all signal components and consequently spectral characteristics of the
transmitted signal are preserved at the receiver. Channel
distortion is simply a multiplicative random process. In
this case multipath components of the signal cannot be resolved. This case is typical for narrowband communication
signals. Flat fading may result in deep fades.
In a case where signalling interval is smaller than the
multipath spread of the channel, consecutive transmitted
symbols interfere with each other at the receiver, producing
intersymbol interference. In this case signaling bandwidth
W is larger than the coherence bandwidth of the channel,
and different signal frequencies are subject to different
gains and phase shifts. This type of channel is referred
to as the frequency-selective channel. Frequency-selective
channels are typical for wideband communication systems
such as time-division multiple-access systems or spreadspectrum systems.
For a signalling interval T smaller than the coherence
time of the channel, T (t)c , channel characteristics can
be regarded as xed during the symbol interval. In the frequency domain, signaling bandwidth is much larger than
the Doppler spread of the channel. This condition corresponds to a slow fading channel. However, the rate of
change of the channel depends both on signalling bandwidth and the velocity of the transmitter or the receiver.
For a signaling interval larger than the coherence time,
channel impulse response changes within the duration of
the symbol. This results in spectral broadening which in
turn leads to signal distortion since the signal bandwidth
is smaller than the Doppler spread of the channel. This
type of fading is referred to as fast fading or time-selective
fading.
It should be pointed out that there is a slight inconsistency in terminology. Term fast fading is used to describe
multipath fading which is fast compared to shadowing effects (3). On the other hand, multipath fading is characterized as fast fading when signaling interval is larger than
the coherence time of the channel (15). Although it might
be somewhat confusing, precise meaning of the term can
be deducted from the context in which it is used.
For a physical channel with given multipath delay
spread and Doppler spread, different scenarios may arise
depending on the signal design. Slow fading condition is
desirable from the standpoint of receiver design since it
allows coherent reception, i.e. synhcronization circuits are
able to track changes of the signal. It is important to note
that even in situations which are characterized as slowly
fading i.e. for which Bd T 1, receiver performance may
be signicantly affected by the channel uctuations via
non-perfect synchronization. The amount of degradation
depends on the exact value of Bd T .
If signal bandwidth W cannot be selected independently
of the signalling interval T, for a given fading rapidity, system may be subject to frequency selective of frequency nonselective fading. Slow frequency-nonselective fading introduces constant multiplicative distortion and no ISI. Both
coherent and noncoherent receivers can be employed. In
the case of slow frequency-selective fading, equalization
techniques should be used to compensate for ISI and equalizer operation is possible due to slow variations of the fading. In the case when signal bandwidth can be selected independently of signalling interval (as in spread spectrum
systems due to bandwidth expansion) one can simultaneously achieve no-ISI and slow fading condition.
In the case of fast fading and frequency-nonselective
channels, noncoherent reception techniques may be employed. On the other hand fast frequency-selective fading
presents serious obstacle for a communication system.
The product Bd Tm is called the spread factor of the channel and has a xed value for a given physical channel. For
underspread channels, Bd Tm < 1, system can be designed to
achieve the condition of negligible ISI and slow fading. For
digital systems optimal data rate can be derived to satisfy
conditions for slow frequency-nonselective fading (13)
R=
1
=
T
Bd
Tm
For overspread channels signal design choices are more
restricted and noncoherent reception is usually used.
Statistical Models for Fading Channels
While path loss models play crucial role in power budget,
determining the range and coverage zone of the radio system, statistical models for small-scale fading play important role in receiver design and evaluation. Performance in
the presence of fading is one of the major factors inuencing the receiver sensitivity. Few major fading models for
different radio scenarios will be summarized in this section.
With the advances in multi-input multi-output wireless systems and space-time coding temporal characterization of the fading channels has been extended to cover
space-time propagation effects as well. Details on spacetime channel models can be found in reference (29).
Jakes Model. Jakes model of a Rayleigh fading path is
based on scattering and is one of the most widely used models for a fading channel. It effectively models frequencynonselective fading channel with a Doppler spectrum of
the form
S( f ) =
fd
K
1(
f fc
)
fd
Radio System Performance
where K is the constant which depends on polarization
(for detailed model derivation see (12)), fc is the signal frequency and f d = v/ is the maximum Doppler shift, v being the speed of vehicle, and signal wavelength. This type
of Doppler spectrum can be simulated using a model with
N0 = 8 sinusoids where in-phase and quadrature components of the fading signal are given by
xc (t) = 2n = 1 cos(n ) cos(n t) + 2cos cos(d t)
xs (t) = 2n = 1 sin(n ) cos(n t) + 2sin cos(d t)
and Doppler shift is given by d = 2 f d , frequencies of the
sinusoids are n = d cos(2n/N), N0 = (N/2 1)/2, and
phases n = n/N0 and = /4. This model shows excellent agreement with Rayleigh distribution, autocorrelation
Bessel function and Doppler spectrum (12). This technique
can be extended to generate up to N0 independent fading
signals, which is of importance for tapped-delay line fading
channel models for frequency-selective fading.
Tapped-Delay Line Model for Outdoor Mobile Systems.
Tapped-delay line models are commonly used to specify mobile radio channels. The structure of the model is straightforward, and measurements for a particular frequency
band and geographical environment are used to develop
model parameters. Model is specied in terms of number
of multipath arrivals (usually xed), relative delays, relative power of multipath components and Doppler spectrum
for each component. For a given radio standard these parameters may widely vary not only due to different measurements but also due to different signal characteristics
considered.
As an example of tapped-delay line model wideband
propagation model for GSM digital cellular standard is
summarized. Discrete number of taps (6 and 12) is determined by their time delay and their average power, amplitudes of each tap are Rayleigh distributed and varying
according to a Doppler spectrum S( f ) (19). Two types of
Doppler spectrum are dened: classical to be used in all
but one case
S( f ) =
f 2
(1
)
fd
f or
[ f d , f d ]
where f d = v/ is the maximum Doppler shift, and Rice
which is the sum of a classical Doppler spectrum and one
direct path
0.41
S( f ) =
2 f d
f or
(1
f 2
)
fd
+ 0.91
( f 0.7 f d )
[ f d , f d ]
Depending on an environment few typical cases are specied: rural area, hilly terrain and typical urban environment. Example of six tap models are summarized in the
Table 4.
With introduction of wider band systems such as Wideband Code Division Multiple Access the model of the propagation channel changes due to different structure of resolvable paths. Moreover, in addition to models where tap delay
is xed as summarized in Table 5, new models with variable
13
tap delay has been introduced (32). The dynamic propagation condition consists of two paths of equal strengths and
equal phases, rst one is static while the time difference
between the two paths is according to
= B +
A
(1 + sin( t))
2
With parameters B = 1 s, A = 5 s and = 40
103 s1
Other system related test require further sophisticated
models such as birth-death model (20) where two paths appear and disappear randomly according to the given statistical model.
Rummler Fading Model for Fixed Microwave Systems.
Three path model developed for a line-of-sight microwave
radio channels is based on a channel measurements on a
typical link in 6 GHz band. Since the differential delay of
two multipath components is small, channel transfer function is given by
H( f ) =
[1
exp( j2( f f min )0 )]
where is the overall attenuation, is a shape parameter
resulting from multipath propagation, fmin is the frequency
of the fade minimum and 0 is the relative time delay between the direct and multipath components. By tting the
model to the measurement data parameter is modeled
as log-normal distributed, while distribution for is given
as (1 )2.3 . Two random variables are statistically independent. For > 0.5 the mean of 20 log is 25 dB, with
standard deviation of 5 dB. For smaller values of the
mean decreases to 15 dB. Delay parameter was estimated
to be 0 = 6.3 ns.
Saleh-Valenzuela Indoor Statistical Model. SalehValenzuela model is based on statistical modeling of the
measurements of indoor propagation with multipath
resolution of 5 ns. Maximum reported multipath delay
spread was on the order of 100200 ns within the rooms
in building and 300 ns in hallways. This model is based
on multipath components arriving in clusters. Amplitudes
of the components are independent Rayleigh distributed,
with variance that decay exponentially with the cluster
delay as well as with excess delay within a cluster. The
difference from tapped delay line model is that clusters
and multipath components within cluster are modeled
as Poisson arrival processes with different rates and
exponentially distributed interarrival time (3).
Spatial Channel Models. Spatial channel models have
been developed over last decade to take into consideration increasingly important use of multiple antennas in radio systems, both at the transmitter and the receiver. The
notion of space-time communications channels deserves a
whole chapter both in terms of importance and complexity
of the material. In addition to small-scale fading spatial
channels take into consideration other effects such as angle spread due to space selective fading, scatterers both in
local zone and remote, polarization, antenna array topologies etc (29). The most complex models are for Multiple
14
Radio System Performance
Input Multiple Output (MIMO) channels example of one
used in system evaluation has been presented in (33).
Effects of Fading on System Performance
Analog radio systems are relatively narrowband, even for
high-capacity microwave links, and major factor that is
contributing to the degradation of system performance is
the fading depth (8). Therefore, at-fading margin should
be provided to insure robust performance. It denes the additional loss that can be tolerated before the system SNR
reaches an unacceptable level.
For a digital system BER is the performance measure
and it should be specied for particular fading condition.
Important note is that performance degradation of the digital signal is not graceful as in the analog one. For a narrowband system performance may be specied in terms of atfading margin. For wideband systems, effects of frequencyselectivity are more emphasized and use of equalization
techniques and/or other forms of diversity are necessary
to insure desired performance. In that case, fading margin
may not be specied explicitly, but the receiver sensitivity
may be specied for fading conditions, and will include the
effects of fading countermeasures.
MAN-MADE NOISE AND INTERFERENCE
Radio systems are either noise or interference limited.
Characteristics of the noise greatly inuence radio receiver
design, however most of the radio systems nowadays are interference limited due to coexistence with other radio systems and/or multiple-access schemes employed. Interference level in the radio system is dependant on frequency,
time, spatial location and signal separation. Some of the
major interference mechanisms for radio systems will be
described in this section.
Impulse Noise
Nature and characteristics of the noise are important for
the radio reception in two ways. Understanding the noise
phenomena leads to more efcient system design. Characterizing noise is crucial in performance prediction of any
radio system since it directly affects the receiver sensitivity. Radio systems are subject to different type of noise
sources. Effects of thermal noise and receiver noise are
well understood and are described by noise density and
noise gure. This type of noise is characterized as Gaussian. Atmospheric noise has non-at spectral density and
decreases rapidly with frequency. Galactic noise is due to
energy radiation from stars and planets, however it is usually below the level of thermal noise.
The type of noise referred to as man-made noise is impulsive in nature and is generated by radiation of the electric equipment of various kinds such as vehicle ignition systems, alternator, power lines, neon lights, industrial noise
form current switches and from the various domestic appliances. This type of noise can signicantly affect the performance of radio system. Its spectral characteristics are
very irregular, and also it varies in level with location and
time. Impulsive noise may be viewed as a combination
of successive impulses, with random amplitudes and random time spacing. Different parameters have been established to qualitatively describe this type of noise including
mean or average voltage, peak voltage, amplitude probability distribution, pulse height distribution, level crossing
rate, pulse duration distribution, etc. (2).
The median value of the man-made noise power is given
as a function of frequency up to 200 MHz by (10)
Fam = c d log( f [MHz])
where c and d take values form Table 6. In the frequency
range 200 MHz to 900 MHz the relationship is given by
Fam = 44.3 12.3 log( f [MHz])
As mentioned earlier, ignition impulses from motor vehicles produce the signicant component of man-made noise,
in particular in VHF band. Noise amplitude distribution
in this case can be determined by
A[dB(V/MHz)] = 106 + 10 logV [vehicles/km2 ] 28 log( f [MHz])
where V is the trafc density.
Predicting the performance of the receiver in the presence of impulse noise is extremely difcult compared to
relatively straightforward method for Gaussian noise. Different techniques are presented in (2).
Cochannel Interference
Cochannel interference is the consequence of better spectrum utilization. It is generated in a scenario where two or
more communication signals are assigned to the same frequency and operate at the same time. CCI may be caused
by a number of scenarios depending on the radio service:
bad route planning for microwave radio links, failure to
carry out co-ordination procedures for the same type of
service, cross-polarization degradation in cross-polar frequency plan, cochannel interference due to frequency reuse
in mobile radio systems, and cochannel interference due to
multiple-access method like in CDMA.
In any case result of the cochannel interference is the
presence of the same-type modulated signal at the receiver
input, which may be lower (in cellular TDMA and microwave systems) or on the same level (in satellite and
cellular CDMA systems) as the desired signal. In most of
the cases cochannel interference problem can be avoided
by careful system planning. This will reduce the cochannel interference to the acceptable level such that system
performance is not signicantly degraded. Technical measures include proper cell design for cellular systems including cell sectorization, power control techniques, and use of
orthogonal codes on the forward link in CDMA cellular systems (note that orthogonal codes in CDMA system can be
used only to eliminate intracell interference, the interference from neighbouring cells will still be present). In the
case when planning methods fail to reduce the amount of
interference, solution may be found in more sophisticated
receiver design using interference suppression techniques.
Radio System Performance
Adjacent Channel Interference
Adjacent channel interference is the results of closely
spaced radio channel in frequency, so the spectral content
of the adjacent channels is spilling into the bandwidth
of the desired channel. Factors determining the level of
the interference are signal power and spectral distribution, power and spectral distribution of the interfering signal intercepted by the receiver and distance dependence
of the transmission losses between interfering transmitter and the desired signal receiver. ACI can be somewhat
controlled by tight ltering, however limiting factors are
spectral utilization of the system and, on the other hand,
nite slope that can be achieved for ltering functions. Another way to improve adjacent channel performance is the
design of spectrally efcient modulation techniques that
have largely concentrated power in a limited bandwidth
with good roll-off properties, which in turn requires more
elaborated demodulation schemes. Also adjacent channel
interference in cellular systems can be avoided by careful frequency planning, so adjacent channels are not used
within the same cell or in the neighbouring cells.
Near-Far Interference
Near-far interference in a radio system occurs when two
different signals arrive at the receiver input with largely
different power levels and may be interpreted as a special form of cochannel interference. This may be result of
difference in distance of different transmitters from the
same receiver, or in some cases in may be the result of
fading conditions on the link, so that closer transmitter
may have lower power at the receiver when subjected to
a deep fade. The consequence of near-far problem is the
performance degradation for the weaker signal reception.
Although near-far problem is general for the radio reception, it is emphasized in direct-sequence CDMA systems
where all signals are present at the receiver input, and
may signicantly reduce the capacity of CDMA system.
Countermeasures for near-far problem include tight power
control which should insure that all signals arrive to the receiver with the power level required for desired operation.
Another possibility for cellular systems is to develop a frequency management plan that can greatly reduce the possibility of near-far problem. Specic techniques for radio
design include multi-user detection techniques described
in (22).
RADIO RECEIVER DESIGN
Radio receiver consists of a channel interface followed by
the demodulation part (23). Channel interface of the radio
receiver is characterized by the receiver selectivity and its
ability to process both desired signal and undesired (interfering) signal. In order to achieve specied performance
radio receiver must be designed to operate in the presence
of noise, fading and interference.
Basic Radio Receiver Parameters
Radio receiver sensitivity determines the input level of the
weak radio signal that can be processed by a radio receiver
15
to meet performance requirements. For analog receivers
performance measure is given in terms of required SNR,
usually specied at the input to the detection circuit. For
digital systems performance measure is given in terms of
BER or other parameters that can be derived from BER
requirement, depending on the type of radio service. Detection algorithm includes coding, interleaving, equalization, diversity and other signal processing functions used
to improve radio reception.
Receiver sensitivity is usually expressed by minimum
detectable signal level for a required SNR, which includes
the knowledge of the system bandwidth and is dened as
RS[dBm] = 174dBm + 10 log(NF [dB]) + 10 log(BW[Hz]) + SNR[dB]
where NF is the overall system noise gure referred to the
input of the receiver and BW is the system noise bandwidth. For digital systems system requirement is commonly given in terms of energy per bit to noise density
ratio Eb/No. Relation between Eb/No and SNR is given by
Eb/No = SNR
BW
where T is the bit duration (23). Depending on the particular radio system and targeted operating environment, performance measure of the system (SNR or BER) can be evaluated under fading conditions or under static conditions.
In the rst case fading margin is derived from the statistics of the large-scale signal variation such as log-normal
fading. Small-scale signal variations are accounted for implicitly during BER evaluation. When BER of the system is
specied under static conditions, fading margin is derived
from combined statistics of both large-scale and small-scale
variation. Fading margin effectively denes the increase
in SNR required to preserve the same BER as in the static
case.
Radio sensitivity species the lowest detectable signal
level at the radio input; however, both desired and unwanted signals may be present at the radio receiver input at high levels, and it is important to specify the performance for undesired response rejection. The unwanted
strong signal may reduce a nearby weak (desired) signal
resulting in radio desensitization. Desensitization level is
specied for a 1 dB weak signal reduction and is a consequence of radio front-end saturation.
Another factor that may severely limit the performance
of a radio receiver is intermodulation distortion, where
unwanted strong signals produce a number of intermodulation products due to receiver front-end nonlinearities.
IMD products, particularly third-order products, may fall
directly in the bandwidth of the desired signal. To limit the
level of the IMD products receiver is required to have high
intercept point of the particular order.
Dynamic range (DR) of the receiver denes the range of
input signals that a radio receiver can process. Different
considerations may be taken into account depending on
the radio service; however, the most common spurious-free
dynamic range (SFDR) is dened as
SFDR[dB] =
2
(IIP3[dBm] RS[dBm])
3
where IIP3 is the third-order input intercept point.
16
Radio System Performance
Modulation and Coding
Choice of the modulation method for radio communication
depends on many factors. In all cases the goal of the radio receiver design is to achieve the required quality of the
output signal (voice, video, data) using minimum SNR or
Eb/No. For power-limited systems, where bandwidth is not
of concern (e.g. space communication links) noncoherent
M-ary frequency shift keying (FSK) can be used to reduce
the required Eb/No for a given BER, at the expense of bandwidth expansion. For a bandwidth-limited systems (like
most of the modern radio systems) spectrally efcient, coherent modulation schemes like phase-shift keying (PSK)
or quadrature amplitude modulation (QAM) can be used at
the expense of increasing required Eb/No as the spectral
efciency increases (5). However, additional requirements
for the modulation format in radio communications such
as robustness to fading, interference, and nonlinear distortions, motivated development of constant-envelope or
near constant-envelope modulation such as offset quadrature phase shift keying (OQPSK) and minimum shift keying (MSK). Different forms of such modulation formats can
be found in wireless systems (3). Inherent resistance of a
modulation method to the interference can be measured
by determining required SIR for a specied BER. Performance functions for digital modulations in the presence of
interference are presented for various types and statistical
characteristics of interferers in (23). Further advances in
signalling methods for radio channels include the application of coded modulation techniques (15).
Additional performance gain can be obtained by using
coding techniques at the expense of introducing redundancy to the transmitted digital signal. Channel coding
protects the signal from the effects of noise, fading and/or
interference. Functions of channel codes are error detection and forward error correction. Both block and convolutional codes have been applied in radio communications
(15). Lately increasing consideration has been given to
space-time codes (29) and turbo codes (30).
The complexity of modern radio communications systems requires further means of efciently utilizing channel based on the quality of the link. Different modulationcoding combinations have been developed as well as methods to switch between them as for example in EDGE cellular standard (19). This is one example of more general
adaptive modulation and coding techniques that in principle requires a feedback path between the transmitter and
receiver, with data transmission rates being varied relative
to propagation channel conditions (34).
Diversity Reception for Fading Channels
Diversity reception techniques are used to reduce the effect of fading by exploiting the random nature of the radio
channel. Independent or highly uncorrelated signals paths,
known as diversity channels, are used to improve quality
of the received signal. The concept of diversity is relatively
simple: if one of diversity channels is subject to a deep
fade another channel with higher signal level can be selected or appropriate combining techniques can be applied
to increase the signal level. The probability of all diversity channels being simultaneously below a given level is
much less than the probability of any of them being below
that level. Therefore, a combination of different diversity
channels will be subject to less severe fading conditions
compared to any individual channel.
Diversity techniques may address different types of signal uctuations. Large-scale fading is caused by shadowing
due to variations in the propagation prole (hills, mountains) and the surrounding objects (buildings). To reduce
the effect of the long-term fading, two geographically separated antennas may be used (for transmission or for reception) and the stronger signal may be selected. This type
of diversity, known as macroscopic diversity, is often used
in short-wave radio systems to reduce the effects of ionosphere and in cellular systems to improve the performance
of the forward link (mobile unit is selecting to receive signal
from base station that is not in the shadow) or of the reverse
link (signals from two base stations may be combined). The
selective combining techniques is the most usual approach
for macroscopic diversity (7).
Small-scale fading caused by multipath propagation in
the vicinity of the receiver results in deep and rapid signal uctuations. They occur over a very short distances
of few wavelengths, and microscopic diversity techniques
are employed to reduce the deep fades of the received signal. Different methods may be used to derive the diversity
branches (channels).
Space diversity methods use physically separated antennas that can provide signals with low correlations. Typically, a separation of few wavelengths is sufcient to obtain independently fading channels. Antenna separation
is also dependant on antenna height, and can be reduced
as the signal frequency increases. This type of diversity is
widely used in microwave radio links and in cellular radio,
particularly for base station reception. For base stations in
mobile radio, recommended ratio between antenna height
and antenna separation is about 11, suggesting about 9
feet of antenna separation for a 100 feet antenna in 900
MHz band (7). Antenna diversity at the mobile unit can be
achieved with the antenna separation as low as 17.5% the
signal wavelength. Practical antenna separation for microwave links is on the order 150200 signal wavelengths.
Receiver conguration for space diversity is relatively simple. Number of branches is selectable and no extra bandwidth or power is needed to achieve diversity, making the
space diversity one of the most popular schemes (4).
Frequency diversity relies on the diversity channels being separated in frequency by more than the coherence
bandwidth of the channel, so the signal replicas will have
very little correlation. Typical values used for mobile radio
in 900 MHz band are larger than 50 KHz in urban areas,
and larger than 300 KHz for suburban areas (7). This technique is often applied in microwave links, where diversity
frequencies are available as the backup frequencies in the
case of deep fading. Beside the requirements for additional
bandwidth, disadvantage of frequency diversity is the need
for multiple receivers at different frequencies.
Polarization diversity is based on the fact that horizontally and vertically polarization signal components fade
independently and may be combined (selected) at the receiver. This technique is primarily used for microwave
links, where channel conditions vary slowly in time, al-
Radio System Performance
though recently it has been increasingly considered for
base station reception in mobile radio. Two differently polarized antennas may be at the same place, hence there is
no separation requirement as in space diversity. The drawback of the technique is the 3 dB power reduction at the
transmitter since the power must be split between differently polarized antennas. Also only two diversity branches
are available in polarization diversity (7).
Angle diversity may be used at high operating frequencies (over 10 GHz) by pointing directional antennas in different directions at the receiver site. It is predominantly
used by mobile units in cellular radio, since most of the
multipath is created by local scatterers. Directive antennas
also help in reducing the Doppler spread for each diversity
branch.
Time diversity relies on transmitting the same signal
at different times, so these signals fade independently. instants so the fading characteristics will be changed and
signals will be uncorrelated. The two transmissions should
be separated by more than the coherence time of the channel. This diversity method is effective in applications where
fading mechanism is not related to the movement of the
receiver. However, this method fails in the scenario where
mobile unit may stay still at the given location which is
the subject of the deep fade, and the repeated signal components are highly correlated. While this method requires
more spectrum and large buffer memory to store different
signal replicas, the advantage is in simple implementation.
Interleaving is the techniques used to obtain time diversity effect in modern digital communication systems. The
function of the interleaver is to spread transmitted bits in
time so that a block of bits is not subject to a deep fade at
the same time. This helps to randomize burst errors and
aid channel coding techniques in reducing the overall BER
(3).
Path diversity technique relies on receiver capability to
resolve different multipath components in time domain,
and subsequently combine them. This type of diversity is
also referred to as the implicit diversity as opposed to other
techniques which explicitly dene diversity channels, since
multipath diversity channels are obtained after signal reception. Adaptive equalizers and RAKE receiver (15) are
receiver techniques that provide path diversity, and are
very effective in frequency-selective channels. Diversity
gain largely depends on the delay prole of the channel.
Diversity techniques are usually applied at the receiver
side, although some of the implementation complexity can
be moved to a transmitter. For example, base station in the
mobile radio scenario may use the antenna that can provide better path propagation to a mobile at the specic location. Transmitter diversity allows the receiver to obtain
diversity gain while operating as a standard nondiversity
receiver (4, 31).
Once the diversity channels are created it remains to
be decided how are they going to be combined in order
to utilize diversity effect. While macrodiversity schemes
use almost exclusively selection combining, microdiversity
schemes may employ one of the combining techniques outlined below. The methods of linear combining are derived
based on different optimization criteria, different level of
information available at the receiver and different com-
17
plexity constraints.
Selective combining simply selects the diversity branch
with the highest signal level. This type of combining is easy
to implement, additional requirements are antenna switch
and monitoring circuitry. While in selective combining only
one branch is used for subsequent signal processing and
others are discarded, maximal ratio combining performs
coherent combining of all diversity channels. This is optimal method in a sense that provides the maximal signal
to noise ratio for signal detection. The weights applied to
different diversity branches are proportional to the signal
to noise ratios in each of the branches, and can be obtained
either by utilizing pilot signals or sophisticated channel estimation techniques. When variable weighting is not available at the receiver, equal-gain combining may be utilized
by coherently summing the signals from different diversity branches without weighting. This method still allows
the receiver to exploit all diversity channels. Performance
of the equal-gain combining is better than the selection
combining and worse than maximal ratio combining, thus
representing viable technique as a trade-off between performance and complexity. Performance improvement obtained by different combining schemes is summarized in
Table 7.
Diversity reception may be interpreted as the simple
case of coding i.e. repetitive coding since the information
is simply repeated on each diversity channel. Better spectrum utilization can be achieved by employing more sophisticated coding techniques for fading channels, in particular
concatenated coding (15).
All of the diversity reception techniques discussed so
far rely on the processing on the reception. Alternative approach might be to move complexity of signal processing
to the transmitter side, typically base station in the cellular radio system, and preserve simple design of the handset unit while still obtaining diversity gain. These techniques have been commonly termed as transmit diversity
and have been utilized to improve performance in WCDMA
systems (31).
Interference Suppression Techniques
Every radio system operates in the presence of some form
of interference. Regardless of the origin of interference (intentional or unintentional) some form of interference suppression must be employed in order to preserve system performance. Due to the increased number of services, most
of the radio systems in operation are interference-limited
rather than noise-limited. For power budget calculation,
in addition to radio sensitivity which is a function of the
noise bandwidth of the receiver and the required SNR, interference margin should be specied. Alternatively signal
to interference ratio should be specied to satisfy required
BER. Depending on the type of interference (CCI, ACI) different requirements may be imposed on the receiver.
In order to improve the efciency of a radio system different approaches may be pursued: reducing the channel
spacing (increasing ACI), narrowband and wideband system overlay (creating narrowband cochannel interference
to the wideband system), smaller cell design in mobile radio
(increasing CCI), overlapping footprints of the satellites,
18
Radio System Performance
etc.
Based on the signal observation methods, interference
suppression techniques could be broadly divided into multichannel techniques where multiple sensors (antenna array) are employed for signal reception and single-channel
techniques where only one antenna is employed (25). In
a multichannel scenario, spatial ltering techniques may
realized via adaptive antenna arrays, relying on the different spatial distribution of the interfering signals. Singlechannel techniques rely purely on temporal processing,
usually using adaptive ltering. The most promising path
for the modern receiver design is to incorporate spatiotemporal processing, efciently combining capabilities of
antenna arrays and temporal signal processing (26).
Spatial processing for interference signal suppression
can be carried out using different techniques, including directional antennas, tilted beams and height adjustment.
However, adaptive antenna arrays with capability of adaptive beamforming have shown great potential for congested
radio environments. Different algorithms and optimization
criteria for adaptive beamforming are presented in (27).
Single-channel interference suppression techniques can
be applied both in cases of spread spectrum signalling
and conventional techniques. Although all forms of spread
spectrum signals (direct sequence, frequency hopping, time
hopping) provide certain interference margin due to their
inherent processing gain, additional techniques can be applied to further improve performance. Depending on the
nature of interfering signal, different interference suppression algorithms have been developed. For narrowband
interference rejection in direct sequence systems solutions include adaptive notch lters, decision feedback techniques, adaptive analog-to-digital conversion and other
nonlinear techniques. In the case of CDMA signals, where
interference has the same characteristics as the desired
signal, multi-user detection techniques are developed either in decentralized form (detection of single user of interest) or in centralized form (joint detection of all active
users) (22). Special techniques have been analyzed for frequency hoping systems as well (25).
In nonspread spectrum systems different techniques
have been developed to cope with CCI, ACI and ISI. Adaptive equalization techniques were shown to be efcient
in combating all three types of interference. Other methods include self-optimizing or blind receivers (based on
different techniques like constant modulus, higher-order
statistics, probabilistic approaches etc.), neural network receivers and other nonlinear techniques (25).
Most of the abovementioned interference suppression
algorithms are performed in baseband. In addition to this
different practical interference suppression schemes are
developed at RF/IF stages of the receivers (28).
Signal Losses in Radio Receiver
Due to the inherent complexity of radio receiver number
of signal loss sources are associated with the receiver design. Bandlimiting loss in transmitter and receiver is the
result of nite bandwidth usage, usually restricted by the
radio regulations. Intersymbol interference in which received digital symbols overlap with each other may re-
sult from different effects: bandlimited operation, physical
multipath radio propagation and usage of partial response
modulation formats. Local oscillator phase noise results in
the phase jitter in the receiver and consequently degrades
the performance of the detection algorithm. Other impairments of radio front-end also result in loss of receiver performance for example DC offset, gain and phase imbalance of in-phase and quadrature branches. Nonlinear distortions in radio equipment include AM/AM and AM/PM
conversion, limiter loss and intermodulation distortion.
AM/PM conversion produces phase variations in the signal, mostly after nonlinear amplication. Signal sidelobes
growth is another effects that can produce excessive interference despite tight ltering at the transmitter. Hard
limiting stages in receiver may cause suppression of weak
signal components in the presence of stronger ones. Intermodulation products are results on multiple signals interaction in a nonlinear device and creates additional noise
source which contributes to total noise level. Also receiver
operation can be largely affected by several imperfections
associated with the demodulation part such as imperfections of the synchronization circuits that are producing
noisy estimates of received signal parameters necessary for
the detection process and nite numeric precision effects
associated with programmable of xed logic implementation of the demodulation functions.
BIBLIOGRAPHY
1. Rec. ITU-R SM.1047, National spectrum management, 1994.
2. D. Parsons, The Mobile Radio Propagation Channel, New York:
John Wiley & Sons, 1992.
3. T. Rappaport, Wireless Communications Principles and Practices, Upper Saddle River, NJ: Prentice Hall PTR, 1996.
4. S. Sampei, Applications of Digital Wireless Technologies to
Global Wireless Communications, Upper Saddle River, NJ:
Prentice Hall PTR, 1997.
5. B. Sklar, Digital Communications, Englewood Cliffs, NJ: Prentice Hall, 1988.
6. J. Wait, Radiowave propagation, in Wiley Encyclopedia of Electrical and Electronics Engineering, New York: John Wiley and
Sons.
7. W. Lee, Mobile Communications Design Fundamentals, 2nd
ed., New York: John Wiley & Sons, 1993.
8. J. Doble, Introduction to Radio Propagation for Fixed and Mobile Communications, Boston, MA: Artech House Publishers,
1996.
9. COST 231, Urban transmission loss models for mobile radio
in the 900- and 1800- MHz band, COST 231 TD(90) 119, Rev.2,
The Hague, September 1991.
10. Rec. ITU-R, PI.372-6, Radio noise, (1994).
11. Rec.ITU-R, PN.833-1, Specic attenuation model for rain for
use in predicition methods, (1992).
12. Rec. ITU-R PN.840-1,Attenuation due to fog and clouds, (19921994)
13. S. Stein, Fading channel issues in system engineering, IEEE
J.Select. Areas Commun. 5 (2): 6889, 1987.
14. R. Steele, Mobile Radio Communications, London: Pentech
Press, 1992.
Radio System Performance
15. J. Proakis, Digital Communications, 3rd ed., New York: McGraw Hill, 1995.
16. C. Lewis, Rayleigh channels, in Wiley Encyclopedia of Electrical and Electronics Engineering, New York: John Wiley and
Sons.
17. H. Bertoni, Rician channels, in Wiley Encyclopedia of Electrical and Electronics Engineering, New York: John Wiley and
Sons.
18. W. Jakes (ed.), Microwave Mobile Communications, An IEEE
Press Classic Reissue, Piscataway, NJ: IEEE Press, 1993.
19. 3GPP TS 05.05, Technical Specication Group GSM/EDGE
Radio Access Network, Radio transmission and reception (Release 1999), www.3gpp.org
20. V. Garg and J. Wilkes, Wireless and Personal Communications
Systems, Upper Saddle River, NJ: Prentice Hall PTR, 1996.
21. S. Faruque, Cellular Mobile Systems Engineering, Boston, MA:
Artech House Publishers, 1996.
22. A. Duel-Hallen, J. Holtzman, and Z. Zvonar, Multiuser detection in CDMA systems, IEEE Personal Communications Magazine 2 (2): 4658, 1995.
23. Rec. ITU-R SM.1235, Performance functions for digital modulation systems in an interference environment, (1997).
24. Rec. ITU-R SM.1132, General principles and methods for sharing between radio services, (1995).
25. J. Laster and J. Reed, Interference rejection in digital wireless
communications, IEEE Signal Processing Magazine, 14 (3):
3762, 1997.
26. A. Paulraj and C. Papadias,Space-time processing for wireless
communications, to appear in IEEE Signal Processing Magazine, 1997.
27. R. Monzingo and T. Miller, Introduction to adaptive arrays,
New York: John Wiley and Sons, 1980.
28. Rec. 734, The application of interference cancellers in the xed
satellite service, (1992).
29. A. Paulraj, R. Nabar, and D. Gore, Introduction to space-time
wireless communications, Cambridge, UK: Cambridge University Press, 2003.
30. L. Hanzo and B. Yeap, Turbo coding, turbo equalization and
space-time coding for transmission over fading channels, New
York: John Wiley & Sons, 2002.
31. A. Hottinen, O. Tirkkosen, and R. Wichman, Multiantenna
transceiver techniques for 3G and beyond, New York: John Wiley & Sons, 2003.
32. 3GPP TS 25.101, Technical Specication Group Radio Access
Network, User Equipment (UE) radio transmission and reception (FDD) (Release 1999), www.3gpp.org
33. 3 GPP TR 25.996, Technical Specication Group Radio Access
Network, Spatial channel model for Multiple Input Multiple
Output (MIMO) simulations (Release 6), www.3gpp.org
34. A. Goldsmith, Wireless Communications, Cambridge: Cambridge University Press, 2005.
ZORAN ZVONAR
Analog Devices, 804 Woburn
Street, Wilmington, MA,
01887
19
RADIOWAVE PROPAGATION GROUND EFFECTS
187
RADIOWAVE PROPAGATION GROUND EFFECTS
Radiowave propagation plays an important role in modern
communication, radar, and navigation systems. The mathematical theory of radiowave propagation is based on Maxwells equations (1), which were formulated by James Clerk
Maxwell in the 1860s (2). The first free-space radiowave
transmission experiments between a pair of antennas were
performed by Heinrich Hertz in the 1880s (3). In 1897 Marconi first patented a wireless telegraphy system based on
long-distance, radiowave propagation.
RADIOFREQUENCY SPECTRUM
Table 1 summarizes the frequency range, propagation characteristics, and applications of the letter-designated bands of
the radiofrequency spectrum. The lower and upper bounds of
the radiofrequency spectrum in Table 1 are somewhat arbitrary, but the indicated frequency range, 3 Hz to 300 GHz,
encompasses the portion of the electromagnetic spectrum for
which conventional antennas are used to transmit and receive
radio waves. Some of the characteristics of the individual
bands are as follows.
Extremely Low Frequency
Because free-space, extremely low frequency (ELF) wavelengths are extremely long (greater than 100 km), antennas
are very inefficient radiators because they are electrically
small. The other major disadvantage of ELF is lack of bandwidth for information transmission. Despite these disadvantages, ELF is useful for worldwide communication with submarines because the long wavelengths have a useful
penetration depth (also called skin depth) of several tens of
meters in sea water. In addition, the earth and the ionosphere
support a low-attenuation waveguide mode at ELF (4) so that
a wave, once launched, will propagate around the world with
little loss of intensity. The ionosphere is the part of the upper
atmosphere where sufficient ionization exists to affect radiowave propagation, and these effects are covered thoroughly in Ref. 5. The earth-ionosphere waveguide can actually support cavity modes with resonances (6) in the ELF
range, and these resonances (called Schumann resonances)
Contribution of the National Institute of Standards and Technology,
not subject to copyright in the United States.
J. Webster (ed.), Wiley Encyclopedia of Electrical and Electronics Engineering. Copyright # 1999 John Wiley & Sons, Inc.
188
RADIOWAVE PROPAGATION GROUND EFFECTS
Table 1. The Radiofrequency Spectrum
Frequency Range
Band
Characteristics
Applications
3 Hz3 kHz
ELF
Submarine communications; underground mine communications; geophysics
330 kHz
VLF
30300 kHz
300 kHz3 MHz
330 MHz
LF
MF
HF
30300 MHz
VHF
300 MHz3 GHz
UHF
Long wavelength; inefficient antennas; earth-ionosphere waveguide; penetration of ground and sea
water
Large transmitting antennas; earth-ionosphere
waveguide
Earth-ionosphere waveguide; high atmospheric noise
Ground wave; ionospheric reflections (at night)
Long-distance ionospheric propagation; maximum usable frequency
Line-of-sight propagation; ionosphere scatter; meteor
scatter
Line-of-sight propagation
330 GHz
30300 GHz
SHF
EHF
Line-of-sight propagation; atmospheric absorption
Line-of-sight propagation; severe atmospheric absorption
further enhance the field strength worldwide. Unfortunately,
these resonances also enhance the competing atmospheric
noise caused by thunderstorms.
ELF waves have an even larger skin depth (typically
greater than 100 m) in ground or rock (which have lower electrical conductivities than sea water) and have been found useful for communication to and within underground mines (7).
Because of this ability to penetrate rock, ELF waves have also
been used in direction-finding applications for location of miners trapped in underground mines (8).
Very Low Frequency
The free-space wavelengths at very low frequency (VLF) are
still very long (10 to 100 km), and this generally dictates the
use of large vertical transmitting antennas with large ground
systems to reduce ground current losses. Smaller antennas,
such as vertical whips, loops, and grounded horizontal wires,
can be used for receiving. VLF antennas and propagation are
thoroughly covered in Ref. 9.
Even though the bandwidth is limited, VLF systems are
useful for long-range reliable communications, long-range dissemination of standards for frequency and time, long-range
navigation (including the U.S. Navy Omega system), and geophysical probing of the ground and the ionosphere. For short
ranges, VLF propagates predominantly by the ground wave.
For long ranges, the reflections from the ionosphere become
important, and multiple reflections between the earth and the
ionosphere can be described by earth-ionosphere waveguide
modes (9).
Low Frequency
The low frequency (LF) band is also characterized by low attenuation of ground wave propagation and earth-ionosphere
waveguide propagation. Thus the LF band is useful for longrange communication and for marine and aeronautical radio
navigation beacons. The LORAN-C navigation system is
based on the arrival time of a ground-wave pulse with a carrier frequency of approximately 100 kHz so that later-arriving ionospheric reflections do not affect the system. The wavelengths are long enough (110 km) that large transmitting
antennas are needed.
Long-range communication; navigation, and time-frequency dissemination; geophysics
Navigational beacons
AM broadcasting; maritime communications
Maritime and aeronautical communications; citizens
band and amateur radio
Television and FM broadcasting; air traffic control;
navigation
Television broadcast; radar; satellite communication;
mobile communication; global positioning system
Radar; satellite communication; microwave links
Radar; secure communication; satellite links
Medium Frequency
The most popular use of the medium frequency (MF) band is
AM broadcasting, which uses the frequency range from 535
to 1705 kHz. The propagation mode is generally vertically polarized ground wave, although ionospheric reflections (sky
wave) can extend the range at night. Maritime communications is another application.
High Frequency
In the high frequency (HF) band, ionospheric reflections provide the possibility of long-distance communications. The
maximum usable frequency (MUF) for ionospheric reflection
generally occurs in the upper part of the HF band, but the
MUF is a complicated function of the incidence angle and the
state of the ionosphere (5). HF applications include maritime,
aeronautical, amateur, and citizens-band communications.
Very High Frequency
At frequencies above 30 MHz, propagation is predominantly
by line of sight with some refraction (bending) caused by the
atmosphere. However, in some cases ionospheric reflections
can occur at frequencies up to 60 MHz. Scattering from ionized meteor trails (which can be up to 25 km in length) can
be used for meteor burst communications (10). Very high frequency (VHF) applications include television and FM radio
broadcasting and air traffic control and navigation.
Ultra High Frequency
Ultra high frequency (UHF) propagation is essentially by line
of sight with some atmospheric refraction and some scattering by ionospheric irregularities. Applications include UHF
television broadcasting, various radars, satellite communications, and personal communications (11). The global positioning system (GPS) uses transmitting satellites in the UHF
band.
Super High Frequency
Super high frequency (SHF) propagation is primarily line of
sight, and atmospheric absorption becomes significant, partic-
RADIOWAVE PROPAGATION GROUND EFFECTS
ularly above 10 GHz. A water vapor absorption line exists
at approximately 21 GHz, and rain absorption and scattering
increase with frequency throughout the band. SHF applications include radar, satellite communication, and microwave
links.
Extremely High Frequency
The extremely high frequency (EHF) band is also called the
millimeter-wave spectrum since the wavelength ranges from
1 to 10 mm. Atmospheric absorption becomes extreme in this
band. An oxygen absorption band is centered at approximately 60 GHz, and numerous absorption peaks due to oxygen and water vapor occur above 100 GHz. Rain attenuation
is significant throughout the entire EHF band. Satellite-tosatellite links are not affected by atmospheric absorption, and
secure short-range communication systems take advantage of
high attenuation to limit propagation range.
LINE-OF-SIGHT PROPAGATION
Line of sight is the dominant mode of propagation for elevated
antennas and high frequencies, as indicated Table 1. The
mathematical description of line-of-sight propagation is most
easily obtained by considering simple antennas in a freespace environment. In this section two classical cases are discussed. First a Hertzian dipole source is considered to illustrate radiation of a spherical wave. Then transmission
between a pair of antennas is analyzed to derive the classical
expression for basic free-space transmission loss.
Radiation from a Hertzian Dipole
This article covers only steady-state, time-harmonic sources
and fields (12) with time variation exp( jt), where the angular frequency 2f and f is the radio frequency. The time
dependence is suppressed in the equations. The basic source
is a current I extending over an incremental length l. This
elementary dipole source is called a Hertzian dipole and has
a moment Il. If the dipole is directed along the z axis, as
shown in Fig. 1, the radiated electric field has two compoz
Er
H
E
0, 0
189
nents, E and Er, and the radiated magnetic field has only a
single component H . The expressions for these field components can be derived from scalar and vector potentials (12):
1
jk0
1
+ 2 +
sin
r
r
jk0 r3
Il0 jk r 1
1
e 0
cos
Er =
+
2
r2
jk0 r3
1
Il jk r jk0
e 0
+ 2 sin
and H =
4
r
r
E =
Il0 jk r
e 0
4
(1)
(2)
(3)
where the free-space wave number k0 00, the freespace impedance 0 0 / 0, 0 is the magnetic permeability
of free space, and 0 is the dielectric permittivity of free space.
The equations for radiation by the dual source (a magnetic
dipole or small loop) are given in Ref. 12.
Very close (k0r 1) to the Hertzian dipole, the electric field
is dominated by the r3 inverse cube term, and Eqs. (1) and
(2) can be approximated by
E
Il sin
Il cos
and Er
4 j
0 r3
2 j
0 r3
(4)
Equation (4) gives the electric field components of charges
I/j separated by a distance l. The magnetic field of an electric dipole has no quasi-static term. At intermediate distances (k0r 1), the inverse-square terms dominate Eqs. (1)
(3), and the field is called the induction field.
For most practical applications, such as communications
or radar, the far fields are of interest. At large distances
(k0r 1), the inverse-distance terms dominate Eqs. (1) and
Eq. (3), and the fields are approximated by
H
jk0 Il jk r
e 0 sin and E 0 H
4r
(5)
Even though Eq. (5) applies to a Hertzian dipole, it illustrates
the more general far-field properties that the electric and
magnetic fields are related by the free-space impedance and
they are orthogonal to each other and to the radial direction
of propagation. Hence the radial electric field Er in Eq. (2) has
no inverse-distance term.
The power density of the electromagnetic field is called the
Poynting vector S and can be written
S = E H
(6)
where boldface denotes vectors and * denotes complex conjugate. The far-field expression for S can be obtained by substituting Eq. (5) into Eq. (6):
Il
S rr
0
x
Figure 1. Geometry for radiation from a short electric dipole.
k0 |I|l sin
4r
2
(7)
where r is the unit vector in the radial direction. The sin2
factor in Eq. (7) is specific to the radiation pattern of a Hertzian dipole, but the inverse-square dependence applies to the
far field of any radiator. The total radiated power P can be
obtained by integrating Eq. (7) over a far-field sphere (12):
190
RADIOWAVE PROPAGATION GROUND EFFECTS
P=
d
0
d r2 sin r S =
0 (k0 |I|l)2
6
(8)
The result for the total radiated power in Eq. (8) is independent of the radius r at which the integration is evaluated, but
the evaluation is simplest in the far field, where S can be
approximated by Eq. (7).
is incident on a half space with permittivity , electrical conductivity , and permeability . The angle of reflection r
equals the angle of incidence, r 0. The reflection coefficient
depends on the polarization of the incident field.
For horizontal polarization (electric field perpendicular to
the plane of incidence), the reflection coefficient h is given by
(1)
k0 cos 0 0 k2 k20 sin2 0
h =
k0 cos 0 + 0 k2 k20 sin2 0
Free-Space Transmission Loss
Now consider free-space transmission between a pair of antennas in the far fields of each other. The received power Pr
can be written (13)
P G Gr 2
P G Gr
Pr = t t 2 = t t 2 0
(2k0 D)
(4D)
(9)
where Pt is the transmitted power, Gt is the gain of the transmitting antenna, Gr is the gain of the receiving antenna, D is
the separation distance between the antennas, and the freespace wavelength 0 2/k0. The D2 factor represents the
inverse-square dependence of the radiated power density
available at the receiving antenna.
The antenna and propagation effects can be separated by
writing Eq. (9) in the following form:
2
0
Pr
= Gt Gr
(10)
Pt
4D
The squared factor on the right side of Eq. (10) is dimensionless and does not involve the antenna gains. The reciprocal is called the free-space transmission loss L0 and is usually
expressed in decibels:
2
4D
4D
L0 = 10 log10
dB = 20 log10
dB
(11)
0
0
Normally the antenna gains and the power ratio in Eq. (10)
are also expressed in decibels.
where k ( j /). For vertical polarization (electric
field parallel to the plane of incidence), the reflection coefficient v is
0 k2 cos 0 k0 k2 k20 sin2 0
v =
0 k2 cos 0 + k0 k2 k20 sin2 0
h =
v =
;;;
Reflected
field
r = 0
0 , 0
,,
Figure 2. Geometry for plane-wave reflection from a homogeneous
half space.
cos 0
cos 0 +
(k/k0 )2 sin2 0
(k/k0 )2 sin2 0
(14)
and
Plane-Wave Incidence
Incident
field
(13)
The reflection coefficients in Eq. (12) and (13) apply to both
the reflected electric and magnetic fields. The power reflection
coefficients are obtained by taking the squares of the magnitudes of the field reflection coefficients, v2 and h2. In general, the reflection coefficients in Eqs. (12) and (13) are complex because k is complex. Thus the reflected field undergoes
phase shift as well as reduction in amplitude. In the limiting
case of grazing incidence (0 /2), both h and v equal 1.
For the common case where the reflecting medium is nonmagnetic ( 0), the reflection coefficients in Eqs. (12) and
(13) simplify to
REFLECTION FROM A PLANAR INTERFACE
Radio waves are often reflected from smooth, flat surfaces,
such as building walls or ground. When the reflecting surface
is not a perfect conductor, part of the radio energy penetrates
the surface, and part of the energy is reflected. Reflection of
a plane wave from a uniform half space with a planar interface can be analyzed exactly by matching boundary conditions
at the interface, and the derived reflection coefficients can
then be used in other practical applications.
Consider the idealized geometry in Fig. 2. A free-space
plane wave propagating at an angle 0 to the surface normal
(12)
(k/k0 )2 cos 0
(k/k0 )2 cos 0 +
(k/k0 )2 sin2 0
(k/k0 )2 sin2 0
(15)
For the further simplification to a dielectric ( 0) reflecting
medium, Eqs. (14) and Eq. (15) reduce to
r sin2 0
h =
cos 0 +
r sin2 0
(16)
r sin2 0
v =
r cos 0 +
r sin2 0
(17)
cos 0
and
r cos 0
where r / 0.
As incidence angle 0 varies from 0 (normal incidence) to
/2 (grazing incidence), h varies smoothly from (1 r)/
(1 r) to 1. However, for vertical polarization, the reflection coefficient v equals 0 at the Brewster angle, B
tan1(r). At this angle, all of the incident energy is refracted
into the dielectric. An examination of Eq. (15) reveals that the
presence of nonzero conductivity (which yields a complex
k) prevents v from going to zero. However, if the imaginary
RADIOWAVE PROPAGATION GROUND EFFECTS
e
Dir
ct
ray
a
dr
y,
cient equals 1, and Eq. (20) reduces to
;;;
fl
Re
ec
te
E | = =
Il
,,
0 , 0
Figure 3. Far-field radiation from a vertical electric dipole over a
homogeneous half space. The reflection coefficient v is a function of
the incidence angle .
part of k is small, there is nevertheless a pseudo-Brewster
angle (14) where v goes through a minimum.
Dipole Sources
Consider now a vertical electric dipole source located at a
height h over a reflecting half space, as in Fig. 3. In the far
field the electric field has only a component E, which can
be written as the sum of a direct and a reflected ray:
E =
j0 Il sin jk r jk h cos
e 0 (e 0
+ v ejk0 h cos )
4r
(18)
where v is given by Eq. (13). For the special case of a perfectly conducting ground plane ( ), the reflection coefficient equals 1, and Eq. (18) reduces to
E | =0 =
j0 Il sin jk r
e 0 cos(k0 h cos )
2r
(19)
Equation (19) has a maximum at the interface, /2.
The dual case of a vertical magnetic dipole source is shown
in Fig. 4. The source is a small loop of area A and current I,
and the loop axis is in the vertical direction. The electric field
is horizontally polarized, and in the far field the component
E is
E =
0 k20 IA sin jk r jk h cos
e 0 (e 0
+ h ejk0 h cos )
4r
191
j0 k20 IA sin jk r
e 0 sin(k0 h cos )
2r
(21)
Equation (21) has a null at the interface, /2.
For realistic (finite) ground parameters, the reflection coefficients for both vertical polarization v and horizontal polarization h equal 1 at grazing incidence ( /2). Hence the
direct and reflected rays cancel in Eqs. (18) and (20), and the
electric field is 0:
E | = /2 = 0 and E | = /2 = 0
(22)
In reality, only the space wave (the inverse-distance field that
occurs for 0) is 0 at the interface. The ground wave is the
dominant field component near the interface, and it will be
discussed in detail later. It has a more rapid decay with distance, but it does not equal 0 at the interface.
PLANE-WAVE REFRACTION
Dielectric Medium
Consider a plane wave incident on a dielectric half space, as
in Fig. 5. For simplicity, the dielectric is taken to be lossless
( 0) and nonmagnetic ( 0). The incident field propagates at an angle 0 to the normal, and the transmitted field
is refracted at an angle t to the normal. The angle of refraction is determined by requiring that the phases of the refracted field and the incident field maintain the same relationship at all points along the interface. This requirement
is met if the tangential wave numbers in the two media are
equal: k0 sin 0 k sin t. For the case of a dielectric medium,
this reduces to
sin t
=
sin 0
0
or t = sin1
sin 0
n
(23)
where n / 0 r is the refractive index. Equation (23)
is called Snells law (14). As with the reflection coefficient,
(20)
;;;;
;;;
;;;
where h is given by Eq. (12). For the special case of a perfectly conducting ground plane ( ), the reflection coeffi-
Incident
field
Reflected
field
e
Dir
ct
ray
Re
fle
r
ted
ay
,h
0 , 0
, 0
IA
0 , 0
,,
Figure 4. Far-field radiation from a vertical magnetic dipole over a
homogeneous half space. The reflection coefficient h applies to horizontal polarization.
Transmitted
field
Figure 5. Reflection from and refraction into a dielectric half space.
The diffraction angle t is determined from Snells law.
192
RADIOWAVE PROPAGATION GROUND EFFECTS
the transmission coefficient depends on the polarization of the
incident field.
For horizontal polarization (electric field perpendicular to
the plane of incidence), the electric-field transmission coefficient Th is given by (1)
where h is given by Eq. (12). The transmitted electric field
Et in the lossy medium (z 0) is
E t = T exp(jkz), where k = (
j /)
(28)
The transmission coefficient T is given by
2 cos 0
Th =
cos 0 +
r sin2 0
(24)
For vertical polarization (electric field parallel to the plane of
incidence), the electric field transmission coefficient Tv is
given by (1)
2
r cos 0
Tv =
r cos 0 +
r sin2 0
Transmission into a lossy medium at oblique incidence is
complicated by the fact that the planes of constant phase do
not coincide with the planes of constant amplitude. Such a
transmitted field is called an inhomogeneous plane wave (1).
To simplify the mathematics, the case of normal incidence, as
shown in Fig. 6, will be considered.
For normal incidence the electric field is transverse to the
z direction. The incident electric field Ei in the free-space region (z 0) has unit amplitude:
(26)
0 =0
exp(jk0 z)
;;;
;;;
(27)
0 , 0
,,
Transmitted field
Z=0
Reflected field
Transmitted field
Figure 6. Reflection from and transmission into a lossy half space
for normal incidence.
2
+ 0
(30)
where 0 is the free-space impedance and /( j /)
is the impedance of the lossy medium. In general, T is complex, and the transmitted field undergoes both phase shift
and reduction in amplitude.
As the transmitted field propagates into the lossy medium,
it undergoes further attenuation and phase delay, as indicated by Eq. (28). The attenuation and phase shift can be isolated by normalizing the field to the value at the interface:
Et
= exp[Im (k)d] exp[ jRe(k)d]
T
(31)
where Re indicates real part, Im indicates imaginary part,
and d z is the depth in the medium. The first exponential
factor on the right side of Eq. (31) represents attenuation, and
the second factor represents phase shift.
The skin depth is defined as the distance over which the
amplitude of the field decreases to 1/e times its initial value.
The expression for is
1
=
=
Im(k)
The reflected electric field Er in the free-space region is
(29)
Equation (29) can be written compactly in terms of impedances:
(25)
Lossy Medium
E r = h |
2k0
k0 + 0 k
T=
The magnetic field transmission coefficients are obtained by
multiplying the electric field transmission coefficients in Eqs.
(24) and (25) by r.
E i = exp(jk0 z)
T=
, where a = 2
and b =
a2 + b 2 a
(32)
For a lossless medium ( b 0), the skin depth is
(no attenuation).
For the case where conduction currents dominate displacement currents ( ), Eq. (32) simplifies to
2
=
1
f
(33)
Equation (33) applies to high-conductivity metals, such as
copper, and to soil and rock at low frequencies.
THROUGH-THE-EARTH PROPAGATION
The antenna and propagation issues are similar for subsurface communications (7) and geophysical probing of the earth
(15). Both applications require transmission of electromagnetic waves through the earth, and both face the problem of
high attenuation. To penetrate the earth to depths of 100 m
or more, it is necessary to employ frequencies below about
3 kHz (ELF). The reason for this is that the skin depth is
proportional to 1/ f, as indicated in Eq. (33). Although plane-
;
;
;;
RADIOWAVE PROPAGATION GROUND EFFECTS
depth h and a horizontal distance d, Hz is given by (16)
;;;
a
Air
193
(d,-h)
Observer
Figure 7. Circular loop on a conducting half space with a subsurface observer.
wave propagation as discussed in the previous section gives
valid results for distant sources, it does not give valid quantitative results for the typical case where the transmitting antenna is located at or below the air-earth interface. So this
section will deal with the fields of surface or buried antennas.
At frequencies below 3 kHz, the free-space wavelength is
greater than 100 km. Consequently, ELF antennas are electrically small even though they could be physically large. The
methods and antennas used in geophysical probing are too
varied to give a complete description. However, if we limit the
applications to deep, subsurface probing and to through-theearth communication, then the most useful antennas are of
two types: wire loop antennas and straight wire antennas
grounded at the ends.
Fields of Loop Antennas
Loop antennas are commonly used in geophysical sounding
and subsurface communications, and they have the advantage that no grounding is required. In geophysical sounding,
loop antennas transmit a time-varying magnetic field into the
earth, and eddy currents are excited in conducting bodies.
These eddy currents generate a secondary magnetic field that
can be received by a second loop antenna. In mine communications, transmitting loops can be used either at or below the
earth surface. Horizontal transmitting loops are typically a
large, single turn of wire laid out on the earth. Various
shapes, such as circular or rectangular, are used depending
on the application. When the loop dimensions are small compared with the skin depth in the earth and the observer distance, the horizontal loop radiates as a vertical magnetic
dipole.
A circular loop of radius a located at the earth surface
(z 0) is shown in Fig. 7. The earth conductivity is , and
the magnetic permeability of both the air and the earth is
0. For the low frequencies considered here, displacement currents are negligible, and the fields are independent of permittivity. This is called the quasi-static approximation and is
obtained by setting the free-space wave number k0 equal to 0.
The wave number in the earth is approximated by k
0 /j, where the square root is taken so that the imaginary part is negative.
When the circular loop carries a uniform current I, the
nonzero field components are Hz, H, and E . The vertical
magnetic field Hz in the earth (z 0) is of primary interest
for downlink communication between horizontal loops. At a
IA
Q
2h3
(34)
where
Earth
Hz =
Q=
0
2
2
2J (ga/h)
g3 e g +jH
dg
J (gD) 1
2 0
ga/h
2
g + g + jH
(35)
H 0h, D d/h, and J0 and J1 are the zero- and firstorder Bessel functions (17). The integration variable g is a
normalized wave number; so the integration in Eq. (35) can
be interpreted as the result of a superposition of all the waves
that the loop antenna transmits into the earth. The exponential factor is consistent with skin depth attenuation because
for small g it is given by exp(jH2) exp(h/ ), where
is given by Eq. (33).
When a/h 1, the factor 2J1(ga/h)/(ga/h) approaches 1
over the significant range of g. In that case the dependence
on a enters only on through the loop area A(a2), and the
loop radiates like a magnetic dipole of moment IA. If the loop
is buried at a depth h and the observer is located at the surface (as in uplink communications), the result for Hz is identical.
For the special case of H a/h 0, the quantity Q reduces to the following result for a static magnetic dipole:
Q|H=a/h=0 =
2 D2
2(1 + D2 )5/2
(36)
If, in addition, D 0, then Q 1. Thus Q is the vertical
magnetic field normalized to the on-axis magnetic field of a
static magnetic dipole. For D 0, both H and E are 0.
In general, the integration in Eq. (35) must be performed
numerically. The infinite upper limit presents no practical difficulty because of the exponential decay for large g. Figure 8
shows the normalized vertical magnetic field magnitude Q
on the loop axis (D 0) as a function of normalized depth H
for a magnetic dipole source (a/h 0). For large values of
H, the field strength decays exponentially just as a plane
wave does. For geophysical probing, vertical sounding (18) is
accomplished by varying the frequency, and low frequencies
are required to obtain information on earth conductivity at
great depth.
In mine communication (7) and source location (8), the offaxis (D 0) field is of interest. Figure 9 shows the dependence of the vertical magnetic field strength on normalized
horizontal distance D. For the static case (H 0), there is a
null at D 2, as shown in Eq. (36), and that null can be
useful in source location (8). For nonzero values of H, Q is
complex, and the null is filled in.
The magnetic field results in Figs. 8 and 9 are actually
valid for small loops of any shape that can be represented by
a magnetic dipole. When the loop dimensions are large, the
field depends strongly on shape. The theory has been developed for loops of arbitrary shape in a conducting medium that
is homogeneous or layered (19). Results for a loop of nonzero
radius are shown in Fig. 10 for the static case (H 0). As the
loop radius is increased, the vertical magnetic field is reduced
on the axis (D 0), but is increased at the larger horizontal
distances. A similar behavior occurs for nonzero values of H.
194
RADIOWAVE PROPAGATION GROUND EFFECTS
1.0
Normalized, vertical magnetic-field magnitude (Q)
Normalized, vertical magnetic-field magnitude (Q)
1.0
0.5
0.2
0.1
D=0
a/h = 0
0.05
0.02
0.01
2
4
6
8
Normalized depth (H = 0 h)
10
a/h = 0
0.6
a/h = 0.5
0.4
a/h = 1
0.2
a/h = 2
0
1
Figure 8. Magnitude of the normalized vertical magnetic field of a
small circular loop (a 0) as a function of the normalized depth H
on the axis (D 0). From Ref. 16.
H=0
0.8
Normalized distance (D = d/h)
Figure 10. Magnitude of the normalized vertical magnetic field of a
circular loop of radius a as a function of normalized distance D. From
Ref. 16.
Also, similar results have been calculated for rectangular
loops (19).
Fields of Grounded Wire Antennas
Grounded wire antennas are the other antenna type commonly used to transmit fields through the earth. In addition
Normalized, vertical magnetic-field magnitude (Q)
1.0
H=0
0.8
1
to geophysical sounding, these antennas are used for uplink
and downlink communications in mines and for ELF communications with submarines (4). A typical configuration is
shown in Fig. 11. The antenna is of length 2l, and the current
I is assumed to be constant over the length of the antenna.
This assumption is valid for insulated antennas grounded at
the ends when the length of the antenna is much less than a
free-space wavelength (20). Because of the use of low frequencies, displacement currents are neglected in air and in the
earth. The earth has conductivity and permeability 0.
The subsurface electric and magnetic fields are both of interest in mine communication and in probing of geophysical
features. In mine communication, the subsurface magnetic
0.6
2
z
y
0.4
Air
0.2
Earth
(x , 0, 0)
x
l
l
R
0
0
Normalized distance (D = d/h)
Figure 9. Magnitude of the normalized vertical magnetic field of a
small circular loop (a 0) as a function of the normalized horizontal
distance D. From Ref. 16.
Observer
(x, y ,z)
Figure 11. Electric line source on a conducting half space with a
subsurface observer.
RADIOWAVE PROPAGATION GROUND EFFECTS
field is received with a loop antenna, or the subsurface electric field is received with a grounded wire antenna.
First consider the fields produced by an incremental current source of length dx located at x, as shown in Fig. 11.
Because of the quasi-static approximation, the Sommerfeld
integral forms for an incremental source of current moment
I dx can be greatly simplified (21). As a result, the magnetic
field components are
P
N
x y z2
x y z
3
4N
I dx
P
3P
+
dH y =
+
2 2 z3
x2 z z2 y2
dH x =
I dx
2 2
and
dH z =
Idx
2 2
2N
3P
4N
2
3
y z
y z y z2
(39)
where N I0[(/2)(R z)]K0](/2)(R z)], P R1
exp(R), R (x x)2 y2 z2, j0, and I0 and K0
are modified Bessel functions (17). Similarly, the electric field
components are
dEx =
I dx
2
2P
3N
+ 2
2
z
y z
dimensionless. The specific forms for the normalized fields are
A(H,Y, X , L) =
B(H, Y, X , L) =
h
2
l
l
F (H, Y, X , L) =
4N
3P
3P
+ 2 2
+ 2
3
z
x z z y
2N
3P
4N
2
3
y z
y z y z2
dx
(46)
dx (47)
1
2
l
l
3N
2P
+ 2
2
z
y z
dx
(48)
(40)
0.8
(41)
Y=0
(42)
Although all six field components are, in general, nonzero,
the dominant field components of interest are Hy, Hz, and Ex.
These are the only nonzero components for a line source of
infinite length, and all other field components vanish in the
plane x 0, even for a line source of finite length. To obtain
the fields of the entire line source, Eqs. (38)(40) must be
integrated over the range of x from l to l. For normalization
purposes, it is convenient to write the fields in the following
forms:
Hy =
I
A(H, Y, X , L)
2h
(43)
Hz =
I
B(H, Y, X , L)
2h
(44)
Normalized, horizontal magnetic-field magnitude (B)
0.7
and
I dx 2P
dEz =
2 x z
The integral forms in Eqs. (46)(48) simplify for both the
low-frequency (small H) and high-frequency (large H) cases
(21), but numerical integration is required in general. Typical
numerical results for H 2 are shown in Figs. 1214. Although A, B, and F are complex for H 0, only the magnitudes are plotted. The phases are relatively constant as a
function of L. For very small values of L, the fields are essentially those of a short dipole and are proportional to L, as
indicated by Eqs. (38)(40). For large L, the field components
eventually reach those of an infinite line source (22). The limits for large L are shown for the horizontal magnetic field as
dashed lines in Fig. 12. In some geophysical applications, it
is desirable to make the grounded wire long enough to simulate an infinitely long line source. An examination of Figs.
I dx 3N
dEy =
2 y x z
h
2
and
(37)
(38)
195
Y = 0.3
0.6
0.5
Y = 0.7
0.4
0.3
H = 2,
X=0
0.2
Y=2
0.1
and
j0 I
F (H, Y, X , L)
Ex =
2
0
0
(45)
where H 0h, Y y/h, X x/h, L l/h, and h
z. The normalized quantities A, B, F, H, Y, X, and L are
Normalized Length (L = l/h)
Figure 12. Magnitude of the normalized horizontal magnetic field as
a function of the normalized line length L. The dashed lines indicate
the limit, L . From Ref. 16.
196
RADIOWAVE PROPAGATION GROUND EFFECTS
pendence of the subsurface electric fields. The frequency dependence of all three components of the electric field has been
analyzed (23), but only the dominant component Ex, will be
considered here. Equations (45) and (48) can be recast in the
following equivalent form:
Normalized, vertical magnetic-field magnitude (B)
0.25
Y = 0.7
0.20
Ex =
Y = 0.3
0.15
I
Exn (W, L, X , Y )
2 h2
(49)
where
Exn = h2
0.10
3N
2P
+ 2
2
z
y x
dx
(50)
Y=2
0.05
H = 2, X = 0
Normalized Length (L = l/h)
and W 0h2. Equations (49) and (50) are consistent with
Eqs. (45) and (48), but here the frequency dependence is explicitly displayed through the dimensionless frequency parameter W.
Some numerical results for Exn as a function of W are
shown in Fig. 15. As W approaches 0, Exn approaches the dc
dc
result Exn
, which is obtained from the gradient of a scalar potential (23):
Figure 13. Magnitude of the normalized vertical magnetic field as a
function of the normalized line length L From Ref. 16.
dc
3
Exn
= (L X )R3
1 + (L + X )R2
(51)
where
(11)(14) and other calculations (21) shows that this is approximately achieved for L greater than about 2. This means
that the antenna length 2l should be at least four times the
depth h of interest in a particular geophysical application.
Straight, grounded wires are also used to excite broadband, transient fields in the earth. To illustrate the dispersive
nature of the earth, it is useful to examine the frequency de-
Y=0
0.30
(X L)2 + Y 2 + 1 and R2 =
(X + L)2 + Y 2 + 1 (52)
When W becomes large, Exn decreases exponentially because
the skin depth becomes small.
Straight, grounded wire antennas are also used underground. For example, long wire antennas have been laid out
in mine tunnels for uplink transmission. The surface fields of
such antennas have been computed for the general case
where the antenna is not parallel to the air-earth interface
(24) to model cases where either the tunnel or the earth surface is not level.
Y = 0.3
0.25
Normalized, horizontal electric-field (Exn)
Normalized, horizontal electric-field magnitude (F)
0.35
R1 =
Y = 0.7
0.20
0.15
H = 2, X = 0
0.10
Y=2
0.05
0
0
Normalized Length (L =l/h)
Figure 14. Magnitude of the normalized horizontal electric field as
a function of the normalized line length L. From Ref. 16.
1.4
L=1
X = 0, Y = 0
1.2
1.0
0.8
X = 1,Y = 0
0.6
0.4
X = 1,Y = 1
0.2
0
0.1
1.0
10
100
Normalized frequency (W)
Figure 15. Magnitude of the normalized horizontal electric field of a
horizontal line source as a function of the normalized frequency W.
;;;
z
z=0
Zs
air
Layer 1
o, 1, 1,
Il
RADIOWAVE PROPAGATION GROUND EFFECTS
197
For a layered earth, the phase of Zs can range from /2 to
/2.
At large distances from the source (k0 1), the dominant
vertical electric field component Ez can be written (26)
Ez
Ez =
j0 Il jk
e 0 F (p)
2
(56)
hs
Layer 2
o, 2, 2,
Figure 16. Vertical electric dipole source at the surface of a twolayer half space. The surface impedance Zs depends on the properties
of both layers.
GROUND-WAVE PROPAGATION
When both the transmitting and receiving antennas are located on the ground, the direct and reflected waves cancel.
Consequently, the inverse-distance space wave is 0 as indicated in Eq. (22). In this case the ground wave is the dominant field component, and it can be calculated for either a
flat-earth model or a curved-earth model.
where
F (p) = 1 j pe p erfc ( j p )
(57)
p (jk0 /2) 2, Zs / 0, and erfc is the complementary
error function (17). The function F(p) is called the Sommerfeld
attenuation function and is actually a correction to the field
of a vertical electric dipole located at the surface of a perfectly
conducting plane. The quantity p is dimensionless and called
the numerical distance. For p 1, F(p) 1.
For large numerical distance (p 1), the asymptotic form
of F depends on the phase of p (26):
F (p)
1
, 3/2 phase (p) 0
2p
(58)
or
Flat Earth
The flat-earth model is useful for analyzing propagation along
flat surfaces (25) or curved surfaces with very large radii of
curvature (such as the earth). Consider a vertical electric dipole (VED) source located at an impedance surface, as shown
in Fig. 16. The air region has free-space permittivity 0, and
free-space permeability 0 is assumed everywhere. The nonzero field components are Ez, E and H .
For propagation analysis, the earth can be characterized
by a surface impedance Zs under fairly general conditions.
The surface impedance is the ratio of the horizontal electric
and magnetic fields at the earth surface: Zs (E /H )z0.
For the two-layer model shown in Fig. 16, the surface impedance for grazing incidence (as in ground-wave propagation) is
given by (26)
Zs = Z1 Q
(53)
where
Z1 = 1 1
Q=
12
02
(u2 /u1 ) + (22 /12 ) tanh u1 h1
(22 /12 ) + (u2 /u1 ) tanh u1 h1
0
, u1 = 12 + k20 , u2 = 22 + k20 , 12
1 j1 /
= j0 (1 + j
1 )
1 =
(54)
1
, 0 < phase (p) /2
F(p) 2 j pe p
2p
(59)
The first term on the right side of Eq. (59) corresponds to a
trapped surface wave (26) that can occur when the surface
impedance is highly inductive (phase of /4). It does not
occur for a homogeneous earth, as indicated in Eq. (58). If Eq.
(58) is substituted into Eq. (56), the following result is obtained for the electric field:
Ez
0 Il
ejk0
2 2 2
(60)
The 2 dependence of the field in Eq. (60) is characteristic of
a lateral wave that requires an interface to support it. A similar analysis is possible for horizontal polarization as produced
by a horizontal electric dipole or vertical magnetic dipole
source (26), but it is of less interest for ground-wave propagation because it decays more rapidly with distance than vertical polarization.
If the height of either the transmitting dipole or the observer is raised above the interface, the field is increased by
a factor called the height-gain function G (26). Consider the
case shown in Fig. 17, where both the source and observer are
;;;;
(55)
and 22 j0(2 j2). The layer thickness h1 and the layer
constitutive parameters are defined in Fig. 16. Q is a correction factor to account for the layering, and Q 1 for a homogeneous earth. For any passive earth model, the real part of
the surface impedance must be nonnegative: Re(Zs) 0. For
a homogeneous earth, the phase of Zs ranges from 0 to /4.
Il
Zs
Ez
Figure 17. Elevated source and observer for propagation over a flat
surface with surface impedance Zs.
198
RADIOWAVE PROPAGATION GROUND EFFECTS
elevated. Then the electric field can be approximately written
Ez =
j0 Il jk
e
F (p)G(h)G(z)
2
(61)
where
G(h) = 1 + jk0 h and G(z) = 1 + jk0 z
(62)
By reciprocity, the height-gain function G is the same for the
source and observer. Equation (62) is valid only for small
heights (k0h 1).
Spherical Earth
The spherical-earth model is used to account for diffraction
loss that occurs for ground-wave propagation into the shadow
region. The mathematical theory is complicated, but it has
been well developed (26) and has been used to generate extensive numerical results for field strength as a function of distance, frequency, and earth surface properties (2729). For
short paths, the flat-earth and spherical-earth models give
similar results, but for long paths the spherical-earth model
must be used for accurate predictions.
The geometry for a radial electric dipole source located at
the surface of a sphere of radius a is shown in Fig. 18. The
surface is characterized by a surface impedance Zs, which can
be determined by Eqs. (53)(55) for a layered earth. The nonzero field components are Er, E, and H . The radial electric
field Er is the component of most interest, and its value at the
surface (r a) is
Er =
j0 Il jk d
e 0 W
2d
(63)
where d is the arc distance along the surface and W is the
spherical-earth attenuation function. It is a correction to the
field of a vertical electric dipole on a perfectly conducting
plane in the same manner as F for the flat earth in Eq. (56).
For large k0a and k0d, W can be written as the classical residue series (25):
W=
Il
x
exp( jxts )
j s=1 ts q2
where x (k0a/2)1/3(d/a), q j(k0a/2)1/3 , and Zs / 0.
The roots ts satisfy the equation
(64)
d
Er
Zs
w1 (ts ) qw1 (ts ) = 0
(65)
where w1 is an Airy function and w1 is the derivative with
respect to the argument. The Airy function w1 is defined by
w1 (t) = [Bi(t) jAi(t)]
(66)
where Ai and Bi are the Airy functions defined by Miller (30).
A systematic method for determining the roots of Eq. (65) has
been presented for arbitrary values of the magnitude and
phase of q (27).
The residue series in Eq. (64) is valid for any value of x,
but converges slowly for small values of x. Two alternative
methods are available for the calculation of W for small x. The
power series in x1/2 is most useful for small q, and the smallcurvature expansion is most useful for large q.
The power series representation for W is given by (27,31)
W=
Am (e j /4 qx1/2 )m
(67)
m=0
where
A0 = 1, A1 = j 1/2 , A2 = 2
1
1
4
A3 = j 1/2 1 + 3 , A4 =
1+ 3 ,
4q
3
2q
3
1
j 1/2
8
7
1 + 3 , A6 =
1+ 3 +
A5 =
2
4q
15
q
32q6
1/2
5
3
j
1
16
27
1 + 3 + 6 , A8 =
1+ 3 +
A7 =
6
4q
2q
105
2q
32q6
1/2
7
j
5
21
1+ 3 + 6 +
A9 =
24
4q
4q
64q9
and
A10 =
64
32
11
7
+
+
+
3
6
945 945q
189q
270q9
In the numerical results to follow, the series has been truncated at m 10 because higher-order coefficients Am are not
available in the literature.
The small-curvature expansion for W is given by (31,32)
1
W = F (p) + 3 [1 j p (1 + 2p)F(p)]
4q
2
p
1
5p2
+
1 F (p)
+ 6 1 j p(1 p) 2p +
4q
6
2
(68)
Figure 18. Vertical electric dipole source at the surface of a spherical
earth with a surface impedance Zs.
where p jxq2 jk0d2 /2 and the flat-earth attenuation
function F(p) is given by Eq. (57). When the radius a approaches (zero curvature), q approaches , and W approaches F. An additional term in Eq. (68) proportional to
q9 has been obtained (31), but numerical results (27) indicate
that it adds little improvement.
To illustrate the range of applicability of the various methods for computing W, a specific example for propagation at a
frequency of 10 MHz is considered. The effective earth radius
RADIOWAVE PROPAGATION GROUND EFFECTS
a is taken to be 4/3 times the actual earth radius of 6368 km
in order to account for normal atmospheric refraction (33).
The magnitude of the normalized surface impedance 0.1;
this yields a value of 9.62 for the magnitude of q. For the
phase of , two values, 30 and 75, are considered. The 30
value (phase of q 60) lies in the range encountered for a
typical homogeneous earth. The 75 value (phase of q 15)
lies in the highly inductive region, where a trapped surface
wave is important. Numerical results for the magnitude of
W as a function of distance are shown in Fig. 19, where the
computations were carried out by four different methods. The
oscillations in the vicinity of d 10 km for the phase ()
75 curve are due to interference between the trapped surface
wave and the usual ground wave.
Although the curve for phase () 75 contains more
structure than that for phase () 30, the range of validity
of the various computational methods depends only weakly on
the phase of . For graphical accuracy in Fig. 19, the different
methods have the following ranges of validity. The residue
series calculation from Eq. (64) using 50 terms is valid for d
greater than about 10 km (x 0.11). This value of d or x can
always be decreased by increasing the number of terms. The
small-curvature formula in Eq. (68) is valid for d less than
about 50 km (x 0.57). The power series in Eq. (67) is valid
for d less than about 1 km (x 0.011). The flat-earth approximation in Eq. (57), which is the first term in Eq. (68), is valid
for d less than about 15 km (x 0.17). The characteristic of
the flat-earth approximation for large d is an algebraic decay
(d2) as opposed to the correct exponential decay of the first
term of the residue series. The small-curvature formula is far
superior to the power series in this example because of the
large value of q 9.62.
Further calculations (27,34) reveal the following general
criteria for the method of calculation. The residue series is
most useful for x 0.2 regardless of the magnitude of q. For
x 0.2, the power series is better for q 1, and the smallcurvature formula is better for q 1. The phase of q does
not have to be considered in deciding which method of calculation to use for W
102
Phase ( ) = 90
89
101
85
100
60
101
0
15
30
102
= 0.1
f = 10 MHz
103
0.1
10
Figure 20. Magnitude of the spherical earth attenuation function for
different values of the phase of the surface impedance. From Ref. 27.
The magnitude of W as a function of d is shown in Fig. 20
for a larger value of 0.2 and for numerous values of
phase (). The results for 0 phase () 45 show the expected smooth decay of the ground wave, but the trapped surface wave has a dominant effect for phase () 45. The root
t0 for the trapped surface wave, also called the Elliott mode
(35), has the following asymptotic expansion (27) for large q:
1
5
11
539
1
+
+
+
+
2q 8q4
32q7
32q10
512q13
4
7
31
397
j2q2 exp q3 1
3
12q3
48q6
288q9
t0 q 2 +
f = 10 MHz
x
exp(jxts )
Gs (y0 )Gs (y)
j s=1 ts q2
Gs (y) =
30
0.1
w(ts y)
w(ts )
(71)
y0 k0h0(2/k0a)1/3, and y k0h(2/k0a)1/3. Since Gs depends on
ts, it is a function of s. Gs as given by Eq. (71) can be expanded
in a power series in y (28,29):
Residue series
Small curvature
Power series
Flat earth
75
104
(70)
where
101
103
(69)
Part of the reason for interest in inductive surface impedances is in modeling propagation over a layer of sea ice on the
ocean (29).
The residue series in Eq. (64) can be generalized to the
case of an elevated source and an elevated observer. For a
source height of h0 and an observer height of h, W takes the
following form (28,29):
Phase ( ) = 30
102
100
d (km)
= 0.1
45
W=
100
75
101
Phase ( ) = 75
199
10
100
Gs (y) = 1 qy +
300
d (km)
Figure 19. Magnitude of the spherical earth attenuation function as
computed by various methods. (Results are shown outside their regions of validity for comparison purposes.) From Ref. 27.
ts 2 1 + ts q 3
y
y +
2
6
(72)
For low heights and small values of s, Gs can be approximated
by the first two terms in Eq. (72):
Gs 1 qy = 1 + jk0 h
(73)
200
RADIOWAVE PROPAGATION GROUND EFFECTS
o, o
d
Il
h
Zs
Observer
d1
h1
Zs1
Figure 21. Propagation along a spherical earth with a change in
surface impedance from Zs to Zs1.
Gs as approximated by Eq. (73) is independent of s and is
equal to the height-gain function for the flat-earth model in
Eq. (62).
VARIABLE TERRAIN
Although the flat-earth and spherical-earth models are useful
for analyzing ground-wave propagation, many propagation
paths involve variable terrain where the ground properties or
surface height vary along the path. The analysis of such paths
usually requires approximate or numerical methods (3639).
Mixed Path
Mixed-path theory has been developed to analyze smooth terrain that has a change in the earth properties (conductivity
and permittivity) between the source and the receiver. The
mathematical theory has been developed for both a flat earth
(40) and a spherical earth (41), but only the more general
spherical-earth model will be considered here.
The geometry for a two-section path is shown in Fig. 21.
Propagation along a three-section path (41) has also been analyzed, but will not be covered here. As in the previous section, the radial electric field Er is normalized to the case of a
flat, perfectly conducting plane:
Er =
j0 Il jk d
e 0 W (x, q, q1 )
2d
sents no difficulty. A useful alternative to Eq. (75) can be obtained by reciprocity (41):
x
W (x, q, q1 ) = W (x, q1 ) +
(q q1 )
j
xx
1 W (x x,
q1 )W (x,
q)
dx (77)
x(
x x)
0
Equations (75) and (77) have been used to calculate W for
a variety of paths and parameters (42). Results for W are
shown in Fig. 22 for propagation along a land-to-sea path for
various lengths of the land section. The frequency is 10 MHz,
the land constants are / 0 15 and 0.01 S/m, and the
sea constants are 1 / 0 80 and 1 4 S/m. The increase in
field strength that occurs in crossing the land-sea boundary
has been called the recovery effect, and it has been observed
experimentally by Millington (43). A similar drop in phase
has been calculated (42) and observed experimentally by
Pressey et al. (44). In Fig. 23, the length of the land section
is fixed at 5 km, and the frequency is varied from 1 to 30
MHz. The recovery effect is most prominent at 30 MHz because the attenuation over land is so rapid at that frequency.
When either height, h or h1, is nonzero, Eqs. (75) and (77)
are still valid, but W needs to be modified by the appropriate
height-gain function:
x
exp(jxts ) w1 (ts y)
,
W (x, q) =
j s=1 ts q2
w1 (ts )
where y = (2/k0 a)1/3 k0 h
and
W (x, q1 ) =
(79)
(74)
All sea
d d1 = 0.4 km
1 km
2 km
5 km
101
10 km
20 km
Al
l
la
nd
x
(q q)
j 1
x1
W (x x,
q)W (x,
q1 )
dx,
d1 > 0
x(
x x)
0
x
exp(jxts ) w1 (ts y)
,
j s=1 ts q21
w1 (ts )
where y1 = (2/k0 a)1/3 k0 h1
where W is the mixed-path attenuation function, q
j(k0 /a)1/3 , q1 j(k0 /a)1/3 1, Zs / 0, 1 Zs1 / 0, and a
is the earth radius. Consider first the case, where the source
and observer are located at the surface (h h1 0). In this
case, the following form of W is most useful (42,43):
W (x, q, q1 ) = W (x, q) +
(78)
(75)
50 km
102
where
W (x, q) =
x
exp(jxts )
j s=1 ts q2
(76)
x (k0a/2) (d/a), and x1 (k0a/2) (d1 /a). The sphericalearth attenuation function in Eq. (76) is the same as that for
the uniform path in Eq. (64), but the arguments are shown
explicitly for use in Eq. (75). The square-root singularity in
Eq. (75) is integrable, and its numerical evaluation (42) pre1/3
1/3
103
0.1
1.0
10
100
d (km)
Figure 22. Magnitude of the attenuation function for a land-to-sea
path for various lengths of the land section. Parameters: f 10 MHz,
h h1 0, / 0 15, 102 S/m, 1 / 0 80, and 1 4 S/m. From
Ref. 42.
RADIOWAVE PROPAGATION GROUND EFFECTS
1
201
2
1 MHz
100
3 MHz
101
200m
10 MHz
102
100m
f = 30 MHz
50m
101
h1 = 0
103
0.1
1.0
10
100m
100
d (km)
Figure 23. Magnitude of the attenuation function for a land-to-sea
path for various frequencies. Parameters: d d1 5 km, h h1
0, / 0 15, 102 S/m, 1 / 0 80, and 1 4 S/m. From Ref. 42.
It is possible to cast the integral forms of Eqs. (75) and (77)
into a modal sum of the following form (42):
W (x, q, q1 ) =
x
(q1 q)
j
s=1 t=1
exp[ j(x x1 )ts jx1tr(l) ] w1 (ts y) w1 (tr(l) yb )
(tr(l) ts )(ts q2 )(tr(l) q21 ) w1 (ts )
w1 (tr(l) )
(80)
where the roots t(1)
r satisfy the mode equation that is similar
to Eq. (65):
w1 (tr(l) ) q1 w1 tr(l) = 0
(81)
The double summation in Eq. (80) converges slowly when either x1 or x x1 is small, but the roots ts and t(1)
r can be computed rapidly (27).
Results for W are shown in Fig. 24 for propagation along
a land-to-sea path for various observer heights. The length of
the land section is 10 km. As many as 600 roots, ts and t(1)
r ,
were used in computing the results (42) in Fig. 24. A good
confirmation of the validity of the numerical results in Fig. 24
as computed by Eq. (80) is that the curve for h1 0 agrees
with the corresponding curve in Fig. 22 as computed by Eq.
(75). As the observer height h1 is increased, the recovery effect
decreases in magnitude and begins at a range beyond the
land-sea boundary.
Irregular Terrain
For many propagation paths, the ground parameters and terrain height vary as a function of position along the path. Analytical methods are not general enough to handle such variations, but the integral equation approach (45,46) has been
found useful for ground-wave propagation over irregular, inhomogeneous terrain. Otts program WAGNER has been thoroughly tested and used on a wide variety of paths (47), and
102
1
10
d (km)
100
Figure 24. Magnitude of the attenuation function for a land-to-sea
path for various observer heights. Parameters: f 10 MHz, d
d1 10 km, h 0, / 0 15, 102 S/m, 1 / 0 80, and 1
4 S/m. From Ref. 42.
Hill has generalized it to allow for an anisotropic layer over a
homogeneous earth (48). The anisotropic layer is intended to
model a forest layer, snow cover, or a layered earth.
The general terrain model is shown in Fig. 25. The terrain
height y, the slab thickness D, and the slab and ground constitutive parameters are functions of the horizontal distance
from the vertical electric dipole source. The terrain is represented by a normalized surface impedance (referred to the
top of the slab), which is a function of the slab and ground
parameters and is thus a function of . The problem is first
solved by considering the case where the source and observer
are located at the surface (ha hr 0). The vertical electric
y
Receiver
0, 0
Il
ha
hr
Origin
y(x)
x
Slab
Integration
point
y( )
D( )
Ground
x
Figure 25. Geometry for the integral equation solution of propagation over irregular, inhomogeneous terrain. The slab and ground parameters can vary as a function of x, but are constant in y.
202
RADIOWAVE PROPAGATION GROUND EFFECTS
field Ey is normalized to the field of a vertical electric dipole
on a conducting plane:
j0 Il jk x
e 0 f (x)
Ey =
2x
(82)
gain functions Gs for a slab medium (48):
f h (x) = f (x)Gs (ha )Gs (hr )
(86)
The source and receiver can be located either in air (positive
height) or in the slab (negative height). For the source or receiver in air, the height-gain function is
where f(x) is the attenuation to be determined from the integral equation. The integral equation is (47,48)
Gs (h) = 1 + jk0 h, h 0
(87)
jk0 x
f (x) = F(x, 0)
f ( )ejk0 (x, )
where is evaluated at the appropriate source or receiver
2 0
location. For the source or receiver in the slab, the heighty(x) y( )
+ [( ) a ]F (x, )
y ( )F(x, )
gain function is
x
x
1 ev 0 z + Rev 0 (2D+h)
d (83)
Gs (h) =
, D < h < 0
(88)
(x )
vc
1 + Re2v 0 D
where
y ( ) y (x)
[y(x) y( )]
+
2x(x )
2
2x
F (x, ) = 1 j peu erfc ( j u )
2
y(x) y( )
p = jk0 2 ( )(x )/2, u = p 1
( )(x )
(x, ) =
and y() is the slope dy/d. The normalized surface impedance () is a function of the slab and ground parameters (48):
( ) = 1
2 + 1 tanh(v0 D)
1 + 2 tanh(v0 D)
(84)
where
gc 1
hc
, 2 =
, v0 = jk0
hc
hc
gc
=
h + h /( j
0 ),
gc =
g + g /( j
0 ),
1 =
hc
vc =
v + v /( j
0 ), =
hc /
vc
v and v are the vertical conductivity and relative permittivity of the slab, h and h are the horizontal conductivity and
relative permittivity of the slab, and g and g are the conductivity and permittivity of the ground. The ground and the slab
have free-space permeability 0. The normalized surface impedance a is evaluated at the source.
Before discussing the numerical solution of Eq. (83), it is
useful to examine the special case of a uniform path: y()
y(x) y() 0 and () a. Thus the integrand is 0, and
f(x) is
f (x) = F (x, 0) = 1 j pa e p a erfc( j pa )
(85)
where pa jk0a2x/2. Equation (85) agrees with the flatearth result in Eq. (56). When either y or varies along the
path, the integral equation must be solved numerically. A forward-stepping solution (49) in x is used to obtain values of f
at discrete values of x along the path. Since Eq. (83) is a Volterra integral equation of the second kind, the value of f(x)
depends only on the previously computed values of f() for
x. Physically this means that backscatter is neglected.
To account for nonzero source and receiver heights, ha and
hr, an attenuation function f h is determined by use of height-
where R (1 2)/(1 2). The limit of Gs as h approaches
0 from above is 1:
Gs (0+ ) = 1
(89)
The limit of Gs as h approaches 0 from below is
Gs (0 ) = 1/
vc
(90)
This is a consequence of the continuity of normal current flow.
The numerical solution of Eq. (83) has been thoroughly
studied, and comparisons with experimental results have
been shown (47,48). Also, a comparison with an approximate
analysis for propagation from a forest to a clearing has been
made. This comparison provides an independent check because the approximate solution uses a Kirchoff integration
over the vertical aperture above the forest-clearing boundary
(48) in contrast to the integral equation solution, which integrates over the terrain surface. The numerical comparison
for f h is shown in Fig. 26. The frequency is 10 MHz, the
ground parameters are g 102 S/m and g 10, and the
slab parameters are D 10 m, h 104 S/m, h 1.1, v
2.5 104, and v 1.25. Both the source and receiver are
located at a height of 10 m. The approximate aperture theory
shows a nonphysical jump for the receiver at the forest boundary, but the more accurate integral equation solution smooths
out the jump. Away from the boundary the two solutions
agree well.
For typical terrain profiles and moderate path lengths (less
than 100 km), the numerical solution of Eq. (83) is reasonably
accurate and efficient for frequencies up to about 10 MHz. At
higher frequencies the numerical solution can become unstable and predict a nonphysical oscillatory field strength as a
function of distance. Also, the computer run time, which is
roughly proportional to frequency squared, can become large.
To improve the stability at higher frequencies, Ott (50) has
modified the integral equation and computer code and has obtained stable numerical results for frequencies up to about
100 MHz.
Knife-Edge Diffraction
At high frequencies where the wavelength is small compared
to terrain features and other obstacles such as buildings,
knife-edge diffraction models (51) can provide useful field predictions. The simplest model is based on a single knife edge,
RADIOWAVE PROPAGATION GROUND EFFECTS
203
decibels) due to the presence of the knife edge is
100
Transmitter
Receiver
Gd = 20 log10 |F (v)|
Forest
Ground
(92)
To avoid computation of the Fresnel integral in Eq. (92), Lee
(51) has provided simple approximations for F. The geometrical theory of diffraction (52) provides a more rigorous highfrequency method for analyzing diffraction from knife-edges
and wedges, and it predicts results that depend on the polarization of the incident field.
A number of extensions to the single knife-edge model in
Fig. 27 have been studied. The knife edge can be replaced by
a rounded diffracting obstacle (53,54), which is a better model
for broad-terrain features in some cases. Multiple knife edges
(5557) can be used to model paths with multiple diffracting
obstacles. Voglers analysis (58) is valid for an arbitrary number of knife edges, and his computer code will handle up to
10 knife edges.
101
fh
102
BIBLIOGRAPHY
Aperture
theory
Integral equation
103
0
Distance (km)
Figure 26. Propagation from a clearing to a forest at a frequency of
10 MHz. From Ref. 48.
as shown in Fig. 27. If the Kirchhoff approximation that the
field in the semi-infinite aperture above the knife edge is
equal to the incident field is made, then the aperture integration can be approximated with the following result for the
diffracted field Ed at the receiver location R (14):
1+ j
Ed = E0 F (v) = E0
exp( jt 2 /2) dt,
2
v
2(d1 + d2 )
where v = h
(91)
0 d1 d2
E0 is the free-space field in the absence of the knife edge and
F(v) is the complex Frensel integral. When the knife edge extends above the line connecting S and R (as in Fig. 27), h and
v are positive. In this case, F 1/2. When the top of the
knife edge is below the line connecting S and R, h and v are
negative. In this case, F 1/2. As v approaches , F approaches 1 in an oscillatory manner. The diffraction gain (in
1. J. A. Stratton, Electromagnetic Theory, New York: McGraw-Hill,
1941.
2. R. S. Elliot, Electromagnetics: History, Theory, and Applications,
New York: IEEE Press, 1993.
3. J. H. Bryant, Heinrich Hertz: The Beginning of Microwaves, New
York: IEEE Press, 1988.
4. M. L. Burrows, ELF Communications Antennas, Stevanage, UK:
IEE/Peter Peregrinus Ltd., 1978.
5. K. Davies, Ionospheric Radio, London, UK: IEE/Peter Peregrinus
Ltd., 1989.
6. P. V. Bliokh, A. P. Nicholaenko, and Y. F. Fillippov (translated
by S. Chomet), Schumann Resonances in the Earth-Ionosphere
Cavity, Stevenage, UK: IEE/Peter Peregrinus Ltd., 1980.
7. J. N. Murphy and H. E. Parkinson, Underground mine communications, Proc. IEEE, 66: 2650, 1978.
8. J. R. Wait, Electromagnetic induction technique for locating a
buried source, IEEE Trans. Geoscience Electronics, 9: 9598,
1971.
9. A. D. Watt, VLF Radio Engineering, Oxford, UK: Pergamon
Press, 1967.
10. D. L. Schilling (ed.), Meteor Burst Communications, Theory and
Practice, New York: Wiley, 1993.
11. K. Siwiak, Radiowave Propagation and Antennas for Personal
Communications, Norwood, MA: Artech House, 1995.
12. R. F. Harrington, Time-Harmonic Electromagnetic Fields, New
York: McGraw-Hill, 1961.
13. C. H. Liu and D. J. Fang, Chap. 29, Propagation, in Y. T. Lo and
S. W. Lee (eds.), Antenna Handbook: Theory, Applications, and
Design, New York: Van Nostrand Reinhold, 1988.
14. E. C. Jordan and K. G. Balmain, Electromagnetic Waves and Radiating Systems, Englewood Cliffs, NJ: Prentice Hall, 1968.
15. G. V. Keller and F. C. Frischnecht, Electrical Methods in Geophysical Prospecting, Oxford, UK: Pergamon Press, 1966.
Knife
edge
d1
S
16. D. A. Hill, Chap. 23, Antennas for geophysical applications, in Y.
T. Lo and S. W. Lee (eds.), Antenna Handbook: Theory, Applications, and Design, New York: Van Nostrand Reinhold, 1988.
h
d2
R
Figure 27. Geometry for diffraction by a single knife edge.
17. M. Abramowitz and I. A. Stegun, Handbook of Mathematical
Functions, Washington, DC: National Bureau of Standards, AMS
55, 1968.
18. A. A. Kaufman and G. V. Keller, Frequency and Transient Sounding, Amsterdam: Elsevier, 1983.
204
RADIOWAVE PROPAGATION IN MULTIPATH CHANNELS
19. J. R. Wait and D. A. Hill, Fields of a horizontal loop of arbitrary
shape buried in a two-layer loop, Rad. Sci., 15: 903912, 1980.
20. E. D. Sunde, Earth Conduction Effects in Transmission Systems,
New York: D. Van Nostrand, 1949.
21. D. A. Hill and J. R. Wait, Subsurface electromagnetic fields of a
grounded cable of finite length, Can. J. Phys., 51: 15341540,
1973.
22. J. R. Wait and K. P. Spies, Subsurface electromagnetic field of a
line source on a conducting half-space, Rad. Sci., 6: 781786,
1971.
23. D. A. Hill and J. R. Wait, Subsurface electric fields of a grounded
cable of finite length for both frequency and time domain, Pure
Appl. Geophys., 111: 23242332, 1973.
24. D. A. Hill, Electromagnetic surface fields of an inclined buried
cable of finite length, J. Appl. Phys., 44: 52755279, 1973.
25. A. Banos, Dipole Radiation in the Presence of a Conducting HalfSpace, New York: Pergamon, 1966.
26. J. R. Wait, Electromagnetic surface waves, in J. A. Saxton (ed.),
Advances in Radio Research, London: Academic Press, pp. 157
217, 1964.
27. D. A. Hill and J. R. Wait, Ground wave attenuation function for
a spherical earth with arbitrary surface impedance, Rad. Sci., 15:
637643, 1980.
28. D. A. Hill and J. R. Wait, Excitation of the HF surface wave by
vertical and horizontal antennas, Rad. Sci., 14: 767780, 1979.
29. D. A. Hill and J. R. Wait, HF radio wave transmission over sea
ice and remote sensing possibilities, IEEE Trans. Geosci. Rem.
Sens., 19: 204209, 1981.
30. J. C. P. Miller, The Airy Integral, Cambridge, UK: Cambridge
University Press, 1946.
31. H. Bremmer, Applications of operational calculus to ground-wave
propagation, particularly for long waves, IRE Trans. Antennas
Propagat., 6: 267272, 1958.
32. J. R. Wait, Radiation from a vertical antenna over a curved stratified ground, Nat. Bur. Stand. J. Res., 56: 237244, 1956.
33. H. Bremmer, Terrestrial Radio Waves, New York: Elsevier, 1949.
34. K. P. Spies and J. R. Wait, On the calculation of antenna patterns
for an inhomogeneous spherical earth, Rad. Sci., 11: 1361
1378, 1967.
35. R. S. Elliott, Spherical surface wave antennas, IRE Trans. Antennas Propagat., 4: 422428, 1956.
36. G. A. Hufford, A. G. Longley, and W. A. Kissick, A Guide to the
Use of the ITS Irregular Terrain Model in the Area Prediction
Mode, National Telecommunications and Information Administration Report 82-100, April 1982.
37. N. DeMinco, Ground-wave Analysis Model for MF Broadcast Systems, National Telecommunications and Information Administration Report 86-203, September 1986.
38. N. DeMinco, Automated Performance Analysis Model for Groundwave Communication Systems, National Telecommunications and
Information Administration Report 86-209, 1986.
39. T. S. M. Maclean and Z. Wu, Radiowave Propagation over Ground,
London: Chapman & Hall, 1993.
40. J. R. Wait, Mixed path ground wave propagation: 1. short distances, J. Res. Nat. Bur. Stand., 57: 115, 1956.
41. J. R. Wait, On the theory of mixed-path ground-wave propagation
on a spherical earth, J. Res. Nat. Bur. Stand., 65D: 401410,
1961.
42. D. A. Hill and J. R. Wait, HF ground wave propagation over
mixed land, sea, and sea-ice paths, IEEE Trans. Geosci. Rem.
Sens., 19: 210216, 1981.
43. G. Millington, Ground wave propagation over an inhomogeneous
smooth earth, Part 1, Proc. Inst. Elec. Engr., 96: 5364, 1949.
44. B. G. Pressey, G. E. Ashwell, and C. S. Fowler, Change of phase
of a low-frequency wave propagated across a coast line, Proc. Inst.
Elec. Engr., 103B: 527541, 1956.
45. G. A. Hufford, An integral equation approach to the problem of
wave propagation over an irregular terrain, Quart. J. Appl. Math.,
9: 391404, 1952.
46. R. H. Ott and L. A. Berry, An alternative integral equation for
propagation over irregular terrain, Rad. Sci., 5: 767771, 1970.
47. R. H. Ott, L. E. Vogler, and G. A. Hufford, Ground Wave Propagation over Irregular, Inhomogeneous Terrain, National Telecommunications and Information Administration Report 79-20, 1979.
48. D. A. Hill, HF Ground Wave Propagation over Forested and Builtup Terrain, National Telecommunications Administration Report
82-114, December 1982.
49. C. Wagner, On the numerical solution of Volterra integral equations, J. Math. Phys., 32: 289401, 1953.
50. R. H. Ott, RING: an integral equation for HF-VHF radio wave
propagation over irregular, inhomogeneous terrain, Rad. Sci., 27:
867882, 1992.
51. W. C. Y. Lee, Mobile Cellular Telecommunications, New York:
McGraw-Hill, 1995.
52. G. L. James, Geometrical Theory of Diffraction for Electromagnetic
Waves, Stevenage, UK: Peter Peregrinus, 1976.
53. J. R. Wait and A. M. Conda, Diffraction of electromagnetic waves
by smooth obstacles for grazing angles, J. Res. Nat. Bur. Stand.,
63D: 181197, 1959.
54. L. E. Vogler, Radio wave diffraction by a rounded obstacle, Rad.
Sci., 20: 582590, 1985.
55. J. Epstein and D. W. Peterson, An experimental study of wave
propagation at 850 Mc, Proc. IRE, 41: 595611, 1953.
56. G. Millington, R. Hewitt, and F. S. Immirizi, Double knife-edge
diffraction in field-strength predictions, Proc. IEE Monogr., 507E:
419429, 1962.
57. J. Deygout, Multiple knife-edge diffraction of microwaves, IEEE
Trans. Antennas Propagat., 14: 480489, 1966.
58. L. E. Vogler, An attenuation function for multiple knife-edge diffraction, Rad. Sci., 17: 15411546, 1982.
DAVID A. HILL
National Institute of Standards and
Technology
204
RADIOWAVE PROPAGATION IN MULTIPATH CHANNELS
RADIOWAVE PROPAGATION IN
MULTIPATH CHANNELS
With the rapid growth in wireless communications, there is
considerable interest in radiofrequency (RF) propagation and
RF channel modeling in the 400 MHz to 2 GHz band, where
multipath propagation is a characteristic property. The topic
of RF signal propagation and the factors that affect the RF
signals have been extensively studied over the last few decades (13). This article presents a comprehensive overview
of the RF signal variations related to propagation in
multipath fading channels. The diverse phenomena that
cause signal variations are described via mathematical models. The different types of fading and their salient features
are discussed in detail. The goal of this article is to provide a
mathematical and an engineering-oriented treatment of
multipath fading, thereby providing the reader with the necessary tools and the information to understand the different
RF propagative issues and how they impact wireless communications.
J. Webster (ed.), Wiley Encyclopedia of Electrical and Electronics Engineering. Copyright # 1999 John Wiley & Sons, Inc.
RADIOWAVE PROPAGATION IN MULTIPATH CHANNELS
The different propagative effects are classified in two major categories as shown in Fig. 1:
small-scale effects
large-scale effects
The small-scale effects are those caused by multipath propagation, the phenomenon by which the transmitted signal
reaches the receiver via multiple paths. The received signal
is a superposition of the different multipath signals, which
add constructively or destructively, thereby producing signal
fluctuations. The variations of the signal level are typically
produced by movement over short distances (1020 times the
wavelength of the RF carrier) during which the mean signal
level remains constant. The different types of fading and their
mathematical characterization and modeling are discussed in
the following sections.
On the other hand, large-scale effects are those caused by
variations in the distance between transmitter and receiver,
which are manifested in the form of path loss, and by environmental factors, such as terrain, which produce log-normal
shadowing about the local mean signal level. These issues are
also discussed in detail.
A schematic of a communication system is shown in Fig. 2.
The term channel is used to imply different portions of the
communication link. Two terms used in this article, namely,
the propagative channel and the baseband channel need to be
defined. The propagation channel is the physical medium that
supports RF propagation between the transmitting and receiving antennas. This includes the objects that cause the
multipath and time dispersion. This channel is time-varying.
The baseband equivalent channel consists of a complex-valued baseband representation of the channel. If the nonlinearities in the IF/RF stages of both transmitter and receiver are
neglected, then the baseband channel can also be represented
by a linear, time-varying channel model, which includes the
appropriate models for the propagation channel.
BASIC DEFINITIONS
Static Channels
The simplest type of communication channel is a static channel, in which the only signal impairment caused by the channel is additive white Gaussian noise (AWGN). The noise is
assumed to have a flat, power-spectral density and is uncorrelated in time. These characteristics are typical of the thermal
noise generated in communication equipment.
The AWGN channel is widely used to evaluate the performance of communication systems (4) (modulation, coding,
etc.). In RF communications, a static environment is unlikely,
except in a few situations when there is a dominant line-ofsight (LOS) component between the transmitter and receiver,
and there is essentially no multipath, such as in microwave
LOS links and in microcells. In practice, the AGWN channel
serves the useful role of providing performance bounds for RF
communication channels because the practical channels (typically Rayleigh and Ricean) always produce signal impairments more severe than the AGWN channel.
Signal variations
due to propagation
Small-scale effects
Large-scale effects
Multipath fading
Free space
Rayleigh fading
Propagative loss
Indoor
Ricean fading
Outdoor
Flat fading
Reflection
Freq.-selective fading
Propagative modes
Diffraction
Fast fading
Scattering
Slow fading
Time dispersion
205
Shadowing
Random FM
Figure 1. Classification of the small-scale effects and the large-scale effects that cause signal
variability in an RF environment.
206
RADIOWAVE PROPAGATION IN MULTIPATH CHANNELS
Propagative
channel
Input
bits
Encoder
Modulation
RF/IF
stages
RF/IF
stages
Demodulation
Outputs
bits
Decoder
Equivalent baseband channel
(complex valued)
Figure 2. Definition of propagative channel and baseband channel.
Fading Channels
Some of the commonly encountered communication channels,
such as that encountered in wireline communications over
telephone networks, can be characterized as AWGN channels
that have a linear time-invariant (LTI) filter channel model.
However, this representation does not apply to radio frequency (RF) channels because their transmission characteristics change with time and hence require statistical characterization using linear, time-varying channel models. Typically
RF channels are characterized by the following time-varying
phenomena:
1.
2.
3.
4.
5.
multipath propagation and signal fading
multipath time dispersion
Doppler shift
random frequency/phase modulation
shadowing
can be expressed as
x(t) =
n (t)s[t n (t)]
(1)
where s(t) is the modulated signal, n(t) and n(t) are the attenuation factor and the propagative delay, respectively, for
the signal received via the nth path. The modulated signal
s(t) is typically represented as (4)
s(t) = m(t) cos[2 f ct + (t)] = Re[u(t)e j2 f c t ]
(2)
with carrier frequency f c, and u(t) m(t) cos(t) - jm(t)
sin(t). Substituting Eq. (2) in Eq. (1), the signal received
x(t) is given by
x(t) = Re
n (t)e
j2 f c n (t )
u[t n (t)] e
j2 f c t
(3)
In this article, these salient features of RF fading channels
and their statistical characterization are described. Some examples of RF channels that fall under this category are
mobile cellular channels
mobile satellite channels
channels with ionospheric and tropospheric propagation
Multipath and Time Dispersion. Multipath is the phenomenon by which a transmitted signal reaches a receiver via multiple propagative paths, each of which has an associated propagative delay. For instance, in Fig. 3, the signal from the base
station to the mobile station has three paths with propagative
delays 1, 2, and 3, respectively. One of them is the direct
line-of-sight path between the transmitter and receiver (with
delay 1). The other two paths (with delays 2 and 2) are
caused by scatterers/reflectors. If a narrow pulse is transmitted in this channel (with 1 2 3), then three copies of the
pulse are received, as shown in Fig. 4. Channels exhibiting
this property are called time-dispersive.
Doppler Spread. Another distinctive characteristic of radio
channels is the time variation of the channel. This can be
caused by mobility of transmitter or receiver or a change in
the propagative environment (change in the scatterers). The
received signal x(t) in this time-varying multipath channel
Figure 3. An example of multipath propagation due to multiple scatterers.
RADIOWAVE PROPAGATION IN MULTIPATH CHANNELS
Transmitted pulse
Time
Received pulses
Time
Figure 4. Effect of time dispersion due to multipath propagation.
207
change (i.e., a 2 change occurs if n(t) changes by 1/f c). Using
the central limit theorem, r(t) can be modeled as a complexvalued Gaussian random process (4), which implies that
c( ; t) is also a complex-valued Gaussian random process in
the t variable. This random amplitude variation is a fundamental characteristic of multipath channels, and hence its
understanding is crucial to the use of these channels. In Rayleigh fading channels, c( ; t) is modeled as zero-mean, and corresponds to the case when there are random variations in the
scattering environment (usually caused by the movement of
the transmitter or receiver) and there is no line-of-sight (LOS)
path. In Ricean fading channels, c( ; t) is non-zero mean because of the presence of a LOS component or because of fixed
scatterers. Typically Rayleigh fading channels are encountered in all cellular and land mobile radio communications,
whereas Ricean fading channels are encountered in mobile
satellite communications. A detailed characterization of these
channels is presented in the subsequent sections.
In some cases, it is more suitable to have a continuous
multipath model. In this case, the received signal x(t) is given
by
x(t) =
c( ; t)s(t ) d
(7)
c( ; t) =
n (t)e j2 f c n (t ) [t n (t)]
(5)
Consider the transmission of an unmodulated carrier s(t) A
cos 2 f ct. Then, using Eq. (4), with u(t) 1, the received
signal is given by
j2 f c n (t )
(6)
r(t) = A
n (t)e
n
which also represents the response of the channel to a complex exponential ej2fct. Even though s(t) is a monochromatic
signal, the output of the channel r(t) contains many different
frequency components, which are generated as a result of the
time variations of the channel response. The range of frequencies over which the spectrum of r(t) is non-zero is called the
Doppler frequency spread (Bd) of the channel, which yields a
measure that is directly proportional to the rate of variations
of the channel response.
Signal Fading. In Eq. (5), c( ;t) is the time-varying (complex-valued) channel impulse response (CIR) and is the discrete multipath channel model, wherein represents the time
dispersion and t denotes the time index for the variations as
a function of time. The CIR c( ; t) can be viewed as the superposition of phasors, each with a time-varying gain/amplitude
n(t) and a time-varying phase n(t) 2f cn(t). With random variations in n(t) and n(t) (caused by the time-varying
multipath), these phasors can add constructively or destructively. The fading phenomenon, which results in signal amplitude becoming very small, is caused by instantaneous destructive combining of the phasors. An important observation
is that a small change in n(t) produces a significant phase
which is similar to Eq. (1) with the summation replaced by
an integral. This model is typically used in tropospheric scatter and ionospheric propagation. In the sequel, we focus on
the discrete model for multipath channels, which is typical for
most terrestrial RF channels.
Fading in a Mobile Communication Environment. If the transmitter and receiver were stationary and all the scatterers
were fixed, then the signal fading variations, described in the
previous section, would not be observed. In mobile communications (such as cellular), the receiver or transmitter is usually moving. So the scattering environment is constantly
changing. A typical fading signal envelope (characterized by
Rayleigh fading) for a mobile receiver using a carrier frequency f c 1.9 GHz and moving at 100 km/h, is shown in
Fig. 5. It can be seen that the signal envelope can vary as
10
5
Magnitude of envelope (dB)
The corresponding baseband (complex) signal r(t) and the
equivalent baseband channel c( ; t) are given by
r(t) =
n (t)e j2 f c n (t ) u[t n (t)]
(4)
0
5
10
15
20
25
30
35
40
0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 0.18
Time (s)
0.2
Figure 5. Typical envelope variations in Rayleigh fading at a vehicle
speed of 100 km/h and carrier frequency 1.9 GHz. The rms value of
the signal envelope is 0 dB.
208
RADIOWAVE PROPAGATION IN MULTIPATH CHANNELS
much as 40 dB (30 dB below and 10 dB above) relative to the
rms value.
A spatial viewpoint of multipath fading is very useful. Consider a stationary transmitter and fixed scatterers. The signals from the scatterers can be visualized as forming a spatial
standing wave pattern with peaks (points in space where the
multipath signals add constructively) and troughs (where the
signals add destructively). The spatial separation between
troughs and peaks is roughly /2, where is the wavelength
of the carrier. If a mobile receiver were to move through this
spatial standing wave pattern, then it experiences the signal
fading every time the receiver has moved by /2. In Fig. 5,
the x-axis is the time in seconds. As the receiver moves, it can
experience multiple signal fades per second.
Example 1: Consider a mobile station moving at 100 km/h
and communicating at a carrier frequency of 1.9 GHz. The
corresponding wavelength 0.15 m. Because the vehicle is
moving at 100 km/h ( 27.77 m/s), the vehicle covers a distance corresponding to 176 in 1 s. Hence, the envelope of
signal received by the mobile can experience as many as 350
signal fades per second, as shown in Fig. 5. The extent of
signal impairment depends on the extent of signal amplitude
reduction, known as the depth of the fade. This is quantified
via the level crossing rate and the duration of fades discussed
in a subsequent section.
Shadowing. Shadowing is the term that refers to the longterm variations in the local mean of the received signal power
at a distance d from the transmitter. These variations are
typically caused by the terrain, buildings, foliage, and other
obstructions in the signal path. If the instantaneous power of
the signal received at the receiver is averaged to remove the
effect of fast fading, the local mean signal level at that
particular distance d from the transmitter is obtained. The
local mean remains constant over a small distance (of the order of multiples of the wavelength of the RF carrier). The local mean is a random variable because of the signal level
variations caused by the terrain and other environmental factors. Through empirical studies, the RV has been characterized by the log-normal distribution (see the next section)
with a standard deviation in the typical range of 6 dB to 9 dB.
FADING DISTRIBUTIONS
The two most commonly encountered types of fading environments are
1. Rayleigh fading
2. Ricean fading
In this section, the statistical properties of Rayleigh and Ricean fading are described. This mathematical framework is
very useful in providing tools for analyzing and simulating
the performance of communications over fading channels and
also for providing insight into the design of robust communication systems in a fading environment (see Ref. 16). Ricean
fading covers a very broad spectrum of environments from
static AWGN channels on the one end to Rayleigh fading
channels on the other. So, it represents an entire family of
fading channels. However, because Rayleigh channels are the
most frequently encountered environment in wireless (mobile) communications, it forms an important class for study.
Both Rayleigh and Ricean distributions refer to the statistics
of the received signals envelope.
RICE DISTRIBUTION
Let X and Y be two independent Gaussian random variables
(RVs) with means m1 and m2, respectively, and common vari
X2 Y2, is distribance 2. Then the RV V, defined as V
uted according to Ricean distribution (8), whose pdf is given
by
2 2
v I sv e v 2+s2
0
2
2
fV (v) =
v>0
(8)
v0
where s2 m12 m22 and I0 is the zeroth order modified Bessel
function of the first kind given by
I0 (x) =
1
2
ex cos u du =
0
1
2
ex cos u du
(9)
V2 has a noncentral chiIt can be verified that the RV V
square distribution whose noncentrality parameter is s2. The
cdf of a Ricean RV V is given by (4)
FV (v) = Pr(v V ) =
V
0
fV (v) dv = 1 Q
s v
,
(10)
where Q is the generalized Q-function defined as
Q(a, b) = e
a 2 +b 2
2
k
a
k=0
Ik (ab),
b>a>0
and Ik(x) is the kth-order modified Bessel function of the first
kind.
Ricean fading is typically encountered in environments
where there is a dominant, nonfading signal component (i.e.,
a line-of-sight path between transmitter and receiver, and
this causes the nonzero mean of X and Y). This is commonly
true in mobile satellite channels. The signal received can be
modeled as the superposition of the scattered multipath signals (with random angles of arrival) and the line-of-sight signal. The effects of the scattered signals depends on their
strength relative to the strength of the nonfaded signal, which
is characterized by the parameter called Ricean factor K, de
fined as K
s2 /22. It is quite common to specify the Rice
10 log10 K dB. As mentioned prefactor in decibels, KdB
viously, the Ricean distribution spans a broad spectrum of
fading environments, including the following special cases:
K 0 ( s2 0) Rayleigh fading channel
K 1 ( s2 2) static channel (no fading)
RADIOWAVE PROPAGATION IN MULTIPATH CHANNELS
Rayleigh Distribution
0.7
As shown in the preceding section, the Rayleigh distribution is a special case of the Rice distribution (K 0 dB). The
Rayleigh distribution can also be derived independently from
the central chi-square distribution, as shown herein. Let X
and Y be two independent identically distributed (iid) zeromean Gaussian RVs with variance 2. Then the RV V defined
X2 Y2, is distributed according to the Rayleigh disas V
tribution, whose pdf is given by
Rayleigh
0.6
pdf of RV fv(v)
0.5
Rice, K = 1
Rice, K = 4
0.4
Rice, K = 10
0.3
v e 2v 2
2
fV (v) =
0.2
0.1
0
209
Amplitude of RV V
Figure 6. Plot of the Rayleigh pdf and the Ricean pdf for different
values of the Rice factor K 1, 4, and 10.
(11)
v<0
A brief derivation of the pdf f V(v) is given in Appendix 1.
Equation (11) can be obtained from Eq. (8) by setting s2 0.
V2 has a central chiIt can also be shown that the RV V
square distribution. The cdf of V is given by
FV (v) = Pr(v V ) =
A plot of the Ricean pdf for different values of the Ricean
factor K is given in Fig. 6. As the value of K increases, the
mean (and also the variance) of V increases. In practice, this
implies that the LOS signal component is stronger relative to
the scattered signal components, and as a result the magnitude of signal fades is reduced. Figure 7 shows segments of
Rayleigh and Ricean fading for different values of K. For K
10, in Fig. 7 the fade depth is typically 10 dB, whereas for
K 1, signal fades of up to 30 dB are observed.
v0
V
0
fV (v) dv = 1 e
2
v 2
2
,v 0
(12)
The mean and variance of V are given by
E[V ] =
and
E[V 2 ] = 2 2
10
0
10
20
Magnitude of envelope (dB)
30
0
Rayleigh
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0.18
0.2
0.06
0.08
0.1
0.12
0.14
0.16
0.18
0.2
0.06
0.08
0.1
0.12
0.14
0.16
0.18
0.2
0.06
0.08
0.1
Time (s)
0.12
0.14
0.16
0.18
0.2
10
0
10
20
30
0
Rice, K = 1
0.02
0.04
10
0
10
20
30
0
Rice, K = 4
0.02
0.04
10
0
10
20
Rice, K = 10
30
0
0.02
0.04
Figure 7. Plot of a typical segment of
Rayleigh and Ricean fading waveforms.
210
RADIOWAVE PROPAGATION IN MULTIPATH CHANNELS
and the higher order moments of V can be obtained from the
expression (4)
k
E[V k ] = (2 2 )k/2 1 +
2
and
2 = E[2 ] 2
CHARACTERIZATION OF RAYLEIGH FADING
Nakagami Distribution
The Nakagami-m distribution (5) was designed to fit empirical data and in some cases provides a better match than Rayleigh or Rice distributions. Hence, it is frequently used to
characterize the statistical fluctuations of signals transmitted
over multipath fading channels. An RV V with Nakagami-m
distribution has a pdf given by
fV (v) =
2 m m 2m1 mv 2
v
e , m 1/2
(m)
(13)
where E[V2], (m) is the gamma function, and the parameter m is called the fading figure, defined as
m=
2
,
)2 ]
E[(V 2
m 1/2
The following are special cases of the Nakagami-m distribution:
m 1; the Nakagami-m distribution reduces to the Rayleigh pdf.
m 1/2; f V(v) is the one-sided Gaussian pdf (more severe
than Rayleigh fading).
m ; it is a static channel (no fading).
The Nakagami-m distribution also provides a close approximation of the Rice distribution with the following
relationship between the Rice factor K and the Nakagami parameter m:
m=
Extensive measurements of the envelope variations of signals
have been carried out in the 100 MHz to 2000 MHz frequency
band, which covers commercial cellular, public safety, and
special mobile radio applications. The measured data confirm
that the envelope of the received signal is Rayleigh distributed when measured over short distances (10 20) over
which the mean signal level is constant. In this section, the
key results of the mathematical model of Rayleigh fading
channels and their characterization are presented. For a complete derivation of the results, the reader is referred to Refs.
13. Herein, the model proposed in Ref. 1 is considered.
Mathematical Model for Rayleigh Fading
In Fig. 8, the XY plane is assumed to be the horizontal
ground plane. Consider a mobile receiver moving with velocity v along the X-axis. The model (1) is that the signal received consists of a number of horizontally traveling plane
waves each with random amplitude (but equal on average)
and random angle of arrival. The phases of the waves are
uniformly distributed in [0, 2] and are assumed to be statistically independent of the amplitudes.
Let the angle of the nth incoming wave be n w.r.t. the Xaxis. We assume a fixed transmitter with a vertically polarized antenna and a mobile receiver with a whip/monopole antenna. The vehicular motion introduces a Doppler shift given
by
fn =
v
cos n (Hz)
(15)
where v is the velocity in m/s and is the wavelength (in
meters) of the transmitted carrier signal. From Eq. (15), the
maximum Doppler shift is given by f n,max v/ Hz.
(K + 1)2
(2K + 1)
Because of its versatility in modeling a wide range of fading
conditions and its analytical tractability, the Nakagami-m
distribution is widely used.
Log-Normal Distribution
The variations in the local mean of the signal power received
in a fading environment is referred to as shadowing. These
variations are caused by terrain and other environmental factors. Shadowing is characterized via the log-normal distribution, as defined herein. An RV is said to have a log-normal
distribution if its pdf is given by
1
f () =
e
2
( ) 2
2 2
nth incoming wave
(14)
n
where
= 20 log10
= E[]
Mobile direction
(velocity v )
Figure 8. Reference figure for deriving the mathematical model for
Rayleigh fading.
RADIOWAVE PROPAGATION IN MULTIPATH CHANNELS
The electric and magnetic field components received at the
mobile receiver are Ez, Hx, and Hy respectively. Next, the expressions for Ez are derived. The superposition of the different
waves is expressed as
Ez = E0
N
Cn cos(2 f ct + n )
(16)
n=1
where n 2f nt n, where f c is the carrier frequency, f n is
the Doppler shift, E0Cn is the amplitude of the nth wave, and
n are the random phase angles uniformly distributed in [0,
N
2]. The Cn are normalized such that n1 Cn2 1, where
indicates the ensemble average. For large N, using the central
limit theorem, the field components Ez, Hx, and Hy can be approximated by Gaussian random processes. The following derivation assumes that the mean power of the signal is constant
with time, which is typically the case as the mobile receiver
traverses short distances (i.e., the shadowing is a constant).
Equation (16) can be expressed as
Ez = Tc (t) cos 2 f ct Ts (t) sin 2 f ct
(17)
and
s
sin =
211
f f 2
c
(21)
fm
Assuming a large number of scatterers and that the mobile
receiver has a vertical whip antenna, then the total incoming
power is uniform as a function of the incident angle . Using
the derivation in Refs. 1 and 3, the power spectrum of the
electric field is given by
3b
1
s
2 f m
2 , | f f c | f m
f fc
SE z ( f ) =
(22)
1
fm
0
| f fc| > fm
where b is a scale factor denoting the mean power received
by an isotropic antenna. A normalized plot of the power spectrum is shown in Fig. 9. The peak of the power spectrum occurs at f m. The power spectra for the magnetic field components are given in Ref. 1.
Envelope Correlation for Rayleigh Fading
where
Tc (t) = E0
N
Cn cos(2 f n t + n )
n=1
Ts (t) = E0
N
Cn sin(2 f nt + n )
(18)
n=1
Both Tc(t) and Ts(t) are Gaussian random processes. The corresponding RVs Tc and Ts have zero mean and equal variance
given by
An important result obtained by using the power spectrum
SEz( f) is the envelope correlation RTT() T(t)T* (t )
where T(t) Tc(t) jTs(t) is the baseband representation of
the E-field components given by Eq. (18) and the envelope
correlation is expressed as a function of the time separation
. Using the Fourier relationship,
fc + fm
RTT ( ) =
SE z ( f )e j2 ( f f c ) df
=
fc fm
fc + fm
fc fm
3b
2 f m
Tc2
= Ts2
= E02 /2
and the ensemble average is evaluated over n, n, and Cn.
The Gaussian RVs Tc and Ts are uncorrelated and hence independent. Their common variance is E02 /2. The envelope of Ez
is given by
Tc2 + Ts2
(19)
Using the results from the section on Raleigh distribution, V
is a Rayleigh distributed random variable, whose pdf is given
by Eq. (11).
Power Spectrum of Rayleigh Fading
If the transmitted signal is a sinusoid of frequency f c, the instantaneous frequency of the received signal arriving at angle
is given by [using Eq. (15)]
f f 2 df
(23)
fm
Doppler power spectrum
8
7
Power spectrum (magnitude)
|Ez | = V =
e j2 ( f f c )t
6
5
4
3
2
1
f () = f c + f m cos
(20)
0
where f m is the maximum Doppler frequency. Hence
cos =
f f
c
fm
1 0.8 0.6 0.4 0.2
0.2 0.4 0.6 0.8
Normalized value ((f fc)/fm)
Figure 9. Power spectrum of the received E-field in a Rayleigh fading environment.
212
RADIOWAVE PROPAGATION IN MULTIPATH CHANNELS
It can be shown that
RTT ( ) = Tc (t)Tc (t + )
+ Ts (t)Ts (t + )
fc + fm
3b
1
s
=
2 cos(2 ( f f c )t) df
f c f m 2 f m
f fc
1
fm
(24)
From Ref. 20, we have the result
1
1 x2
cos(ax) dx = /2J0 (a)
where J0 is the zeroth order Bessel function of the first kind.
Using this result,
RTT ( ) = 1.5bJ0 (2 f m )
(25)
This shows that the envelope of a Rayleigh faded signal has
a Bessel autocorrelation function, which plays an important
part in designing effective error-correcting coding and interleaving techniques for Rayleigh fading channels.
Two statistical properties that help characterize Rayleigh fading are
1. the level crossing rate (LCR)
2. the duration of fades
The derivation of these two parameters is based on the work
of Rice (1,7,8).
Level Crossing Rate. The level crossing rate (LCR) is defined
as the expected rate at which the Rayleigh fading envelope
V(t), normalized to the local rms signal level, crosses a specified level (Vref in a positive direction. Figure 10 depicts a typical envelope variation which has three crossings of the level
Vref in the duration of T seconds. The LCR NV, the number of
level crossings per second is given by (1)
Signal strength (dB)
NV =
Vref
2
vp(V,
v)
dv = 2 f m e
(26)
Example 2: Consider a signal with carrier frequency of 1.9
GHz and a receiver moving at a velocity of 100 km/ph. The
corresponding value of the Doppler frequency f m 176 Hz. At
a normalized value of 0 dB, using Eq. (26), we obtain the
value NV 162 crossings/s. Hence the LCR gives a characterization of the fluctuation rate of the signal envelope.
Duration of Fades. The duration of fades is the expected
value of the time at which the signal level is below a specified
value Vref . In Fig. 10, consider a time interval T (seconds), and
let i be the duration of the ith fade below the level Vref . Then
P [v V ] =
Time (s)
Figure 10. A method to compute the level crossing rate and fade
duration for Rayleigh fading.
1
T i i
(27)
Using the expression for the pdf of V,
Rayleigh Fading Statistics
where v is the rate of change (the time derivative) of the envelope V(t), p(V, v) is the joint pdf of v and v at a specified value
of V, and V/Vrms is the normalized value of the envelope.
It can be verified that the peak value of NV occurs at
3 dB.
P [v V ] =
p(v) dv = 1 e
(28)
where V/Vrms. Using Eqs. (27) and (28), we obtain the
expression for Tfade, the average duration of a fade below v
V (as given in (1)),
T fade =
2
1
1 [e 1]
i =
TNV i
2 f m
(29)
Example 2 (contd.): Consider the same example as in the
preceding section. For the same parameters, using Eq. (29),
the average duration of a fade 20 dB below the rms value of
the envelope 1.43 ms.
Observations Based on Fading Statistics.
NV f m and Tfade 1/f m
At low Doppler frequency, the LCR is low, and hence the
duration of fades is long. This observation is helpful in
choosing diversity and error-correction schemes over fading channels.
The value of Tfade yields an estimate of the number of
symbols that may be lost because of a fade (burstiness of
errors caused by fading).
In particular, if a mobile communication system is designed
with a specified fade margin, then the performance of the receiver can be estimated by using the parameters f m, , and
Tfade to relate the instantaneous signal to noise ratio (SNR) to
the instantaneous bit error rate (BER).
Random Frequency/Phase Modulation. The baseband representation of the E-field components, as given in Eq. (18), can
be expressed as
T (t) = Tc (t) + jTs (t) = A(t)e j (t )
(30)
RADIOWAVE PROPAGATION IN MULTIPATH CHANNELS
where Tc(t) is the in-phase component and Ts(t) is the quadrature component. The effect of the multipath fading channel is
represented by T(t), which is a complex-valued, multiplicative
scaling of the transmitted signal. The time-varying nature of
Tc(t) and Ts(t) manifests itself as a random phase modulation
of (t). In FM receivers, the effect of T(t) produces a baseband
noise component, which can be evaluated (1). The baseband
noise is characterized by the frequency deviation it produces.
It is interesting to note that the frequency deviation (and
hence, the rms value of the baseband noise) depends on the
depth of the fade. As the signal envelope V experiences deep
fades, the frequency deviation of the random FM increases
proportionally. The effect of T(t) can also be viewed as random
phase rotation given by arctan [Ts(t)/Tc(t)]. Hence, multipath
fading causes impairments of the received signal that have
the characteristics of random FM and random phase shifts.
3. The output of the Doppler filters yields the in-phase and
quadrature component of the Rayleigh fading signal.
It can be verified that the output signal Tc jTs has a
Rayleigh envelope with the time correlation specified by the
maximum Doppler frequency f m. The filtering in step 2 can be
done in the frequency domain via FFT. The steps for using
the FFT for the filtering are outlined in Ref. 9.
Jakes Method
In this section, the Jakes approach (deterministic) for generating Rayleigh fading is discussed (Fig. 12). The main advantage of this method over the stochastic method is that segments of time-correlated fading can be generated without
having to generate the fading in between the desired segments, as is typically the case when simulating the performance of Time Division multiple access (TDMA) signals.
The details of the model (Fig. 12) are as follows:
SIMULATION OF RAYLEIGH FADING
There are N0 oscillators with frequencies given by
In the preceding sections, the power spectrum and the envelope correlation of a Rayleigh faded signal have been discussed. Using these results, two methods are obtained for
generating Rayleigh fading:
f n = f m cos
2n
N
, n = 1, 2, . . ., N0
(31)
and one oscillator with frequency f m, where f m is the maximum Doppler frequency. The total number of oscillators
is (N0 1), where N0 (N/2 1). The parameter N is
chosen such that Tc and Ts are Gaussian and the power
spectrum of T(t) Tc(t) jTs(t) closely approximates the
condition in Eq. (22). In Ref. (1), it is shown that a good
approximation is obtained for N 34 ( N0 8).
In Fig. 12, the phases and n, n 1, 2, . . ., N0 are
chosen such that the probability distribution of the resultant phase (arg[T(t)]) is close to a uniform distribution in
[0, 2]. These phases are introduced by the respective
multipliers 2 cos n and 2 sin n. The uniform distribution of arg[T(t)] is ensured if
1. the stochastic filtered noise approach (9,10)
2. the Jakes method (deterministic) (1)
Both methods are easily modified to generate Ricean fading.
Let T(t) Tc(t) jTs(t) be the complex baseband representation of Rayleigh fading. The main requirements on Tc and Ts
for generating Rayleigh fading are that
they have a Gaussian pdf
they be zero mean and have equal variance (typically
normalized to 0.5, such that the variance of T(t) is unity
Tc and Ts must be uncorrelated
the autocorrelation of T(t) must satisfy the condition in
Eq. (25); This condition also implies that the power spectrum closely approximates Eq. (22).
Tc2
= Ts2
and
Tc Ts
= 0
Filtered Noise Approach
(32)
Choices of values that satisfy Eq. (32) are
=0
The method for implementing the stochastic filtered noise approach is shown in Fig. 11. The steps are as follows:
n
N0 + 1
1
Scale1 =
2N0
1
Scale2 =
2(N0 + 1)
n =
1. Generate two independent Gaussian (white) noise
sources.
2. Each of the sequence of random variables (samples of
the noise) is filtered by a baseband Doppler filter whose
frequency response is given by SEz( f).
Gaussian
noise source
(real)
213
Baseband
doppler filter
Tc
In-phase component
Tc + jTs
Gaussian
noise source
(real)
Baseband
doppler filter
Ts
Quadrature component
Figure 11. Stochastic filtered noise method for generating Rayleigh fading.
214
RADIOWAVE PROPAGATION IN MULTIPATH CHANNELS
cos(2 f1t + 1)
2 sin 1
...
2 cos 1
cos(2 fNo t + No 1)
2 cos No
2 sin No
1
cos(2 fnt)
2
2 sin
2 cos
...
...
+
Scale2
Scale1
Tc(t)
Ts(t)
f No / f m
f1 / fm
1.0
1/ 2
...
1.0
fn / fm
Oscillator frequencies
Figure 12. Jakes method for generating Rayleigh fading.
Other choices are also possible.
The oscillator phases n in Fig. 12 are randomly chosen
in the range [0, 2]. The purpose of n is to provide a
means of randomly initializing the fading waveform generator. They do not affect the statistical properties of
Tc(t) and Ts(t).
CHARACTERIZATION OF MULTIPATH CHANNELS
Consider the equivalent, complex-valued, low-pass impulse
response of a time-varying multipath channel c( ; t) as given by
Eq. (5), where c( ; t) is a zero-mean, complex-valued Gaussian
random process in the variable t. Assume that c( ; t) is wide
sense stationary (WSS). Then the autocorrelation of c( ; t) is
given by
Generating Multiple Uncorrelated Fading Waveforms
In many practical applications, it is necessary to generate
multiple uncorrelated Rayleigh fading waveforms. In Ref. 1,
a method of modifying Fig. 12 to generate multiple uncorrelated fading is mentioned. However, it has been observed (11)
that the method in Ref. 1 produces waveforms with nonzero
correlation when more than two waveforms are generated. An
attractive and computationally efficient alternative is presented in (11) based on the WalshHadamard transform. Using this method, the multiple fading waveforms are guaranteed to be uncorrelated.
c (1 , 2 ; t) =
1
E[c(1 ; t)c (2 ; t + t)]
2
(33)
Assuming uncorrelated scattering, that is, that the scattering/
multipath signal at delay 1 given by c(1; t) is uncorrelated to
the signal at delay 2 given by 2 given by c(2; t), hence,
c (1 , 2 ; t) = c (1 ; t)(1 2 )
(34)
When t 0, c(, 0) c(), which is called the multipath
intensity profile, or the power-delay profile. There are different
RADIOWAVE PROPAGATION IN MULTIPATH CHANNELS
experimental methods for measuring the power-delay profile
of a channel such as direct pulse measurement, spread spectrum sliding correlator measurement, and frequency-domain
channel sounding (10). A typical power-delay profile is given
in Fig. 14. The range of over which c() is nonzero is called
the multipath spread of the channel m.
Time-Dispersion Parameters. The mean excess delay, rms
delay spread, and the excess delay spread are channel parameters that are obtained from the power-delay profile c().
The mean excess delay of the power-delay profile is given
by
c (k )k
k
=
c (k )
(35)
where c(k) is the strength of the multipath component
with delay k (obtained from the power-delay profile).
Typically the k are measured relative to the first arriving multipath signal.
The rms delay spread is defined as 2 ()2, where
2 is given by
c (k )k2
k
2 =
c (k )
(36)
Suppose that 0 is the first arriving multipath and K is
the delay beyond which the power drops below X dB of
the maximum strength multipath component, then,
(K 0) is called the maximum excess delay (X dB).
In practice, the value of , 2, and depend on the value of
the threshold for detecting the presence of a signal component.
Coherence Bandwidth. The coherence bandwidth Bc is an
important parameter in time-dispersive channels. Bc can be
viewed as the range of frequencies over which there is a
strong envelope correlation. There is, however, more than one
formal definition of Bc. In Ref. 4, the coherence bandwidth is
defined as the reciprocal of the multipath spread of the channel m,
1
Bc
m
(37)
1
1
2
6
is said to be frequency-selective. If Bc is greater than the BW
of the signal, the channel is said to be frequency-nonselective.
Coherence Time. The rms delay spread () and the coherence BW (Bc) describe the time-dispersive nature of the channel. Analogously, the Doppler spread (BD) and the coherence
time (Tc) describe the time-varying nature of the channel in a
local area (small-scale region). The Doppler spread BD, is defined as the range of frequencies over which the Doppler spectrum is nonzero. In Ref. 4, the coherence time is defined as
Tc
1
BD
(39)
Coherence time is a measure of the length of time over which
the channel impulse response is essentially constant. Alternatively, the coherence time can be viewed as the length of time
over which two received signals have a strong potential for
envelope correlation. Just as coherence BW has a statistical
definition, coherence time Tc can be defined (12) as the time
over which the time correlation function is 0.5. In Ref. 12 it
is shown that
Tc
9
16 f m
(40)
where fm is the maximum Doppler frequency.
Channel Classification
Multipath fading channels are classified on the basis of the
way the channel appears to the transmitted signal. The coherence bandwidth Bc and the coherence time Tc, two of the
main properties that influence the effect of the channel on the
transmitted signal, are classified as follows:
If the BW of the transmitted signal (W) is smaller than
the coherence BW (Bc), then the fading channel appears
as frequency-nonselective or frequency flat. This implies
that all the frequencies of the transmitted signal experience the same channel effects. If W Bc , then it is classified as frequency selective fading channel.
If the duration of the received waveform [a symbol duration (Ts) for digital transmission) is less than the coherence time, then the channel is said to be time flat. Otherwise, it is classified as time-selective.
Based on these definitions, we have the classification depicted
in Fig. 13.
A Tapped-Delay Line Channel Model
A more widely used engineering definition of Bc is given in
Refs. 1, 7, 10, and 12. Wherein the coherence bandwidth is
defined as the bandwidth over which the frequency correlation is 0.5. In Ref. 12 it is shown that to satisfy this condition,
Bc
215
(38)
where is the rms delay spread given by Eq. (36). Another
interpretation of Bc is as follows. If two sinusoids with frequency separation Bc are transmitted through this channel,
then they are affected differently. For modulated signals, if
Bc is smaller than the BW of the signal(W), then the channel
In addition to providing channel information such as the rms
delay spread, the power-delay profile c(), also provides a
means for modeling the channel using a tapped-delay line
(FIR) model. From Eq. (5), n(t) is the amplitude/gain coefficient for a path arriving with delay n(t). A typical power-delay profile is shown in Fig. 14, which in the second figure, is
uniformly sampled into equal delay bins. In general, the different bins contain a number of received signals (corresponding to different paths) whose times of arrival lie within the
particular delay bin. These signals are represented by an impulse at the center of each delay bin that has an amplitude
with the appropriate statistical distribution (Rayleigh, Ricean, etc.). In deriving this model, two assumptions are made:
216
RADIOWAVE PROPAGATION IN MULTIPATH CHANNELS
Signal bandwidth (W)
those samples below the threshold. This model can be implemented by using a tapped-delay line or FIR model, thereby
allowing us to model any arbitrary channel.
Time flat
Time selective
Frequency selective
Frequency selective
LARGE-SCALE EFFECTS
Bc
Frequency flat
Time selective
Time flat
Frequency flat
Tc
Symbol duration (Ts)
Figure 13. A typical power delay profile and the method of sampling
the power delay profile to generate a tapped-delay line model.
there are sufficient number of rays clustered together in
each delay bin;
the statistical distribution of the envelope is known.
Relative power c( ) (dB)
The rate of sampling the power-delay profile is affected by the
time resolution desired and also the bandwidth of the transmitted signal. The next step after sampling the power-delay
profile is to use a threshold (say X dB below the peak of the
power-delay profile), and using the threshold to truncate
Time delay
Radio Signal Propagation
The salient features of RF propagation are briefly described
in this section. For a detailed treatment of this subject, the
reader is referred to Refs. 7 and 10. The three basic propagative mechanisms, illustrated in Fig. 15, are reflection, diffraction, and scattering. Together, these three modes enable us to
estimate the signal level received by a transmitter for a given
RF propagative channel.
Reflection: occurs when a radio wave propagating in one
medium is incident upon another medium that has different electrical properties and a part of the energy is
reflected back into the first medium, depending on the
specific electrical properties of the second medium. If the
second medium is a perfect conductor, all of the incident
energy is reflected. If the second medium is a dielectric,
then the energy is only partially reflected. The reflection
coefficient is a function of the mediums properties, the
signal frequency, and the angle of incidence. Reflections
of RF signals typically occur from objects in the propagative path whose size is larger than the wavelength () of
the RF carrier, such as buildings and walls. In the case
of cellular/PCS signals at 1.9 GHz, the wavelength
;
;;;
X (dB)
Understanding and characterizing the effects of the RF propagative channel are essential to designing RF communication
systems. A wide range of channel conditions are encountered
in RF communications, all the way from LOS channels to severely obstructed channels. Further, the channel may also be
time-varying. Hence modeling is based on statistical and experimental information. This is an area of extensive research
and measurements, over the past two decades, and even until
the present time (1,7,10,1620). In this section, the two main
components of signal variability due to the large-scale effects
of RF propagation, namely, path loss and shadowing, are discussed.
Scattering
;;
;;
Reflection
Diffraction
Receiver
Transmitter
Figure 14. Classification of multipath fading channels.
Figure 15. The different modes of RF signal propagation, reflection,
diffraction, and scattering.
RADIOWAVE PROPAGATION IN MULTIPATH CHANNELS
15 cm 6 in. Hence, a variety of objects act as reflectors.
Signals are also reflected from the ground. A model commonly used to characterize RF channels is the two-ray
ground reflection model (10).
Diffraction: can be viewed as the bending of RF signals
around an obstruction, as shown in Fig. 15. Diffraction
occurs when the obstruction between the transmitter and
receiver has sharp edges. As explained by Huygens principle, when a wavefront impinges on an obstruction, then
secondary wavelets are produced, which give rise to
bending of waves around the obstruction. The field
strength of the diffracted wave in the shadowed region
is the vector sum of the electric field components of the
secondary wavelets. The knife-edge diffraction model (10)
can be used to characterize the diffraction caused by a
single object, such as building in the path of an RF signal.
Scattering: occurs when the RF signal is incident on a
surface that has a certain degree of roughness (7,10).
Scattering in an RF channel is commonly caused by objects, such as buildings. The critical height hc /8 sin
i, where i is the angle of incidence. This implies that
the maximum to minimum level of the surface must be
greater than hc.
ing the propagative path loss (because the criteria for freespace propagation are not met). This topic has been very extensively studied. Detailed information can be found in Refs.
7, 10, 12, and 16. Computation of path loss is of particular
interest to communication systems designers. Because the actual RF communication environments encountered in practice
are so numerous, a unified theoretical/analytical framework
for estimating path loss is not feasible. Most system designers
resort to empirical approaches and semianalytical methods,
which have been validated by experimental/measured data,
to estimate the path loss. The work of Okamura and Hata
(13) is very widely used for path loss estimation. Okamuras
work is based purely on measured data, and Hata provided
the empirical model to fit that data. The advantage of using
empirical models and curve fitting to measured data is that
it accounts for both known and unknown sources of path loss.
On the other hand, the disadvantage is that the validity of
the empirical model, derived from a set of data, is not guaranteed for a different environment.
Let d be the distance between the transmitter and receiver. Both theoretical and measurement-based models show
that the average path loss [Lp(d)] increases directly proportional to dn, where n is called the path loss exponent. Typically n 2, as summarized in the Table 1. By contrast, in
free space, N 2. The path loss Lp(d) is given by
Propagative Path Loss Models
Free-Space Propagative Loss. When there is a LOS path between the transmitter and receiver, the free-space propagation model can be used to predict the signal strength. Such
conditions occur in some satellite and terrestrial microwave
communication links. Suppose that the distance between the
transmitter and receiver is d meters, where d is in the far
field. Then the free-space model (based on the Friis formula)
gives
Pt Gt Gr 2
Pr (d) =
(4 )2 L
1
d2
(41)
where Pt and Pr are the transmitting and receiving power,
respectively, with transmitting antenna gain Gt, receiving antenna gain Gr, and is the wavelength of the carrier. The
term L represents the losses in the system (L 1). The path
loss Lp is the difference between the transmitting and receiving power expressed in decibels.
L p (dB) = 10 log10
(4 )2 L
Pt
= 10 log10
+ 20 log d
Pr
Pt Gt Gr 2
(42)
As the signal propagative distance d increases, the received
power decreases at 20 dB/decade, as seen from Eq. (42). Another commonly used method to compute the signal power received Pr(d) is by measuring it relative to the received power
Pr(d0) at a reference point (distance d0 from the transmitter)
as given by
Pr (d) = Pr (d0 )
d 2
0
217
Lp (d) d n
Lp (d) = L p (d0 ) + 10n log10
(44)
d
(45)
d0
The reference point d0 is chosen such that Lp(d0) can be computed using the free-space path loss model.
Hata and COST-231 Models. This is one of the most widely
used models for estimating path loss in RF communication
channels. Based on extensive measured data, Okamura generated sets of curves that characterize the median attenuation (50th percentile) Lp,50, for a wide range of environments
(range of carrier frequency ( fc), effective height of transmitting antenna (ht,eff ), and distance d from transmitter. Hata
(12) provided an empirical formulation from Okamuras data,
which shows good agreement (between the model and the
measured data) for f c 1.5 GHz. An extension of Hatas
model for frequencies up to 2 GHz is provided in Ref. 14. The
Hata model and COST-231 models are given below: Hata
Model
Lp,50 = 69.55 + 26.16 log10 f c 13.82 log10 ht,eff a(hr,eff )
+ (44.9 6.55 log10 ht,eff ) log10 d
(46)
Table 1. Path Loss Exponents for Different Environments
, d d0
(43)
Outdoor Propagative Loss. In dealing with non-LOS environments, which is typical of most RF communication links,
such as cellular/PCS, we need appropriate models for comput-
Environment
Free space
Urban cellular
In-building (non-LOS)
Shadowed urban cellular
Path Loss Exponent n
2
2.7-4.0
3.0-6.0
4.0-6.0
218
RADIOWAVE PROPAGATION IN MULTIPATH CHANNELS
Pt
COST-231 Model
Lp,50 = 46.3 + 33.9 log10 f c 13.82 log10 ht,eff a(hr,eff )
+ (44.9 6.55 log10 ht,eff ) log10 d + CM
Path loss (Lp)
(47)
where hr,eff is the effective height of the receiving antenna,
a(hr,eff ), is a correction factor based on hr,eff and CM is a correction factor based on the propagative environment. The details
regarding a(hr,eff ), and CM are provided in Ref. 10. The range
of values for which the Hata and COST-231 models are valid
are summarized in Table 2.
The Hata model has a correction factor for rural environments. In general, the Hata model and the COST-231 model
provide an example of the path loss computation in a outdoor,
non-LOS environment. A variation, called the COST231-WalfishIkegami model can be used for transmitting antennas
above or below rooftops and is accurate for d 20 m. A number of models similar to these discussed in this section are
also used in practice. So, the choice of path loss models must
take into account all aspects of the propagative environment,
including transmission frequency, distance of transmission,
polarization, antenna heights, surface refractivity, terrain irregularity, foliage, climate, ground conductivity, and ground
dielectric constant (10).
Indoor Propagative Loss. An increasing number of wireless
communications applications are designed for indoor environments. Hence, there is considerable interest in indoor propagation and in models for it. Although the characteristics of
indoor propagation vary slowly [quasi-static behavior (15)] as
compared with outdoor propagation, a key difference is that
propagation within a building is strongly influenced by a
number of factors, such as building type, layout, construction
material (amount of metal used), types of partitions, and
height and placement of antennas. As a result, the variability
in signal propagation and hence the path loss is quite significant. The model best suited for characterizing path loss in
indoor propagation is similar to that for log-normal shadowing. The path loss at a distance d from the transmitter is
given by
d
Lp (d) = Lp (d0 ) + 10n log10
+ (dB)
d0
Transmit power
Log-normal shadowing
Shadow in margin (Ls)
Rayleigh/ Ricean fading
Fast fading margin (Lf)
Pmin
Receiver sensitivity
Thermal noise floor
Figure 16. Different components of link budget:propagative loss,
shadowing margin, and fading margin.
Shadowing
As discussed earlier, shadowing is caused by terrain and
other environmental factors, such as foliage. The effect of
shadowing causes the variations in the mean of the received
signal. Let L p(d) denote the path loss (including the effect of
shadowing) at a specified distance from the transmitter.
Based on extensive measurements, it has been verified that
L p(d) can be characterized as a random variable with a lognormal distribution about the mean value. When expressed
in dB, the RV perturbing the local mean value of the path
loss is a normal RV, as given by
p (d) = [Lp (d) + ](dB)
L
(49)
where Lp(d) is given by one of the path loss models in the
preceding section and is a normal RV with standard deviation . The RV is obtained from the log-normal RV ,
whose pdf is given earlier. Equation (49) describes the path
loss for a specified value of d but with different values of
shadowing/obstructions between the transmitter and receiver. In practice, the path loss exponent n and the standard
deviation , are used to characterize any environment. In
most cases, n and can be calculated from measurements.
(48)
where is a normal RV with standard deviation and n is
the path loss exponent. It was reported in Ref. 15 that the
typical range of n is 3 to 4. A comprehensive list of the typical
values of n and for a variety of indoor environments is provided in Ref. 10.
Table 2. Range of Validity of Hata and COST-231 Models
Range of Validity
Parameter
Hata
COST-231
Carrier frequency f c
Effective transmit height h t,eff
Effective receive height h r,eff
Distance d from transmitter
Correction factors
150-1500 MHz
30-200 m
1-10 m
1 km
a(h r,eff )
1500-2000 MHz
30-200 m
1-10 m
1-20 km
a(h r,eff ), C M
A Practical Design Model
The goal of this section is to provide a framework for combining the various results in this article relating to small-scale
signal variations and large-scale signal variations into a set
of relevant parameters that may be used by communication
systems designers for link budget calculations. The components of the different effects and their impact on the link budget are shown pictorially in Fig. 16.
Predicting the expected mean received signal power is an
important step in designing a communication link and in estimating the coverage area for a specific transmitter. In Fig.
16, Pt is the transmitted signal power and Pmin is the minimum signal strength that must be received for the receiver to
operate satisfactorily, that is, the signal strength to produce
the minimum carrier-to-noise ratio (C/N) needed for acceptable communications. This is called receiver sensitivity and is
expressed in dBm. There are three margins, one for each of
the following practical effects:
RADIOWAVE PROPAGATION IN MULTIPATH CHANNELS
propagative path loss
small-scale effectsfading margin
large-scale effectsshadowing margin
f ( ) =
In the previous section, the methods of determining the
path loss Lp for different environments were presented. The
large-scale effect due to shadowing is modeled as a log-normal
random variable. Hence, a shadowing margin Ls must be included in the link budget (Fig. 16). The small-scale effects,
characterized by Rayleigh/Ricean fading cause significant
amplitude variations. Hence a fading margin Lf is also included in the link budget. From Fig. 16, the total transmitted
power is given by
Pt = Pmin + Lf + Ls + Lp
(50)
The path loss Lp is deterministic (based on the distance between the transmitter and receiver), whereas the fading and
shadowing are probabilistic. The amount of margin must be
judiciously chosen so that the net margin is minimized but
still meets the minimum signal strength requirements. If the
amount of losses exceeds the margin of a communication system, then an outage occurs, which implies that the communication link cannot be used until the channel conditions improve. This summarizes the tradeoffs that must be handled
by the designers of communication systems.
CONCLUSION
A comprehensive overview of the RF signal variations related
to propagation in multipath fading channels has been presented. The diverse phenomena that cause signal variations
are described via mathematical models. The different types of
fading and their salient features are discussed in detail. This
article provides a mathematical and an engineering-oriented
treatment of multipath fading, thereby providing the reader
with the necessary tools and the information to understand
the different RF propagative issues and the way they affect
wireless communication.
APPENDIX
Derivation of Rayleigh pdf
Let X and Y be two independent identically distributed (iid)
zero-mean Gaussian RVs with variance 2. Because X and Y
are independent, their joint pdf is the product of their marginal pdf s:
FX ,Y (x, y) =
2
2
1
x +y
2 2
e
2 2
(51)
X2 Y2 and
Using a change of variables, V
tan1
(Y/X), the joint pdf f V,(v, ) is given by
fV, (v, ) =
2
f X ,Y (x, y)
v
v
=
e 2 2
2
|det J (x, y)|
2
(52)
where J(x, y) is the Jacobian matrix for the transformation of
variables. From Eq. (52), we obtain the marginal pdf of V and
as
fV, (v, ) dv =
v e 2v 2
2
fV (v) =
1
2
v0
219
(53)
(54)
v<0
Hence, is uniformly distributed, and V is Rayleigh distributed.
BIBLIOGRAPHY
1. W. C. Jakes, Microwave Mobile Communications, New York: Wiley, 1974.
2. R. H. Clarke, A statistical theory of mobile-radio reception, The
Bell Syst. Tech. J., 47: 9571000, 1968.
3. M. J. Gans, A power-spectral theory of propagation in the mobileradio environment, IEEE Trans. Veh. Technol., VT-21: 2738,
1972.
4. J. G. Proakis, Digital Communication, 3rd ed., New York:
McGraw-Hill, 1995.
5. M. Nakagami, The m distribution: A general formula of intensity
distribution of rapid fading, in W. G. Hoffman (ed.), Statistical
Methods in Radio Wave Propagation, New York: Pergamon, 1960,
pp. 336.
6. I. S. Gradshteyn and I. M. Ryzhik, Table of Integrals, Series, and
Products, Orlando, FL: Academic Press, 1980.
7. W. C. Y. Lee, Mobile Communications Engineering, New York:
McGraw-Hill, 1982.
8. S. O. Rice, Statistical properties of a sine wave plus random
noise, The Bell Syst. Tech. J., 27: 109157, 1948.
9. J. I. Smith, A computer generated multipath fading simulation
for mobile radio, IEEE Trans. Veh. Technol., VT-24: 3940, 1975.
10. T. S. Rappaport, Wireless Communications: Principles and Practice, New York: IEEE Press, 1996.
11. P. Dent, G. E. Bottomley, and T. Croft, Jakes fading model revisited, Electron. Lett., 29: 11621163, 1993.
12. R. Steele, Mobile Radio Communications, New York: IEEE
Press, 1995.
13. M. Hata, Empirical formula for propagation loss in land mobile
radio services, IEEE Trans. Veh. Technol., VT-29: 317325, 1980.
14. COST 231, Urban Transmission Loss Models for Mobile Radio in
the 900- and 1800-MHz Bands, COST 231 TD(90) 119, Revision
2, The Hague, September 1991.
15. A. Saleh and R. A. Valenzuela, A statistical model for indoor
multipath propagation, IEEE J. Selected Areas Commun., SAC5: 128137, 1987.
16. G. L. Stuber, Principles of Mobile Communications, Boston:
Kluwer, 1996.
17. G. A. Arredondo and J. I. Smith, Voice and data transmission in
a mobile radio channel at 850 MHz, IEEE Trans. Veh. Technol.,
VT-26: 8893, 1977.
18. C. Loo and N. Secord, Computer models for fading channels with
applications to digital transmission, IEEE Trans. Veh. Technol.,
VT-40: 700707, 1991.
19. G. A. Arredondo, W. H. Chriss, and E. H. Walker, A multipath
fading simulator for mobile radio, IEEE Trans. Veh. Technol., VT22: 241244, 1973.
20. GSM Series 03.30.
R. D. KOILPILLAI
Ericsson, Inc.
270
RECEIVING ANTENNAS
z
I(s')
RECEIVING ANTENNAS
An antenna can be used for both reception and transmission.
This article discusses the properties of an antenna when it is
used for the reception of an electromagnetic (EM) plane wave
(1,2). Figure 1(a) shows an example of a receiving antenna, in
which a half-wavelength dipole is used. A receiver, expressed
as antenna load ZL, is connected to the center terminals of the
dipole. The arrows in this figure show the flow of the power
density (Poynting power) of an incident EM plane wave,
which propagates toward the dipole antenna from the right
side. It is observed that the power of the incident EM plane
wave moves toward the center terminals and is absorbed in
the antenna load ZL.
r(s')
s1
R
s'
y
x
s2
Figure 2. Coordinate system for radiation field.
The point of interest of a receiving antenna is the power
WL delivered to a receiver or antenna load ZL, as shown in an
example of Fig. 1(a). To calculate WL, the induced current I0
at the antenna terminals must be obtained. For this, an
equivalent circuit for the receiving antenna is introduced. The
maximum value of WL is discussed on the basis of the vector
effective height h.
The receiving antenna is recognized as an electrical net for
collecting an EM plane wave. For example, the power of the
EM plane wave in Fig. 1(b) is collected by many elements on
a circular cavity of area Aap (aperture) and transferred to the
center port to which a receiver (ZL) is connected. Generally,
the collected power WL at the center port is less than the
power Wap given by Aap times the power density at the receiving antenna aperture. In other words, 100% of Aap is not used
for the reception of the EM plane wave. In the final section,
the aperture efficiency ap as a measure of receiving antenna
performance is defined after the discussion of the receiving
cross section Ar. (Note that some fundamental relationships
used in the discussion of receiving antennas are summarized
in the last part of this article.)
(a)
y
Aap
VECTOR EFFECTIVE HEIGHT
Consider an antenna isolated in free space specified by permittivity 0 and permeability 0, as shown in Fig. 2, where
spherical coordinates (R, , ) are used with unit vectors (R,
, ). The antenna is driven by a voltage source of frequency
f. The current I(s) flows along the antenna conductor of
length L s2 s1, radiating the electric field E expressed as
Einc
Z
x
ZL
E = j 30k
e jkR
I h
R 0
(1)
where h, called the vector effective height, is defined as
(b)
Figure 1. Reception of an electromagnetic plane wave. (a) Halfwavelength dipole antenna. (b) Array antenna on a circular cavity.
h=
1
+ (S )
}
{(S )
I0
J. Webster (ed.), Wiley Encyclopedia of Electrical and Electronics Engineering. Copyright # 1999 John Wiley & Sons, Inc.
(2)
RECEIVING ANTENNAS
with
S=
s2
I(s )s e jkr(s
)R
ds
(3)
s1
The notations in Eqs. (1) to (3) are as follows: k( 00
2/ with 2f, where is wavelength) is the phase constant, I0 is the input-terminal current [i.e., I0 I(0)], s is the
distance from the driving point to a source point along the
antenna conductor, s is the unit vector tangential to the antenna conductor at the source point, and r(s) is the position
vector directed toward the source point from the coordinate
origin.
Example. Let us consider an infinitesimal dipole antenna
(kr 0) on the z axis, assuming that the current has constant
amplitude and phase over the antenna length l. Equation (3)
is calculated to be S I0ls. From Eq. (2), h l(s ) l
sin . In the direction normal to the dipole axis ( 90),
h l lz hd.
nal current and V01 is the open terminal voltage (open-circuit
voltage).
Antenna 2 is an infinitesimal dipole antenna with vector
effective height hd lz, which is used as receiving and transmitting antennas in (a) and (b), respectively.
The radiation field E1 from antenna 1 induces an opencircuit voltage at antenna 2
V02 = E1 z
= ( j 30 k
e jkR
I h) hd
R 01
It should be noted that the vector effective height h is defined
for the situation where the antenna is used for transmission.
Let us consider how h is related to receiving antenna operation.
Figure 3 shows two antenna systems in which antenna 1
with vector effective height h is used as a transmitted antenna in (a) and as a receiving antenna in (b). I01 is the termi-
(4)
Using Eq. (1), the radiation field from antenna 2 with terminal current I02 is written as
E2 = j 30 k
e jkR
I h
R 02 d
(5)
The open-circuit voltage V01 at antenna 1 induced by the radiation field from antenna 2 satisfies the relationship according
to the [reciprocity theorem (1)]
V01 I01 = V02 I02
OPEN-CIRCUIT VOLTAGE
271
(6)
Substituting Eq. (4) into Eq. (6) yields
V01 = j 30 k
e jkR
h I02 hd
R
(7)
Using Eq. (5) and replacing V01 and E2 with V0 and Einc, respectively, Eq. (7) is written as
V0 = Einc h
(8)
Equation (8) means that the open-circuit voltage is given by
the inner product of the incident wave field and the antenna
vector effective height.
I01
E1
Ant. 1, h
V02
Ant. 2, hd
(a)
Example. Let us obtain the maximum open-circuit voltage
for a half-wavelength dipole antenna (L /2) located on the
z axis. The maximum open circuit voltage is obtained when
the polarization of an incident wave Einc is parallel to the dipole; that is, the incident wave illuminates the dipole from
the 90 direction. When the half-wavelength dipole has a
current distribution of I(z) I0 cos kz over the antenna conductor from z /4 to z /4, Eq. (3) yields S I0
/z using r R 0. Hence, h /z. From Eq. (8), the maximum open-circuit voltage is V0 Einc /.
EQUIVALENT CIRCUIT OF A RECEIVING ANTENNA
V01
I02
E2
Ant. 1, h
Ant. 2, hd
(b)
Figure 3. Determination of open-circuit voltage at antenna 1. (a) Antenna 1 for transmission. Antenna 2 is an infinitesimal dipole for reception. (b) Antenna 1 for reception. Antenna 2 is an infinitesimal
dipole antenna for transmission.
The original receiving antenna problem shown in Fig. 4(a) can
be handled by superimposing two cases, shown in Fig. 4(b)
and (c). In Fig. 4(b) an EM plane wave illuminating the antenna induces an open-circuit voltage of V0 Einc h. In Fig.
4(c) the antenna acts as a transmitting antenna with terminal current I. Superimposing these two cases leads to a relationship of
Vrec = V0 Vtrans
(9)
where Vrec ZLI and Vtrans ZAI, with ZA defined as the antenna input impedance. From Eq. (9), the terminal current I
272
RECEIVING ANTENNAS
XA)], Eqs. (13) and (14) become
I
Einc
Vrec
ZL
Einc
(a)
+
V0
WR =
Vtrans
(b)
(c)
Figure 4. A receiving antenna. (a) Original receiving antenna problem. (b) Antenna with open terminals. (c) Transmitting antenna.
is given as
I=
V0
ZA + ZL
(10)
Therefore, the equivalent circuit for Eq. (10) is as shown in
Fig. 5, where the open-circuit voltage V0 is used as a voltage
generator with an internal impedance ZA.
Let us express the load impedance ZL and the antenna input impedance ZA as
ZL = RL + jXL
(11)
ZA = (RA + rA ) + jXA
(12)
where RA and rA are the radiation resistance and the loss resistance (not contributing to the radiation), respectively.
Then, the power delivered to the antenna load is given as
|V0 |2 RL
WL =
|ZA + ZL |2
(13)
|V0 |2 (RA + rA )
|ZA + ZL |2
(14)
When an impedance matching condition is satisfied [i.e.,
when ZL is complex conjugate of ZA(RL RA rA and XL
ZA
I
V0 = Einc h
(RA + rA ) + jXA
RL + jXL
(15)
|E h|2
|Einc h|2
= inc
4(RA + rA )
4 RL
(16)
ZL
It follows that half of the power provided by the generator is
delivered to the antenna load. Note that WL max in Eq. (15) is
the maximum power that can be delivered to the load ZL, because the antenna is perfectly matched to the load. Also, note
that WR is recognized as the scattered (reradiated) power from
the receiving antenna (1,4,5).
Example. The relationship WR WL obtained when the impedance is matched can be checked by a numerical technique
called the method of moments (3), where the incident power
Winc and the received power WL are calculated. The scattered
power is obtained from Winc WL. For a center-load dipole
antenna (4), the relationship WR WL is obtained when the
antenna length L is less than 0.8. Limits on the validity of
the equivalent circuit shown in Fig. 5 are discussed elsewhere (1,4,5).
RECEIVING CROSS-SECTION AND APERTURE EFFICIENCY
The receiving cross-section Ar of an antenna is defined as
the ratio of the power WL [Eq. (13)] received by the antenna
to the Poynting power (power density of the incident wave,
Pinc Einc2 /Z0, where Z0 120):
Ar =
|Einc h|2 RL Z0
|ZA + ZL |2 |Einc |2
Using an impedance mismatch factor Mimp defined as WL
WL maxMimp, Eq. (17) is rewritten as
|Einc h|2
Z0
Mimp
4(RA + rA )
|Einc |2
|E h|2
Z0
= inc
Mimp
4RA
|Einc|2
(18)
where RL RA rA RA / (Appendix Eq. (a1)) is used. Using
the absolute gain Ga [see Eq. (a5)], Eq. (18) is written as
Ar =
2
|Einc h|2
Ga
M
4
|Einc |2 |h|2 imp
(19)
Because the received power is proportional to the square of
the open-circuit voltage, that is, Einc h2, the third factor
Einc h2 /Einc2h2 in Eq. (19), is the reduction factor of the received power from polarization mismatch. Let the third factor
be denoted as Mpol, which has a maximum value of 1 for the
case in which h is a real constant multiplied by the complex
conjugate of Einc. Eq. (19) now can be written as
Ar =
Figure 5. Equivalent circuit of a receiving antenna.
(17)
Ar =
and the power consumed in the generator is given as
WA =
|Einc h|2
WL max
4 RL
WL =
2
Ga MpolMimp
4
(20)
The maximum receiving cross-section, which is called the effective area Aeff , is obtained when Mpol Mimp 1.
RECEIVING ANTENNAS
Aeff =
2
Ga
4
(21)
Let the receiving antenna gain be defined as the ratio of the
effective area of the antenna, Aeff , to the effective area of an
isotropic antenna, Aeff, iso 2 /4. Then, from Eq. (21), the receiving antenna gain is equal to the absolute gain when the
same antenna is used as a transmitting antenna.
Example. An isotropic antenna has a gain of Ga 1 (0 dB)
by definition. Using Eq. (21), the effective area is calculated
to be Aeff 0.07962. An infinitesimal dipole, whose vector
effective height is h lsin , has a gain of Ga 1.5 (1.76
dB) at 90 and an effective area of Aeff 0.1192. A halfwavelength dipole antenna with h / yields an effective
area of 0.1302, because Ga 1.64 (2.15 dB). Note that all
Gas are calculated using Appendix (a6) and (a7) with 1.
When a receiving antenna has an aperture Aap which is
much larger than the wavelength , the performance of the
receiving antenna is evaluated by how efficiently the aperture
is utilized for reception. Since the receiving antenna of Aap has
the potential to collect an EM wave power of Wap PincAap, the
ratio of the received power Pinc Aeff to Wap is defined as the
aperture efficiency ap, where
ap =
Aeff
1
Aap
273
ation power density from a reference antenna that has the
same input power as the test antenna:
|E(R, , )|2 /Win
|E0 |2 /Win,0
G(, ) =
(a4)
where E0 is the far-field radiated from the reference antenna,
and Win, 0 is the power input to the reference antenna. Note
that the maximum value of the gain is conventionally used
for the antenna gain if the coordinates (, ) for the direction
of interest are not specified.
When an isotropic antenna (hypothetical antenna radiating with uniform radiation power density in all directions) is
chosen as the reference antenna, the gain is called the absolute gain and denoted as Ga. Equation (a4) becomes
Ga =
4R2 |E(R, , )|2
Z0
Win
4R2 |E(R, , )|2
=
Z0
Wrad
(a5)
Using Eq. (1) and Eq. (a2),
Ga = D
(a6)
4|h(, )|2
|h(, )|2 d
(a7)
where
(22)
D=
Example. A parabolic reflector antenna has an aperture efficiency of less than 70% (68). Flat antennas, such as curl
and helical element array antennas (9,10), have aperture efficiencies ranging from 70 to 90%, where the elements are
arrayed using a circular cavity, as shown in Fig. 1(b).
D is called the directivity.
BIBLIOGRAPHY
APPENDIX
The input power Win[ I(0) (RA rA)] to a test transmitting
antenna is transformed to the radiation power Wrad( RA
I(0)2) with a relationship of Wrad Win, where is called the
radiation efficiency:
2
Wrad
RA
=
Win
RA + r A
(a1)
The radiation power Wrad can be calculated by integrating the
Poynting power P E(R, , )2 /Z0 over the surface of a
sphere of radius R
RA |I(0)|2 = R2
|E(R, , )|2
d
Z0
(a2)
where d is the solid angle subtended by the area dA R
sin d d.
Using Eq. (1), the radiation resistance RA is given as
RA =
k
4
2
Z0
|h|2 d
(a3)
The gain G(, ) is defined as the ratio of the radiation power
density from a test antenna in a direction of (, ) to the radi-
1. J. D. Kraus, Antennas, 2nd ed. New York: McGraw-Hill, 1988,
Ch. 2.
2. R. E. Collin and F. J. Zucker, Antenna Theory. New York:
McGraw-Hill, 1969, Ch. 1, 4, 17.
3. R. F. Harrington, The Field Computation by Moment Methods.
New York: Macmillan, 1968.
4. K. Hirasawa, M. Shintani, and M. Morishita, Received and scattered power of receiving antenna. IEEE Antennas and Propagation Symposium Digest. New York: IEEE, 1994, pp. 205208.
5. R. F. Harrington, Electromagnetic scattering by antennas. IEEE
Trans. Antennas Propag., AP-11 (3): 595596, 1963.
6. W. L. Stutzman and G. A. Thiele, Antenna Theory and Design.
New York: Wiley, 1981, Ch. 8.
7. R. C. Johnson, Antenna Engineering Handbook, 3rd ed. New
York: McGraw-Hill, 1993, Ch. 17.
8. H. Nakano, J. Yamauchi, and H. Mimaki, Backfire radiation from
a monofilar helix with a small ground plane. IEEE Trans. Antennas Propag., AP-36: 13591364, 1988.
9. H. Nakano et al., Low-profile helical array antenna fed from a
radial waveguide. IEEE Trans. Antennas Propag., AP-40: 279
284, 1992.
10. H. Nakano et al., A curl antenna. IEEE Trans. Antennas Propag.,
AP-41: 15701575, 1993.
HISAMATSU NAKANO
Hosei University
274
RECORDERS
RECEPTION, DIVERSITY. See DIVERSITY RECEPTION.
RECLOSERS. See CIRCUIT BREAKERS.
RECOGNITION, FACE. See FACE RECOGNITION.
RECOGNITION OF HANDWRITING. See ONLINE
HANDWRITING RECOGNITION.
RECOGNITION, SPEAKER, SPEECH, VOICE. See
SPEAKER RECOGNITION.
RECONFIGURABLE ARCHITECTURES. See CONFIGURABLE COMPUTING.
360
REFLECTOR ANTENNAS
REFLECTOR ANTENNAS
Reflector antennas have been of importance for decades in
several areas of electrical engineering, ranging from telecommunications and radars to deep-space exploration and radio
astronomy. This is due to the high gain of reflector antennas,
typically above 30 dBi. If we extend the concept of a reflector
antenna to a reflecting mirror and view the human eye as the
feed antenna operating in receiving mode, reflector antennas
have been known for centuries. Optical astronomers have
long been using reflecting mirrors in telescopes to enhance
the visibility of stars, planets, and other celestial bodies.
The basic principle of operation of a parabolic reflector is
that all rays emanating radially from a point source located
at the focal point are reflected as a concentrated bundle of
parallel rays, which can propagate for very long distances
without loss due to speading. Inversely, incident rays parallel
to the axis of symmetry of the paraboloid are all reflected toward its focal point, which concentrates the received signal at
a single point. In that case, if the human eye or camera is
placed a little bit behind the reflector focal point, an image
with enhanced luminosity and definition is formed (Fig. 1).
However, reflector antennas can be designed to be wideband devices, not limited to operation at frequencies covered
by the spectrum of visible light. Radio telescopes, for example,
search for celestial radio sources over a wide range of frequencies (e.g., 300 MHz to 40 GHz). In this case, the radio sources
and corresponding frequencies are marked on charts ac-
Image
Figure 1. Basic principle of operation of a parabolic reflecting mirror. The paraboloid surface is formed by rotating the parabolic curve
about its axis of symmetry (s axis).
J. Webster (ed.), Wiley Encyclopedia of Electrical and Electronics Engineering. Copyright # 1999 John Wiley & Sons, Inc.
REFLECTOR ANTENNAS
(a)
(b)
S
(c)
S
(d)
Figure 2. The evolution of reflector antenna systems: (a) single axisymmetric reflector, (b) dual axisymmetric reflector, (c) single offset
reflector, and (d) dual offset reflector. The main reflectors are parabolic.
cording to their physical locations in the sky, forming maps
similar to the ones elaborated by optical astronomers. Feed
antennas are employed to receive the signals from the celestial radio sources at different bandwidths.
One of the first reflector antennas operating at radio frequencies was built by Hertz in 1888 and consisted of a sheet
of zinc of about 2 m by 1.2 m, molded as a parabolic cylinder
and illuminated by a dipole feed (1). Since then, reflector antenna technology has gradually evolved toward the state of
the art known today for the purpose of improving electrical
performance and/or simplifying mechanical structure (Fig. 2).
The most basic form is the single axisymmetric parabolic reflector shown in Fig. 2(a), which is still in widespread use
primarily at low frequencies and for low-cost applications.
Large reflectors frequently use an axisymmetric dual reflector
system with a parabolic main reflector, as shown in Fig. 2(b).
The subreflectors are hyperbolic (Cassegrain system) or elliptical (Gregorian system). These systems offer a shorter transmission line (or waveguide) run to the feed antenna and are
often used as earth terminal antennas in satellite communication networks.
Axisymmetric single and dual reflectors suffer from aperture blockage due to the presence of feed/subreflector and
supporting mechanical structures in front of the main reflector aperture. This problem is solved by using an offset system
with a main reflector that is a section of a parent reflector,
normally a paraboloid of revolution, as shown in Fig. 2(c) and
(d). Design and construction of offset reflectors are more elaborate than for their symmetrical counterparts.
361
Remarkable technological advancements were achieved
during World War II, as reflectors were widely employed in
radar and communication systems (2). However, it was only
with the proliferation of digital computers in the late 1960s
that the most accurate analysis and synthesis algorithms
were developed, especially the ones related to the configurations of Fig. 2(c,d) and (37). Closed-form analysis algorithms
are generally only applied to symmetrical reflectors (4,8).
In addition, substantial improvements on the electrical
performance of both axisymmetric and offset dual reflector
configurations were obtained with shaping algorithms, an effort only possible with efficient numerical processing combined with a solid knowledge of differential geometry and
electromagnetics (5,9,10). The axisymmetric dual shaped reflector was introduced in the 1970s (9) and is popular for large
earth station antennas. The offset dual shaped reflector has
reportedly achieved aperture efficiencies of about 85% (11)
and has been enjoying an increase in popularity. As a consequence, the analysis and design of reflector antennas is nowadays a specialized and unique area in applied electromagnetics, occupying many distinguished workers in industry and
academia.
For the past two decades, reflector antennas have been applied primarily to satellite communications and networks,
deep-space exploration, and electronics defense. The reflector
antenna carried by the Voyager spacecraft, for example, is a
dual-reflector antenna shaped for high gain (Fig. 3). Besides
specific designs applied to unique purposes, such as spacecrafts and radiotelescopes, reflector antennas are also being
produced on a very large scale for commercial applications, a
multimillion-dollar market directly related to the globalization of communication currently underway. In particular,
VSAT systems (very small-aperture terminals) are proliferating and connecting together branches of large corporations,
such as chains of stores, banks, and car manufacturers. The
VSAT market is expected to grow at a rate of 20% per year
(12).
Other examples of substantial economic importance are
the satellite-based cellular communication systems, such as
the Motorola IRIDIUM, in which well-defined multibeam coverage is required, and direct-to-home (DTH) satellite TV systems, such as Hughes DirecTV and others, which employ
small offset parabolic antennas to receive satellite signals.
Thus, reflector antennas are present in our lives as major
gateways for the exchange of information at home and, less
conspicuously, in defense systems. Reflector antennas can
therefore be considered one of the most successful electrical
devices of all time, in view of their importance in many modern engineering systems and applications, such as cellular
communications, satellite TV, and electronic defense, as well
as in the exploration of our galaxy and beyond.
THE PARABOLIC REFLECTOR ANTENNA AND
OTHER SINGLE-REFLECTOR SYSTEMS
Preliminary Considerations and Geometry
Single-reflector systems, such as the parabolic reflector antenna, consist of a reflecting surface illuminated by a feed
antenna, usually a horn. It is necessary to know the radiation
characteristics of the feed antenna in order to evaluate correctly the electrical performance of the reflector system. A
362
REFLECTOR ANTENNAS
Figure 3. Full-scale model of one of the twin Voyager spacecraft. Note at the center the dual-shaped, high-gain reflector
antenna employing a main reflector with a diameter of 3.66
m (12 ft). (Courtesy Jet Propulsion Laboratory. Copyright
California Institute of Technology, Pasadena, CA. All rights
reserved. Based on government-sponsored research under
contract NAS7-1260.)
more detailed discussion is presented in the section entitled
Feed Antennas. In the next few subsections we employ simple analytical models to describe the radiation properties of
feed antennas. Once the feed pattern is known, the total radiation pattern of the reflector system can be obtained using
the techniques described in the section entitled Analysis
Methods and Evaluation, in combination with the geometrical properties of the reflector itself, which is the main subject
of this and following subsections.
The general geometry of a parabolic reflector is shown in
Fig. 4, and all associate symbols are listed in Table 1. Figure
4 is a cross-section view of the three dimensional paraboloid,
which is formed by rotating the parabolic curve shown in Fig.
4 about its axis of symmetry (s axis). If the rotation is performed with H 0, an axisymmetric paraboloid of diameter
Dp is formed. Otherwise, an offset reflector is generated. We
limit our analytical analysis to parabolic reflectors with circular projected diameters. However, other shapes, such as elliptical, are also used in practice to some extent, especially as
earth-station antennas in communication links with synchro-
nous satellites. An effort is made to discuss this and other
types of reflector antennas.
Basic Equations
First we consider the axisymmetric parabolic reflector, which
is obtained from Fig. 4 for H 0 and f 0 (i.e., the feed
Table 1. Definitions of Symbols for Single Configuration
Symbol
Diameter of the projected aperture of the parabolic main
reflector (D Dp for an axisymmetric paraboloid)
Dp
Diameter of the projected aperture of the parent paraboloid
Offset of reflector center (H 0 for an axisymmetric paraboloid)
Paraboloid focal length
Point F
Focal point
Point A
Apex of the parent paraboloid
Point B
Point on main reflector that bisects subtended angle
viewed from focal point
Point C
Point on main reflector that projects to the center of the
projected aperture
Point P
Point on main reflector corresponding to the ray arising
from the peak of the feed pattern
Angle of feed antenna pattern peak relative to reflector
axis of symmetry s (feed is aimed at point P)
Value of f that bisects the reflector subtended angle (i.e.,
feed is aimed at point B)
Value of f when the feed is aimed at the reflector point
C corresponding to the aperture center
U L
Angle subtended by the parabolic main reflector as
viewed from the focal point
U
Zf
D/2
C
B
Dp/2
D/2
H
L
L
Xf
S
F
Figure 4. Geometry for the axisymmetric (H 0) and offset parabolic reflector. See Table 1 for definitions of parameters.
Definition
REFLECTOR ANTENNAS
antenna main beam peak is aimed at the reflector apex, point
A). For this particular case, Dp is the diameter of the projected
aperture and the aperture plane is the xf yf plane, in which we
define polar coordinates (f , f ). Furthermore, we associate
the spherical coordinates (rf , f , f ) with the rectangular system xf yf zf shown in Fig. 4. The parabolic curve can then be
expressed at any f as
rf =
2F
= F sec2 f
1 + cos f
2
(1)
or
F rf cos2
f
=0
2
(2)
and the projection of rf onto the aperture plane is
f = rf sin f = 2F tan f
(3)
which yields f 0 at the reflector apex (f 0) and f
Dp /2 at the reflector edge (f 0 L U).
The axisymmetric paraboloid is completely specified in
terms of its diameter Dp and curvature rate F/Dp. The greater
is F/Dp, the flatter is the reflector. Common values are usually
between 0.25 and 1.0. At the reflector edge, Eq. (3) becomes
Dp /2 2F tan 0, which yields
0 = 2 tan1
1
4F/Dp
(4)
source located at the focal point of a parabolic reflector will
be constant across the aperture plane after reflection. This
result yields another very important property of parabolic reflectorsthat is, a parabolic reflector illuminated by a feed
antenna with a unique phase center located at the focal point
produces a uniform phase distribution across the aperture
plane. As already seen, the beam is also collimated, forming
a section of a plane wave. Nevertheless, the amplitude distribution is not uniform. In general, reaching a maximum at the
center of the projected aperture and decreasing toward the
edges of the axisymmetric paraboloid.
These basic properties make parabolic reflectors so widely
used as reflector antennas. Although herein derived for the
axisymmetric paraboloid, they are also valid for the offset
case (i.e., H 0 in Fig. 4). Offset reflectors offer significantly
reduced aperture blockage, as the feed is not directly in front
of the reflector, although it is still located at the focal point
F, yielding higher gains than do axisymmetric reflectors of
similar aperture sizes. From Fig. 4 we see that the feed needs
to be tilted by an angle f in order to direct its pattern toward
the offset reflector; otherwise large spillover (i.e., feed radiation missing the reflector) and associated gain loss are introduced. In many systems, the feed pointing angle f is set
equal to the angle that bisects the reflector, B, or to the angle
pointed toward the center of the projected aperture, C. The
influence of the feed pointing angle f on the electrical characteristics of offset parabolic reflector antennas is discussed in
the section entitled Design of Axisymmetric and Offset Parabolic Reflector Antennas. The angles shown in Fig. 4 are obtained from the following relations:
The unit vector n
normal to the parabolic surface, can be
found by normalizing the gradient of Eq. (2) and is given by
L = 2 tan1
4H Dp
4F
n = r f cos f + f sin f
2
2
U = 2 tan1
1
4F/Dp
(5)
L + U
2
H
C = 2 tan1
2F
B =
The angle i between the surface normal, given by Eq. (5),
and an incident ray coming from the focal point can then be
calculated from
cos i = r f n = cos
f
2
(6)
cos r = cos
f
= z n
2
(7)
where z rf cos f f sin f . Equation (7) shows that all
rays coming from the focal point F are reflected by the parabolic surface as a collimated beam parallel to the z axis, which
is coincident with the s axis for the axisymmetric reflector.
Thus the total path length from all rays coming from the focal
point F to the aperture plane is given by
rf + rf cos f = rf (1 + cos f ) = 2F
(8)
where Eq. (1) was employed in the derivation. Equation (8)
shows that the total path length is constant, and we conclude
that the phase distribution of a wave coming from a point
(9)
(10)
(11)
(12)
with
Dp = D + 2H
Finally, the angle r between the correspondent reflected ray
and surface normal can be determined by enforcing the law
of reflection on the reflector surface; that is, r i. Thus
363
(13)
As we see, there are many parameters necessary to specify
completely a parabolic reflector antenna. However, before we
design reflector systems, it is necessary to study a few of their
characteristics and properties. This knowledge is essential for
selecting appropriate configurations for specific applications.
Long-distance and frequency-reuse communication systems,
for example, require antennas with high gain and low cross
polarization. Reflector antennas are particularly suitable for
such applications due to their high gain. However, an indepth understanding of their depolarization characteristics is
necessary in order to achieve designs that guarantee a suitable isolation between orthogonally polarized channels in frequency-reuse systems. This and other properties that are almost exclusive to reflector antennas are discussed next.
Cross Polarization, Beam Squint, and Beam Deviation
Cross Polarization. Polarization is a basic characteristic of
an electromagnetic wave and describes the motion of the elec-
REFLECTOR ANTENNAS
(14)
where k is the free-space wave number 2/ and
C(f ) = G0 10 (FT/20)( f / f0)
(15)
where FT is the feed taper in decibels at f f0 and the gain
normalization constant G0 is found by numerical integration
of Eq. (15); see Ref. 14 for further details. We first examine
the axisymmetric reflector of Table 2 using a balanced feed as
described by Eq. (14), which is purely linearly polarized (LP)
along the x axis, with the Gaussian pattern of Eq. (15) yielding a 10 dB beamwidth of 70 (i.e., FT 10 and f0 35). A
yf -polarized feed pattern can be obtained from Eq. (14) by replacing the argument f with f /2. In addition, a circu-
Table 2. Axisymmetric and Offset Parabolic Reflector
Configurations
Reflector Configuration
Axisymmetric
Offset
Shape:
Projected diameter D:
Parent reflector diameter Dp :
F/Dp :
Offset of reflector center, H:
Parabolic
171
171
0.3
0
Parabolic
85.5
171
0.3
42.75
Feed Configuration (On Focus)
Polarization:
Pattern shape:
Gain Gf (dBi):
10 dB beamwidth (deg):
Feed angle f (deg):
Linear (xf)
Gaussian; Eqs.
(14) and (15)
14.04
70
0
Linear (xf)
Gaussian; Eqs.
(14) and (15)
14.04
70
39.81
0.02
70
0.01
0.00
70
3
90
90
70
70
6
90
0.01
10
e jkr f
C(f )[f cos f f sin f ]
rf
XPOL
Ef =
COPOL
0.03
66
tric field vector at a fixed point in space as a function of time.
The polarization of an antenna is the polarization of its radiated wave when operating in the transmitting mode. Generally, the polarization of any antenna system can be decomposed into two orthogonal components in the far field,
referred to as copolarization and cross polarization. In the
particular case of reflector antenna systems, the copolarization is usually taken to be the polarization presented by the
feed antenna illuminating the reflector. As a consequence, the
cross polarization is orthogonal to the feed-antenna main polarization. This agrees with Ludwigs third definition of cross
polarization (13) and is the one herein employed. The crosspolarization (XPOL) level is defined quantitatively as the ratio of the peak in the cross-polarized radiation pattern to the
peak value of the copolarized pattern (i.e., the main beam
peak), usually expressed in decibels.
As previously mentioned, reflector antennas cannot be
properly evaluated without first describing the feed antenna.
A detailed discussion about modeling feeds is presented in the
section titled Feed Antennas. Here we employ an analytical
model that, despite its simplicity, approximates reasonably
well the copolarized radiation properties of feeds usually encountered in practice, such as conical corrugated horns. The
radiation pattern Ef of an idealized balanced feed (i.e., the
primary radiation is symmetric in f ) with a fixed phase center can be described by (3)
v = sin sin
364
20
0.02
40
0.03
0.03
0.01
0.01
u = cos sin
0.03
Figure 5. Contour plots in decibels of the computed co- and crosspolarized patterns of the 171 diameter axisymmetric parabolic reflector specified in Table 2.
larly polarized feed is obtained by combining the xf -polarized
pattern in Eq. (14) with a yf -polarized pattern that is in phase
quadrature (i.e., multiplied by a factor of j). The physical optics portion of the commercial code GRASP (general reflector
antenna synthesis package, TICRA Engineering) is used to
compute the radiation patterns (14). The physical optics formulation applied to the analysis of reflector antennas is discussed in the section entitled Analysis Methods and Evaluation.
For the axisymmetric configuration of Table 2, the computed level of XPOL displayed in Fig. 5 is very low (a maximum of 65.35 dB below the main beam peak of 48.62 dBi).
Typically, the feed assembly and supports, although not
taken into account by the computer simulations, will create
more XPOL than this. Although illustrated for a particular
case, this is a general result (9), and we conclude that reflector-induced XPOL in axisymmetric reflectors illuminated by
balanced feeds is often negligible. In addition, according to
Fig. 5, the XPOL peaks are all located in the 45 planes.
The cross-polarization behavior of offset reflectors is illustrated with a derivative of the 171 axisymmetric parent reflector of Table 2. A portion of the upper half of the axisymmetric reflector is retained, so that the offset reflector with a
85.5 projected aperture diameter is just fully offset (i.e., the
bottom of the reflector just touches its axis of symmetry). If
the feed remains pointed at the apex of the parent paraboloid
(i.e., f 0), negligible XPOL is generated (15). However,
this leads to large spillover and associated gain loss. Therefore, in practice the feed is tilted to direct its pattern toward
the reflector, resulting in the introduction of high XPOL.
The offset configuration of Table 2 is not symmetric about
the yz plane, and therefore XPOL is not canceled in this plane
as in the axisymmetric case. In fact, it is exactly in the yz
plane that the peak XPOL levels occur. However, reflector
REFLECTOR ANTENNAS
50
stantial XPOL. However, beam squint does occur (9,16); that
is, the main beam peak squints off of the reflector axis in the
plane perpendicular to the plane of symmetry (i.e., beam
squint occurs in the yz plane of Fig. 4). The beam squints to
opposite sides depending on the sense of the CP. Beam squint
can be a major problem in satellite and deep-space communications if not carefully taken into account. A practical formula
for the prediction of the beam squint angle S in offset parabolic reflectors with on-focus CP feeds is (16)
COPOL
XPOL
50
0.02
0.00
3
2
v = sin sin
2
0.01
x
50
3 6
35
2
0.01
S = sin1
10
35
0.02
20
50
50
0.03
0.03
0.01
0.01
u = cos sin
0.03
Figure 6. Contour plots in decibels of the computed co- and crosspolarized patterns of the 85.5 diameter just fully offset parabolic
reflector specified in Table 2.
symmetry is still present about the xz plane, and no substantial XPOL occurs in that plane. These results are demonstrated using the GRASP code for the offset reflector example
with the XPOL contour plot shown in Fig. 6, for which the
feed has a pointing angle of f B 39.81, computed according to Eq. (11). The feed again has the pattern given by
Eqs. (14) and (15), with a 10 dB beamwidth of 70. The computed peak XPOL is 22.4 dB relative to the copolarized
beam maximum of 47.39 dBi. Figure 6 indicates that the copolarized pattern is still symmetric, and the XPOL peaks are
located at the copolar 6 dB contour line.
This example is typical of single offset reflectors and shows
that single offset paraboloids illuminated by conventional
feeds are limited by XPOL performance (9). It is worth noting
that cross polarization arises from the reflector curvature and
from the tilting of the feed. A planar reflector, for instance,
does not depolarize an incident field coming in a direction perpendicular to the reflector. Thus we see that the XPOL decreases as the reflector curvature rate, F/Dp, increases. However, this reduction is not significant in offset reflectors, on
account of the substantial feed tilting normally encountered
in practice (15).
A XPOL level above 22 dB is often unacceptably high
(3,7). In the section titled Conditions for Minimizing Cross
Polarization in Offset Cassegrain and Gregorian Systems we
discuss procedures to reduce XPOL in offset parabolic reflectors. Next we discuss an important property of parabolic reflector antennas that is inherently related to XPOL.
Beam Squint. As we have seen, offset reflectors offer significantly reduced aperture blockage but introduce high
XPOL when illuminated by a LP feed. On the other hand,
offset parabolic reflectors fed by a circularly polarized (CP)
feed presenting a balanced radiation pattern do not have sub-
sin
f
(16)
2Fk
where F is the focal distance and k is the free-space wave
number 2/ . A negative S means that the beam is squinted
toward the left (left-hand CP feed), and a positive S means a
squint to the right (right-hand CP feed). Equation (16) shows
that the amount of squinting is inversely proportional to the
focal distance F. That is, longer-focal-length reflectors experience less beam squint. If the feed is displaced from the focal
point, an equation similar to Eq. (16) is derived, as in Ref. 17.
In offset parabolic configurations illuminated by off-focus CP
feeds, beam squint occurs simultaneously with an effect
called beam deviation, treated in the next subsection. Due to
sense reversal encountered upon reflection from the main reflector, the sense of the far-field radiation is opposite to that of
the feed (18). For example, a right-hand circularly polarized
(RHCP) feed produces a left-hand circularly polarized (LHCP)
radiation in the reflector far field, as illustrated next.
We consider as an example a just fully offset configuration
with a diameter D 18.8 , F/Dp 0.25, and H 9.4 ,
illuminated by a CP feed with a pattern of 10 dB beamwidth
70. This geometry was selected because it is used in VSAT
applications at 18.5 GHz and measured data are available
(3,9). Figure 7 shows copolarized pattern cuts computed by
GRASP (14) in the yz plane with opposite-sense CP feeds.
For a RHCP main beam (the feed is LHCP) the squint is to
the left, as observed in Fig. 7. Likewise, the LHCP main beam
squints to the right. From Fig. 7 we note that the angle be-
RHCP beam
LHCP beam
0
Relative pattern (dB)
0.03
365
10
0.7
20
30
40
10
2
2
Angle (deg)
10
Figure 7. Computed RHCP and LHCP far-field patterns of an 18.8
just fully offset parabolic reflector. Note the beam squint effect. The
LHCP (circles) and RHCP (crosses) cross-polarized patterns are associated with the RHCP and LHCP beams, respectively.
366
REFLECTOR ANTENNAS
tween the two beams (total beam separation) is 0.700, which
is in agreement with the value of 0.686 (2 s) from Eq. (16).
The reported measured value (3,9) is 0.750. Finally, Fig. 7
also shows that circular XPOL is low (maximum of 42.71 dB
below the gain of 33.88 dBi for any of the feed polarizations).
The absence of circular XPOL in offset paraboloids with onfocus feeds is a general result, not limited to just fully offset
paraboloids (17).
Although circular XPOL is low, there are substantial LP
cross-polarized fields present at any given instant of time, in
both the aperture distribution and the far-field pattern of the
offset reflector (19). This and the fact that the orthogonal components of the incident field are not in phase (which is the
case for a circularly polarized feed antenna) are the two necessary and sufficient conditions to generate beam squint (17).
We now present a brief explanation of the beam squint generation mechanism.
The electric field components on the left side of the reflector (y 0 in Fig. 4) always lead or lag in phase relative to
the ones on the right side, depending on whether the primary
field is LHCP or RHCP (19). This leads to a phase slope condition across the aperture, which squints the main beam to the
left (negative angles in Fig. 7) or to the right (positive angles
in Fig. 7). To illustrate the process, consider any two points
in the projected aperture of the offset paraboloid that are
equidistant from the reflector plane of symmetry. If the feed
is LHCP, the electric field at those points rotates counterclockwise (RHCP main beam), as shown by Fig. 8. Thus, the
electric field vector to the left is leading the one to the right,
and as a final result (considering the influence of all points)
the beam squints to the right (the view in Fig. 8). This is
equivalent to a negative S in Eq. (16), or to a squint to the
left in the yz plane of the reflector coordinate system (negative
angles in Fig. 7).
Beam Deviation. When a feed is laterally displaced from
the focal point of a reflector, either axisymmetric or offset, the
pattern main beam is scanned to the opposite side of the reflector axis. This is referred to as beam deviation, and it
arises from a tipping of the aperture field phase plane relative
to the reflector aperture plane. The main beam direction determined from these considerations is herein referred to as
the reference axis; it is tilted from the reflector axis (i.e., the
z axis in Fig. 4) according to the amount of feed displacement.
Note that for an axisymmetric configuration the reflector and
reference axes intersect at the apex of the parent paraboloid
(point A in Fig. 4). If a CP feed is displaced from the focal
point of an axisymmetric or offset paraboloid, both beam
squint and deviation are present at the same time. Thus the
amount of beam squint should be added to the reference axis
in order to determine accurately the final position of the
main beam.
For a feed displacement f along the positive yf direction in
Fig. 4, the reference axis is tilted in the opposite direction
from the reflector axis by an angle D, computed according to
D = BDF tan1
f
F
(17)
where BDF is the beam deviation factor, which can be approximately determined for small feed displacements f , in axisymmetric and offset paraboloids, by (3,20)
1 + 0.36 4
BDF =
1+ 4
F
Dp 2H
F
Dp 2H
2
2
(18)
The nonuniform aperture phase distribution introduced by
the feed displacement is responsible for the beam deviation,
as already commented, but also leads to pattern deterioration, which includes beam broadening and null filling. These
effects increase as f becomes larger, resulting in substantial
gain loss. Nevertheless, scanning of the beam by feed displacement is a technique widely used in practice, especially
when it is difficult or impossible to move the reflector itself.
Summary of Parabolic Reflector Properties. There are a large
number of possible reflector geometries, feed types, locations,
and polarizations. Representative configurations were examined in the previous subsections to provide specific values as
well as general conclusions. The many possible configurations
employing a parabolic reflector are summarized in Table 3
together with XPOL, beam squint, and beam deviation effects. Table 3 shows that unbalanced feeds (i.e., the primary
radiation pattern is not symmetric) usually generate substantial XPOL independent of the feed polarization or reflector
configuration. Beam squint normally occurs with circularly
polarized feeds, except for small reflector antennas (i.e., D
12 and F/Dp 0.25), where it can also be present with a
linearly polarized illumination (21). Also, displacing the feed
from the focal point normally generates XPOL and beam deviation. Table 3 presents a complete overview of the various
depolarization and beam-pointing properties of single parabolic reflector antennas, which is of fundamental importance
for designing effective reflector configurations.
Design of Axisymmetric and Offset
Parabolic Reflector Antennas
S
Z
Figure 8. Beam squint generation mechanism.
Design of reflector antennas presents a challenge to the antenna engineer, especially in that so many parameters are
available for adjustment. The main purpose of this subsection
is to present a complete procedure to design axisymmetric
and offset reflectors, as well as to provide some insights into
the basic tradeoffs inherent in the process.
REFLECTOR ANTENNAS
367
Table 3. Polarization and Beam Pointing Characteristics of Single Parabolic Reflectors
Reflector
Geometry
Axisymmetric
Location
On Focus
Feed Type
Balance
Unbalanced
Off Focusa
Balanced
Unbalanced
Offset ( f 0)
On Focus
Balanced
Unbalanced
Off Focusa
Balanced
Unbalanced
Cross
Polarization
Beam
Squint
Linear
Circular
Linear
Circular
Linear
Circular
Linear
Circular
No
No
Yes
Yes
Yes
Yes
Yes
Yes
No
No
No
No
No
Yes
No
Yes
Linear
Circular
Linear
Circular
Linear
Circular
Linear
Circular
Yes
Nob
Yes
Yes
Yes
Yes
Yes
Yes
Nob
Yes
Nob
Yes
Nob
Yes
Nob
Yes
Beam deviation also occurs; see the subsection entitled Beam Deviation.
Except for small reflector antennas (i.e., D 12 and F/Dp 0.25); see Ref. 21 for further details.
Within this context, we start by examining the influence of
the feed pointing angle f on the gain G, sidelobe level (SLL),
and cross polarization of offset reflectors having H D/2 (i.e.,
general offset reflectors). The feed pointing angle is a parameter of significant influence on the electrical behavior of reflector antennas and provides many insights into XPOL behavior.
Scattering from supporting structure (struts) is not included,
but for an offset configuration it is typically negligible. An
offset reflector is chosen with a diameter D 85.5, F/Dp
0.3, and offset distance H 5D/8, corresponding to a geometry that is popular in VSAT applications. The balanced feed
pattern employed to illuminate the reflector is xf -polarized, as
modeled by Eqs. (14) and (15), with a 10 dB beamwidth of 70.
Figure 9 shows the gain, SLL, and XPOL computed with
GRASP (14) as the feed pointing angle, f , is varied for the
selected configuration. We note that the gain curve has a
broad peak, and the sidelobe level is not sensitive to feed
pointing except at very small angles (f 30). Only the nearin sidelobes were considered in this analysis, and therefore
spillover from the feed, which is particularly high for f 40
and f 60, was not included in Fig. 9. The XPOL, however,
decreases with decreasing f . Although illustrated for a particular case example, this is a generic and important result,
showing that in the limiting case where the feed is pointed at
the apex of the parent paraboloid (i.e., f 0), negligible
XPOL is generated (15). However, this leads to large spillover
and associated gain loss. Therefore, in practice the feed is
tilted to direct its pattern toward the reflector, which introduces high XPOL. For approximate designs, such as are often
sufficient in practice, the feed can be aimed within the range
40 f 60 in order to keep spillover losses (and consequent gain loss) reasonable. For the particular configuration
herein considered, peak gain operation is achieved with f
47, which yields G 47.52 dBi.
A classical design scenario has now emerged. The feed
pointing angle f is reduced until desirable cross-polarization
performance is achieved or until the gain is reduced as far as
can be accepted. If, on the other hand, the SLL is a critical
parameter, f can be optimized to yield nearly the lowest SLL
over a practical range of angles, with only small reductions in
G and the XPOL (15). This is discussed in more detail in the
procedure for designing parabolic antennas, presented next.
We now have a reasonable understanding of the basic concepts of reflector antennas. The following steps summarize a
procedure to design axisymmetric and offset reflector antennas:
1. Determination of Reflector Diameter. The following
equation is very useful to estimate a value of D to
15
49
48
20
47
46
25
45
44
30
43
Gain (dBi)
XPOL and near-in SLL (dB)
Polarization
42
35
Gain
XPOL
SLL
40
41
40
39
30
40
50
60
Feed pointing angle f (deg)
70
Figure 9. Gain, sidelobe level (SLL), and cross-polarization level
(XPOL) as a function of the feed pointing angle f for an offset parabolic reflector with a 85.5 diameter.
368
REFLECTOR ANTENNAS
decibels at a direction f as
achieve a required gain G (20):
g[not in dB] =
ap
4Ap
=
ap
2
D 2
(19)
where Ap is the physical area of the antenna aperture
and ap is the aperture efficiency, typically 0.65 (65%) for
many parabolic reflector systems used in practice. Note
that the gain in decibels is G 10 log g.
2. Determination of Offset Distance. The offset distance H
controls the amount of blockage caused by the feed and
supporting structure on the reflector projected aperture.
Many reflectors nowadays are just fully offset paraboloids (H D/2) (i.e., the bottom of the reflector just
touches its axis of symmetry). This configuration avoids
the blockage from the feed supporting structure (struts)
and waveguide, although part of the feed aperture is
still directly in front of the reflector. Nevertheless, the
total blockage area is still significantly smaller than the
one presented by axisymmetric configurations (H 0).
Values of H larger than D/2 can overcome blockage, but
also increase the total volume occupied by the reflector,
which in some cases is undesirable. In addition, the
manufacturing process and associated adjustments become more difficult as H increases.
3. Selection of Reflector Curvature. Values usually encountered in practice for the reflector curvature, F/Dp, are
between 0.25 and 1.0, where Dp is given by Eq. (13).
Higher values ease the manufacturing process (i.e., the
reflector is flatter) but require a narrower feed pattern
to illuminate the reflector, which results in larger feed
antennas. A typical value nowadays is F/Dp 0.3,
which yields a compact design.
4. Determination of Feed Pattern. An important parameter
for determining the necessary feed pattern is the reflector illumination RI in the aperture plane, which, in decibels, is given by
RI = 20 log cos2
f + f
+ 20 log cosq f
2
(20)
where the first term in the right is normally referred to
as spherical spreading loss and takes account of the
power spreading due to spherical propagation of the
wave between the focal point and the parabolic reflector
surface. The second term in the right side of Eq. (20) is
the normalized feed pattern in decibels of
C(f ) = cosq f
(23)
A usual value for RI at the reflector edges, often referred to as the edge illumination EI, is 11 dB and
assures optimal gain performance for axisymmetric paraboloids (f 0 and f L U); see Ref. 20. In
offset reflectors, L U, but an specified value of EI at
both edges can still be obtained by solving for f the
equation formed by imposing that q(f L) q(f
U). Once f is determined, the parameter q can be calculated directly from Eq. (23) for the specified value of
RI (i.e., EI), at either f L or f U, since they now
should yield the same result. Under this condition, the
edge illumination is called balanced and yields nearminimum sidelobe levels over practical ranges of feed
pointing, with only small penalties in gain and XPOL
(15). A graphical technique to determine f for the same
condition was introduced in Ref. 15 and is especially indicated when only measured feed patterns are available. In practice, however, it is common to find offset
systems employing f B, Eq. (11), or f C, Eq. (12).
For either case the edge illumination is in general unbalanced. As a consequence, different values of q are
obtained with Eq. (23), depending on whether f L
or f U. A simple arithmetic mean can then be taken
to specify the required feed pattern. Although approximate, this simple procedure yields reasonably good results in practice and is well suited for our purposes.
Once the parameter q is determined, Eqs. (22) and (19)
can be used to estimate the aperture diameter, or area,
of the required feed antenna. A typical value of ap
0.55 (55%) can be used for feed horns. A more exact
approach is available when the feed antenna is an openended rectangular waveguide of wide and narrow dimensions a and b, or an open-ended circular waveguide
of radius a. For those cases, the waveguide dimensions
can be determined from Eq. (22) with gf 32ab/2
(ap 0.81) or gf 10.5a2 / 2 (ap 0.84). If the result
indicates a feed antenna with an aperture considered
too large, a higher value of RI (i.e., EI at the reflector
edges) should be employed to avoid unnecessary
blockage. For an offset reflector, a larger value of H can
also be tried and the whole procedure needs to be repeated.
(21)
which is a pattern model widely used in practice with
Eq. (14), as an alternative to the one given by Eq. (15).
The main advantage of Eq. (21) over Eq. (15) is that the
directivity of the feed, or its gain if ohmic losses are not
taken into account, can be found analytically in closed
form with (4)
gf [not in dB] = 4q + 2
RI
+ f
log cos2 f
2
q = 20
log cos f
(22)
where gf is the gain of the balanced feed modeled by
Eqs. (14) and (21). The parameter q can be obtained
from Eq. (20) for a required reflector illumination RI in
The aforementioned design procedure was successfully employed to obtain the preliminary design of a 1.6 m just fully
offset paraboloid, built and tested at the University of Brasilia for satellite TV reception at C band (Fig. 10). Nevertheless, the use of a suitable computer code before the manufacturing process is highly recommended to confirm the electrical
performance of the reflector system. Techniques often implemented in numerical codes for the analysis of reflector antennas are discussed in the section entitled Analysis Methods
and Evaluation.
As a final note, we mention that surface distortions from
ideal parabolic shapes are normally introduced in any manufacturing process. Random reflector surface errors can be al-
REFLECTOR ANTENNAS
Figure 10. Just fully offset parabolic reflector antenna for satellite
TV reception at C band, built and tested at the University of Brasilia.
(Courtesy of Carlos Muller, University of Brasilia.)
lowed for by augmenting the aperture efficiency in Eq. (19) to
include a random surface error efficiency, rs (20):
g[not in dB] =
ap
rs
D 2
(24)
rs = e
inder, for example, is generated by displacing the parabolic
curve along the y axis. This yields a focal line in contrast to
the focal point of the parabolic reflector. A feed line or a linear
array of feed antennas must be placed along the focal line for
proper illumination of the reflector. Another example is the
parabolic torus, formed by rotating the parabolic curve with
respect to an axis perpendicular to the s axis of Fig. 4. The
axis of revolution is, in general, placed at a distance from the
apex greater than F and is confined to the plane shown by
Fig. 4. Thus the parabolic torus possesses a focal arc and can
be visualized as a curved parabolic cylinder. Multiple feeds
are normally employed to illuminate different sections of the
reflector, producing independent beams with a single reflector
antenna, a configuration widely used in satellite communications.
An example of a reflector antenna not generated by a parabolic curve is the spherical reflector, which is a section of a
sphere. The Arecibo Observatory, located in Puerto Rico, employs an axisymmetric spherical main reflector with a diameter of 305 m for radio astronomy, ionospheric research, and
radar investigation of celestial bodies. A line feed is used because parallel rays coming from space are reflected along the
reflector axis (22). This is in contrast to the parabolic cylinder,
where the focal line is perpendicular to the reflector axis of
symmetry. However, the spherical reflector does not merely
possess a focal line, but rather a focal region where feeds can
be placed (23). Dual-offset shaped reflectors have also been
proposed and used as feed systems in order to correct the nonuniform phase distribution characteristic of spherical reflectors (22). Dual- and multiple-reflector systems are treated in
the next section.
MULTIPLE-REFLECTOR ANTENNA SYSTEMS
Cassegrain, Gregorian, and Multiple-Reflector Systems
where
(2 s ) 2
369
(25)
The parameter S is the root mean square (rms) surface deviation and is approximately one-third of the peak-to-peak error
(20). Detailed information on other efficiency factors left in
ap can be found in the literature (20). Mass production of reflectors using presses and molds to shape glass fibers and
other components normally yields S 0.01 (rs 0.98 or
98%). Errors larger than that can cause significant gain loss
and pattern deterioration due to the introduction of phase errors. Machined metal reflectors are very accurate and have
S 0.04 mm. Reflector antennas are wideband devices by
nature and are limited at the upper frequencies by the
smoothness of the reflector surface and at the lower ones by
the reflector electrical size. The operating bandwidth of a reflector antenna in practice is normally set by the bandwidth
of the feed antenna employed to illuminate the reflector. Although herein illustrated for parabolic reflectors, random surface errors occur in any type of reflector antenna, such as the
ones discussed next.
Other Single-Reflector Systems
A few other types of parabolic reflectors can also be obtained
from Fig. 4, including offset configurations. The parabolic cyl-
Dual-reflector systems, such as the ones in Fig. 2(b, d), can
be formed by adding to the parabolic reflectors previously
studied a hyperbolic reflector (Cassegrain system) or an elliptical reflector (Gregorian system). Hyperbolic and elliptical reflectors have two focal points, F1 and F2, and are normally
referred to as subreflectors because they are smaller than the
parabolic main reflector.
It is known (3,24) that any ray coming from one of the focal
points, say F2, is reflected by an ellipsoid surface toward the
other focal point, F1. For the hyperbolic surface, the reflection
occurs such that the reflected ray appears to come from F1.
Thus, any circular cone of rays (i.e., a section of a spherical
wave) emanating from F2 and directed to an elliptical, or hyperbolic, subreflector will then be reflected as another circular
cone of rays with vertex at the other focal point F1. Furthermore, if F1 is coincident with the paraboloid focal point F, and
the feed is located at F2, a perfect circular cone of rays originated at F1 will illuminate the paraboloid. Due to the paraboloids reflecting property, a section of a plane wave will then
appear at the paraboloid aperture plane.
There are a few basic advantages of dual configurations
over single ones. They provide a shorter waveguide run to the
feed antenna. In addition, dual configurations present lower
noise when used as satellite earth terminals. This is due to
the limited noise introduced by the feed spillover beyond the
subreflector, given that it is now directed to cold sky, rather
370
REFLECTOR ANTENNAS
than hot earth as in the single-reflector case. Many satellite
communication networks employ dual configurations as earth
terminal antennas. Finally, the inclusion of the subreflector
introduces another degree of freedom, which can be used to
enhance electrical performance, such as by canceling XPOL
in offset systems and/or prescribing the main aperture amplitude and phase distributions in dual shaped reflectors.
Within this context, dual shaped reflectors can be used to
illuminate larger reflectors, such as the Arecibo spherical reflector (22), forming a multiple-reflector system. In this case,
they are shaped to correct the phase aberration characteristic
of spherical reflectors. They can also be used to enhance the
scanning properties of spherical reflectors (25).
A multiple-reflector system can also be formed with a parabolic main reflector, which is illuminated by a sequence of
hyperboloids and/or ellipsoids employed as subreflectors. The
subreflectors must be properly arranged so that a spherical
wave is formed after each reflection. It can be shown (26) that
such a multiple-reflector system is always equivalent to a single parabolic reflector, normally referred to as the equivalent
paraboloid. This concept also applies to Cassegrain and Gregorian systems (24,26), and is especially useful to determine
the conditions for canceling XPOL in offset systems, as discussed next.
signs often require that an existing single-offset reflector
mold be used to construct the main reflector of a dual configuration. However, many such existing molds in industry are
for just fully offset geometries, which justifies the recent preference for Gregorian configurations. [To upgrade an existing
mold to a dual-offset Cassegrain system, a main reflector
other than just fully offset is normally required to avoid
blockage, because the hyperbolic subreflector, in contrast to
the elliptical one, is located above the axis of symmetry of the
parent main reflector, as shown by Fig. 2(d).] In addition, the
Gregorian configuration allows the main reflector also to be
used as a single focused configuration without the need of removing the subreflector; see the next subsection for further
details. Nevertheless, Cassegrain configurations have also
been widely used in many practical systems, and all conditions for minimizing XPOL herein discussed also apply to
them.
The general geometry of a dual offset Gregorian configuration is shown in Fig. 11, and the symbols are defined in
Table 4. Although not shown in Fig. 11, the main reflector
projected aperture is circular, such as the one in Fig. 4. The
subreflector employed in a Gregorian offset design is a section
of a parent ellipsoid described by the following expression:
y2S + z2S
(xS c)2
+
=1
2
( f S + c)
( f S + c)2 c2
Conditions for Minimizing Cross Polarization
in Offset Cassegrain and Gregorian Systems
The Cassegrain and Gregorian offset configurations of Fig.
2(d) can be optimized to cancel reflector-induced XPOL. We
focus our discussion on the Gregorian system, but all main
results herein presented are also valid for the Cassegrain system (3,2324). Although less compact, the Gregorian configuration has been increasily used in practical applications, especially due to the fact that it allows the main reflector to
have a just fully offset geometry (i.e., the bottom of the main
reflector just touches its axis of symmetry). Cost-effective de-
(26)
where all variables and coordinates are as defined in Fig. 11
and Table 4. It is worth mentioning that the projections of the
subreflector onto the ySzS and xSyS planes are ellipses.
As mentioned in the previous subsection, the geometry of
Fig. 11 is equivalent to a single parabolic system. Furthermore, we saw previously that if the feed pointing angle f is
coincident with the reflector axis of symmetry, no substantial
XPOL is generated. This condition can be satisfied for the
equivalent single paraboloid, provided that the original dual
D/2
Dp/2
z
ZS
fS
D/2
Xf
2c
A
AS
F1
F2
XS
DS
B
Zf
Figure 11. General geometry of the dual
offset Gregorian reflector antenna. The
symbols are defined in Table 4.
REFLECTOR ANTENNAS
taneously minimizes XPOL and spillover loss (i.e., feed radiation missing the subreflector):
Table 4. Definitions of Symbols for Dual Configuration
Symbol
D
Definition
Diameter of the projected aperture of the parabolic main
reflector
Dp
Diameter of the projected aperture of the parent paraboloid
Offset of reflector center
Paraboloid focal length
Point F1
Common focal point of the parabolic main reflector and
ellipsoidal subreflector
Point F2
Ellipsoid focal point; feed antenna location
Point A
Apex of the parent paraboloid
Point AS
Apex of the ellipsoidal subreflector
Point B
Point on subreflector that bisects subtended angle
viewed from F2 . Point B also results from the intersection of the ray coming from point C on the main reflector and the feed axis (zf)
Point C
Point on main reflector that projects to the center of the
circular projected aperture
Angle of feed antenna pattern peak after reflecting on
the subreflector relative to the main reflector axis of
symmetry (s)
U L
Angle subtended by the parabolic main reflector as
viewed from the focal point F1
DS
Height of the ellipsoidal subreflector
Subreflector eccentricity (0 e 1 for an ellipsoid)
Half of the ellipsoid interfocal distance
FS
Distance between a focal point and the closest ellipsoid
apex
Feed pointing angle measured relative to the ellipsoid
axis of symmetry (xs)
Angle between the ellipsoid and parent paraboloid axes
of symmetry (xS and s, respectively)
Angle between the main reflector and feed axes (s and zf)
Half the angle subtended by the subreflector as viewed
from the feed antenna location (ellipsoid focal point F2)
|e2 1| sin
(1 + e2 ) cos 2e
(27)
where e is the subreflector eccentricity (0 e 1 for an ellipsoid and e 1 for a hyperboloid), e c/( fS c). Equation (27)
is generally referred to as the Mizugutch condition and has
the following alternative form, known as the Dragone condition (24,26):
tan
e+1
=
tan
2
|e 1|
2
tan
=
2
e 1 2
e+1
tan
+ C
2
(28)
where the factor (e 1)/(e 1) is normally referred to as
the subreflector magnification M. Rusch (24) gave a condition
based on the same equivalent-paraboloid concept that simul-
(29)
where C is the angle subtended to the center of the main
reflector and is given by Eq. (12). The Rusch condition, Eq.
(29), can only be applied to dual systems employing a parabolic main reflector with a circular projected aperture, in contrast to the Mizugutch and Dragone conditions, Eqs. (27) and
(28), which can be applied to reduce XPOL in systems with
arbitrary projected apertures. Although more restrictive, the
Rusch condition, in addition to XPOL, also minimizes spillover loss, given that the resulting equivalent paraboloid is
constrained to be always axisymmetric (24). This yields the
feed axis of the original dual configuration pointing in the
direction that bisects the subreflector subtended angle, as
shown in Fig. 11. Enforcement of Mizugutch or Dragone conditions, in general, does not result on an axisymmetric equivalent paraboloid, which leads to high spillover loss even
though XPOL is kept to a minimum. In fact, the result in the
Rusch condition, Eq. (29), can be visualized as the one particular solution of Eq. (27) or (28) that yields an axisymmetric
equivalent paraboloid with f 0, thus simultaneously minimizing XPOL and spillover loss.
It is important to note that Eqs. (27) to (29) are effective
only in reducing the reflector-induced XPOL. A simple worstcase model for predicting the influence of feed XPOL in reflector systems is (18)
XPOLS = XPOLF + XPOLR
configuration of Fig. 11 satisfies the following relation (27):
tan =
371
(30)
where XPOLS, XPOLF, and XPOLR, are, respectively, the
cross-polarization levels of the total system, the feed, and the
reflector(s). The XPOL here is expressed as a field ratio (not
in decibels; it is 10(value in dB)/20). The simple result in Eq. (30)
shows that either the feed or the reflector XPOL can dominate
the system XPOL. Since dual offset configurations satisfying
any of the conditions in Eqs. (27) to (29) yield low reflector
XPOL, system XPOL is usually limited by feed XPOL. We
use, as an example, a low-cross-polarization dual offset Gregorian reflector antenna, employing a just fully parabolic
main reflector with a 2.4 m diameter. When a feed XPOL
value of 32 dB is included, the system XPOL computed by
GRASP (14) increases from 48.19 dB to 31.75 dB. Equation (30) yields 30.75 dB, which is in good agreement for
such a simple formula. In addition, Eq. (30) can be used to
predict a feed XPOL level required to attend a given specification of system XPOL.
As an illustration of using the minimum XPOL conditions
expressed in Eqs. (27) to (29), we consider the following offset
parabolic system.
The Green Bank Radio Telescope
The Green Bank Radio Telescope (GBT) will be the largest
fully steerable radio telescope in the world. It is currently under construction (as of January 1998) and is expected to be
completed by 1999 (Figs. 12 and 13). Its offset design provides
a clear 100 m diameter projected circular aperture. The GBT
structure can be pointed to view the entire sky down to a 5
372
REFLECTOR ANTENNAS
Figure 12. Construction site of the
Green Bank Radio Telescope reflector antenna. The 100 m main reflector consists
of 2000 solid panels. The structure can be
pointed to view the entire sky down to a
5 elevation angle and will be the largest
fully steerable radio telescope in the
world. (Courtesy of George Behrens, National Radio Astronomy Observatory.)
elevation angle, using a wheel-and-track mechanical design.
The reflecting surface consists of 2000 solid panels that can be
positioned using actuators behind the panels. A laser ranging
system will be used to determine the positions of the panels,
adjusting the surface accuracy with closed-loop control.
The GBT is connected to radiometers that can receive signals in several frequency bands. From 290 MHz to 1230 MHz,
the GBT operates as a single offset reflector using a feed assembly aimed directly at the main reflector. From 1 GHz to
45 GHz, it operates as a Gregorian dual offset reflector using
the feeds in the receiver room that are aimed at the ellipsoidal subreflector. The Gregorian configuration has the focal
points in the area between the subreflector and the main reflector, allowing the subreflector to remain fixed even when
the telescope operates in the single-offset-reflector mode. This
is not possible with a Cassegrain configuration (28).
We start by examining the GBT single-offset configuration,
with the characteristics listed in Table 5, employing the codes
Figure 13. Artwork of the Green Bank Radio Telescope reflector antenna. The dual offset Gregorian configuration employs an offset parabolic main reflector with a 100 m projected aperture
diameter. (Courtesy of George Behrens, National Radio Astronomy Observatory.)
REFLECTOR ANTENNAS
Table 5. GBT Single Offset Reflector Configuration and
Computed Performance Values
Main Reflector Configuration
Shape:
Projected diameter D (m):
Parent ref. diameter Dp (m):
Focal length F (m):
Offset of reflector center, H (m):
Computation:
Offset parabolic
100
208
60
54
PRAC
GRASP
373
tions and approximations inherent in each of these formulations, as well as on their intrinsic differences.
The GO technique yields the aperture fields, assuming
equal angles of incidence and reflection. The far-field patterns
can then be calculated using a Fourier transformation directly, which is equivalent to obtaining equivalent currents
and then integrating, as described later in this subsection.
With the use of image theory, it is necessary to know only the
electric field distribution over the reflector projected aperture, Er, which is computed from the incident electric field Ei
(i.e., the feed radiation), with (20)
Feed Configuration (On Focus)
Polarization:
Pattern shape:
Linear (xf)
cos4.58; Eqs.
(14) and (21)
13.08
77.92
42.77
Gain Gf (dBi):
10 dB beamwidth (deg):
Feed angle f (deg):
Linear (xf)
Gaussian; Eqs.
(14) and (15)
13.14
77.92
42.77
System Performance
Gain G (dBi)
Cross-polarization level,
XPOL (dB):
Sidelobe level, SLL (dB)
Aperture efficiency ap , %
82.87
21.54
82.79
21.56
26.72
78.48
27.21
77.05
GRASP (14) and PRAC (7). Further information on PRAC
(parabolic reflector analysis code) is presented in the section
entitled Numerical Implementation and Accuracy Evaluation. The performance values, also listed in Table 5, were
computed at 15 GHz in the plane normal to the plane of symmetry (i.e., the yz plane of Fig. 4). Note that the two codes
yield very similar results for this geometry. We note from Table 5 that the gain is 82.87 dBi and the XPOL is 21.54 dB
(61.33 dBi), as computed by PRAC.
To lower the XPOL, we upgrade the GBT single-offset system of Table 5 to a low-cross-polarization dual-offset Gregorian antenna according to Eq. (29). Design parameters,
such as the desired subreflector size, were obtained from Ref.
28. The resulting configuration is listed in Table 6 and agrees
with Ref. 28. New dual configurations employing the same
GBT offset main reflector of Table 5 can be obtained using
different design parameters, such as a new subreflector size
or feed configuration, as discussed in Ref. 29.
Table 6 also presents the performance values at 15 GHz
computed with GRASP in the same plane considered for the
single-offset configuration previously discussed. We note that
the XPOL is now 43.01 dB, more than 20 dB lower than the
XPOL of the single configuration in Table 5. However, a feed
antenna with high XPOL will likely degrade the total system
XPOL performance, as addressed in the previous subsection.
ANALYSIS METHODS AND EVALUATION
E r = 2(n E i )n E i
where n
is the unit vector normal to the surface; see Eq. (5).
Equation (31) assumes that at the point of reflection the reflector is planar and perfectly conducting. In addition, the incident wave from the feed antenna is treated locally as a
plane wave. These same assumptions are also used by the PO
technique to determine the surface currents, Js, over the reflector as follows:
J s = 2n H i
(32)
where Hi is the incident magnetic field from the feed antenna
and can be computed from Eq. (14), recalling that in the far
field H (r E)/ (where is the free-space characteristic
impedance). The PO approximation assumes that currents exist only over the side of the reflector directly illuminated by
the feed antenna.
Table 6. GBT Dual Offset Reflector Configuration and
Computed Performance Values
Main Reflector Configuration
Shape: Offset paraboloid
Projected diameter D: 100 m
Parent reference diameter Dp : 208 m
Focal length F: 60 m
Offset of reflector center, H: 54 m
Angle , 5.58
Subreflector Configuration
Shape: Offset ellipsoid
Projected height DS : 7.55 m
Parameter c of ellipse: 5.9855 m
Parameter fS of ellipse: 5.3542 m
Eccentricity e: 0.5278
Feed Configuration (On Focus; GRASP Calculation)
Polarization: Linear (xf)
Pattern shape: Gaussian, Eqs. (14) and (15)
Gain Gf : 21.31 dBi
10 dB beamwidth: 30
Angle : 17.91
Angle : 12.33
Geometrical and Physical Optics Formulations
In both the geometrical optics (GO) and physical optics (PO)
formulations, the ultimate goal is to determine equivalent
currents, which can then be integrated to obtain the far-field
patterns, a process well described in the literature on aperture antennas (20). We focus our attention on the assump-
(31)
System Performance (GRASP Calculation)
Gain G: 82.83 dBi
Cross-polarization (XPOL) level: 43.01 dB
Sidelobe level (SLL): 22.56 dB
Aperture efficiency ap : 77.76%
374
REFLECTOR ANTENNAS
The far-field pattern can then be determined by summing
the individual contributions of each current point over the
surface, taking into account the different amplitudes and
phases due to the excitation and spatial location. Antenna
theory shows that a unit point source of current radiates a
spherical wave, which is normally referred to as the freespace Greens function (ejkr /4r); see Ref. 20 for further details. In the limit as the current distribution becomes continuous, such as the one given by Eq. (32), the weighted sum of
spherical waves becomes an integral, yielding the radiated
patterns.
Note that the integration process for obtaining the patterns is the same as the one employed by the GO technique,
given that once the aperture distribution is determined from
Eq. (31), equivalent currents can then be obtained and integrated over the reflector aperture. This process is equivalent
to computing the Fourier transform of the aperture distribution given in Eq. (31). One difference between GO and PO is
that PO currents are determined over the reflector curved
surface and the GO equivalent currents over the planar projected aperture, with the latter already in a format more appropriate for integration through a Fourier transform. However, the use of a Jacobian transformation (3,4) maps the PO
currents over the reflector curved surface to the planar aperture, yielding the possibility of also using Fourier transformations for performing the integration. Analytical integration is
only possible for symmetrical reflectors (4,8), and numerical
techniques are normally required to evaluate offset reflectors,
as discussed in the next subsection.
The PO formulation is generally considered more accurate
than GO to evaluate offset reflectors, especially if XPOL assessment is a main concern. However, pattern accuracy as
determined from both techniques degrades beyond the main
beam and near-in sidelobes. The pattern in the far-out region
is dominated by diffraction effects, especially scattering from
the reflector and/or subreflector edges. This is taken into account by augmenting GO with the geometrical theory of
diffraction (GTD) or augmenting PO with edge currents
through the physical theory of diffraction (PTD); see Refs. 20
and 30 for details. However, the near-in pattern region is,
most of the time, the region of interest when analyzing highgain antennas such as the reflector antennas considered
here.
Numerical Implementation and Accuracy Evaluation
In this subsection we discuss one of many possible numerical
implementations of the PO formulation previously addressed;
see Ref. 4 for alternative procedures. Reflector surface currents are computed from Eq. (32) for the balanced feed model
given by Eqs. (14) and (21). A set of coordinate transformations, rotations, and translations, is necessary in order to describe the far-field patterns as a function of the reflector local
coordinate system xyz, given that the feed pattern is described as a function of the feed local coordinate system
xf yf zf . Although not shown, Eulerian angles (31) are employed for generality, and we mention that a solid background
in geometry and vector calculus is normally required for the
analysis of reflector antennas.
The procedure employs a Jacobian transformation, as discussed in the preceding subsection, and evaluates numerically the following integral (4) using a numerical procedure
based on the GaussZirnike integration method (32):
E (rr ) = j
e jkR
(I rr )
2 R
J (rr ) J e jkr rr ds
(33)
s
where rr a is shorthand for r(r a), and I rr is included to
remove the radial component (far-field approximation) (4).
The unit dyad I is equal to the identity matrix for our purposes, and the Jacobian transformation J (4,31) is employed
to allow the integral to be evaluated over the reflector planar
projected aperture s. However, the currents are still defined
over the reflector curved surface. In addition, the Jacobi
Bessel method (31) is used to express part of the kernel in
Eq. (33) as a sum over a set of orthogonal functions defined
on the antenna aperture. Within this context, numerical integration is necessary only to evaluate the coefficients of the
series expansion, which employs the modified Jacobi polynomials in the radial direction and a Fourier series in the circumferential direction.
The aforementioned procedure was implemented in the
code PRAC (7). PRAC is a user-friendly code developed by the
author to analyze axisymmetric and offset parabolic reflectors, and it yields the co- and cross-polarized radiated fields
with high accuracy and efficiency. PRAC is currently being
used by many universities and major industries worldwide,
and a freeware version of the code is expected to be distributed with the electronic version of this encyclopedia.
To evaluate the accuracy of the code, we select as a baseline configuration for analysis a just fully offset paraboloid
with a diameter D 85.5, F/Dp 0.3, and offset distance
H 42.75. The reflector illumination is modeled by the balanced feed described by Eqs. (14) and (21), with a 10 dB
beamwidth of 78 (q 4.57 yielding a feed gain of 13.07 dBi).
The offset reflector choice corresponds to a 1.8 m diameter
VSAT earth terminal antenna operating at 14.25 GHz, similar to the one shown in Fig. 14.
Figure 15 shows the computed co- and cross-polarized patterns and measured data for the example offset parabolic reflector in the plane normal to the plane of symmetry (i.e., the
yz plane). The XPOL is expected to be maximum at this plane,
as discussed in the subsection titled Cross Polarization. We
note from Fig. 15 that the results obtained with PRAC are in
good agreement with the measured data. The measured gain
of 46.78 dBi is about 0.8 dB below the computed gain of 47.60
dBi due to losses and system imbalances. The system XPOL
is also a little overestimated by the computer simulations for
this example. In fact, the measured system XPOL is 22.00
dB, whereas PRAC yields 21.27 dB. Nevertheless, PRAC
yields a valuable estimate on how the reflector system behaves electrically, showing the necessity of a numerical evaluation previous to the manufacturing process. It is worth mentioning that analysis of this same baseline configuration with
the physical optics portion of GRASP (14) yielded almost identical results (7), with the exact same locations for the nulls
and peak sidelobes and XPOL lobes, confirming the accuracy
of PRAC and PO analysis for evaluating offset reflectors.
As final notes on the analysis of reflector antennas, we
mention that the lower integration limit in Eq. (33) can be set
so as to allow for a circular area of blockage equivalent to
that normally caused by the feed and supporting structure in
axisymmetric reflectors. In addition, the integral in Eq. (33)
REFLECTOR ANTENNAS
375
can also be evaluated over areas s other than the circular, in
order to analyze reflectors with projected apertures such as
the elliptical one. Reflectors with elliptical apertures present
a far-field pattern with a main beam that is narrower in the
plane containing the major axis of the ellipse and are used in
practice to transmit signals to synchronous satellites. The
main advantage is a more compact design than for the full
circular aperture, offering less resistance to wind and saving
material during manufacture. The lower gain (i.e., wider
beam) in the plane containing the minor axis of the ellipse
does not degrade system performance for this particular application, given that only a single belt of synchronous satellites exists and therefore beam resolution is required only in
one plane. Finally, both GO and PO formulations can also
be used to evaluate dual and multiple-reflector systems. The
simplest procedure is first to determine the radiation pattern
of the system formed by the feed antenna and subreflector,
and then to use this result as the incident field on the next
subreflector or main reflector. It is also common to employ GO
for the subreflector analysis and then PO in the final step to
evaluate the main reflector for better accuracy. This combination saves computer time, as GO analysis is generally faster
than PO (20).
FEED ANTENNAS
Figure 14. Just fully offset parabolic reflector antenna with a projected aperture diameter of 1.8 m. (Courtesy of Nick Moldovan, Prodelin Corporation.)
COPOL (PRAC)
XPOL (PRAC)
COPOL (meas.)
XPOL (meas.)
Ef =
10
Relative pattern (dB)
We start by discussing analytical models that approximately
describe the electrical behavior of feed antennas usually encountered in practice. The simplest model is the balanced feed
given by Eq. (14), normally used with Eq. (15) or (21). Balanced radiation patterns can be obtained in practice with the
use of multimode horns, such as the Potter horn, and hybridmode horns, such as the conical corrugated ones (23,33). An
alternative version of Eq. (14) can be obtained for feeds presenting different pattern cuts in the E-plane (f 0) and Hplane (f 90). The feed patterns in these two planes are,
most of the time, all that is known. As in Eq. (14), we assume
that the feed is purely linearly polarized in the xf direction,
yielding (3)
e jkrr f
[ f CE (f ) cos f f CH (f ) sin f ]
rf
where CE(f ) and CH(f ) denote the feed-pattern cuts in the Eand H-planes, respectively. A yf -polarized feed pattern, as
well as circularly polarized ones, can be obtained from Eq.
(34) by introducing the modifications already suggested in the
subsection entitled Cross Polarization. Note that Eq. (34)
reduces to Eq. (14) for CE(f ) = CH(f ) C(f ).
Finally, we can approximate even further the electrical behavior of feed antennas, although still ideally modeled with a
fixed phase center, using the complex polarization ratio pr,
defined as
20
30
40
pr = XPOLF (cos + j sin )
50
3
(34)
0
Angle (deg)
Figure 15. Computed and measured radiation patterns at 14.25 GHz
of a 1.8 m single offset parabolic reflector antenna.
(35)
The quantity XPOLF (not in decibels) determines the feed
XPOL peak relative to the peak copolarized beam, and is the
difference in phase between the cross- and copolarized feed
patterns defined as
= phase(XPOLF ) phase(COPOLF )
(36)
376
REFLECTOR ANTENNAS
The final result, assuming a linear polarization in the xf direction, is
E CO (rr f ) = {[CE (f ) cos2 f + CH (f ) sin2 f ]
+ pr [CE (f ) CH (f )] cos f sin f }
e
( f cos f f sin f )
jkr f
(37)
rf
and
E CROSS(rr f ) = {[CE (f ) sin2 f + CH (f ) cos2 f ]pr
+ [CE (f ) CH (f )] cos f sin f }
e jkr f
( f sin f f cos f )
rf
(38)
where ECO and ECROSS are the co- and cross-polarized radiation
patterns, respectively. It is worth mentioning that Eq. (37)
reduces to Eq. (14) for CE(f ) CH(f ) C(f ). In addition, Eq.
(38) yields the cross-polarized peaks in the 45 planes, which
is consistent with measured data and more sophisticated theoretical models (33). Nevertheless, it is not our intention to
analyze feed antennas completely, but rather to present simple analytical models that can be useful to antenna engineers
as a first approximation to real feed patterns.
As discussed in the subsection titled Summary of Main
Results, balanced feeds yield improved performance when illuminating reflector systems. A type of balanced feed antennas widely used in practice is the corrugated conical horn,
also often referred to as a scalar horn (Fig. 16). Corrugated
conical horns present a phase center that is reasonably stable
with changing frequency (23,33), in addition to a copolarized
pattern that is nearly balanced for practical purposes and can
be well modeled by Eqs. (14) and (15) or (21).
The main purpose of using corrugations is to obtain the
same boundary conditions around the inside of the horn. For
corrugation depths of a quarter wavelength, the short circuit
at the bottom is transformed to an open circuit at the top of
the corrugation, yielding boundary conditions that appear to
be more uniform as the number of corrugations per wavelength increase (20). This yields a symmetric radiation pattern down to as low as 25 dB over a reasonably wide operational bandwidth, typically of 1.6 : 1 or more.
The cross-polarized pattern, however, may contain peaks
in the 45 planes, similarly to the unbalanced model of Eq.
(38). As discussed in the section entitled Multiple-Reflector
Antenna Systems, and as approximately modeled by Eq.
Figure 16. Conical corrugated horn. (Courtesy of Emilio Abud Filho,
Brazilian Telecommunications Center for Research and Development.)
(30), feed XPOL peaks appear in the far-field patterns of dualreflector systems, even with the enforcement of the conditions
given by Eqs. (27) to (29). To minimize feed XPOL, a careful
design and construction process must be performed. The corrugation depths must be set properly to achieve resonance,
taking into account all dimensions related to the corrugations
as well as the general geometry of the horn (23). The process
has been successfully accomplished in practice, often also employing sections of tapered and/or dual-depth corrugations,
yielding horns presenting balanced copolar patterns and
XPOL levels below 35 dB over practical operational bandwidths (23,34).
Finally, other types of feed antennas are also used in practice, such as wire antennas and pyramidal and sectoral horns
(20). The latter are used for illuminating reflectors with elliptical projected apertures, which were discussed previously. In
addition, it is common to have an array of horns or other feed
antennas illuminating shaped reflectors or large reflectors,
such as radio telescopes. This is because the array yields better control of the phase distribution employed to illuminate
the reflector, enhancing beam contour and beam scanning
performance. A contoured beam is required to illuminate
properly a specified region of the earth, as seen from a satellite, and can be accomplished with the shaping of the reflector, one of the topics addressed in the following section.
ADVANCED TOPICS AND RESEARCH
Reflector Antenna Upgrading
During the past few decades reflector antenna designs have
evolved through several configurations to increase performance and/or reduce structural complexity. Electrical parameters that are of prime interest are aperture efficiency, SLL,
and, more recently, XPOL. All topics herein discussed apply
to the various types of reflectors previously addressed. However, the offset configuration is likely to retain, in the near
future, the largest percentage of the reflector antenna
market.
We first focus our attention on XPOL. Reflector antennas
presenting low XPOL (e.g., XPOL 35 dB) are necessary
for frequency-reuse applications, in which an overlap of orthogonally polarized channels is permitted. Many efforts are
being conducted to develop these kinds of antennas for mass
production (7,35). Dual offset reflectors can be designed for
low-cost construction, provided that specific manufacturing
constraints are carefully taken into account (7), an effort only
possible due to increased interest from industry. Single offset
reflector systems illuminated by a matched feed (23) or a feed
with a lens (35) can also be designed to satisfy stringent requirements on XPOL, yielding very compact designs. In the
latter case, the lens is designed to replace the subreflector,
and in both cases bandwidth performance is not as straightforwardly obtained as with the dual reflector configuration.
Research continues to be conducted within the area, yielding
innovative solutions that provide satisfactory XPOL performance while attending to practical manufacturing specifications. Cost-effective solutions normally require that attendance to a particular specification, such as low XPOL, be
achieved with minimal capital outlay, which implies using the
maximum amount of infrastructure and technology already
implemented. As mentioned previously, there is a tendency to
REFLECTOR ANTENNAS
employ existing molds for the main reflector, a concept referred to as reflector upgrading (7).
Within this context, existing single offset reflector molds
are normally used to construct the main reflector of a dual
configuration. However, many such molds are for just fully
offset geometries, which, in general, produce a dual reflector
configuration that is Gregorian with the feed axis zf intersecting the main reflector (Fig. 11). The same problem may
also occur in certain Gregorian configurations even when the
main reflector is not just fully offset. The final design should
provide suitable clearance between the bottom of the main
reflector and the feed axis in order to access the feed antenna
with a straight section of waveguide, thus reducing the complexity and cost of the manufacturing process. This setup is
achieved by rotating the parent ellipsoid (i.e., the conical surface from which the subreflector is generated) until the desired clearance is obtained. The rotation is performed in such
a way that the feed remains pointed toward the intersection
of the new subreflector and the ray coming from the center of
the main reflector, thus avoiding the introduction of spillover
and phase errors. The amount of rotation R that yields a desired angle between the main reflector and feed axes can
be determined from (7)
1+e
f
1 e cos(180 R C ) S
1+e
f
2(c + f S )
1 e cos(180 R C ) S
sin(180 R C ) = sin( + R + )
(39)
Given the initial configuration and the desired angle ,
Eq. (39) can be solved to determine R. In general, values for
smaller than the one used in the original configuration
bring the feed axis away from the main reflector. However,
the nonconventional design obtained after the rotation of the
parent ellipsoid may lead to a XPOL degradation due to the
fact that the minimum-XPOL conditions, Eqs. (27) to (29), are
no longer satisfied. A simple solution to this problem is to
alter the value of the subreflector eccentricity while keeping
all orientation angles constant. In general, eccentricity values
greater than the one employed before the rotation will reduce
the system XPOL (7), yielding a low-cross-polarization dual
offset Gregorian antenna that has adequate clearance between the feed axis and the bottom of the main reflector. In
addition, the resulting configuration has the ability to operate
with either a linearly polarized or a circularly polarized feed
over a wide bandwidth without the need to be repositioned
(no substantial beam squint).
Compact designs for reflector systems have been investigated for years. It is desirable, for example, to upgrade a main
reflector with a subreflector that is as small as possible. In
addition, with the proliferation of satellite TV at Ku band employing single offset reflector systems for reception, there is
now interest in minimizing the size of the reflector while
maintaining required gain performance. This is only possible
by increasing aperture efficiency through the reduction of diffraction effects and feed blockage. High-performance feeds are
also necessary, especially if they are located close to the reflector, as in a very compact design requiring a complete nearfield analysis. Finally, microelectronics technology is integrating both low- and high-frequency hardware into the feed
377
system. It is common nowadays to find feeds that already include low-noise amplifiers, downconverters, and other electrical devices in a single unit.
Shaped, Deployable, and Frequency-Selective Reflector Surfaces
Reflector antennas can be shaped to improve gain and aperture efficiency (e.g., Fig. 3). The basic concept is to shape the
subreflector of a dual-reflector system to obtain the desired
amplitude distribution at the main reflector projected aperture, normally nearly uniform for maximizing gain, and then
shape the main reflector to recover the uniform phase distribution, which was disturbed by shaping the subreflector. Resulting aperture efficiencies are usually in the range of 70%
to 80%, although higher values have been reported for dual
shaped offset systems (11). The fundamental tradeoff that antenna designers must face, in this case, is the achievement of
such efficiencies while still obtaining satisfactory sidelobe
levels.
Another topic certain to continue receiving attention in the
coming millenium is the shaping of single-reflector systems
for contoured beam applications. A contoured beam is necessary to illuminate specific regions of the earth efficiently from
a satellite. This avoids unnecessary coverage of regions outside the satellite main service area. Reflector shapes for contoured beam and high-gain applications, especially for the offset case, are normally obtained numerically through
elaborate synthesis and optimization processes (35,9,10),
which include the feed antenna or array. Equation (17) can
be used to set the initial positions for the feed antennas (3).
Also important is mathematical and numerical modeling of
the surfaces to yield results that can be implemented in practice (36). Feed arrays are also used with reconfigurable mesh
surfaces and deployable reflectors to enhance performance (1).
Reconfigurable shapes, deployable reflectors, and polarization- or frequency-selective surfaces require an in-depth study
of electrical and magnetic materials, which show that reflector antenna engineering is not possible without large interdisciplinary activities. Frequency-sensitive surfaces, for example, reflect only radiation in specific frequency bandwidths (1).
A four-frequency reflector system has been used for the ESA
NASA Cassini mission to Titan, the largest moon of Saturn
(1). The subreflector carried by the Voyager spacecrafts shown
in Fig. 3 is another example; it is transparent to radiation at
S band and reflecting at X band. The subreflector was manufactured with X-band resonant aluminum crosses etched on a
Mylar sheet, yielding reflecting and transmitting losses lower
than 0.1 dB.
Reflectors that can be deployed in space deserve distinct
attention. The classical solution is the so-called umbrella reflector (1), which works well for axisymmetric configurations.
An idea that seems promising for the offset case is a configuration that opens like a manual fan. In addition, inflatable
reflectors have been investigated over the past few years,
with the successful construction and testing of a few prototypes.
Neural Networks, Fuzzy Logic, and Genetic Algorithms
Applied to the Synthesis of Reflector Antennas
As we have seen, reflector antenna applications range from
very specific and unique systems, such as deployable reflectors and radiotelescopes (see the subsection on the Green
378
REFLECTOR ANTENNAS
Bank Radio Telescope), to large-scale production for commercial systems, such as earth terminals in VSAT networks and
small receiving antennas for satellite TV. In addition, reflector antennas are also used in radar systems and other devices
directly related to electronic warfare and defense. Due to the
wide range of applications and commercial importance, it is
essential to have reliable alternative algorithms to design effective reflector configurations. Analytical and numerical tools
that have recently been used with applied electromagnetics
and appear to work well for the synthesis of reflector antennas, although they are most familiar in other areas such as
controls and signal processing, include neural networks, fuzzy
logic, and genetic algorithms (3740).
The basic idea of using neural networks for designing reflector antennas is as follows: First one relates a few radiation
patterns, or other parameters of interest, directly to the reflector geometries that generated them according to a previously selected analysis algorithm. This process is normally
referred to as training the neural network. Once the training
is completed, the desired radiation pattern is employed as the
input parameter, so that the neural network yields the corresponding reflector geometry. Finally, the selected analysis
technique is used on the resulting geometry to validate the
process. This type of synthesis has been successfully implemented to determine reflector shapes for contoured beam applications (37). Reflector antenna synthesis based on neural
networks is normally very fast, because analysis techniques
are required only for the training of the network.
Also currently under investigation is the effectiveness of
using neural networks at a real-time level to reduce XPOL,
noise, and interference in single offset reflector systems.
Fuzzy logic can be applied to the synthesis and enhancement
of reflector antenna systems in a similar manner, although
there are certain inherent differences not addressed herein
(38).
Genetic algorithms (GAs), on the other hand, rely on an
optimization search to elect a suitable design. The initial set
of configurations, referred to as the basis population, is
formed by relating random designs to chromosomes, each one
formed by a sequence of binary numbers that define the corresponding reflector geometry. A figure of merit is then associated with each chromosome, becoming higher as the electrical
performance of the configuration comes closer to the desired
one. Chromosomes with the highest figures of merit are selected to cross with the remaining ones, a process called crossover (39,40). In addition, the best chromosomes (i.e., the ones
with the highest figures of merit) are often duplicated before
crossover, eliminating the worst ones, a procedure referred to
as natural selection (39,40). After the crossover is completed,
a few binary numbers in a subset of chromosomes can be randomly altered, simulating the process of mutation in biological evolution. All figures of merit are then recomputed, and
the whole process is repeated if the desired performance has
not been achieved.
The aforementioned procedure is a basic description of a
genetic algorithm, although variants are often encountered in
practice (39,40). A major advantage of GAs over other optimization techniques is that once the process has converged,
more than one configuration attending the desired performance is often encountered. It is also common to find configurations that are very different from classical solutions and
could not be easily conceived through straightforward reasoning. Practical manufacturing constraints can then be consid-
ered for choosing the most adequate reflector antenna configuration.
ACKNOWLEDGMENTS
The author is deeply indebted to George Behrens (National
Radio Astronomy Observatory) and Nick Moldovan (Prodelin
Corporation), as well as to their institutions, for providing
many illustrations and photographs reprinted in this work.
The author likewise thanks NASA/JPL and the Brazilian
Telecommunications Center for Research and Development
(CPqD/TELEBRAS).
BIBLIOGRAPHY
1. W. V. T. Rusch, The current state of the reflector antenna art
entering the 1990s, Proc. IEEE, 80: 113126, 1992.
2. S. Silver (ed.), Microwave Antenna Theory and Design, New York:
McGraw-Hill, 1949.
3. Y. Rahmat-Samii, Reflector antennas, in Y. T. Lo and S. W. Lee
(eds.), Antenna Handbook, New York: Van Nostrand Reinhold,
1988.
4. C. Scott, Modern Methods of Reflector Antenna Design, Norwood,
MA: Artech House, 1990.
5. P. S. Kildal, Synthesis of multireflector antennas by kinematic
and dynamic ray tracing, IEEE Trans. Antennas Propag., 38:
15871599, 1990.
6. K. W. Brown, Y. H. Lee, and A. Prata, Jr., A systematic design
procedure for classical offset dual reflector antennas with optimal
electrical performance, in IEEE Antennas Propag. Soc. Symp.
Dig., Ann Arbor, MI, 1993, pp. 772775.
7. M. A. B. Terada and W. L. Stutzman, Computer-aided design of
reflector antennas, Microw. J., 38 (8): 6473, 1995.
8. W. V. T. Rusch and P. D. Potter, Analysis of Reflector Antennas,
New York: Academic, 1970.
9. A. W. Love (ed.), Reflector Antennas, Piscataway, NJ: IEEE
Press, 1978.
10. B. S. Westcott, Shaped Reflector Antenna Design, London: Wiley, 1983.
11. A G. Cha, Preliminary announcement of an 85 percent efficient
reflector antenna, IEEE Trans. Antennas Propag., 31: 341342,
1983.
12. A. H. Rana, J. McCoskey, and W. Check, VSAT technology,
trends, and applications, Proc. IEEE, 78: 10871095, 1990.
13. A. C. Ludwig, The definition of cross polarization, IEEE Trans.
Antennas Propag., 21: 116119, 1973.
14. TICRA Eng., GRASP7Single and Dual Reflector Antenna Program Package, Copenhagen, Denmark.
15. M. A. B. Terada and W. L. Stutzman, Design of offset-parabolicreflector antennas for low cross-pol and low sidelobes, IEEE Antennas Propag. Mag., 35 (6): 4649, 1993.
16. A. W. Rudge and N. A. Adatia, Beam squint in circularly polarized offset parabolic reflector antennas, Electron. Lett., 11: 513
515, 1975.
17. D. W. Duan and Y. Rahmat-Samii, Beam squint determination
in conic-section reflector antennas with circularly polarized feeds,
IEEE Trans. Antennas Propag., 39: 612619, 1991.
18. W. L. Stutzman, Polarization of Electromagnetic Systems, Norwell, MA: Artech House, 1993.
REFRACTION AND ATTENUATION IN THE TROPOSPHERE
19. M. A. B. Terada and W. L. Stutzman, Cross polarization and
beam squint in single and dual offset reflector antennas, Electromagnetics J., 16: 633650, 1996.
20. W. L. Stutzman and G. A. Thiele, Antenna Theory and Design,
2nd ed., New York: Wiley, 1998.
21. A. G. P. Boswell and R .W. Ashton, Beam squint in a linearly
polarized offset parabolic reflector, Electron. Lett., 12: 596597,
1976.
22. P. S. Kildal, Diffraction analysis of a proposed dual-reflector feed
for the spherical reflector of the Arecibo observatory, Radio Sci.,
24: 601617, 1989.
23. A. W. Rudge et al. (eds.), The Handbook of Antenna Design, Stevenage, UK: Peregrinus, 1982.
24. W. V. T. Rusch et al., Derivation and application of the equivalent paraboloid for classical offset Cassegrain and Gregorian antennas, IEEE Trans. Antennas Propag., 38: 11411149, 1990.
25. B. Shen and W. L. Stutzman, A scanning tri-reflector antenna
with a moving flat mirror, IEEE Trans. Antennas Propag., 43:
270276, 1995.
26. C. Dragone, Offset multireflector antennas with perfect pattern
symmetry and polarization discrimination, Bell Syst. Tech. J., 57:
26632684, 1978.
27. Y. Mizugutch, M. Akagawa, and H. Yokoi, Offset dual reflector
antenna, IEEE Antennas Propag. Soc. Symp. Dig., Amherst, MA,
1976, pp. 25.
28. S. Srikanth, Comparison of spillover loss of offset Gregorian and
Cassegrain antennas, IEEE Antennas Propag. Soc. Symp. Dig.,
London, Canada, 1991, pp. 444447.
29. M. A. B. Terada and W. L. Stutzman, Computer-aided design of
reflector antennas: The Green Bank radio telescope, IEEE Trans.
Microw. Theory Tech., 46: 250253, 1998.
30. P. H. Pathak, Techniques for high-frequency problems, in Y. T.
Lo and S. W. Lee (eds.), Antenna Handbook, New York: Van Nostrand Reinhold, 1988.
31. Y. Rahmat-Samii and V. Galindo-Israel, Shaped reflector antenna analysis using the JacobiBessel series, IEEE Trans. Antennas Propag., 28: 425435, 1980.
32. W. L. Stutzman, S. W. Gilmore, and S. H. Stewart, Numerical
evaluation of radiation integrals for reflector antenna analysis
including a new measure of accuracy, IEEE Trans. Antennas Propag., 36: 10181023, 1988.
33. P. J. B. Clarricoats and A. D. Olver, Corrugated Horns for Microwave Antennas, Stevenage, UK: Peregrinus, 1984.
34. S. Ghosh, E. Kuhn, and A. Prata, Simplified high-performance
dual-banded feed comprising a dual-depth corrugated launcher
and a conventional horn, Electron. Lett., 20: 532533, 1984.
35. E. Lier and S A. Skyttemyr, A shaped single reflector offset antenna with low cross polarization fed by a lens horn, IEEE Trans.
Antennas Propag., 42: 478483, 1994.
36. J. R. Bergmann et al., Synthesis of shaped-beam reflector antennas patterns, IEE Proc., part H, 135: 4853, 1988.
37. G. Washington, Aperture antenna shape prediction by feedforward neural networks, IEEE Trans. Antennas Propag., 45: 683
688, 1997.
38. J. S. R. Jang and C. T. Sun, Neuro-fuzzy modeling and control,
Proc. IEEE, 83: 378406, 1995.
39. D. S. Weile and E. Michielssen, Genetic algorithm optimization
applied to electromagnetics: A review, IEEE Trans. Antennas Propag., 45: 343353, 1997.
40. J. M. Johnson and Y. Rahmat-Samii, Genetic algorithms in engineering electromagnetics, IEEE Antennas Propag. Mag., 39 (4):
725, 1997.
MARCO A. B. TERADA
University of Brasilia
REFLECTOR RADAR ANTENNAS. See RADAR ANTENNAS.
379
666
SATELLITE ANTENNAS
tion is paid to three important parameters: weight, power,
and volume. Likewise, reliability is a key requirement because an anticipated lifetime exceeding ten years is expected
without the possibility of repair or servicing. Spaceborne antenna designs must be sufficiently rugged to survive the rigors of launch environments. In addition to these physical requirements, the rf requirements differ from common antenna
specifications. Spaceborne antennas provide performance over
prescribed coverage areas containing the user segment; the
pattern of the antenna projected on the earths surface is referred to as its footprint. Antenna systems are specified by
their minimum performance within the specified coverage
area rather than their peak performance levels. This discussion uses different coverage area requirements to describe
space segment antennas.
Earth Coverage Antennas
SATELLITE ANTENNAS
Antennas for satellite systems have developed and increased
in performance over the last 40 years and greatly contribute
to satellite capabilities. The preponderance of this antenna
development has been for communication systems, and other
development has been for remote sensing applications. Early
satellite systems provided not only proof of concept but also a
useful initial operating capability. The payload constraints of
early satellites together with primitive attitude control systems limited antennas to simple narrow-bandwidth, broadcoverage designs, a far cry from todays sophisticated antenna
systems with substantial throughput capabilities. Similarly,
early users needed large ground terminal antennas to achieve
the required link performance, and terrestrial links were provided to other system users. Today, systems are being developed to communicate to individual users with hand-held terminals (1).
Both spaceborne and user antenna systems have requirements and development issues that are unique to satellite
systems. In addition, satellite system antennas have testing
requirements and objectives that differ from other antenna
applications. Antenna technology, requirements and development issues, and test methods and objectives are described.
SPACE SEGMENT ANTENNAS
Spaceborne antennas have unique requirements and development issues. Like all satellite subsystems, much design atten-
The simplest spaceborne antenna design is one that matches
its coverage area with the available field of view from the
satellite. These antennas are generally referred to as earth
coverage designs, and their beamwidth subtends the field of
view plus the angular uncertainties of the satellites attitude
stability. Early satellite attitude control systems had both
limited attitude control and uncertain reliability so that simple antenna designs with broad coverage were used. However,
todays attitude control systems provide stabilities on the order of one-tenth of a degree, so that earth coverage antennas
can be essentially matched to the field of view. Thus, for geosynchronous satellites, an antenna with an 18 beamwidth is
required; lower altitude LEO and MEO satellites require a
correspondingly larger beamwidth for their larger field of
view.
A typical earth-coverage design for geosynchronous altitudes is a corrugated horn antenna (2) which can be easily
constructed and has a rotationally symmetric, low sidelobe
pattern for an efficient fit to the design coverage area. These
horns are typically wide flare designs chosen to maintain the
same beamwidth and coverage over a wide bandwidth. Thus,
a single horn together with a diplexer to isolate the receiver
and transmitter can be used for both uplink and downlink
frequencies. A design alternative at higher EHF frequencies
is separate uplink and downlink horn antennas because these
horns are very small and lightweight. Since the uplink and
downlink frequencies are widely separated at EHF, the transmit diplexing filter is inherent because the transmit waveguide is cutoff at the receive frequencies and a modest physical separation may provide adequate isolation to meet the
receive diplexing filter requirement of avoiding receiver saturation at the transmit frequency. Thus, the weight of a second
horn antenna is offset because a diplexer is not required and
performance is not degraded by the insertion loss of the
diplexer.
At lower LEO and MEO altitudes, the broader field of view
requirements can be met with an open-ended waveguide surrounded by corrugations to produce low sidelobes and beam
symmetry. At these lower altitudes, the range differences between the subsatellite point and the edge of the earth become
more significant and thus users at the edge of the coverage
experience less sensitivity from the satellite antenna. In these
cases, a shaped beam that provides more gain at the edge of
the earth than the subsatellite point may be desired. Such
J. Webster (ed.), Wiley Encyclopedia of Electrical and Electronics Engineering. Copyright # 1999 John Wiley & Sons, Inc.
SATELLITE ANTENNAS
shaped-beam antennas can be realized by an array of horns
with properly selected excitation to meet the beam-shaping
requirement. However, such shaped-beam designs have more
design complexity than a simple horn, and weight and volume
are important satellite parameters. A well-known example of
shaped beam designs is the array antenna used by the GPS
satellites to broadcast navigational information. The GPS satellites are in a 12-h orbit, and the antenna array (3) is designed to deliver a uniform flux density to users within the
earths field of view.
Tracking, Telemetry, and Command Antennas
Every satellite requires a TT&C (tracking, telemetry, and
command) subsystem to assist in determining the satellites
orbital position, reporting the satellites health and status to
the ground-based satellite control segment, and providing a
means to command and validate changes to the satellites operation. The TT&C antennas have two distinct coverage requirements. One requirement is to provide complete spatial
coverage during launch operations so that corrective commands or, in the worst case, a command for destruction can
be delivered irrespective of the satellites orientation. The second requirement is to provide coverage over the earths field
of view once on-orbit so that different terminals within the
ground-control network can access the satellite. TT&C antennas must be both simple in design and highly reliable since
the control of satellite operation is critical.
The broad coverage requirements of the TT&C antennas
provide several design challenges. Aside from the impossibility of an ideal isotropic antenna with uniform sensitivity in
all directions, the presence of the spacecraft blocks the antenna coverage. As a result, two antennas that strive for
hemispheric coverage are used with one hemisphere being
earth facing and the other hemisphere facing away from the
earth. During launch operations, the antenna having the
higher signal strength is used and once the satellite is stable
in its orbital position, the earth facing antenna is used. In
principle, somewhat better on-orbit performance could be
achieved with an earth-coverage antenna, but the additional
antenna and required switching increase weight and volume
and add switch reliability risks. Since TT&C data rate requirements are low, ample link performance is readily
achieved by ground-based control terminals.
Achieving and quantifying the performance of hemispheric
coverage antenna elements remain challenges. Satellites have
irregular shapes and appendages such as solar arrays whose
positions vary to sun track and maximize their power output.
The hemispheric coverage required by TT&C antennas is degraded by blockage and diffraction effects from the satellite
structure. Controlling wide-angle sidelobes and backlobes of
a simple, electrically and physically small antenna is difficult
to do but is needed to minimize antenna interactions with
satellite structure. Extending the antenna from the spacecraft to reduce interactions results in deployment risk. Analytic projections of the antenna coverage in the presence of
the spacecraft are challenged by the design complexity and
the wide variety of satellite designs. Measurements of the antenna performance have all the problems of testing broad coverage antenna designs. Certainly, measurements using flight
hardware not only have practical limitations such as the inability to deploy solar arrays but also moving and positioning
667
flight hardware for the measurements have unacceptable
risks. Generally, these problems are addressed by a combination of selecting a satellite antenna location with a clear field
of view, analytic estimates of the effects of nearby satellite
structure, scale model measurements, and ample margin in
the ground segment.
Typically, TT&C systems operate at the lower microwave
frequencies. Generally, a single uplink and downlink antenna
is used to minimize weight and volume and avoid mutual
blockage. The difference between uplink and downlink frequencies often spans a significant percentage bandwidth.
These considerations typically result in selecting a frequency
independent antenna design such as a spiral to achieve the
broad coverage requirements.
Spot Coverage Antennas
Spot coverage antennas service only a portion of the available
field of view. These antenna designs provide higher gain levels than earth coverage antennas because of their narrower
beamwidths. This higher gain performance may be traded for
higher data rate transfer, reduced user antenna dimensions,
or increased link margin. In addition, spot beam antenna designs can provide coverage that is confined to geographical
areas (4) and avoid worldwide licensing requirements. Thus,
these designs are widely used to serve national or regional
requirements. Several different design implementations have
been developed to satisfy these needs.
The easiest design solution is to select a sufficiently large
antenna so that its beamwidth matches the required coverage
area and any variations caused by attitude uncertainty. Such
a design might consist of a reflector whose size is selected to
meet the desired beamwidth and a single feed horn, a simple
design approach. In such a design, any variations in coverage
requirements during the satellite lifetime can be accomplished by mechanically repositioning the antenna to serve a
different coverage area. Such a capability is commonly used
in military satellites so that coverage to different geographical areas within the field of view can be provided to service
changing political needs.
A wide variety of designs has been developed to service
more irregular coverage areas. These coverage areas may be
bounded by geopolitical borders, different time zones, etc. A
standard means of providing such coverage is to synthesize
the coverage area by combining different beams generated by
a cluster of feed elements in the focal region of a reflector
antenna. Each feed has its own pointing direction and the
collection of beams subtends the desired coverage area. The
individual beams being combined are required to have a common phase center so that grating lobes do not degrade coverage characteristics. Thus, the collection of beams is generated
by a common aperture. The arrangement of individual beams
and their corresponding feed elements mimic the desired coverage area. The reflector antenna size is much larger than
that needed to provide coverage over the composite area.
When active devices are used in the feed elements prior to
combining, design attention must be paid to their amplitude
and phase tracking. The precision of fitting to the design coverage area can be increased by using a larger number of
beams at the expense of design complexity and a larger overall aperture size.
668
SATELLITE ANTENNAS
In addition to the complexity of the additional number of
feeds and the larger reflector required by this approach, the
design complexity also increases because the individual feeds
are no longer located at the focal point of the reflector but
instead are distributed within the focal region. The off-axis
feeds are not ideally focused, and suffer gain loss and pattern
degradation. These problems are shared with multiple beam
designs and, as will be discussed, result in dual reflector designs to limit beam degradation when a large number of beam
are combined.
Another design approach has developed in recent years to
provide coverage over irregular areas. This design uses a single feed, and the irregular coverage area is generated by deforming the reflector surface (5). Synthesis procedures have
been developed (6) to specify the required deformation of the
reflector surface. This design approach has the advantage of
a simple feed design rather than a cluster of feed elements
and combining, but requires developing and maintaining controlled reflector surface distortion. If the coverage requirements change, a new surface deformation is required. This
approach also precludes changing the coverage areas on-orbit
by using beam switching techniques.
A final design alternative for spot coverage is an adaptive
uplink antenna. The pattern within the coverage area is
adaptively controlled to service desired users within the design coverage area to the extent practical while reducing received interference by forming pattern nulls in the direction
of interference sources. This capability is principally required
for military systems whose users desire protection from intentional interference. When interference is not present, a quiescent pattern provides coverage over the design region. When
interference is initiated, the antenna responds under adaptive control to form pattern nulls.
Adaptive antennas require a means to distinguish desired
signals from interference, to detect the initiation of interference, and to determine a set of complex weighting values that
combine antenna elements to satisfy an optimization criteria.
Several different discriminates have been employed to distinguish desired signals from interference: spectral characteristics, power levels, and signal arrival directions. When systems also use spread spectrum modulation for additional
interference protection, the properties of the spread spectrum
codes provide a basis to distinguish interference from desired
signals. Interference initiation can be detected by increased
received power levels that do not have spread spectrum modulation components; correlation processes are normally used
to indicate interference power. An optimization criterion such
as maximum SINR (signal to interference plus noise ratio) is
used to establish complex weighting coefficients to combine
antenna elements. Generally, a recursion relation based on
measured correlation values and the optimization criteria is
derived to form a control algorithm. The combination of the
complex weighted antenna elements produces a pattern containing null in the direction of interference, and equivalently,
the received interference power is minimized.
Two types of antenna designs have been investigated for
these adaptive systems (7,8). One approach uses a thinned
array where the elements are combined to generate pattern
nulls when interference is present and a set of weighting coefficients for interference-free cases. The potential advantage of
this approach is that the resolution between desired users
and interference is enhanced by increasing the array element
spacing. However, the separation between array elements
produces additional nulls within the coverage area referred to
as grating nulls that degrade coverage characteristics. The
separation of the array elements results in the inherent frequency scanning properties of an array creating dispersion
that limits cancellation performance over the required bandwidth. In addition, the coverage of the array elements generally extends beyond the desired coverage area resulting in
susceptibility to interference beyond the coverage area. Finally, the thinned array design generally results in reduced
G/T performance within the coverage area.
The second design approach uses an offset reflector with a
cluster of feeds in the focal region. In interference-free conditions, a set of quiescent weighting values for the individual
feeds maintains performance over the design coverage area.
The initiation of interference produces correlation products
associated with the interference that are used to derive complex weighting values. The advantage of this design approach
is that all beams generated by the feed cluster have the same
phase center location so that the dispersion inherent in array
designs is not present and broad bandwidth cancellation performance is achieved. The resolution between interference
and desired signals is limited by the beamwidth of the individual beams. This resolution depends on the amount of G/T
margin that the user can sacrifice and the interference
sources location on the earth, as illustrated in Fig. 1. A comparison between the performance of a thinned array design
and a multiple feed design is presented in Fig. 2. The grating
nulls are apparent in the thinned array contours, which reduces the coverage available to users.
Several performance measures are used with adaptive antenna designs. The threshold SNIR establishes a bound for
acceptable communication performance that can be measured
using BER (bit error rate) values by injecting both desired
signals and residual interference into the receiver. The null
depth performance and its variation over the required bandwidth as measured between the quiescent pattern and the
pattern after adaption are sometimes used to judge nulling
performance. The amount of time the adaptive process takes
to convert the quiescent pattern to the nulled pattern is referred to as the convergence time and measures the transient
performance of the system. Finally, for this uplink application, the amount of the design coverage area that exceeds the
threshold SNIR value is another measure for space segment
antennas. These performance measures depend on the number of interference sources and their locations, and their
power levels and spectral characteristics; this description of
the interference is referred to as a scenario. Adaptive systems
are commonly developed using a simulation that varies the
scenario parameters on a Monte Carlo basis to provide statistical measures of the performance parameters. The simulation is then validated by using hardware measurements on a
limited number of scenario cases in order to avoid the impractical amount of time required to test on a Monte Carlo basis.
Multiple Beam Antennas
Multiple beam antenna systems (MBA) (9) greatly increase
the on-orbit satellite capabilities. These antennas simultaneously produce more than one antenna beam from a single aperture. For space applications, a single aperture is a significant savings in space required on the satellite and the overall
SATELLITE ANTENNAS
1000
800
600
669
20 Elevation
Separation (nmi)
400
Gain loss
200
100
80
60
Interference
source
40
Separation
Subsatellite
20
Desired
source
3 dB gain loss
10 dB gain loss
2
1 0.8 0.6 0.4
Antenna beamwidth (deg)
weight compared with multiple aperture alternatives. When
the coverage area is subdivided by different beams, the gain
available to users increases permitting higher data rates or
reduced user terminal performance. A further advantage of
multiple beam antennas is that beams directed to different
geographic locations are isolated by their spatial separation.
Thus, the same frequency bands can be used simultaneously
at both beam locations increasing the information that can be
transferred by a single satellite. These advantages are the
reasons for the wide application of multiple beam antennas.
The antenna design has several different requirements. Independent beams must be efficiently generated with high pattern fidelity and minimal gain degradation over the required
field of view. Isolation between different beams must be maintained so that signals using the same frequency subband are
not degraded by cochannel interference. Beam-forming networks that combine signals from the antenna beam outputs
to and from the satellite transponder must be developed. The
transponder must provide a means to map the uplink information in each beam position to the downlink beam positions.
One design issue is to efficiently generate independent
beams with high pattern fidelity over the required field of
view. When optical designs (i.e., reflector and lens antennas)
are used, each feed produces an independent beam in a different direction, but not all of the feeds are ideally focused. As a
consequence, some of the feeds experience phase distortions
that degrade the pattern characteristics and reduce the gain
compared to the feeds that are well focused (10). Several different techniques in the design of the optics are used to minimize the degradation. One technique is to select a long f /D
ratio to reduce off-axis phase distortions. The popularity of
dual reflector designs in these applications results in part
from their increase of the effective f /D and careful attention
to the optics can produce excellent results (11,12). The dual
reflector designs in an offset configuration also provide the ability to achieve very good polarization purity. Other techniques
have been applied to selecting lens surface contours (13).
In addition to achieving good pattern and gain performance over the required field-of-view for an individual feed,
Figure 1. Adaptive antenna resolution
(after Ref. 8).
0.8
0 10
16
0.6
22
25dB
0.4
Latitude (deg)
0.2
0
U3
10
0
16
0.2
J7
22
0.4
0.6
0.8
0.8 0.6 0.4 0.2
0 0.2 0.4
Longitude (deg)
0.6 0.8
(a) Multiple beam contours
0.8
0.6
0.2
0
9
6
3
3
0
36
3
6
3
6
9dB
9
J7
0.2
0
3
0.4
6
0.6
6
3
0.4
Latitude (deg)
10
0
3
30
6
3
6
6
0.8
0.8 0.6 0.4 0.2 0 0.2 0.4 0.6 0.8
Longitude (deg)
(b) Thinned array contours
Figure 2. Comparison of multiple beam antennas and thinned array
performance (after Ref. 8).
SATELLITE ANTENNAS
another set of design challenges exists with the cluster of
feeds. The problem is a conflict among beam isolation, the
crossover level between adjacent beams, the sidelobe levels of
the beams, and the antenna efficiency. On the one hand, low
sidelobe patterns are desired to enhance the beam isolation
which requires the feed to produce an amplitude taper over
the aperture resulting in a relatively large feed. The physical
interference between large adjacent feeds results in a relatively low tangential crossover level (the pattern level relative
to the beam boresight at an equal distance between beam centers) between beams. The consequence of the low crossover
level is that the minimum gain within the field of view is
reduced. On the other hand, high beam crossover levels result
a higher minimum gain level within the field of view, but
beam patterns with high sidelobes having reduce beam isolation. High beam crossover levels also result in beam coupling
loss (14) that degrades antenna efficiency.
These conflicts have resulted in several different design
approaches. One technique is to couple signal samples from
adjacent feed elements (15) to achieve high crossover levels,
low sidelobe patterns, and reasonable antenna efficiency. Another technique (16) uses a underilluminated aperture with
the feeds displaced from the focal region. This approach requires a somewhat larger reflector than the focused case, but
at EHF frequencies where compact antennas result from the
small wavelength, the size increase may be unimportant.
The required isolation between beams depends on the susceptibility of the signal modulation to cochannel interference
and the dynamic range of the system users. Thus, the required isolation level must be derived for each application.
Isolation between adjacent beams can be achieved in two
ways. One way, referred to as frequency reuse, divides the
allotted bandwidth into subbands. A frequency plan is devised to adequately separate the beams using the same subband to meet the isolation requirement. Since the same frequency subband is simultaneously used in different beam
positions, the plan is referred to as a frequency reuse plan.
A second way of increasing isolation referred to as polarization reuse uses orthogonal polarizations together with frequency subbands in applications requiring high isolation. Orthogonal polarization is also used in applications demanding
as much capacity as possible because the same beam position
can transfer twice the information. These techniques require
antennas with high polarization purity; a typical requirement
is to suppress cross polarization levels at least 27 dB lower
than the principal polarization. Generally, satellite systems
use circular polarization so that the user does not have to
align to a linear polarization. However, inclement weather
can cause coupling between polarizations reducing the isolation. This effect is less severe for linear polarization, so that
orthogonal linear polarization is sometimes used. This combination of polarization and frequency reuse is commonly used
by commercial systems to obtain as much capacity as possible
from a single satellite.
Multiple beam antennas are generally considered as designs covering the available field of view. As the beamwidth
of the individual beams decreases, the required number of
beams greatly increases as shown in Fig. 3, larger aperture
sizes are required and the complexity of beam routing increases. In some applications, multiple beams are used in a
limited geographical region (4,17). The Ka-band multiple
beam system described in Ref. 17 has such small beamwidths
103
Number of beams required
670
102
3 dB
crossover
10
5 dB
crossover
Earth
coverage
1
0.5
10
20
Half-power beamwidth (deg)
Figure 3. Number of beams to cover FOV (after Ref. 9).
that active tracking of a ground beacon is required to compensate for satellite attitude variations.
The ability to process and direct the information on the
collection of multiple beams requires beamforming networks
that separate information on the uplink beams and combine
information on the downlink beams. A wide variety of beamforming networks has been developed. The simplest design is
a fixed beam-former network that simply combines beams in
a fixed pattern. This beam former together with transponder
channelization maps the uplink beam patterns into the downlink patterns. other beam-former designs use cascaded networks on switches or variable power dividers so that the beam
patterns can be changed on-orbit. Still other beam formers
use adaptive processing on the uplink beam former to negate
interference. Yet other systems route the uplink information
to a terminal on the ground, referred to as a gateway, where
the uplink beams are mapped into the downlink collection.
The beam formers have different requirements for the
uplink beams and downlink beams. The system noise temperature on the uplink can be established by preamplifiers which
are followed by the beam formers. In this way the system sensitivity is not degraded by the insertion loss of the beam formers, and lightweight networks and diode switching can be
used in the beam-former implementation. The downlink
beam-former network must be placed between the transmitters and the beam ports, so that power handling and insertion
loss are important design parameters. In addition, switching
for redundant transmitters is also required. These requirements depend on the application.
The development of multiple beam antenna systems has
been principally pursued for geosynchronous satellites, and
future design challenges exist to develop multiple beam designs for LEO and MEO satellites. While offset reflector an-
SATELLITE ANTENNAS
tennas are commonly used in geosynchronous satellites, these
antennas are examples of limited scan antennas (14), that are
not appropriate for the wider field of view at the lower altitudes. These lower altitudes also need wider beamwidths
than geosynchronous designs to maintain reasonable footprint dimensions, and thus the overall antenna size is electrically smaller. Array antennas are commonly used to meet
performance requirements for these orbits. With their smaller
electrical size compared with geosynchronous designs, a reasonable number of array elements is required.
The trend toward using digital technology in the transponder also naturally leads to considering digital beam-forming
techniques in the array design. On the uplink, the received
signal at each element is preamplified, downconverted to
baseband, and transformed into the digital domain by an analog-to-digital converter (ADC). Digital samples from each
array element are combined using a complex matrix multiplier containing the complex coefficients that perform the
beam steering. Parallel complex multipliers are required to
form multiple beams. The same constraints on beam isolation
exist for multiple beam operation. The downlink antenna follows the same procedure in reverse. The input digital data
stream is complex matrix multiplied to obtain the properly
weighted samples for each array element for the beam steering, and a digital-to-analog converter (DAC) provides conversion to the analog domain. This analog baseband is upconverted to the transmit frequency and suitably amplified to the
required element power level. Experience with digital array
designs is described in Ref. 18; an adaptive array using digital technology to increase beam isolation is described in Ref.
19.
The downlink array designs for multiple beam operation
also present challenges in transmit power control. The element transmit power is shared with each of the beams transmitted by the array. The relative amplitude and phase
between the array elements must be maintained to achieve
the desired downlink pattern. In addition, the total power in
the array elements must be maintained near saturation to
achieve the design ERP performance and maximum power
efficiency. However, the total power must not extend into the
transmitters nonlinear region that would degrade performance by producing intermodulation products and signal suppression. The challenge is to maintain the required amplitude
and phase excitation for each beam while allotting each beam
an equable share of the downlink transmit power.
Remote Sensing Antennas
Satellite antennas also include designs for remote sensing applications. Two types of sensors are used: active, or radar, and
passive, or radiometric, sensors. These systems are used in
low-orbiting satellites to provide both global coverage and
high resolution of terrain features.
Radar sensitivity is commonly measured by the power aperture product of the design, and thus, in addition to high
peak power levels, a physically large aperture size is required
(20). Antenna designs for active applications generally require deployment to achieve the required aperture size. A
conflict exists in these designs. Large apertures can be obtained from low frequency arrays where a reasonable number
of elements are required and the mechanical tolerance is not
overly stringent, but the resolution of these systems is lower
671
than higher frequency designs. The better resolution of the
higher frequency designs is accompanied by the requirements
of additional array elements and more stringent mechanical
tolerance. The size, weight, and prime power requirements
for these sensors make their development an ambitious effort,
but future systems designs, particularly those using synthetic
aperture processing, can be anticipated.
Passive sensors are commonly used for both meteorological
and terrain remote sensing (21). Two types of antenna designs are used in these applications. One type of antenna,
generally an offset reflector, is mechanically rotated at a constant angle from the satellites nadir to generate a conical
trace on the ground at a constant elevation angle. As the satellite proceeds in its orbit, a swath on the ground is swept
out. The requirements for these antennas include design attention to minimize insertion loss to achieve high sensitivity,
good polarization purity to discern the radiometric contrast
for oblique incidence angles used by these sensors, low sidelobes so that the emission received by the main beam is not
degraded by sidelobe returns, and a narrow beamwidth to obtain good resolution of terrain features. Commonly, these antennas are specified by their solid beam efficiency performance. This parameter is defined by the power received for
the specified polarization over a required angular region to
the total power received from all space assuming the antenna
is uniformly illuminated from all directions. This parameter
is defined to provide a means to quantify the degree to which
the power received by the sensor is isolated to its specified
coverage area. This system design commonly operates at several different frequency bands using the same aperture.
A second type of antenna has a much simpler design and
is used in sensors referred to as sounders. In this design, the
antenna boresight is aligned with the satellite nadir. These
designs operate at frequencies selected from peak of molecular absorption resonances, which occur at EHF frequencies.
Several different IF bandwidths are used in these designs. As
the bandwidth straddles more of the molecular resonance, the
absorption toward the bandwidth edges reduces from the
peak value at the resonant frequency. This reduced absorption at the bandwidth edges permits the sensor to view emission closer to the earths surface. Thus, by examining the
emission received at different IF bandwidths, a profile for different altitudes may be observed. The antennas used in these
sounder designs are generally corrugated horn designs selected for their low insertion loss and rotationally symmetric,
low sidelobe patterns.
USER SEGMENT ANTENNAS
User segment antennas also have requirements specific to
satellite applications. In addition to meeting G/T and ERP
requirements for link closure, the user antennas generally
need to align their main beams with the appropriate satellite.
Other design issues include the control of interference to and
the reduction of interference from other systems. With the
increased number of satellite users, terminal costs including
the antenna, transportability to alternative locations, and in
future personal communication systems, techniques to reduce
interactions with the surrounding environment form the principal development issues.
672
SATELLITE ANTENNAS
Radio Frequency Issues
The most common user antenna is a reflector design capable
of reasonable RF performance with a modest cost. Other antenna technologies, such as phased arrays, have more complexity which results in higher manufacturing costs, and
therefore are used when compelling reasons exist for their design. For example, aerodynamic constraints for high performance aircraft may dictate the development of conformal
array designs.
The trend toward increased numbers of users results in
the terminal costs having a significant fraction of the total
system costs. This trend in turn requires sufficient performance from the satellite to minimize terminal performance
requirements. An excellent example is provided by direct
broadcast satellite TV services where both the user terminal
size and costs have dramatically shrunk. More generally, the
present development of systems for VSATs (very small aperture terminals) illustrates this trend. A very important part
of system planning is to insure that adequate satellite performance is available to minimize user performance requirements, and thereby control the total system cost as well as
provide a minimum amount of equipment, particularly for
those users with mobility requirements.
The increased number of satellites on-orbit has also resulted in additional requirements for the antennas. Concern
regarding interference to and from satellites that are closely
spaced in orbit has resulted in sidelobe envelope requirements to reduce interference levels. These requirements are
mandated on antenna manufacturers. A typical example of
these requirements has been levied by the CCIR [International Radio Consultative Committee, technical branch of International Telecommunication Union (ITU)] (22), which results in a sidelobe envelope constraint that depends on the
electrical size of the antenna.
Sidelobe control is one antenna technique that reduces interference for both reception and transmission. Another technique for reducing received interference is to construct an
adaptive cancellation system. A typical design for ground terminal applications called a sidelobe canceller (23) assumes the
desired signal is received by the main beam and that the interference arrives through the antenna sidelobes. The antenna hardware consists of a main reflector antenna and a
set of auxiliary antenna elements which are combined with
complex weighting values with the main reflector antenna to
cancel interference. The principal issue for this design approach is the dispersion inherent in the main reflectors sidelobe response. The sidelobe response of reflector antenna includes radiation from edge diffraction, direct feed radiation
and spillover, scattering and blockage contributions, and
leakage from the panels composing the reflector surface. The
complex sum of these individual radiation mechanisms varies
over the operating bandwidth so that effective cancellation
requires frequency-dependent weighting values. Such frequency-independent weighting values can be achieved by
adaptive equalization circuitry using a transversal filter with
adaptive weighting coefficients at each time delay tap. An example of this interference reduction approach may be found
in Ref. 24, which describes the design and the measured results of a demonstration of cancellation capabilities.
The trend toward increased microwave users and systems
extends beyond satellite applications. This trend also in-
creases the possibility of interference to user terminals from
nearby terrestrial systems. A particular concern is the possibility of out-of-band receiver compression from high-level
pulse systems such as radars which suppresses desired signal
levels and creates intermodulation products degrading receiver performance. Design attention must be paid to the receivers dynamic range and RF filtering to protect the receiver
from high-level interference. However, RF filtering while necessary has insertion loss that degrades sensitivity by reducing
the received signal and increasing the system temperature.
Inexpensive, low noise receiver front ends and relatively
low sky temperature values (25) provide the potential of low
system temperatures that increase system performance and
reduce the required antenna size. Controlling antenna spillover and wide angle sidelobe levels by both efficient rotationally symmetric feed designs and reflector geometry choice
produces low antenna temperature values (26). Low antenna
temperature values also require minimizing feed and transmission line losses. Filter loss even in a high Q media becomes the dominant insertion loss, but such receiver protection is necessary. These steps permit development of
terminals that meet the required G/T with a minimum size
antenna. These factors result in the popularity of offset reflector antennas for user applications. The efficiency of offset
reflectors is enhanced by the absence of feed blockage, their
high level spillover lobes are directed toward the lower sky
temperature instead of the ambient earth temperature, and
their low sidelobe levels minimize the antenna temperature
while also complying with sidelobe envelope requirements.
User antennas for future personnel communication systems face additional challenges. These antennas must provide
broad coverage so that antenna pointing is not required, particularly since the LEO and MEO orbits used in these systems produce ever-changing angular positions between the
user and the satellite. One problem inherent in broad-coverage antennas, as previously discussed for TT&C antennas, is
the sensitivity of the antenna performance to the environment surrounding the antenna. For the personal communication application, there exists a wide variety of environments
with varying degrees of multipath (i.e., scattering from humanmade and terrain features). In addition, link performance
can be degraded by foliage attenuation and blockage from terrain features. These development issues are being actively
pursued for both satellite systems and terrestrial cellular networks that operate at somewhat lower frequencies. Development of techniques to mitigate these degradations is also being actively pursued; techniques such as Rake receivers that
provide adaptive diversity combining and equalization are being evaluated.
Antenna Pointing Techniques
User antennas must also be aligned with the desired satellites for signal reception. The satellites orbital position is
specified by its ephemeris which describes the satellites altitude, eccentricity of orbit, inclination, longitude of ascension,
and time of epoch. The satellite ephemeris and the user location are required to determine the antenna pointing angles.
Typically, this information is transferred to the antenna control unit where the required pointing angles are computed
and the necessary commands to the antenna positioner are
issued. The requirements for antenna pointing depend on the
SATELLITE ANTENNAS
antenna beamwidth, the uncertainty in the satellites location, and the orbital dynamics. For communication applications, a pointing accuracy of one-tenth of a beamwidth is typically specified to minimize signal loss caused by pointing
errors.
In some cases, antenna pointing is trivial. For example,
the antennas for personal communication systems are purposely designed with broad coverage antennas so that the
user has no pointing requirements. Another common example
is a small user antenna for direct broadcast satellite reception
from geosynchronous satellites. These antennas are roughly
aligned using the users geographic location, verified by signal
peaking techniques, and rigidly secured. The combination of
the relatively wide beamwidth from the small antenna and
the stationary position of the geosynchronous satellite and its
station keeping results in a fixed pointing requirement.
In other applications where satellites have inclined orbits
or orbits that are not geosynchronous, the satellite undergoes
dynamic motion with respect to the user, and a means must
be provided to follow the satellite in orbit. For users with relatively broad beamwidths compared to the orbital uncertainty,
the knowledge of the satellite ephemeris and the user location
may be adequate to simply command the users antenna position. This open loop procedure is commonly referred to as program tracking.
If some uncertainty exists in the antenna pointing, such
as ephemeris data that are not current, another open loop
technique referred to as step track is commonly used to verify
correct alignment with the satellites position. The users antenna is pointed at the nominal position of the satellite as
might be obtained from program track. The antenna is then
commanded to move in equal and opposite angular positions
from this nominal pointing direction. If the antenna is correctly aligned with the satellite, the received signal level
should be reduced by the same amount at both angular offsets. If the signal levels at both positions are not identical,
the difference in the power level provides the angular correction to the nominal position. This process is repeated in the
orthogonal plane to properly position the antenna in two angular coordinates.
These two open loop antenna pointing techniques are commonly used together. The step track procedure is periodically
exercised to validate the correctness of the program track
pointing. A significant advantage of these techniques is that
minimal equipment is required. The antenna positioner is required and simple software commanding is needed to execute
the angular offsets. The power measurements at the different
angular offsets may be obtained from a simple measurement
of the receivers automatic gain control (AGC) voltage with
the appropriate calibration and linearity verification.
When user antennas have beam widths that approach ten
times the uncertainty in satellite position or a significant
amount of orbital dynamics exist, closed loop tracking techniques are required. These techniques are commonly used in
radar systems to locate targets and are referred to as monopulse designs. This term is derived from radar applications
because the tracking information is obtained from each radar
pulse. In radar applications, the received signal level varies
as the target changes aspect angle and typically has a significant dynamic range. Thus, tracking information is derived
from each radar pulse so that the dynamics of the target return do not degrade antenna pointing performance. Mono-
673
pulse systems for communication applications have easier requirements than those for radar systems. The received signal
level has a relatively constant amplitude, so that the monopulse signals can be sequentially sampled reducing hardware
requirements. The antenna pointing accuracy depends on
aligning the antenna with sufficient accuracy to minimize signal loss as compared to locating a radar target with the maximum precision practical.
Monopulse systems operate by forming two types of antenna beam shapes: a sum beam that receives the desired signal and a difference beam that has a null on the antenna
boresight axis. The signal power in the sum beam is maximized by positioning the antenna to the null of the difference
pattern. The ratio of the signal levels in the sum and difference beams is independent of the power density of the received signal, and provides a measure of the angular displacement from the antennas axis. This ratio can be used in a
closed loop system so that the antenna will align to the received signal and will follow any variations in the signal location.
Two different types of feed designs are commonly used to
generate the sum and difference beams. One feed design (27)
uses a conventional feed for the sum beam and an additional
four small feed elements to produce the difference beam. The
second design (28) uses a higher order waveguide mode to
produce the difference beam which is sampled by couplers. In
both cases, the difference beam outputs are coupled into the
sum channel and switched sequentially. The resulting AM
modulation on the sum signal is processed to derive the antenna pointing measurements.
SATELLITE ANTENNA TESTING
While antenna testing is widely developed for many applications, satellite antenna systems pose some unique testing
challenges. Rigorous testing is required for spaceborne antenna systems because of their reliability requirements and
the inability to service on-orbit antennas. This discussion illustrates the trends toward increased integration of antennas
with payload electronics, and testing where the antenna performance depends on both the antenna hardware and system
electronics further increases testing complexity. The user segment also follows the trend of increased integration with terminal requirements. Finally, the demands for capacity also
result in more stringent than normal testing requirements,
such as validating the polarization purity requirements for
systems using polarization reuse.
Spaceborne antenna testing generally has three distinct
phases:
Development testing that demonstrates design compliance
with system specifications
Qualification testing that ensures flight hardware not only
meets the RF performance demonstrated in development testing but also is capable of surviving the launch
and on-orbit environments and meets weight and power
requirements
On-orbit testing that validates system compliance with
system specifications and provides diagnostic evaluation
of system shortfalls during the orbital lifetime.
674
SATELLITE ANTENNAS
These three test phases have a scope that greatly exceeds the
normal testing to define the component level performance of
antennas. This situation is particularly true with the trend
toward incorporating payload electronics into the antenna
system which results in the antenna system performance that
depends not only on the antenna components but also the
electronics performance.
The development testing has the objective to demonstrate
the designs fully comply with the system specifications. The
testing in this phase of the program is most closely related
to conventional antenna testing and quantifies gain values,
pattern characteristics, polarization purity, impedance properties, bandwidth variations, etc., as is typically performed.
These RF characteristics can generally be performed using
conventional RF test facilities and general purpose instrumentation. The antenna components are designed on a prototype or engineering model basis, so that weight projections
and compliance with thermal and mechanical requirements
for launch can also be evaluated.
The qualification testing has the objective of determining
whether or not the flight hardware is capable of the performance established for the design whose compliance has been
demonstrated in the development testing. An important part
of the program test planning is a requirements flowdown that
identifies the testing necessary to demonstrate overall system
compliance and assurance of flight worthiness. The principal
concern in this phase of the testing is how to perform the
testing without risk to flight hardware. Moving flight hardware to conventional antenna test facilities poses an unacceptable risk in many cases. Testing very lightweight antenna
designs may be precluded because they can be destroyed by
wind or become contaminated and the effects of gravity can
degrade the ability to project on-orbit performance. The development of portable near field test facilities is viewed as a need
to permit detailed antenna testing in payload assembly areas.
Another important part of the test planning in the early part
of the program is the identification of test ports for system
evaluation and the required development and calibration of
any specialized test fixture. Qualification testing must also
evaluate performance variations in the thermal variations
and vacuum conditions, and the ability to survive the shock
and vibration levels experienced during launch. Other issues
such as the reliability of deployment mechanisms if used
must be demonstrated.
On-orbit testing is conducted using specialized test terminals, which are often a part of the satellite control networks.
The testing at this phase is generally conducted shortly after
the satellite arrives at its orbital location. This testing generally concentrates on system level performance parameters,
such as uplink G/T, downlink ERP, and payload antenna
pointing alignment, to establish overall performance compliance. After this initial performance evaluation, these test
facilities are used to periodically assess on-orbit performance,
and together with the telemetry information provide a means
for on-orbit diagnostics. An important part of the test planning is an examination of the adquacy of the telemetry and
test terminal information for identifying failed components
and their potential substitution with redundant components.
Satellite systems also have some specialized testing requirements peculiar to their environment. One problem with
satellite systems is the static charging of components on-orbit. Typically, this charging occurs on dielectric materials
used for thermal protection not only for the antenna but also
other satellite subsystems. The charge builds up on these
components until discharge occurs. This phenomenon is referred to as ESD (electrostatic discharge), and the spectral
components of the discharge may have sufficient intensity to
interfere with payload receivers. The ESD spectral content is
typically measured by embedding a sample of the material in
an electron beam and using a pickup probe antenna and a
spectrum analyzer or very wide bandwidth oscilloscope.
Another problem which results from high power operation
in vacuum conditions is multipaction. In this case, the RF
energy may be sufficient to strip electrons from the surfaces
of materials exposed to the high power, which will damage
the component. The potential for multipaction is typically
evaluated by testing the components associated with high
power in an evacuated bell jar and using a spectrum analyzer
to observe any RF noise associated with multipaction. Such
testing can also evaluate any power handling and microarcing
limitations of the components by examining the components
after the test for damage. Temperature increases from highpower operation can also be measured.
A third problem is also associated with high-power operation and is referred to as PIM (passive intermodulation). This
problem results from exposing junctions with high power. Dissimilar metals or contamination forms weak diode junctions
whose nonlinearity generates intermodulation products. Such
junctions can occur in transmission line components, filters
and diplexers, and joints within the antenna structure. Those
components exposed to high power are tested by injecting two
tones at high-power levels (higher power transmitters than
the operational ones are sometimes used to insure the observed intermodulation does not result from products generated by a saturated transmitter) and observing any intermodulation products generated by the components under test
with a spectrum analyzer.
User antenna testing also has some unique issues for satellite systems (29). One problem results from testing on ground
terminal antennas, where the physical size of the antenna
precludes testing in conventional antenna test facilities. A
test technique using the emission from astronomical sources
is used to measure the G/T (30). The flux density of these
emissions has been well established, and a variety of sources
exist. The emission from the sun has a high level, but the
angular width of the solar disk limits this technique to relatively wide beam widths. Other sources such as the moon and
radio stars such as Cassiopeia A are used for narrower beam
widths. The positions of these sources are also well known,
and thus, the opportunity to validate the accuracy of the antenna positioning systems is also available. The antenna gain
in the transmit bandwidth can also be established by connecting a low noise preamplifier in place of the transmitter, measuring the G/T, separately measuring the system temperature, and multiplying to obtain the transmit antenna gain.
While this technique was developed for large ground terminal
antennas (31), with the available low noise preamplifiers, useful measurements can be made on much smaller antennas.
This technique is particularly useful for those antennas that
follow the trend of integrating the antenna feed and receiver
front end without connectors for test purposes.
Other electromagnetic measurements are required in the
development of both user and spaceborne systems to validate
EMI/EMC (electromagnetic interference/electromagnetic
SATURABLE CORE REACTORS
675
compatibility) issues. Part of the system planning for satellite
systems involves frequency planning to avoid such problems;
measurements at the assembly level are performed to assure
on-orbit performance will not be degraded by interference between subsystems. User terminal designs are also evaluated
to insure the design is not susceptible to outside interference
as well as not creating interference to other systems that may
be located nearby.
16. P. Ingerson and C. A. Chen, The use of non-focusing aperture for
multibeam antennas, 1983 IEEE AP-S Symp. Dig., pp. 330333.
SUMMARY
20. Special issue on SIR-C/X-SAR, IEEE Trans. Geosci. Remote Sens.,
33: 1995.
Satellite antenna development is a significant part of present
day antenna technology. Much progress in antenna systems
for both the space and user segments has been made. However, future requirements such as the development of satellite
systems for personal use provide a rich opportunity for further development and challenges to meet the increased performance levels demanded in future systems and applications.
21. K. Tomiyasu, Remote sensing of the earth by microwaves, Proc.
IEEE, 62: 8692, 1974.
BIBLIOGRAPHY
1. F. Ananasso and F. D. Priscoli, The role of satellites in personal
communication services, IEEE Selected Areas Commun., SAC-13:
180195, 1995.
2. A. W. Love, Electromagnetic Horn Antennas, New York: IEEE
Press, 1976.
3. C. T. Brumbaugh et al., Shaped beam antenna for the global positioning satellite system, 1976 IEEE AP-S Symp. Digest, Amherst,
MA: pp. 117120, October 1115, 1976.
4. K. Ueno et al., Design and characteristics of a multiband communication satellite antenna system, IEEE Trans. Aerosp. Electron.
Syst., AES-31: 600607, 1995.
5. T. Katagi and Y. Takeichi, Shaped beam horn reflector antenna,
IEEE Trans. Antennas Propag., AP-23: 757763, 1982.
6. O. M. Bucci et al., Antenna pattern synthesis: A new general approach, Proc. IEEE, 82: 358371, 1994.
7. J. T. Mayhan, Area coverage adaptive nulling from geosynchronous satellites: Phased arrays versus multiple beam antenna,
IEEE Trans. Antennas Propag., AP-27: 410419, 1986.
8. K. M. SooHoo and R. B. Dybdal, Resolution performance of an
adaptive multiple beam antenna, IEEE MILCOM 89 Symp. Dig.,
Session 7, Boston, MA.
9. R. B. Dybdal, Multiple beam communication satellite antenna
systems, IEEE ICC Conf. Dig., 27D-1-5, Minneapolis, MN: June
1719, 1974.
10. A. W. Love, Reflector Antennas, New York: IEEE Press, 1978.
11. C. M. Rappaport, An offset bifocal reflector antenna design for
wide angle beam scanning, IEEE Trans. Antennas Propag., AP32: 11961204, 1984.
12. R. Jorgensen, P. Balling, and W. J. English, Dual offset reflector
multibeam antenna for international communications satellite
applications, IEEE Trans. Antennas Propag., AP-33: 1304
1312, 1985.
13. C. M. Rappaport and A. I. Zagloul, Optimized three dimensional
lenses for wide angle scanning, IEEE Trans. Antennas Propag.,
AP-33: 12271236, 1985.
14. R. J. Mailloux, Phased Array Antenna Handbook, Norwood, MA:
Artech, 1994.
15. K. S. Rao et al., Development of a 45-GHz multiple beam antenna
for military satellite communications, IEEE Trans. Antennas Propag., AP-43: 10361047, 1995.
17. F. Carducci and M. Francesi, The ITALSAT satellite system, Int.
J. Satellite Commun., 13: 4981, 1995.
18. R. Miura et al., Beamforming experiment with a DBF multibeam
antenna in a mobile satellite environment, IEEE Trans. Antennas
Propag., AP-45: 707713, 1997.
19. T. Gebauer and H. G. Gockler, Channel individual adaptive
beamforming for mobile satellite communications, IEEE J. Selected Areas Commun., SAC-11: 439448, 1995.
22. CCIR Recommendation 465-3.
23. P. W. Howells, Intermediate frequency sidelobe canceller, U.S. Patent 3.202,990, August 24, 1965.
24. K. M. SooHoo and W. Masenten, Adaptive sidelobe canceller designs for large earth terminals, 1984 IEEE MILCOM Symp. Dig.,
paper 40.5 (Classified).
25. E. K. Smith, Centimeter and millimeter wave attenuation and
brightness temperature due to atmospheric oxygen and water vapor, Radio Sci., 17: 14551464, 1982.
26. R. C. Rudduck and K. Lambert, Calculation and verification of
antenna temperature for earth-based reflector antennas, Radio
Sci., 27: 2330, 1992.
27. J. DiTullio, A high performance feed for earth station satellite
communications, 19th Annu. USAF Antenna R&D Symp. Dig.,
Robert Allerton Park, IL: October 1517, 1968.
28. Y. H. Chong, K. R. Goudey, and L. O. Bryans, Theory and design
if a Ku-band TE21 mode coupler, IEEE Trans. Microw. Theory
Tech., MTT-30: 18621866, 1982.
29. R. B. Dybdal, Measurement of satellite ground terminals, 1991
AMTA Symp. Dig., Boulder, CO: 69, 13, October 711, 1991.
30. J. W. M. Baars, Measurement of large antennas with cosmic radio sources, IEEE Trans. Antennas Propag., AP-21: 461473,
1973.
31. D. F. Wait, Precision measurement of antenna system noise using radio stars, IEEE Trans. Instrum. Meas., IM-32: 110116,
1983.
R. B. DYBDAL
The Aerospace Corporation
SATELLITE SYSTEMS, MOBILE. See MOBILE SATELLITE
COMMUNICATION.
SCANNING ANTENNAS
SCANNING ANTENNAS
Electronically scanning antennas are usually arrays of elements, with the elements excited in a progressive phase
across the array in a given direction; this scans the beam in
that direction. Such an array is called a phased array. The
array usually produces a narrow (pencil) beam, with modest
or low sidelobes. Arrays that produce a shaped beam are usually not scanned. (Related articles are ANTENNA ARRAYS,
MULTIBEAM ANTENNAS, and ACTIVE ANTENNAS.) Advantages of
phased arrays include the ability to scan and track at rapid
rates, the ability to easily and quickly modify beam shape,
and the flat physical form of the planar array. Mechanically
steered arrays or dishes require a much larger volume. Applications are eclectic, including communication and tracking of
nonstationary satellites, earth mapping synthetic aperture
radar from aircraft or satellite, airport surveillance radar,
ship and aircraft radar and electronic countermeasures, and
finally missile defense.
Important array factors for the systems designer are pattern, gain versus angles, element input impedance, and efficiency. In general, each element of an array will have a different input impedance. For a fixed beam array these are called
embedded impedances; the obsolete and misleading term active impedance is deprecated. Not only does a scanning array
have different element impedances, but each of them varies
with scan angle. These element input impedances are called
scan impedances. The pattern of array gain versus angles is
called scan element pattern; this term replaces active element
pattern. The scan element pattern (SEP) is an extremely useful design factor. The element pattern and mutual coupling
effects are subsumed into the scan element pattern; the over-
685
all radiated pattern is the product of the scan element pattern, and the pattern of an isotropic array of elements that is
scanned to the proper angle. The isotropic array factor incorporates the effects of array size and array lattice, while the
scan element pattern as mentioned incorporates element pattern, backscreen if used, and mutual coupling. Since the scan
element pattern is an envelope of array gain versus scan
angles, it tells the communication system or radar designer
exactly how the array performs with scan, whether blind
angles exist, and whether matching at a particular scan angle
is advantageous. At a blind angle the array has zero gain.
Scan element pattern is used for antenna gain in the conventional range equations. Use of infinite array scan element patterns allows array performance to be separated into this SEP
and edge effects.
Element options for scanning arrays are mostly low gain:
waveguide slots, dipoles, patches, flat TEM horns, and open
waveguides. This is because element spacing of roughly halfwave is needed to obviate grating lobes. Elements such as the
spiral, helix, log-periodic, Yagi-Uda, horn, and backfire are
too large for scanning arrays, except for arrays with very
small scan range. It is important to note that array elements
may be modified for scan impedance compensation, as discussed below.
This article presumes some basic knowledge about antennas; the reader may wish to consult these articles: ANTENNA
THEORY, ANTENNAS, APERTURE ANTENNAS, RADAR ANTENNAS, and
WAVEGUIDE ANTENNAS.
This article comprises four sections: phased array basics,
scanning techniques, mutual coupling effects, and finite
arrays. A more extensive treatment of all of these subjects is
given in the book by Hansen (1).
PHASED ARRAY BASICS
Notation used here is that common in the literature: is
wavelength, k 2/ , and angular variables are u and v
where
u = sin cos sin 0 cos 0
v = sin sin sin 0 sin 0
(1)
Here the spherical coordinates are , , and the beam position
is 0, 0. It is sufficient to consider a planar array with even
numbers of elements, N and M, along the x and y axes. The
pattern is then
F (u, v) =
M/2
N/2
Anm cos n 12 kdx u cos(m 12 kdy v
(2)
n=1 m=1
with the elements spaced dx and dy in a rectangular lattice.
Equilateral triangular lattices are sometimes used, because
they reduce the element density by 17%. The element pattern,
which multiplies Eq. (2), has been omitted. For uniform excitation the pattern becomes simply
F(u, v) =
sin 12 Nkdx u
N sin
1
2 kdx u
sin 12 Mkdx yv
M sin 12 kdy v
(3)
Use of low sidelobe designs, such as the Taylor n, or the Taylor one-parameter, requires the Anm coefficient amplitudes to
be properly chosen (2). The coefficients contain the interele-
J. Webster (ed.), Wiley Encyclopedia of Electrical and Electronics Engineering. Copyright # 1999 John Wiley & Sons, Inc.
686
SCANNING ANTENNAS
ment scan phase:
The bandwidth is then given by
exp[ jkd(nu + mv)]
(4)
BW
This is the progressive phase shift needed to scan the beam.
0.866
L sin 0
(9)
for a uniform array. When the beam angle is 30, the commonly used formula for fractional bandwidth results:
Beamwidth
The half-power points on a uniform array principal plane pattern are found by putting Eq. (3) equal to 0.5. This gives,
for all but very small arrays, the 3 dB beamwidth 3 as
arc sin(sin 0 0.4429/Nd)
(5)
For large N, this reduces to
3 0.8858/Nd cos 0
(6)
Cross-plane beamwidth is nearly constant for large N.
Interelement phase shift is necessary to provide beam scan;
the devices that produce this phase shift are called phasers.
Row and column phasing is the simplest, even for circular
planar arrays. Thus each phaser is driven by a command to
produce a specified x and a specified y-axis phase. The steering bits affect the precision of the beam steering, control
bandwidth, and produce phase quantization lobes. Each of
these will be discussed.
The smallest steering increment is related to the smallest phase bit: / 3 1/2M, where 3 is the half-power beamwidth. Note that in this case the largest bit is . Thus four
bits gives a steering least count of 0.0625 beamwidth, or 1/16
of a beamwidth. Adding a bit, of course, decreases the steering increment by a factor of two.
Scan Bandwidth
Bandwidth of an array is affected by many factors, including
change of element input impedances with frequency, change
of array spacing in wavelengths that may allow grating lobes,
change in element beamwidth, and so on. When an array is
scanned with fixed units of phase shift, provided by phasers,
there is also a bandwidth limitation as the position of the
main beam will change with frequency. When the array is
scanned with true time delay, the beam position is independent of frequency to first order. But with fixed phase shift,
the beam movement is easily calculated. Beam angle is simply related to scan angle 0 by sin ( f0 /f) sin 0. For example, a beam at 45 moves from 42.2 to 47.9 over a 5% frequency excursion.
To calculate steering bandwidth assume that the main
beam has moved from scan angle 0 to the 3 dB points for
frequencies above and below nominal. Let subscripts 1 and 2
represent the lower and upper frequencies. Fractional bandwidth is then given by
f2 f1
sin 0 (sin 2 sin 1 )
=
f0
sin 1 sin 2
As a result, long arrays have smaller bandwidth in terms of
beam shift at the band edges.
Adding bits of time delay increases this bandwidth; each
bit slightly more than doubles the bandwidth as it must, since
the number of bits that provide kL sin 0 gives infinite scan
bandwidth (1).
The array pattern given above allows the inference that a
maximum pattern value of unity occurs whenever (d/ )u
n. If d/ and 0 are chosen properly, only one main beam will
exist in visible space, which is for 90 90. Large
spacings will produce additional main beams called grating
lobes (GL); this is because the larger spacing allows the wave
from each element to add in phase at the GL angle as well as
at the main beam angle. The equation for GL is easily determined:
n
d
=
sin 0 sin gl
(11)
For half-wave spacing, a GL appears at 90 for a beam
scanned to 90. A one wavelength, spacing allows GL at
90 when the main beam is broadside. The onset of GL versus scan angle is shown in Fig. 1. The common rule that halfwave spacing precludes GL is not quite accurate, because part
of the GL may be visible for extreme scan angles.
The pattern of a rectangular lattice with scanning in either
principal plane behaves exactly like that of a linear array,
just described. For other scan angles the situation is less simple. The uv plane, sometimes called the direction cosine
1
Grating lobe angle
0.9
0.8
30
45
0.7
60
75
90
(7)
For large arrays we have
1
3
BW 2
=
sin 0
sin 0
(10)
Grating Lobes
Scan Accuracy
BW =
2
L
d/
3 = arc sin(sin 0 + 0.4429/Nd)
BW
0.6
0.5
0
(8)
10
20
30
40
50
60
Scan angle (deg)
70
80
90
Figure 1. Grating lobe appearance versus element spacing and scan.
SCANNING ANTENNAS
Elements
Phasers
in
Parallel phasers
Series phasers
Figure 2. Linear series feeds.
plane, was developed by Von Aulock (3) and is extremely useful for understanding GL behavior. The GL positions can be
plotted in the uv plane; they occur at the points of an inverse
lattice, that is, the lattice spacing is /dx and /dy. All real
angles, representing visible space, are inside or on the unit
circle. The latter represents 90. Angles outside the unit
circle are imaginary, or in invisible space. When the main
beam is scanned, the origin of the uv plot moves to a new
value, and all GL move correspondingly. However, the unit
circle remains fixed. For further details on this and on hexagonal lattices, see Hansen (1).
GL can be partially suppressed in one plane of scan for
a limited scan range, and over a narrow frequency band, by
pseudorandomizing the position of each element. However,
the complete two-dimensional pattern, over even a modest
bandwidth, will have GL.
SCANNING TECHNIQUES
Phaser Scan
Linear series feeds may be scanned by use of phasers in either the feed line or adjacent to the elements; these are series
phasers and parallel phasers. Figure 2 sketches these arrangements. With waveguide slot array sticks there is no
room for phaser devices. Many other configurations do allow
phasers to be incorporated. An advantage of parallel phasers
is that each phaser handles only 1/N of the transmitted
power, and each element path has the phaser loss. A disadvantage is that the basic interelement phase shift kd sin
0 must be multiplied by the number of the element. Because
almost all phasers are now digitally controlled, this is a minor
disadvantage. Modern digital controls are inexpensive. For
series phasers the advantage is that all phasers have the
same setting equal to . Disadvantages are that the first
phaser must handle almost the entire power and that the
phaser losses are in series. In both topologies there are systematic errors due to impedance mismatch. Amplitude taper
for sidelobe control is provided through coupling adjustments.
A detailed treatment of feed systems is given by Hansen (1).
A word about phasers is appropriate, although a full discussion is outside the scope of this article. Phasers, often
called phase shifters, are primarily of two basic types:
switched line and ferrite. Switched line phasers use semiconductor switches to switch in or out lengths of transmission
line, where the latter is usually coax, microstrip, or stripline.
687
Discrete components may be used, but more often the
switches, loads, and controls are in integrated circuit packages that are integrated into microstrip or stripline. MMIC
configurations are becoming more common, especially above
10 GHz. Of course, these phasers utilize time delay, but because the largest bit is usually 180, they are classed as
phasers. Digital phasers are binary: a 5-bit phaser has phases
of 180, 90, 45, 22.5, and 11.25. The phaser design must
keep the impedances matched as the bits switch, and in extreme applications it must also match the losses. If amplifiers
are incorporated to offset losses, the phasers become nonreciprocal.
Ferrite phasers are of two types, and although these are
intrinsically analog phasers, current practice has digitized
the phase control. They are almost always embodied in waveguide. Toroidal ferrite phasers use one or several ferrite toroids of rectangular shape placed in a waveguide such that
the axis of the toroid is along the guide axis. A drive wire
threaded through the toroid provides either a positive or negative current pulse. The pulse drives the toroid to saturation,
and the toroid is latched at a positive or negative remanentinduction point. The difference between the electrical lengths
of the two states provides the phase shift. Typically, several
toroids are placed serially with lengths chosen to provide the
N bits of phase shift. Advantages and disadvantages are obvious: These phasers are nonreciprocal, and they must be reset
after each pulse for radar operation; drive power is used only
when switching, so that power and heat dissipation are minimized.
The Faraday rotation ferrite phaser uses a ferrite rod
along the axis of the waveguide, with quarter-wave plates at
each end to convert linear to circular polarization and vice
versa. In practice, the ferrite rod is plated with silver or gold
to form a small ferrite waveguide. This ferrite guide is then
placed inside the regular guide, with nonplated tapered transition sections on each end to match between the two guides.
An external solenoidal coil around the waveguide provides the
magnetization that controls the phase shift. Again, typical
drive circuits are digitally controlled. These phasers are reciprocal and are often used in high-power applications such as
radars. Sophisticated drive compensation circuits have been
developed to adjust the drive as the temperature of the ferrite
increases; the drive power must be applied at all times. Useful references on phasers include a special issue of MTT
transactions on array control devices (4), a chapter on phasers
and delayers (5), and a two-volume monograph (6).
Corporate feeds are common in arrays of dipoles, open-end
guides, and patches and are named after the structure of organization charts, where the feed divides into two or more
paths, then each path divides, and so on. Such feeds are commonly binary, but sometimes the divider tree includes threeway, or even five-way dividers, depending upon the number
of array elements. Figure 3 shows a simple binary parallel
feed with phasers. As in the case of main-line series feeds,
the phase shift needed is progressively larger. Now, however,
each path length has only one phaser. For wideband applications, where modulo 2 phase shift is inadequate, time delay
units (delayers) can be used, as discussed below. Figure 4
shows a planar patch array, where the microstrip corporate
feed lines are visible. Switched-line 3-bit phasers can also be
seen.
In a distributed array each element is connected to its own
receiver/transmitter module. Such models often contain
688
SCANNING ANTENNAS
Elements
1
Phasers
Power
dividers
Figure 3. Parallel (corporate) feed. The numbers are phase shift
units.
diplexers, circulators, filters, preamps, power amplifiers,
phasers, and control components. Sometimes the element is
part of the module. Several advantages accrue. Many lowpower semiconductor sources may be used instead of a single
high-power tube source. Feed network loss, which reduces
S/N, is minimized. Graceful degradation allows system operation with slight gain and sidelobe changes as modules fail.
MTBF is greatly increased. Against these advantages is the
higher cost. The nominal few-hundred-dollar module has been
a long-sought goal, but nonrecurring engineering costs are
high, especially for MMIC, so that a large number of production units is needed to make the modules cost-effective. Solidstate and tube modules are discussed by Ostroff (7); the current art is reviewed by Cohen (8). Figure 5 shows four generations of x-band modules that feed circular waveguide and
TEM horn radiators.
Time Delay Scan
Wideband scanning requires time delay, with the maximum
amount of delay equal to (L sin 0)/c, where the array length
Figure 4. Corporate fed patch array. (Courtesy of Ball Aerospace &
Communications.)
Figure 5. Four generations of modules. (Courtesy of Dr. Chris
Hemmi and Texas Instruments.)
is L, the maximum scan angle is 0, and c is light velocity.
For example, a 100 array scanned to 30 needs up to 50 of
delay, or 7 bits of time delay from to 64. These bits are in
addition to the phaser bits below ; 5 bits of phase gives bits
of 11.25 to 180, making a total of 12 bits to control the array
scanning. As mentioned above, many phasers utilize switched
lengths of transmission line; time delayers use switched coax
or stripline or microstrip. The major problems with microwave delayers are size and loss. Size and weight are important factors, but loss is critical. When a long delay bit is
switched in, an amplifier must be controlled to offset the delay loss, so that the array amplitude distribution is not degraded (see article entitled APERTURE ANTENNAS). Long-term
and environmental stability are prime concerns with such
long delays. An alternate to microwave transmission line delayers is photonic delay, where the delay in a fiber-optic line
is used.
Photonic delay lines are used much as microwave delay
lines are, with switches to select various delay segments. Of
course, optical modulators and detectors are needed; these are
discussed in Zmuda and Toughlian (9) and Hansen (1). In the
binary delay chain, semiconductor optical switches are used
to switch in or out delay segments that are binary multiples
of a minimum delay, say 1. An n-bit delay chain then provides 1, 2, 4, . . ., 2n1 delays. If phasers are not used for
delays below 1, the photonic delay chain can include these
also. An efficient use of photonic delay is for subarrays only,
where each antenna element is connected to a conventional
phaser, with largest phase bit of 180; each subarray is then
connected to a photonic delay chain.
Problems encountered in the utilization of photonic delay
are connected with uniformity, stability and drift, and IP3.
Unfortunately a 1 dB optical nonuniformity produces a 2 dB
microwave error. This is because optical intensity produced
by a modulator is proportional to microwave current or voltage. IP3 is the third-order intercept point, and it applies to
the amplifiers used to offset link loss. An amplifier (extrapolated) gain curve intersects the third-order product gain curve
at IP3. The spur-free dynamic range is from the gain curve to
the IP3 curve, at the noise floor.
SCANNING ANTENNAS
An alternative to the use of switched series delays is an Nway switch that selects one of N delays. The optical switch
is an acousto-optical Bragg cell, in which an applied microwave signal launches a traveling acoustic wave in the
acousto-optic crystal. An incident optical beam is modulated
by the phononphonon interactions and is deflected by an
angle proportional to the microwave drive frequency. The optical beam then impinges on one of a cluster of optical fibers,
each of which has a binary delay value. Outputs of all the
fibers are collected in an optical power combiner, which in
turn feeds a photodetector (10).
Delay can be selected by choosing the proper light wavelength, through use of dispersive fibers or through use of frequency selective gratings. The RF-modulated optical beam is
sent through a length of highly dispersive fiber; the incremental delay is proportional to the wavelength shift of the laser
carrier (11). A representative chromatic dispersion is 50 ps
per millimeter of wavelength shift per kilometer of fiber. This
dispersion is approximately constant over the laser tuning
range. A variable delay can be obtained simply by tuning the
laser wavelength, with the delay produced by a length of dispersive fiber. A simple (but expensive) array feed would have
a laser oscillator for each array element, with each laser output going through a dispersive fiber and then to a photodetector to the element. Dispersive fiber lengths are of the order of
1 km to 10 km. As mentioned elsewhere, the delay must be
stable and accurate to a small part of the smallest phaser bit
in the array.
A more elegant architecture utilizes one tunable laser (per
plane of scan) with the modulated output fanned out to a cluster of dispersive fibers. These would be designed for delays of
, 2, 4, and (2N 1), and each would drive one array element. Tuning the laser would change and thus scan the
beam. For large arrays the fan-out loss can be appreciable.
For two-dimensional scan, two tunable lasers would be used,
and the azimuth and elevation delays would be combined at
the element photodetectors (12,13).
A grating can be made in a length of fiber by focusing ultraviolet energy onto a periodic sequence of spots along the
fiber, thereby producing a grating of spots of different index
of refraction. Such a grating will reflect an optical wave at
the resonance frequency of the grating. A series of Bragg gratings can be spaced along a fiber, with each reflecting at a
different optical wavelength. An optical signal enters the fiber
with gratings through an optical circulator; the reflected signal path length (delay) depends upon which grating is selected by the optical frequency. The circulator output then
goes to an element, through a photodiode. Thus time delay is
controlled by optical wavelength. It should be noted that the
photonic delay art is relatively recent, while the microwave
feed/phaser art is mature.
Frequency Scan
Advantage can be taken of the beam squint with frequency,
by increasing the interelement path length such that a small
frequency change scans the main beam. Such arrays are traveling wave arrays, as sketched in Fig. 6. Assume that the
radiating elements spaced d apart are connected by a serpertine (snake or sinuous) feed of length s and wave number .
The wavefront is defined by
689
Crossguide
directional coupler
Terminators
Sinuous
feed
Slots
s
d
Linear
arrays
Input
Figure 6. Frequency-scanning sinuous feed feeding an array of traveling-wave linear arrays.
kd sin m = s 2m
(12)
To avoid an endfire beam, kd s 2m. More than one
main beam can be avoided by keeping
1
d
<
1 + sin m
(13)
The phase equation is often written as
sin m =
s
m
dg
d
(14)
where g is the serpentine feed wavelength. Clearly larger
s/ , and correspondingly larger m, gives a faster change of
beam angle with frequency. If the frequency scan passes
through broadside, there will be a significant change in impedance due to the addition of all the element conductances.
Element conductances, or couplings, are determined just as
for a TW array. If the waveguide bends needed to make a
snake are not well-matched over the frequency band, it may
be necessary to calculate the input admittance using the techniques developed for resonant arrays. The coupler reflections
tend to produce a sidelobe at the conjugate beam direction;
the level of this reflection sidelobe depends upon the coupling
conductances, and it is higher when the scan is closer to
broadside. The signal instantaneous bandwidth must be sufficiently small to avoid beam broadening. Figure 7 shows an
early airport surveillance radar antenna.
Quantization Effects
Phaser Quantization Lobes
Most phasers are now digitally controlled, whether the intrinsic phase shift is analog or digital. Such phasers have a least
690
SCANNING ANTENNAS
lobes called quantization lobes (QL), which have predictable
amplitudes and positions. An N-element array has an end-toend phase of (N 1) kd sin 0. The number of elements per
phase step, with M bits of phase, is J:
J=
N
(N 1)2M (d/) sin 0
(17)
This allows the largest scan angle for which J 2 to be found.
For large arrays N approximately cancels, leaving sin 0
(1/((d/ ) 2M1). Larger angles produce pattern errors that are
more complex. For J between one and two, there is a transition region between the random sidelobes regime (J 1) discussed and the QL regime.
The QL angle q are governed by the step width W:
1
W
= M
2 sin 0
(18)
This gives a peak sawtooth error of 2M, as mentioned
above. The QL amplitudes are given by sinc ( i), where
i 0 gives the main beam. The first two QL and the main
beam are given by
Main beam:
First QL:
Second QL:
Figure 7. Airport surveillance antenna. (Courtesy of J. J. Lee and
Hughes Aircraft Company.)
phase, corresponding to one bit. An M-bit phaser has phase
bits of 2/2M, 2/2M1, . . ., . The ideal linear phase curve for
electronic scanning is approximated by stair-step phase, producing a sawtooth error curve. Since the array is itself discrete, the position of the elements on the sawtooth are important. There are two well-defined cases. The first case is when
the number of elements is less than the number of steps. In
this case the phase errors assume a random nature. This can
be evaluated by approximating the phase variance by onethird of the peak sawtooth error of /2M. The variance is
2 =
2
3 4M
(15)
For uniform excitation of the array the rms sidelobe level is
2 /N. A modest gain decrease accompanies the increased sidelobes:
G
1 2
G0
sinc
sin
sin
2
(19)
When /2, the main beam and first QL are equal, clearly
a bad case. But this only occurs for one bit phasers, an extreme case. Table 1 gives lobe amplitudes versus number of
phaser bits.
Gain decrease is approximate given by the main beam decrease:
G
sinc2
Gp
(20)
These data allow the array designer to make an intelligent
trade on phaser bits.
For rectangular lattices, the QL appear along the u and
along the v axes at the GL angles derived for linear arrays.
There are several schemes for decollimation of phaser QL,
of which the best is the phase-added technique (14). In this
technique the element phases are calculated to a number of
bits that would provide sufficiently low QL if used. Call this
number of bits N. The phasers are driven by a smaller number of bits M. A set of random numbers, one for each phaser,
with each number N bits long, is generated and stored. The
Table 1. Phaser Quantization Lobe Amplitudes
(16)
At scan angle 0 the main beam is reduced by approximately
cos2 0, so that the rms sidelobe level with respect to the main
beam is increased. The second case has two or more elements
per phase step, and the discrete (array) case is approximated
by a continuous case. This quantization produces a set of
Main Lobe
QL1
QL2
1
2
3
4
5
6
3.92
0.912
0.224
0.056
0.014
0
3.92
10.45
17.13
23.58
29.84
35.99
13.46
14.89
19.31
24.67
30.38
36.26
SCANNING ANTENNAS
random number table is then truncated to M bits and stored
separately. When the set of phaser drive bits is calculated for
a given scan direction, the N-bit random numbers are added,
one to each phaser drive word. The phaser drive words are
then truncated to M bits and the M bit random numbers are
then subtracted, one for each phaser drive word. The resulting words are then used to drive the phasers. The result
of this is decollimation of the QL, along with a small rise in
the average sidelobe envelope. An important advantage is
that this method, unlike some others, has zero mean value.
Subarray Quantization Lobes
An array composed of contiguous subarrays, when scanned,
has a phase consisting of stair-steps, with one step over each
subarray. This, just as in the case of digital phasers, produces
quantization lobes, except that a subarray has at least two
elements. So the QL are always deterministic here. Call the
sawtooth width (subarray width) W and call the peak sawtooth error , as before. For large arrays the discrete subarray
may be replaced by a continuous aperture, but with the sawtooth phase. Consider N equal subarrays, and with uniform
excitation. The pattern is found by integrating over a subarray, and then summing over the subarrays.
The integration and summation can both be performed in
closed form, with the following result:
F (u) =
sin NW
sinc v
N sin W
(21)
This is immediately recognized as the pattern of a uniform
array of N isotropic elements spaced W wavelengths apart,
times the pattern of a uniform line source of length W wavelengths, with main beam at v 0. For W larger than one
wavelength, grating lobes will exist from the first factor.
These GL have unit amplitudes and positions for w 0, 1, 2,
. . .. Of these the first is the main beam. These lobes are
weighted by the sinc beam. The subarray variable is v,
and v (W/ ) sin 0. The QL amplitudes, and that of the main
beam, are formed from Eq. (21), using this subarray . The
QL locations are approximately at the GL angles:
sin gl = sin 0
m
,
W
m = 1, 2, 3, . . .
(22)
For example, (W/ ) sin 0 0.025 gives QL of roughly 32
dB; a value of 0.05 gives 26 dB; a value of 0.1 gives 20 dB.
These results comprise the reasons why these lobes are
called QL rather than GL. Although they occur at the GL
angles, their appearance and amplitude is a function of scan
angle. While GL appear even for zero scan, phaser and subarray quantization lobes do not exist for zero scan. Their amplitude increases with scan. Further GL tend to have high amplitude, diminished from main beam amplitude only by a
slowly varying element pattern. QL, on the other hand, are
typically much lower than the main beam, and each additional QL is significantly lower than the previous one.
MUTUAL COUPLING EFFECTS
Scan Element Pattern
A most important and useful parameter for large arrays is
scan impedance (the obsolete term active impedance is confus-
691
ing and is deprecated); it is the impedance of an element as a
function of scan angles, with all elements excited by the
proper amplitude and phase. From this the scan reflection coefficient is immediately obtained. Array performance is then
obtained by multiplying the isolated element power pattern
(normalized to 0 dB maximum) times the isotropic array factor (power) times an impedance mismatch factor. The isolated
element pattern is measured with all other elements opencircuited. This is not quite the same as with all other elements absent, except for canonical minimum scattering antennas (1). Here it is assumed that the array is sufficiently
large that edge effects are negligible and that scan impedance
is that for an infinite array. This simple performance expression allows the contributions of array lattice and element
spacing, element type, and mutual coupling to be discerned.
Unlike scan impedance, which is difficult to measure because
all elements must be properly excited, scan element pattern
(SEP) is measured with one element excited and all other elements terminated in Z0. It is important to note that scan element pattern provides the radar or communications systems
designer array gain, at the peak of the scanned beam, versus
scan angles. The SEP gs can be calculated from the scan impedance Zs and the isolated element power pattern giso:
gs (, ) =
4Rs (o, o)giso (, )
|zs (, ) + zg |2 giso (o, o)
(23)
For the general case where the generator reactance is not
zero, it is appropriate to use the conjugate reflection coefficient. For the special case where the generator impedance is
real the SEP can be written as
gs (, ) =
Riso giso (, )
|1 (, )|2
Rg
(24)
An often quoted approximation results for SEP needs to be
examined. This purports to show a cos variation:
gs (, )
4Aelem
cos 0 [1 | (, )|2 ]
2
(25)
where Aelem is the unit cell area of the element. Use of spectral
domain results for slots and dipoles, show that this expression is exact, provided that no grating lobes exist, the
elements are thin and straight, and no higher modes are
engendered by the feed. However, the scan behavior of the
mis-match factor must not be overlooked. It can be shown
that the SEP behavior is closer to cos1.5 , the extra power
being contributed by the impedance mismatch. So it is better
to use the generally applicable formula for SEP to get accurate and useful results. Extensive scan impedance data for
dipole arrays are given by Hansen (1).
Floquets theorem for arrays states that an infinite regular
periodic structure will have the same fields in each cell except
for a progressive exponential multiplier. Furthermore, the
fields may be described as a set of orthogonal modes. In essence the boundary conditions are matched in the Fourier
transform domain, resulting in some cases in an integral
equation reducing to an algebraic equation. Each element is
contained in a unit cell, with all cells alike and contiguous.
The periodic cell approach has proved to be the most powerful
and perceptive technique for understanding and for designing
692
SCANNING ANTENNAS
diagonal plane scans are only slightly affected. Adding a
screen replaces the infinite singularities by a roughly 3 dB
dip at grating lobe onset; other planes are not much affected.
Clearly, for wideband operation a screen is essential. Behavior for other lattice spacings is similar about the grating lobe
onset angle(s). The blind angle occurs because the reflection
coefficient there is unity magnitude (1).
Scan element pattern (dB)
E plane
D
10
cos1.5
Waveguide Simulators
15
20
0
10
20
30
40
50
degrees
60
70
80
90
Figure 8. SEP, dipole array. dx dy 0.5 wv, l 0.5 wv.
sophisticated arrays. Because of the Floquet symmetry, the
single unit cell contains the complete admittance behavior of
each element in the array.
Graphs of scan impedance and SEP give insight into how
arrays work as scan angle and lattice spacing are changed.
Figure 8 shows SEP for thin half-wave dipoles on a half-wave
square lattice. The SEP shows more decrease with scan angle
in the E-plane, due to the increasingly poor mismatch. A comparison of the SEP with powers of cos shows that the best
fit is cos1.5 ; this power pattern lies roughly between the Hplane and E-plane curves. Note that slot parameters are obtained from those of the dipole by multiplying by a constant.
Adding a ground plane raises the broadside scan resistance,
but most important it removes the H-plane trend to infinity.
This is offset by the screen pattern factor, with the result that
the SEP for dipoles/screen (Fig. 9) is worse for all planes than
that of the dipole array (Fig. 8). Again, the cos1.5 power pattern is a good fit out to about 50 scan; beyond this the SEP
falls rapidly due to the screen factor.
Behavior at grating lobe onset is illuminating; the SEP
shows a blind angle there for H-plane scan. The E-plane and
Scan element pattern (dB)
5
cos
1.5
10
E plane H
50
80
15
20
0
10
20
30
40
60
70
90
(degrees)
Figure 9. SEP, dipole/screen array. dx dy 0.5 wv, l 0.5 wv,
h 0.25 wv.
A simple microwave tool that allows the measurement of scan
impedance in one scan direction and for one frequency is the
waveguide simulator, developed by Brown and Carberry (15)
and by Hannan et al. (16). There are some symmetry planes
and scan angles that allow perfectly conducting unit cell
walls. These walls can then be a metallic waveguide, terminated in a section of the array containing a small number of
elements. Another way of visualizing the process utilizes the
decomposition of a TE01 waveguide mode into two plane waves
that reflect off the side walls. The angle of the plane wave
with the guide axis corresponds to the 0 scan angle. When
neither mode index is zero, an additional pair of plane waves
reflect off the broad walls (17). For the less often used triangular waveguide simulator, there are three sets of plane
waves involved. In a rectangular waveguide simulator the
dominant modes allow principal plane scans, while diagonal
plane scans require higher modes.
A typical simulator consists of a transition section from
standard guide to the simulator guide, the simulator waveguide, and the element port. Across the end of the simulator
waveguide is placed a thin metallic sheet with the slot elements appearing as holes. Behind this sheet is a waveguide
that encompasses all the complete elements, and that is terminated in a matched load; this is called the element port. For
elements other than slots or for complex element structures,
the element plate is modified accordingly. Careful matching
of the transition section is important. The simulator waveguide must be sufficiently long to allow decay of evanescent
modes. All diagonal plane simulators contain some fractional
elements, because the simulator waveguide is always larger
than the array unit cell. Most simulators use rectangular or
square waveguide as the modes are simple and the connection
to standard guide is easy. Because the waveguide mode constituent plane waves reflect in the H-plane, simulation of Hplane scan, either in a principal plane or in a diagonal plane,
is direct. Conversely, since there is no situation where there
are only two plane wave constituents that reflect in the Eplane, there are no linearly polarized E-plane simulators. Approximate simulation in the E-plane will be discussed below.
From the previous discussion it is clear that each simulator provides limited scan impedance data, although frequency
variation may provide limited additional data. Thus the
choice of the fewest and most economical simulators for a
given application is like a chess game, as observed by
Wheeler (18).
Figure 10 shows simulators for a square lattice array (16).
The H-plane tools use the TE10 mode, and are principal plane
(C) or diagonal plane (IC). The TM11 mode is used in the Eplane tools.
Simulation of a broadside beam is not possible; small
angles require large simulators. Also E polarization produced
by a linear element cannot be simulated exactly. The E-plane
SCANNING ANTENNAS
693
Figure 10. Rectangular lattice waveguide simulators. (Courtesy of Peter Hannan and
Hazeltine Corporation.)
simulators previously shown use circularly polarized or crosslinear polarized elements. E-plane simulation in a diagonal
plane can utilize a pair of modes, TE and TM, with identical
nonzero indices. A detailed review of waveguide simulator design and practice has been given by Wheeler (18). Microstrip
patch arrays have also been simulated. Solbach (19) used two
H-plane waveguide simulators. Here it was important to carefully reproduce the patch feeds and load resistances.
Scan Compensation
Some arrays radiate no power at a certain angle; this is called
a blind angle. This phenomenon can, like array performance
in general, be viewed from the standpoint of either scan reflection coefficient or SEP. The scan reflection coefficient may
develop a magnitude near or equal to unity, or the SEP may
be near or equal to zero. The blind angle occurs when a higher
mode cancels the dominant mode in the element. Since the
relative phase changes rapidly with scan angle, the two
modes produce a resonance, with a resonant phase angle that
can be defined on the decrease of with scan angle. The
higher mode can be produced by external or internal structure. Examples of external structure causing a blind angle are
a dielectric radome covering each element or row of elements,
a dielectric sheet placed on the array face, and a dielectric
plug protruding from the mouth of each element. In all cases
the scan wave impedance at the array face is modified by the
external structure so as to produce a dominant-mode and
higher-mode resonance at a particular scan angle; the higher
mode exists primarily outside the face. Examples of internal
structure that can allow a higher mode that can cause a blind
angle are dielectric plugs flush with the array face, dielectricloaded waveguide elements (as opposed to a short plug), and
a brick array of rectangular guide elements. For these various
structures the waveguide element supports the next higher
mode, which causes the resonance at the blind angle. For the
brick array the higher mode is just below cutoff due to the
large guide size.
A surface wave can be excited on the dielectric substrate
supporting a patch array. The blind angle occurs when the
Floquet scan phase velocity matches the surface wave velocity.
Compensation of scan impedance changes can be made by
any of three different practical methods. These are reduction
of coupling, multimode elements, and external wave transformers.
The simplest way to reduce relative coupling uses H-plane
baffles between slots or dipoles (20). These baffles are simply
metal strips perpendicular to the ground plane and running
between the dipoles or slot rows in the H-plane.
The ideal element power pattern of cos implies that both
TE and TM modes are needed in the array unit cell. The additional modes can be produced by modifying the radiating element in any of several ways. Probably the most effective way
of producing the optimum mix of modes uses open-end waveguide radiators, rectangular or circular, with the waveguide
internal environment suitably modified. A dielectric plug can
be placed at the rectangular waveguide width, and height
steps (in different planes) are also used. The mix of modes
that is needed depends upon the waveguide and lattice dimensions. With a dielectric plug in the waveguide the TE20
694
SCANNING ANTENNAS
mode will likely be propagating; all other higher modes will
be evanescent. The plug should be a half-wave long (in loaded
guide wavelengths) at center frequency to properly locate the
scan admittance on the Smith chart. Then the guide dimensions are adjusted to give a high susceptance for the TE20
mode looking into the matching network. Steps in guide
height can then be used for center angle impedance matching;
additional dielectric plugs can also be used for matching.
Another scheme combines the TE and TM radiators: a longitudinal shunt slot with parasitic monopoles with L-shaped
extensions developed by Clavin et al. (21). The monopoles reduce the E-plane beamwidth and add an electric-type mode
to complement the slot magnetic-type mode. Patterns in Eand H-plane can be made equal. Mutual coupling between elements is reduced, and the array wide-angle sidelobes and
backlobe are significantly reduced.
A different use of a dielectric sheet involves placing a high
dielectric constant thin sheet near the array face. This technique, called wide-angle impedance matching (WAIM), uses
the susceptances of the sheet to equalize part of the aperture
susceptance variation with scan. This is possible because the
sheet susceptance varies differently in the E-plane and Hplane. An optimum match is obtained by adjusting the sheet
thickness t (for a given r) and the sheet spacing from the
array face to minimize SEP variations over the scan volume.
The array SEP may be altered through use of a spatial
filter or transformer placed in front of the array face, but sufficiently far away that the filter does not directly affect element impedances. The transfer function of the spatial filter is
designed to compensate the SEP variations; polarization effects may also be improved. With dielectric sheets the transmission coefficient varies differently with angle in E- and Hplanes, so that in principle one or more sheets could improve
the SEP (22).
FINITE ARRAYS
Edge Effects
Although the infinite array techniques previously discussed
are excellent for system trades and preliminary design, final
design requires a finite array simulation. For thin half-wave
dipoles or narrow slots, for example, sizable planar arrays
may be solved using direct impedance methods; moment
methods are not necessary. When the element current distribution is complicated, the number of moment method expansion functions needed for good convergence will restrict this
method to small arrays. A discussion of other methods as well
as these is given by Hansen (1).
Using the impedance matrix approach, Diamond (23) computed scan impedance and SEP of several small arrays of dipoles. Small arrays exhibit behavior that oscillates around
that of a large array. In general, as might be expected, larger
arrays have more oscillations, but of smaller amplitude.
Gibbsian Model
A computer simulation of dipole and of dipole/screen arrays
showed that the oscillations in scan impedance across the finite array occur even at broadside (no scan) and that they
closely resemble the Gibbs phenomenon. This is the oscillations in a square wave represented by a band-limited function. They are largest at the edges (of the square wave and
of the array) and diminish away from the edge. The Gibbs
amplitudes are given by the Sine Integral, with extrema occurring for arguments of n (24). The period of the oscillations
in scan impedance increases with scan angle, as 1/(1 sin
0), both in the simulated results and in the model. This mode
was derived from the Fourier transform of a single pulse with
a linear phase, and it also utilizes Sine Integrals. Simple finite array models using the Gibbsian form have been constructed, with good comparisons with simulated results (25).
BIBLIOGRAPHY
1. R. C. Hansen, Phased Array Antennas, New York: Wiley-Interscience, 1998.
2. R. C. Hansen, Array pattern control and synthesis, Proc. IEEE,
80: 141151, 1992.
3. W. H. Von Aulock, Properties of phased arrays, Proc. IRE, 48
(10): 17151727, 1960.
4. L. R. Whicker, (ed.), Special issue on microwave control devices
for array antenna systems, IEEE Trans. Microw. Theory Tech.,
MTT-22: 589708, 1974.
5. W. J. Ince and D. H. Temme, Phasers and time delay elements,
In L. Young (ed.), Advances in Microwaves, Vol. 4, New York:
Academic Press, 1969, pp. 1189.
6. S. K. Koul and B. Bhat, Microwave and Millimeter Wave Phase
Shifters, Vols. I and II, New York: Artech House, 1991.
7. E. D. Ostroff, Solid-State Radar Transmitters, New York: Artech
House, 1985.
8. E. D. Cohen, Trends in the development of MMICs and packages
for active electronically scanned arrays (AESAs), In Proc. IEEE
Symp. Phased Array Syst. Technol., Boston, MA, 1996, pp. 14.
9. H. Zmuda and E. N. Toughlian (eds.), Photonic Aspects of Modern
Radar, New York: Artech House, 1994.
10. W. D. Jemison and P. R. Herczfeld, Acoustooptically controlled
true time delays, IEEE Microw. Guided Wave Lett., 3 (3): 72
74, 1993.
11. R. A. Soref, Optical dispersion technique for time-delay beam
steering, Appl. Opt., 31: 78957897, 1992.
12. S. T. Johns et al., Variable time delay of microwave signals using
high dispersion fiber, Elect. Lett., 28: 555556, 1993.
13. M. Y. Frankel, P. J. Matthew, and R. D. Esman, Two-dimensional
fiber-optic control of a true time-steered array transmitter, IEEE
Trans. Microw. Theory Tech., MTT-44: 29692702, 1996.
14. M. S. Smith and Y. C. Guo, A comparison of methods for randomizing phase quantization errors in phased arrays, IEEE
Trans. Antennas Propag., AP-31: 821827, 1983.
15. C. R. Brown and T. F. Carberry, A technique to simulate the
self and mutual impedance of an array, IEEE Trans. Antennas
Propag., AP-11: 377378, 1963.
16. P. W. Hannan, P. J. Meier, and M. A. Balfour, Simulation of
phased array antenna impedance in waveguide, IEEE Trans. Antennas Propag., AP-11: 715716, 1963.
17. R. E. Collin, Foundations for Microwave Engineering, New York:
McGraw-Hill, 1966.
18. H. A. Wheeler, A survey of the simulator technique for designing
a radiating element, In A. Oliner and G. H. Knittel (eds.), Phased
Array Antennas, Proc. 1970 Phased Array Antenna Symp., New
York: Artech House, 1972, pp. 132148.
19. K. Solbach, Phased array simulation using circular patch radiators, IEEE Trans. Antennas Propag., AP-34: 10531058, 1986.
20. S. Edelberg and A. A. Oliner, Mutual coupling effects in infinite
planar arrays. Part I, Slot arrays; and Part II, Compensation effects, IRE Trans., AP-8: 286297, 360367, 1960.
SCHEDULING
21. A. Clavin, D. A. Huebner, and F. J. Kilburg, An improved element for use in array antennas, IEEE Trans. Antennas Propag.,
AP-22: 521526, 1974.
22. B. A. Munk, T. W. Kornbau, and R. D. Fulton, Scan independent
phased arrays, Radio Sci., 14: 979990, 1979.
23. B. L. Diamond, Small arraysTheir analysis and their use for
the design of array elements, In A. A. Oliner and G. H. Knittel
(eds.), Phased Array Antennas, Proc. 1970 Symp., New York: Artech, 127131, 1972.
24. H. S. Carslaw, An Introduction to the Theory of Fouriers Series
and Integrals, New York: Macmillan, 1930.
25. R. C. Hansen and D. Gammon, Standing waves in scan impedance of finite scanned arrays, Microw. Opt. Technol. Lett., 8: 175
179, 1995.
R. C. HANSEN
R. C. Hansen, Inc.
SCANNING ELECTRON MICROSCOPES. See ELECTRON MICROSCOPES.
SCATTERING. See ELECTROMAGNETIC WAVE SCATTERING.
SCATTERING AND PROPAGATION IN REMOTE
SENSING. See MICROWAVE PROPAGATION AND SCATTERING FOR REMOTE SENSING.
SCATTERING AND PROPAGATION, MICROWAVE. See MICROWAVE PROPAGATION AND SCATTERING
FOR REMOTE SENSING.
SCATTERING IN REMOTE SENSING. See MICROWAVE
REMOTE SENSING THEORY.
SCATTERING PARAMETERS. See ATTENUATION MEASUREMENT.
SCATTERING, RADAR. See RADAR REMOTE SENSING OF IRREGULAR STRATIFIED MEDIA.
695
SMITH CHART
391
SMITH CHART
The origin of the Smith Chart as a fast and accurate graphical design tool is rooted in the predigital computer era of the
1930s. The original paper chart format was routinely used to
solve difficult transmission line matching problems in a variety of broadcasting, microwave, circuit design, and telecommunications applications. While it has since evolved almost
exclusively into a software format, the original paper forms
remain a convenient detailed medium for print output of finished work from software implementations of Smiths graphical method.
Phillip H. Smith, inventor and developer of the Smith
Chart, was affiliated with Bell Telephone Laboratories for his
entire engineering career after graduating from Tufts College
in 1928. At Bell Labs, Smith repeatedly demonstrated his incisive, pragmatic, problem-solving abilities on a variety of
technical problems. In his first decade on the job he developed
the Chart to its common present form through an iterative
trial process of several more rudimentary forms. Smiths original intended application was to simplify the tedious work of
installing and matching open wire transmission lines for antennas. During this period he almost single-handedly developed and proved the concept of using a graphical chart as a
viable solution method. One can certainly appreciate his desire to minimize tedious slide-rule calculations required to
solve transmission line equations involving hyperbolic functions. Smith (1) cited his routine need to obtain impedance
solutions using Flemings (2) telephone equation as a driving
force to consider a graphical solution alternative. [Readers interested in greater detail on the life of Phillip Smith are referred to the Foreword section of Ref. 3 for a written transcript of the memorial session in Phillip Smiths honor
presented at the 1990 IEEE MTT/S Symposium. An on-line
IEEE source (4) also provides good detail on Smiths background.]
The Chart was first published in a usable developmental
form in 1939 by Smith (5). A subsequent publication (6) in
1944 described the refined form known and used today. A
standard preprinted paper Chart in this format is shown in
Fig. 1(a), and Fig. 1(b) shows only the enlarged central portion around the normalized point z (1 j0). In each of
Smiths publications the fundamental equations and methods
J. Webster (ed.), Wiley Encyclopedia of Electrical and Electronics Engineering. Copyright # 1999 John Wiley & Sons, Inc.
392
SMITH CHART
tiv
0
0.11
0.39
Ne
0
11
0.40
0.10
0.4
1
0.0
9
100
100
0.10
0.40
110
0.12
0.38
70
0.
0. 18
32
19
0. 31
0.
40
20
0. 30
0.
30
1
0.2 9
0.2
2
0.2
8
0.2
20
20
0.23
0.27
10
0.24
0.26
0
0.25
0.25
10
0.26
0.24
0.27
0.23
0.2
8
0.2
2
0.2
0. 9
2
1
3
0
32
0. 18
0.
50
0.1
0.3 7
3
3
0.3 7
0.1
60
0
6
0.1
6
0.3
4
4
0.3
6
0.1
70
0.15
0.35
0.35
0.15
0.14
0.36
80
0.36
0.14
80
0.13
0.37
0.37
0.13
90
90
( ZR )
0.38
0.12
9
0.0
1
0.4
Resistance component
0.39
0.11
12
20
8
0.0 2
0.4
X
Z0
ga
07
0. 43
0.
0.4
0.0 2
8
0.
0. 06
44
+J
( (
13
0.
0. 43
07
1
30
44
0. 06
0.
0
0.01 Wave l eng
0.49
0.02 t h s to
0
0.49
0.48
war
0.01
0.48
2
d
0.0 d gen
.0
a
o
7
l
0
d
4
ar
3
i
f
c
f
i
e
e
o
n
w
c
t
0.
era
i
n
n
o
o
i
d
t
t
e
c
g
e
l
r
f
ees
0.4
3 ng t h s g l e o f r e
to
0
.
7
0 ve l e
n
0 . 0r
6
160
a 160 A
4
.
W
0 4
0.4 4
6
0.0
0.
15
0
45
0
0. 05
0. 05
15
P
45
osa
X
J
0.
tiv
er
t Z0
ea
en
14
cta
0
on
4
p
0
nc
1
m
e
co
co
ce
m
n
p
on
ta
ac
en
re
t
0.
0. 31
19
0.
0. 30
20
(a)
Figure 1. Full-size Smith Chart grid. (b) Enlarged center portion of the chart.
are provided along with demonstrations of the utility of the
Chart method. Certainly the ability to represent and mathematically manipulate a complete impedance data vector on a
single compact graphical paper chart was a significant step
forward at the time for the telecommunications and emerging
electronics industries at the onset of World War II.
THE ATTRACTION OF A GRAPHICAL SOLUTION METHOD
The essential objective in using the Smith Chart is stated
simply: An impedance (or equivalent admittance) locus over a
frequency range is systematically repositioned on the Chart
in well-defined ways by the progressive effects of series or
1.0
1.1
1.2
1.3
.50
1.0
1.5
2.0
2.5
1.0
1.001
1.002
1.004
1.01
1.02
1.03
1.04
1.05
1.06
90
3.0
Retn Refl.
Loss in db
.20
Vol. Pwr.
.13
.15
.10
.05
.04
Refl. Coef.
.05
1.4
100
0.40
0.10
0.39
0.11
0.38
0.12
0.37
0.13
0.36
0.14
80
0.35
0.15
1.10
0.4
1
0.0
9
0.4
0.0 2
8
0
11
0
70
4
0.3
6
0.1
20
0.
0. 43
07
Capacitive
rea
44
0. 06
0.
,o
iv
ct
du
ri
n
30
(
t
en
pon
ctan
ce c
om
( (
1.5
1.4
1.3
1.2
1.1
1.0
0.9
0.8
0.
0. 06
44
0.3
1.08
0.27
0.23
1.07
0.26
0.24
.22
.14
15
16
17
18
19
20
.04
.03
3.5
( RZ (, or conductance component ( GY (
0.25
0.25
10
0.3
.20
.18
.16
.02
.01
1.5
60
1.10
( +JB
Y (
.14
.12
.10
4.0
0.1
6
0.3
4
3
0.3 7
0.1
70
0.24
0.26
10
0.2
8
0.2
2
0
0.01 Wave l eng
0.49
0.02 t h s to
0
0.49
0.48
war
0.01
0.48
0.0 d ge
7
0.02w ard lo a d
3 ner
i
f
c
f
i
e
e
o
n
c
t
0.4
in degree
to
ction
e
l
f
s
e
ato
0
r
h
s
f
t
3
.47
.0 e l eng Ang l e o
0
0 . 0r
v
6
1
a
0
4
6
0
W 16
0. 4
0.4 4
0
.
6
0
0.
15
45
50
0
0. 05
0. 05
1
45
0.
JB
Ind
uct
Y0
nt
ive
14
ne
rea
0.7
po
40
0
1
m
cta
o
c
nc
e
c
e
n
c
ta
om
ep
po
sc
ne
su
n
e
0.1
0.23
0.27
20
0.4
.03
.02
.01
Figure 1. (Continued )
0.15
0.35
80
.08
.002
.001
(b)
25
30
40
50
Center
.06
.001
Toward load
.04
Toward generator
90
0.
0. 31
19
Sds Loss
steps coef.
20
.02
In
db
2
0.2
8
0.2
0.2
0.2 9
1
3
0
0.3
tiv
0.14
0.36
0.
0. 30
20
uc
omp
ce c
0.13
0.37
32
0. 18
0.
nd
tan
onent
0 0
Vol.
ratio
1
0.2 9
0.2
30
J
X
Z
X ,
+J 0
Z
i
or
es
ep
usc
0.12
0.38
0.2
20
0. 30
Resistance component
1
20
0.11
0.39
100
0.
40
0.1
07
0. 43
0.
0
13
110
0.4
9
0.0
1
0.4
0.10
0.40
8
0.0 2
0.4
19
0. 31
0.
Standing Wave Transm. Loss
SMITH CHART
393
0.1
0.3 7
3
0.
0. 18
32
50
Radially scaled parameters
394
SMITH CHART
shunt elements, thereby placing the passband portion of the
impedance vector in the central area of the Chart and the
stop-band portion at the outer periphery of the Chart. The
circuit elements commonly used include distributed elements
such as transmission lines and quarter-wave stubs, discrete
elements such as inductors and capacitors, and any combination of these and related elements. Resistance elements are
seldom intentionally used.
There is no need for a graphical method as an aide to solve
a straightforward equation such as Ohms law. However, the
more difficult general equation for the input impedance
(sending impedance in early 20th century publication parlance) of a transmission line with an arbitrary value of load
impedance (receiving impedance) is another matter. The
presence of multiple hyperbolic functions in Flemings equation is shown in Table 1, line 1. Figure 2(a) illustrates the
concentric circular arc movement on the interior of the chart
for a value of load impedance, ZL, connected to a length, l/ ,
of transmission line to produce an input impedance of Zi. The
input impedance equations for transmission line stubs (i.e.,
lengths of line typically less than a half-wavelength, with
short-circuit and open-circuit load impedance values) are
given by the equations of Table 1, lines 2 and 3 respectively.
Figure 2(b) illustrates the concentric circular arc movement
along the periphery of the chart for a value of load impedance
(ZL 0 for the short-circuit stub and ZL for the opencircuit stub), connected to a stub with length l/ , to produce
an input impedance of Zi.
The equation of Table 1, line 2 shows the effective parallel
L-C circuit tuning effect for a short-circuit stub, and line 3
illustrates the effective series L-C circuit tuning effect for the
case of an open-circuit stub. At relatively higher frequencies
for which the length of the stub is greater than a half-wavelength, these tuning effects repeat unlike the effect of discrete
(lumped element) parallel or series L-C circuits.
The repetitive nature of these three equations when line
length exceeds a half-wavelength is evident when one looks
at them plotted on the Chart. Likewise, the tuning effects of
discrete elements is also calculable and treated on the Chart.
The inductor and capacitor are especially used in typical
equalizer, filter, and matching circuits. The resistor element,
while it can also be treated on the Chart, is seldom used except in cases such as ultra-broadband (i.e., multi-octave to decade bandwidth) tuning of electrically short antennas to meet
a VSWR specification. The equations for the discrete elements
and the relative motion of their locus on the Chart is given in
Table 1, lines 4, 5, and 6. Figure 2(c) illustrates the movement
on the chart along arcs of constant-resistance for some load
impedance, ZL, when a capacitance or inductance is placed
in series.
The Smith Charts power as a graphical solution method
comes from its versatility to design cascade and parallel circuits made up of any combination of distributed and lumped
circuit elements defined by the equations given in Table 1. In
addition, algorithmic methods such as the design of transmission line impedance transformers can also be implemented
directly on the Chart. Design aids such as a voltage standing
wave ratio (VSWR) goal plotted as a circle concentric to the
center of the Chart, and other geometric constructs such as
forbidden areas and boundary circles make the design process
straightforward. These design aids as well as the line entries
in Table 1 are discussed in greater detail in subsequent sections.
MATHEMATICAL STRUCTURE OF THE CHART
The mathematical basis of the Smith Chart is a conformal
bilinear transformation for which all data values approaching
infinitely large values on the (R,X) or (G,B) complex plane
map to a single uniquely defined finite point. In implementing
the transformation, the simple, familiar, straight-lined, rectangular, impedance grid with resistance (or conductance) on
the positive abscissa and reactance (or susceptance) on both
positive and negative axes of the ordinate map into two families of circles that intersect orthogonally to one another. This
holds true for charts scaled for impedance or admittance.
The equations which uniquely define the bilateral transformation between complex normalized impedance and complex
reflection coefficient are given as
z=
1+k
1k
and k =
z1
z+1
(1)
where z represents normalized impedance (Z/Z0), k is the
complex reflection coefficient at the point where z is measured, and Z0 is the reference or normalizing value of impedance.
The generality of the underlying bilinear transformation is
seen by plotting normalized impedance values (or analogously
admittance values) on the right half of a Cartesian coordinate
plane and evaluating the paired Eq. (1) for reflection coefficient loci which have constant values of magnitude with varying modulus as well as constant modulus (angle) values with
varying magnitudes. Figure 3(a) illustrates normalized impedance plotted in a Cartesian frame, with the dashed lines
indicating constant-resistance and constant-reactance values
Table 1. Distributed and Lumped Element Components Commonly Used in Smith Chart Analysis and the Direction of
Progression with Frequency on the Chart
Element Type
Defining Equation
Transmission line
Short-circuit stub
Z in Z 0(Z L Z 0 tanh l)/(Z 0 Z L tanh l)
Z in jZ 0 tan l
Open-circuit stub
Z in jZ 0 cot l
Inductance
Capacitance
Resistance
E IXL ; XL jL
E IXC ; XC 1/jC
E IR
Motion on the Smith Chart
CW along a circle centered at Z 0 1 j0. See Fig. 2(a).
CW along chart perimeter with region near especially useful
(l /4). See Fig. 2(b).
CW along chart perimeter with region near 0 especially useful
(l /4). See Fig. 2(b).
CW along constant-R circles. See Fig. 2(c).
CCW along constant-R circles. See Fig. 2(c).
To right along constant-X arcs toward R .
SMITH CHART
395
l/
Zi
l/
Zi
Z0
Z L=
l/
Z L=
ZL
Z L= 0
Zi
Zi
Z0
Z0
Z L= 0
l/
ZL
l/
l/
(a)
(b)
(Zi + jXL )
Zi
(Zi jXC )
(c)
Figure 2. (a) Concentric circular movement along the chart interior to a value of input impedance, Zi, resulting from addition of a transmission line to a load impedance ZL. (b) Concentric
circular chart movement along the chart periphery for the input impedance of a short-circuit
(ZL 0) and an open-circuit (ZL ) stub. (c) Movement along circular arcs of constant-resistance for addition of a capacitance (downward) and inductance (upward) to a load impedance
ZL.
(hereafter referred to as constant-R and constant-X, respectively). Figure 3(b) uses the same frame to depict loci of constant-k values as solid lined full circles, and it uses loci of
constant phase angle values for variable-k magnitudes as
solid lined partial circles. The constant angle value loci
all intersect at the single point z (1 j0) for which k
(0 j0).
Alternately, one can plot the complex reflection coefficient
[k K exp( j)] in two-dimensional cylindrical (i.e., polar) coordinates as shown in Fig. 4(a), with k (0 j0) located at
the center and cylindrical angle representing the reflection
coefficient angle . Then the transformation to complex normalized impedance using Eq. (1) results in loci for constant-R
and constant-X values which are again full circles and partial
circles, respectively, as illustrated in Fig. 4(b). These circles
and circular arcs have progressively offset centers located
along the orthogonal axes defined by the horizontal line at
x 0 and the vertical line tangent to the outer circle at the
right side point. Figure 4 illustrates the orthogonality features in both domains of (a) reflection coefficient and (b) normalized impedance. The latter grid of circles and circular arcs
represent exactly the Chart solution method first proposed by
Smith to the radio engineering community.
The radii and centers of the constant-R circles and constant-X circular arcs of Fig. 4(b) are derived relative to the
center of the cylindrical coordinate plot of complex reflection
coefficient. The radii are given by Eq. (2), where the value of
either r or x is held constant. The geometric locations of the
centers of the constant-R circles are given by Eq. (3), and the
positions of the centers of the constant-X circular arcs are
given by Eq. (4).
rk =
[(r2 1 + x2 )2 + 4x2 ]1/2
(r + 1)2 + x2
(kxR , kyR ) =
r
,0
r+1
(kxX , kyX ) = 1,
1
x
(2)
(3)
(4)
where the subscripts R and X denote constant-R circles and
constant-X arcs, respectively, and the superscripts x and y denote the directional components.
NORMALIZATION OF THE CHART
For commonality and other reasons it is useful to deal with
normalized rather than unnormalized Charts. Both full-size
and enlarged central area Charts are normalized by dividing
SMITH CHART
relation of VSWR, return loss, and reflection coefficient, one
is also able to read length (l/ ) along a transmission line segment from the rim of the chart and use boundary circles as a
design aid. These features are reviewed below, but the reader
is referred to Smiths seminal text (7) for added detail on
these and other features. This reference also contains an exhaustive bibliography on publications related to the Smith
Chart through 1969.
x = X/Z0
+j3
+j2
+j1
VSWR Circles
1
j3
r = R/Z0
The relationship between VSWR and reflection coefficient is
given by Eq. (5), where K is the magnitude of the complex
reflection coefficient written as k K exp( j).
j2
VSWR =
j1
x
=
30
+j3
+j2
90
120
60
30
= 0
K =1/3 K =1/2
2
3
4
1 1
j3
j1
60
150
+j1
j2
(5)
On the Smith Chart, loci of constant VSWR are concentric
circles with all centers located at K 0, and radii are nonuniformly distributed between unity and infinity for values of K
ranging from 0 to 1.
(a)
1+K
1K
3
=
00
6
= 360
r
.2
180
.4
.6
.8
K 0
33
150
30
60
120
90
(b)
(a)
x=+j
Figure 3. (a) Plots of normalized constant-R and constant-X on a
Cartesian frame. (b) Loci of constant-k (full circles) and constant-
angle (circular arcs).
x=+j2
396
GRID-BASED FEATURES OF THE SMITH CHART
A number of useful design features result from the final form
of the Chart grid as derived by Smith. In addition to the inter-
2.
r=
r=
r=
j.5
+ jx
1.
j.5
z=
jx
j.5
j
.1
x=
values of resistance (R) and reactance (X) by the real reference value (Z0). This scaling results in all real axis values to
the left of the center point having values between 0 and 1,
and all real axis values to the right of the center point represent values between 1 and infinity. Charts which plot admittance values are analogously normalized using the reference
value Y0 1/Z0. The usage of normalized impedance values
on the Smith Chart grid facilitates the computation of reflection coefficient, VSWR, and return loss from the chart. Design
examples given in subsequent sections for transmission line
problems illustrate the advantage of using normalized values
for chart-based computation.
(b)
Figure 4. (a) Two-dimensional cylindrical coordinate plot of loci of
constant reflection coefficient magnitude and constant reflection coefficient angle. (b) Loci of constant-R circles and constant-X circular
arcs scaled to fit on the reflection coefficient loci plot of part (a).
SMITH CHART
397
Chart Perimeter as Transmission Line Length
Boundary circles
The classical transmission line equation (see Table 1, line 1)
clearly illustrates that impedance, and hence VSWR, varies
in a repetitive fashion every half-wavelength in distance
along a lossless line. On the Smith Chart this is equivalent to
repeated values for reflection coefficient with every complete
rotation around the chart relative to the center of the chart.
The entire perimeter of the Smith Chart calibrates uniformly
as a quarter-wavelength distance relative to a reference
location. Clockwise rotation around the Charts periphery is
equivalent to moving along the line in a direction toward the
source (generator), and counterclockwise rotation around the
Chart is equivalent to moving toward the load.
Impedance and Admittance Locus Movement for Discrete
Circuit Elements
Figure 5. VSWR circle and its associated boundary circle pairs for a
value VSWR 1.25.
Electronic circuit impedance matching is easily performed
utilizing the Smith Chart. Typically one desires to use only
lossless components to accomplish a match. The addition of a
discrete circuit element such as a capacitance or an inductance in a matching circuit has a well-defined effect on moving the locus of a load impedance vector on the Chart grid.
The four key discrete circuit elements commonly used in circuit matching are given as follows along with their effect on
motion of an impedance locus (for series elements) or an admittance locus (for shunt elements):
Series inductance rotates an impedance locus CW on
constant-R circle
Series capacitance rotates an impedance locus CCW on
constant-R circle
Shunt inductance rotates an admittance locus CCW on
constant-G circle
Shunt capacitance rotates an admittance locus CW on
constant-G circle
VSWR
circle
a
a
a
a
Here CW and CCW denote clockwise and counterclockwise
motion, respectively, on the specified circles.
cannot be matched to within a specified value of VSWR with
a single matching element.
OTHER IMPORTANT OPERATIONS EXECUTABLE ON THE
SMITH CHART
Two Element Matching Over the Entire Chart
In contrast to the case just shown where impedance values
over much of the Chart area cannot be matched with a single
element, it can be shown that all single-frequency impedance
values anywhere on the Chart exclusive of the outer rim circle (r 0) can always be perfectly matched to a value of r
(1 j0) by use of a two-element circuit. Smith (8) and a current software Chart implementation (9) provide overlay diagrams of the eight possible ell-type circuits (four L-C, two LL, and two C-C), each with its associated allowed area and
supplementary forbidden area on the Smith Chart. (The same
diagrams are also illustrated in Ref. 10, p. 40.) Any point
within an allowed area can always be perfectly matched using
the associated ell-circuit.
Transmission Line Transformers
Boundary Circles
The Chart-centered circle for a finite value of VSWR leads
directly to the useful design construct known as boundary circles. They serve as a visual aid to determine the correct value
of a circuit element which will cause the proper amount of
rotation of an impedance or admittance locus for acceptable
matching (i.e., which meets the VSWR requirement).
For a specified VSWR there are two related pairs of boundary circles. Each pair consists of two circles which are centered on the x 0 line and which are doubly tangent to both
the VSWR circle and the periphery of the Chart where the
magnitude of the reflection coefficient is unity. Figure 5 shows
the combination of a VSWR 1.25 circle along with its associated pair of boundary circles. Knowing the directions of rotation for an impedance or admittance locus caused by the addition of series or shunt elements, respectively, one
immediately determines that impedance values over a very
large portion of the Chartnamely, the interior portions of
both small boundary circles, as well as the upper and lower
exclusion areas outside both of the larger boundary circles
Frequently the design of an impedance matching circuit must
be implemented using distributed parameter elements such
as transmission lines. Somlo (11) has shown how a singlefrequency design for a transmission line transformer is graphically implemented using a Smith Chart. Narrow bandwidth
matching designs are also achievable using the same method.
The Somlo technique is implemented as a utility tool in the
software implementations of Refs. 9 and 12.
The ability to graphically analyze the matching effects of
a variety of transmission line circuit elements (e.g., open or
short-circuit stubs as well as transmission line sections and
impedance transformers in both series and shunt configurations) gives the Smith Chart more versatility than software
synthesis tools that are typically unable to treat a full array
of transmission line elements.
EXAMPLE
A circuit matching design exercise is described here to illustrate the methods used in obtaining a solution on the Smith
398
SMITH CHART
Table 2. Load Impedance Values for Example
Matching Problem
Frequency (MHz)
850
875
900
925
Load Impedance ()
15.0
19.0
21.5
25.0
j250
j227
j192
j180
Chart. To keep the Charts visually succinct, each of the intermediate plotting steps including Z Y and Y Z inversions
as well as the effects of individual branch-by-branch circuit
constructions are presented on a set of individual Charts. The
traces plotted on the Charts use the convention of ZN or YN
for the driving point impedance or admittance when looking
toward the load from branch N. Readers who wish to study a
variety of antenna matching problems executed in step-bystep detail can consult Ref. 13.
In this problem a device, such as an antenna, has a highly
capacitive input impedance vector as given in Table 2 for the
range of 850 MHz to 925 MHz. The objective is to produce a
matching circuit which meets a VSWR requirement of 2.5 : 1
over the reduced frequency range of 855 MHz to 920 MHz
with the restriction that no lumped circuit inductors be used
in the matching circuit.
The bandwidth requirement of about 8.5%, the highly capacitive values, and the implementation restriction of no
lumped inductors all imply that additional elements beyond a
simple two-element L configuration will be required to implement a satisfactory matching circuit. The two solutions derived by use of a software chart method are described in Table
3, giving all the circuit element parameter values. The first
solution uses transmission lines and short-circuit stub elements exclusively and is illustrated in the set of successive
charts in Fig. 6. The second solution requires fewer elements
but includes a capacitance and a short-circuit stub having a
low value of Z0 which is virtually impossible to implement. It
is presented in the second part of Table 3 for purposes of contrast only.
Table 3. Component Types and Values for the Solutions of
Example Problem
Element
No.
1
2
3
4
1
2
3
4
a
Element Type
Primary
Transmission line
Shunt short-circuit
stub
Transmission line
Shunt short-circuit
stub
Parameter Values
(v/c group velocity)
Solution a
Z 0 234 , 2.60 cm, v/c 0.69
Z 0 10.1 , 4.84 cm, v/c 0.69
Z 0 76.22 , 0.70 cm, v/c 0.69
Z 0 6.53 , 6.25 cm, v/c 0.69
Contrasting Solution
Transmission line
Z 0 100 , 4.91 cm, v/c 0.69
Shunt capacitance
9.4 pF
Transmission line
Z 0 42.8 , 5.84 cm, v/c 0.69
Shunt short-circuit
Z 0 3.12 , 5.91 cm, v/c 0.69
stub
See Fig. 6 for associated chart traces.
The first element in the matching circuit must be either of
the two inductive types: either a short-circuit stub or a transmission line of sufficient length to rotate the load impedance
trace toward the inductive reactance portion (i.e., upper half),
of the Chart. There is, however, a distinct observable difference in how the two candidate initial elements cause the load
impedance trace to rotate on the chart. A series wired shortcircuit stub will move the load trace CW and upward along
constant-R grid lines, causing the rotated impedance vector
to lie closer to the center of the Chart, but extending it over
the Chart to a greater degree than would be caused by a series transmission line. Conversely, the series transmission
line keeps the rotated load trace further out toward the chart
periphery. A compromise for the initial element is to use a
short length of transmission line with high Z0. This has an
impedance transforming effect in bringing the load trace toward the center of the chart, but the short length keeps the
trace located mostly on the capacitive side of the chart. In this
case the characteristic impedance and length of the transmission line element are selected to rotate the impedance locus
such that it approximately straddles the lower-left crescent of
the boundary circles associated with the specified VSWR.
The second element, a shunt position short-circuit stub, rotates the Y1 trace CCW and downward to position the ends of
the Y2 trace on the inductive susceptance side of the chart
with the trace approximately surrounding the center of the
chart. This begins the process of pulling the ends of the trace
inward toward accomplishing the ultimate goal of collapsing
the final trace into a loop configuration and positioning it inside the 2.5 : 1 VSWR circle.
The third element, a length of transmission line, rotates
the Z2 trace CW to an approximately vertical orientation with
about half of the trace in each of the top and bottom halves
of the chart. With the central portion of the Z3 locus inside
the VSWR circle, the sole purpose of the fourth and final element will be to complete the process of collapsing the trace
into the needed loop and place it inside the VSWR circle.
The fourth element, a shunt wired short-circuit stub, adds
sufficient compensating susceptance (inductive susceptance
at the low frequencies and capacitive susceptance at the high
frequencies) to further close the loop. To provide the necessary magnitude of compensating susceptance given by Bstub
(cot l/Z0), the characteristic impedance required for the
shunt short-circuit stub element is a small value. This also
keeps the Y4 trace tightly looped. In practice, this low value
of Z0 is typically accomplished with four equal-length stubs
wired in parallel and each having a Z0 of about 25 , which
is a practical physical lower limit. Frequency interpolation is
used to confirm that the VSWR specification is met for the
frequency range of 855 MHz to 920 MHz.
The contrasting solution listed in Table 3 follows much the
same sequence used in the first solution. The primary difference in the two methods is that use of a capacitance for the
second element places its associated Y2 trace in a more vertical orientation with the extremes of the trace spread further
out from the VSWR circle. That causes the value of characteristic impedance for the short-circuit stub used for the fourth
element to be much lower than that for the first solution
method where more of the admittance locus for the third element is already within the VSWR circle. This second solution
is clearly not a physically realizable situation. Hence, the associated chart traces are not shown.
SMITH CHART
399
Z1
Y1
ZL
(a)
Y2
(b)
Z3
Z2
Y4
Y3
(c)
(d)
REVITALIZATION OF THE SMITH CHART IN SOFTWARE
The original printed paper versions of the Smith Chart were
primarily used in industrial research and World War II development efforts through the 1940s. For the following three decades, these paper Charts became a mainstay tool in every
transmission lines course in academia as well as a broadbased industry design tool. Beginning about 1980 with the
onset of a variety of numerical design tools such as the realfrequency method pioneered by Yarman and Carlin (14) and
implemented in commercial software (15), the paper version
of the Smith Chart fell into relative disuse. However, beginning about 1990, a number of software tools (9,10,12,16,17)
became available which implemented the core operations of
the Smith Chart on a computer screen.
These software implementations of the Smith Chart alleviated most or all of the key difficulties encountered with manual use of the paper charts. These advantages include (1)
elimination of math errors since the software internally performs all the background calculations of branch impedance or
admittance values including normalization, (2) elimination of
errors due to using an incorrect rotation direction on the
chart, (3) elimination of interpolation errors when plotting
data onto or reading values off of the chart, (4) requiring the
user to select a proper value of group velocity for a transmission line (v/c 1), (5) having the software internally perform
spline calculations to plot smoothly contouring line vectors on
Figure 6. Solution to the example impedance
matching problem requiring no lumped element
inductors. (a) ZLoad trace and Z1 trace after adding
series transmission line section. (b) Y1 trace and
Y2 trace after adding shunt short-circuit stub. (c)
Z2 trace and Z3 trace after adding series transmission line section. (d) Y3 trace and Y4 trace after
adding shunt short-circuit stub.
the chart, (6) the dramatic speed improvement factor over
hand-drawn Smith Charts, and (7) toggled screen presentation of VSWR and boundary circles as convenient design aids.
In addition, an attractive feature of the software versions is
their ability to plot the finished work in excellent detail on a
preprinted paper Smith Chart form for delivery or publication purposes.
With the availability of software-based Smith Chart programs, the professional circuit designer, student, and researcher can take full advantage of the clarity and simplicity
of graphical methods with a minimum investment of labor
and time to obtain the needed solution. More important, one
gains the value of personal insight into the process of design
of a matching circuit when using this graphical method.
The value of the Smith Chart as a potent graphical design
tool was well stated by Carlin (18) at a 1983 IEEE MTT/S
professional workshop on broadband matching: So far the
best transfer functions analytic theory has to offer are based
on Chebyshef polynomials and almost invariably these yield
matching structures which can be surpassed in performance
by significantly simpler equalizers. In effect the old-fashioned
procedure of playing around on the Smith Chart may produce better results than sophisticated theory.
As software-based Smith Chart tools evolve to include additional capabilities such as built-in circuit optimization functions and IEEE-488 interfaces to port measured impedance
data from network analyzers, it is expected that they will en-
400
SMOOTHING CIRCUITS
joy continued future usage by a variety of technical practitioners and students
BIBLIOGRAPHY
1. P. H. Smith, Introduction section, Electronic Applications of the
Smith Chart, New York: McGraw-Hill, 1969; subsequent reprints
Malabar, FL: Krieger Publishing, 1983, and Tucker, GA: Noble
Publishing, 1996.
2. J. A. Fleming, The Propagation of Electric Currents in Telephone
and Telegraph Conductors, New York: Van Nostrand, 1911.
3. L. M. Schwab, Advanced Automated Smith Chart, Version 2, Boston: Artech House, 1995.
4. Available on line: http://www.ieee.org/history_center/oral_
histories/abstracts/smith3_abstract.html
5. P. H. Smith, Transmission line calculator, Electronics, January,
29, 1939.
6. P. H. Smith, An improved transmission line calculator, Electronics, January, 130, 1944.
7. P. H. Smith, Electronic Applications of the Smith Chart, New
York: McGraw-Hill, 1969; subsequent reprints Malabar, FL:
Krieger Publishing, 1983, and Tucker, GA: Noble Publishing,
1996.
8. Ref. 7, pp. 116117.
9. L. M. Schwab, Automated Smith Chart, Version 3 (MS Windows
95/NT), Boston: Artech House, 1998.
10. L. M. Schwab, Automated Smith Chart (DOS), Boston: Artech
House, 1991.
11. P. I. Somlo, A logarithmic transmission line chart, IRE Trans.
Microw. Theory Tech., 8: 463, 1960.
12. L. M. Schwab, Advanced Automated Smith Chart, Version 2, (MS
Windows 3.1), Boston: Artech House, 1995.
13. W. N. Caron, Antenna Impedance Matching, Newington, CT:
Amer. Radio Relay League, 1989.
14. B. S. Yarman and H. J. Carlin, A simplified real frequency technique applied to broadband multistage microwave amplifiers,
IEEE Trans. Microw. Theory Tech., 30 (12): 22162222, 1982.
15. Complex Match II, synthesis software by Compact Software, Inc.
Paterson, NJ, 1988.
16. W. Hayward, MicroSmith (DOS), Newington, CT: American Radio Relay League, 1992.
17. Eagleware, winSMITH (MS Windows), Tucker, GA: Noble Publishing, 1995.
18. H. J. Carlin, Keynote speech: That perennial problem of broadband matching, 1983 IEEE MTT/S Symp. Workshop: Broadband
matching and design of microwave amplifiers.
LEONARD M. SCHWAB
Germaine Development Corp.
SPECTRAL-DOMAIN ANALYSIS
105
SPECTRAL-DOMAIN ANALYSIS
Many numerical methods exist for analyzing microwave and
millimeter-wave passive structures. Among them, spectraldomain analysis (SDA) is one of the most popular. It was developed by Itoh and Mitra in 1974 (1). In 1978, Itoh also presented an SDA version for quasistatic analysis (2). SDA is
basically a Fourier-transformed version of the integral equation method. However, as compared to the conventional
space-domain integral equation method, SDA has several advantages. Its formulation results in a system of coupled algebraic equations instead of coupled integral equations. Closedform expressions can be easily obtained for the Greens functions. In addition, incorporation of physical conditions of analyzed structures via the so-called basis functions is achieved,
and the obtained solutions are stationary. These features
make SDA numerically simpler and more efficient than the
conventional integral equation method. SDA has been used
extensively in analyzing planar transmission lines (e.g., Ref.
3). In this article, we present a detailed formulation of SDA
for planar transmission lines in both quasistatic and dynamic
domains. Other applications of SDA, to resonators and antenna and scattering problems, can be found in Refs. 4, 5, and
6, respectively.
J. Webster (ed.), Wiley Encyclopedia of Electrical and Electronics Engineering. Copyright # 1999 John Wiley & Sons, Inc.
106
SPECTRAL-DOMAIN ANALYSIS
FORMULATION OF THE SPECTRAL DOMAIN ANALYSIS
charge density. The boundary conditions are
Figure 1 shows a cross section of the multilayer coplanar
waveguide (CPW) to be analyzed. The central and ground
strips are assumed to be perfect electric conductors of zero
thickness, uniform and infinite in the z direction. The formulation for the case of finite strip thickness is discussed in Ref.
7. The dielectric substrates are assumed to be lossless. The
enclosure or channel is assumed to be a perfect electrical conductor and is used to simplify some of the analysis and computation, but the resulting analysis can be used for an open
structure by letting the appropriate dimensions be large. The
eigenmodes existing in the considered structure consist of
both TE and TM fields. Although SDA can produce results for
all of the real and complex eigenmodes for this structure (8),
in this article we are restricted to the analysis of the real
eigenmodes, including the dominant (CPW) mode, for the purpose of illustrating SDA. Due to the effect of multiple dielectrics surrounding the metallic strips, the dominant CPW
mode is quasi TEM.
Quasistatic Spectral-Domain Analysis
A quasistatic analysis solves the two-dimensional Laplaces
equation for the electric potential in the transverse plane subject to appropriate boundary conditions in the space domain,
2 i (x, y) 2 i (x, y)
+
= 0,
x2
y2
i (0, y) = i (a, y) = 0
(2)
1 (x, b) = 3 (x, 0) = 0
(3)
1 (x, h2 + h3 ) = 2 (x, h2 + h3 ) = V (x)
U
2 (x, h3 ) = 3 (x, h3 )
s (x)
2
1
r2
r1
=
y y=h 2 +h 3
y y=h +h
0
2
3
r3 3 y=h r2 2
=0
3
y
y y=h
(1)
where i(x, y) is the unknown potential in the ith region. The
quasistatic SDA, on the other hand, solves Laplaces equation
by applying a moment method, Galerkins technique, in the
Fourier transform, or spectral, domain (2). The analysis obtains the charge density on the central strip, and from this
the per-unit-length (PUL) capacitance is obtained. The PUL
capacitance can then be used to determine the effective dielectric constant and characteristic impedance of the transmission line. The boundary conditions are derived from the fact
that (x, y) is continuous everywhere and n
(D2 D1) s
at the upper interface (between the first and second layers),
is
where Di, i 1, 2, is the electric flux density in region i, n
the unit vector normal to the interface, and s is the surface
a
r1
G1
h1
S1
r2
r3
S2
(6)
(7)
where 0 is the free-space permittivity and ri, i 1, 2, 3, is
the relative dielectric constant of the ith layer; VU(x) denotes
the potential at the upper interface and can be expressed as
V U (x) = V0 (x) + V (x)
(8)
V0(x) V0 on the central strip and zero elsewhere; we choose
the value of V0. The ground strips are assumed to be at zero
potential. V(x) is the unknown potential on the two slots and
zero on the central and ground strips. The surface charge density, s, can be described as
(9)
where s1(x), s2(x), and s3(x) are unknown charge densities on
the central, left, and right ground strips, respectively, and are
nonzero only on the corresponding strips.
The Fourier transform used is defined as follows:
2
f(n , y) =
a
f (x, y) sin n x dx
(10)
where n n/a, n 1, 2, 3, . . . denoting the spectral order
or term. This choice for the Fourier transform variable n will
cause the boundary conditions of Eq. (2) on i(x, y) to be met
automatically.
Transforming Eqs. (1) to (9) gives
2 i
n2 i = 0
y2
(11)
1 (n , b) = 3 (n , 0) = 0
y
(5)
s (x) = s1 (x) + s2 (x) + s3 (x)
i = 1, 2, 3
(4)
(12)
1 (n , h2 + h3 ) = 2 (n , h2 + h3 ) = V U (n )
(13)
2 (n , h3 ) = 3 (n , h3 )
2
1
s (n )
r2
r1
=
y y=h 2 +h 3
y
0
(14)
(15)
y=h 2 +h 3
G2
r3 3
y
h2 b
2
y=h r2
3
y
=0
V U (n ) = V 0 (n ) + V (n )
h3
x
Figure 1. A cross section of the coplanar waveguide used for illustrating SDA.
(16)
y=h 3
s (n ) = s1 (n ) + s2 (n ) + s3 (n )
(17)
(18)
The solution of Eq. (11) is well known. A judicious choice
yields the following forms, which satisfy Eq. (12):
SPECTRAL-DOMAIN ANALYSIS
1 (n , y) = A sinh n (b y)
(19)
2 (n , y) = B sinh n (y h3 ) + C cosh n (y h3 )
(20)
3 (n , y) = D sinh n y
(21)
where A, B, C, and D are unknown functions of n. Substituting Eqs. (19) to (21) into Eqs. (13)to (16) and solving for
U(n) in terms of s(n) yields
V
n ) s (n ) = V U (n )
G(
The use of the Galerkins technique enables us to eliminate
by applying Parsevals theorem to the
the unknown voltage V
right-hand sides. Specifically, Parsevals theorem eliminates
, because the charge densities and
the summations involving V
voltages are nonzero in complementary regions in the space
domain. Furthermore, two summations involving known voltages are eliminated. Equations (26) to (28) can be solved for
the unknown coefficients of the charge density basis functions. The transmission lines PUL capacitance is given as
(22)
N1
(n) is the spectral-domain Greens function given as
where G
n ) = tanh n h
G(
1
r3
tanh n h2 + tanh n h3
r1
0 n tanh n h2 r3
+ tanh n h1 coth n h2
r2
!1
+ tanh n h3 (r1 coth n h2 + r2 tanh n h1 )
Q
C= 0 =
V0
r2
We now begin to apply the Galerkins technique in the
spectral domain by expressing each strips charge density as
a truncated summation of basis functions in the space domain
as
G 1 +S 1 +W
G 1 +S 1
(29)
which can then be used to calculate the characteristic impedance and effective dielectric constant as
Zc =
p1
(30)
c CCa
and
r eff =
d ji s ji (x)
s1i (x) dx
V0
s (x)
d1i
i=1
(23)
j
3
107
(24)
C
Ca
(31)
j=1 i=1
where sji(x) describes the charge distribution on the jth strip
and is nonzero only on that strip, dji is the unknown coefficient, and Nj denotes the number of basis functions used for
the jth strips charge density. In the spectral domain,
N
s (n )
j
3
d ji s ji (n )
(25)
j=1 i=1
Substitute Eq. (25) into Eq. (22), multiply by sji(n) for j 1,
2, 3 and i 1, 2, . . ., Nj, and sum over n. This results in a
system of coupled linear algebraic equations
N1
s1 j G s1i
n=1
i=1
N2
i=1
d3i =
s3 j G s1i
N
3
i=1
d1i +
n=1
2
V
a 0
N2
s2 j G s3i
n=1
n=1
N2
s3 j G s3i
d2i
G 1 +S 1 +W
G 1 +S 1
s1 j (x) dx,
j = 1, 2, . . ., N1
s2 j G s2i
d3i = 0,
i=1
(26)
s1i (x) =
d2i
n=1
d1i +
s1 j G s2i
n=1
i=1
N3
i=1
s1 j G s3i
s2 j G s1i
n=1
i=1
d1i +
n=1
i=1
N2
i=1
N3
N2
respectively, where c is the free-space velocity and Ca is the
capacitance per unit length with the dielectrics removed.
To obtain numerical results, we need to choose basis functions for the charge densities, sji(x), with j 1, 2, 3 for the
central strip and left and right ground strips, respectively.
These basis functions influence strongly the numerical efficiency of the solution process and the accuracy of the solutions. The computation time can be reduced significantly if
the chosen basis functions describe closely the actual behavior
of the charge distributions and have closed-form Fourier
transforms. In addition, the basis functions should form complete sets, so that the solution accuracy can be enhanced by
increasing the number of basis functions. Furthermore, they
should be twice continuously differentiable to avoid spurious
solutions. The basis functions used for the considered problem
have the form
j = 1, 2, . . ., N2
s3 j G s2i
(27)
d2i
s3i (x) =
j = 1, 2, . . ., N3
(28)
x S 1 G1
W
2(x S
G1 ) W
1
W
;
x G1
cos i 12
G1
cos
n=1
d3i = 0,
s2i (x) =
cos (i 1)
;
1 (x/G1 )2
1
2
x a + G2
G2
a x 2
G2
2
(32)
(33)
(34)
108
SPECTRAL-DOMAIN ANALYSIS
whose Fourier transforms are obtained as
(i 1) + W
W
n
sin
J0
s1i (n ) =
2a
2
W
(i 1)
+ n S 1 + G1 +
2
2
(i 1) W
n
sin
J0
2
!
W
(i 1)
n S 1 + G1 +
2
2
G1 " ;
J0
s2i (n ) = (1)i
i 12 + n G1
2a
;
#
J i 1 n G
0
S1 S2 S. The SDA for the asymmetrical CPW can be
found in Ref. 8. Let ei(n, y) and hi(n, y) represent the scalar
electric and magnetic potentials associated with the TM and
TE modes in the ith region in the spectral domain, respectively. The Fourier-transformed Helmholtz equations of these
potentials can be expressed as
(35)
G2 " ;
J0
s3i (n ) = (1)i+n+1
i 12 + n G2
2a
;
#
J0
i 12 n G2
p = e, h,
i = 1, 2, 3
(38)
where
(36)
2
(n , y) i2 pi (n , y) = 0,
y2 pi
(37)
where J0 stands for the zeroth-order Bessel function of the
first kind. A remark needs to be made at this point, that the
numbers of both basis functions Nj and spectral terms n affect
the accuracy of the numerical results. The larger these numbers, the more accurate the results, but at the expense of the
computation time. For most engineering purposes, three basis
functions and two hundred spectral terms are sufficient.
As a demonstration of the quasistatic SDA, we show in Fig.
2 the calculated values of the characteristic impedance and
effective dielectric constant for the CPW versus the right gap.
i2 = 2 + n2 k2i
(39)
is the propagation constant, and ki ii is the wave
number in region i. Here is the angular frequency, and i
and i are the permittivity and permeability of medium i, respectively.
The boundary conditions are:
For 0 x a, y h3,
Ex3 (x, h3 ) = Ex2 (x, h3 )
(40)
Ez3 (x, h3 ) = Ez2 (x, h3 )
(41)
Hx3 (x, h3 ) = Hx2 (x, h3 )
(42)
Hz3 (x, h3 ) = Hz2 (x, h3 )
(43)
For 0 x a, y h2 h3,
Ex2 (x, h2 + h3 ) = Ex1 (x, h2 + h3 ) = Ex (x)
(44)
Ez2 (x, h2 + h3 ) = Ez1 (x, h2 + h3 ) = Ez (x)
(45)
Dynamic Spectral-Domain Analysis
Hx2 (x, h2 + h3 ) Hx1 (x, h2 + h3 ) = Jz (x)
(46)
The dynamic SDA solves the wave equation in the spectral
domain using Galerkins technique (1). The analysis can obtain the propagation constants, effective dielectric constants,
and characteristic impedances of the transmission line for all
of the eigenmodes. In essence, its formulation is similar to
that for the quasistatic case.
To simplify the analysis without loss of generality, we consider here a symmetrical CPW; i.e., we let G1 G2 G and
Hz2 (x, h2 + h3 ) Hz1 (x, h2 + h3 ) = Jx (x)
(47)
r eff
64
2.7
62
2.6
60
2.5
58
2.4
56
2.3
54
0
10
15
20
25
S2 (mils)
30
35
Effective dielectric constant
Characteristic impedance Z0 ()
66
2.2
40
Figure 2. Calculated characteristic impedance and effective dielectric constant of the coplanar waveguide using the quasistatic SDA.
a 100 mil, W S1 2G1 20 mil, h1 h3 4h2 20 mil, r1 1,
r2 2.2, r3 10.5. (1 mil 0.001 in. 0.0025 cm.)
For 0 x a, y 0 and b,
Ex3 (x, y) = Ex1 (x, y) = Ez3 (x, y) = Ez1 (x, y) = 0
(48)
Here Ei and Hi, i 1, 2, 3, are the electric and magnetic fields
in region i, respectively. Ex(x) and Ez(x) are the x and z components of the unknown electric field on the two slots; they are
nonzero on the slots and zero elsewhere. Jx(x) and Jz(x) are
the total x- and z-directed current densities on the central and
ground strips, and they are nonzero only on those strips. Now
taking the Fourier transform of Eqs. (40) to (48) gives
E x3 (n , h3 ) = E x2 (n , h3 )
(49)
E z3 (n , h3 ) = E z2 (n , h3 )
(50)
(n , h ) = H
(n , h )
H
x3
3
x2
3
(51)
(n , h ) = H
(n , h )
H
z3
3
z2
3
(52)
E x2 (n , h2 + h3 ) = E x1 (n , h2 + h3 ) = E x (n )
(53)
E z2 (n , h2 + h3 ) = E z1 (n , h2 + h3 ) = E z (n )
(54)
(n , h + h ) H
(n , h + h ) = Jz (n )
H
x2
2
3
x1
2
3
(55)
(n , h + h ) H
(n , h + h ) = Jx (n )
H
z2
2
3
z1
2
3
(56)
E x3 (n , 0) = E x1 (n , 0) = E z3 (n , 0) = E z1 (n , 0) = 0
(57)
E x3 (n , b) = E x1 (n , b) = E z3 (n , b) = E z1 (n , b) = 0
(58)
SPECTRAL-DOMAIN ANALYSIS
The fields in each region i are given in terms of the potential
functions as
3 3
Dh sinh 3 y
E x3 (n , y) = jn De sinh 3 y +
(78)
k
sinh 1 (b y)
E z1 (n , y) = jAe 1
(79)
k2 2
ei (x, y)e j z
Ezi (x, y, z) = j i
(59)
k2 2
hi (x, y)e j z
Hzi (x, y, z) = j i
(60)
Eti (x, y, z) = t ei (x, y)
i
z t hi (x, y) e j z
(61)
i
z t ei (x, y) e j z
(62)
Hti (x, y, z) = t hi (x, y) +
where the subscript t indicates the transverse (x or y) component, and
109
k2 2
E z2 (n , y) = j 2
[Be sinh 2 (y h3 ) + Ce cosh 2 (y h3 )]
(80)
k2 2
E z3 (n , y) = jDe 3
sinh 3 y
(81)
(n , y) = jn A cosh (b y) + 1 1 Ae cosh (b y)
H
x1
1
1
h
(82)
(63)
i hi (n , y)
E xi (n , y) = jn ei (n , y) +
(n , y) = jn [B sinh (y h ) + C cosh (y h )]
H
x2
2
3
2
3
h
h
2 2
[Be cosh 2 (y h3 ) + Ce sinh 2 (y h3 )]
(83)
3
3
(n , y) = jn D cosh y
(84)
De cosh 3 y
H
x3
3
h
(64)
(n , y) = jA k1 cosh (b y)
H
z1
1
h
n i
ei (n , y)
hi (n , y) +
E yi (n , y) = j
(65)
k2 2
ei (n , y)
E zi (n , y) = j i
(66)
(n , y) = jn (n , y) + i ei (n , y)
H
xi
hi
(67)
(n , y) = j n i (n , y) + hi (n , y)
H
yi
ei
(68)
t =
x + y
x
y
In the spectral domain, they are
(n , y) = j
H
zi
k2i
hi (n , y)
(n , y) = jD k3 cosh y
H
z3
3
h
(70)
h1 (n , y) = Ah cosh 1 (b y)
(71)
e2 (n , y) = Be sinh 2 (y h3 ) + Ce cosh 2 (y h3 )
(72)
h2 (n , y) = Bh sinh 2 (y h3 ) + Ch cosh 2 (y h3 )
(73)
e3 (n , y) = De sinh 3 y
(74)
1 1
Ah sinh 1 (b y)
E x1 (n , y) = jn Ae sinh 1 (b y)
(76)
E x2 (n , y) = jn [Be sinh 2 (y h3 ) + Ce cosh 2 (y h3 )]
2 2
+
[Bh cosh 2 (y h3 ) + Ch sinh 2 (y h3 )]
(77)
(87)
G 11 (n , )E x (n ) + G 12 (n , )E z (n ) = Jx (n )
(88)
G 21 (n , )E x (n ) + G 22 (n , )E z (n ) = Jz (n )
(89)
ij(n, ), i, j 1, 2, are the Greens functions in the
where G
spectral domain and can be obtained easily by the spectraldomain immitance approach (9). Using Galerkins technique,
we now express the slots electric fields as truncated summations of basis functions in the spectral domain as
E x (n )
M
cm E xm (n )
(90)
dk E zk (n )
(91)
m=0
E z (n )
K
k=1
(75)
where Ae,h, Be,h, Ce,h, and De,h are unknown constants. The
fields in the three regions in the spectral domain can now be
derived by substituting Eqs. (70) to (75) into Eqs. (64) to (69)
as
(85)
Applying the boundary conditions Eqs. (49) to (56) using Eqs.
(76) to (87) yields
(69)
e1 (n , y) = Ae sinh 1 (b y)
2
2
(n , y) = j k2 [B sinh (y h ) + C cosh (y h )]
H
x2
2
3
2
3
h
h
(86)
The solutions of Eq. (38) that satisfy Eqs. (57) and (58) are
h3 (n , y) = Dh cosh 3 y
where cm and dk are the unknown coefficients, and the basis
zk(n) describe the x and z electric field
xm(n) and E
functions E
distributions on the slots in the spectral domain. Substituting
Eqs. (90) and (91) into Eqs. (88) and (89) and taking the inner
xi(n), i 0, 1, . . .,
product of the resultant equations with E
zj(n), j 0, 1, . . ., K, respectively, results in the
M, and E
following system of coupled linear algebraic equations:
M
im
P11
( )cm +
m=0
M
m=0
K
ik
P12
( )dk = 0,
i = 0, 1, 2, . . ., M (92)
jk
P22
( )dk = 0,
j = 0, 1, 2, . . ., K (93)
k=1
jm
P21
( )cm +
K
k=1
110
SPECTRAL-DOMAIN ANALYSIS
where
im
P11
=
E xi (n )G 11 (n , )E xm (n )
(94)
E xi (n )G 12 (n , )E zk (n )
(95)
n=1
ik
P12
=
n=1
jm
P21
=
E z j (n )G 21 (n , )E xm (n )
(96)
E z j (n )G 22 (n , )E zk (n )
(97)
n=1
jk
P22
=
n=1
The unknown current densities Jx,z(n) are eliminated through
the use of Parsevals theorem.
The propagation constant can now be evaluated by setting the determinant of the coefficient matrix of Eqs. (92) and
(93) equal to zero and solving for . The effective dielectric
constant is obtained as r eff (/k0)2, where k0 00. To
obtain numerical results for , suitable basis functions for the
electric field components are needed. These functions should
satisfy the same criteria for basis functions discussed in connection with the quasistatic SDA. Here, we employ the following basis functions (10):
x G S/2
cos
m
[2(x
S/2)/S]
x G W 3S/2
cos m
,
1 [2(x G W 3S/2)/S]2
m = 0, 2, . . .
(98)
Exm (x) =
x G S/2
sin m
1 [2(x G S/2)/S]2
x G W 3S/2
sin m
+
,
1 [2(x G W 3S/2)/S]2
m = 1, 3, . . .
x G S/2
cos k
1 [2(x G S/2)/S]2
x G W 3S/2
k
cos
[2(x
W
3S/2)/S]2
k = 1, 3, . . .
(99)
Ezk (x) =
x G S/2
sin k
S/2)/S]2
1 [2(x G
x G W 3S/2
k
sin
+
,
[2(x
W
3S/2)/S]2
k = 2, 4, . . .
which are defined only over the slots. Their Fourier transforms are found to be
Propagation constant (rad/mm)
3.5
4th mode
3.0
2.5
2.0
1.5
1.0
0.5
0.0
10
3rd mode
2nd mode
1st mode
15
20
25
Frequency (GHz)
30
35
Figure 3. Propagation constants of the first four modes for the coplanar waveguide calculated using the dynamic SDA. a 60 mil, h1
30 mil, h2 10 mil, h3 20 mil, W 19.7 mil, S1 S2 S 23.6
mil, r1 1, r2 9.6, r3 13.
n S + m
S
cos[n (G + S/2)] J0
4
2
n S m
+ J0
n S + m
S
cos[
(G
+
W
+
3S/2)]
J
n
0
4
2
n S m
+ J0
, m = 0, 2, . . .
2
E xm (n ) =
n S + m
S
(100)
sin[
(G
+
S/2)]
J
n
0
4
2
S m
n
0
n S + m
S
+
sin[
(G
+
W
+
3S/2)]
J
n
0
4
2
S m
J0 n
, m = 1, 3, . . .
2
n S + m
S
j
sin[
(G
+
S/2)]
J
n
0
4
n S m
+ J0
n S + m
S
sin[
(G
+
W
+
3S/2)]
J
+
n
0
4
2
n S m
+ J0
, k = 1, 3, . . .
2
E zk (n ) =
n S + m
(101)
cos[
(G
+
S/2)]
J
n
0
4
S m
J0
n S + m
S
cos[n (G + W + 3S/2)] J0
4
2
S m
, k = 2, 4, . . .
J0 n
2
An example of computed results from the dynamic SDA is
shown in Fig. 3, in which the propagation constants of the
SPEECH ANALYSIS
first four modes for a CPW are plotted against frequency. The
first mode, which is the dominant mode, is found to be propagating up to about 26.5 GHz.
SPECTROSCOPY, DEEP LEVEL TRANSIENT. See
SUMMARY AND CONCLUSIONS
SPECTROSCOPY, PHOTOEMISSION. See PHOTO-
A detailed formulation of the SDA for planar transmission
lines has been presented. The SDA obtains the solution of the
integral equation method by applying Galerkins technique in
the Fourier transform or spectral domain with high efficiency.
SDA is simpler and numerically more efficient than the conventional integral equation method, from which SDA was derived. It has been used extensively for various microwave
problems such as planar transmission lines, resonators, antennas, and scattering. The quasistatic SDA produces accurate results only at low frequencies, whereas the dynamic process results in accurate calculations at both low and high
frequencies. Because of SDAs attractive features, it has become one of the most popular numerical methods for analyzing microwave and millimeter-wave passive structures, and
is expected to remain so.
SPECTRUM EFFICIENCY. See CHANNEL CAPACITY.
DEEP LEVEL TRANSIENT SPECTROSCOPY.
SPECTROSCOPY, NONLINEAR. See NONLINEAR
OPTICS.
EMISSION.
BIBLIOGRAPHY
1. T. Itoh and R. Mittra, Spectral-domain approach for calculating
the dispersion characteristic of microstrip line, IEEE Trans. Microw. Theory Tech., MTT-21 (7): 498499, 1973.
2. T. Itoh and A. S. Hebert, A generalized spectral domain analysis
for coupled suspended microstriplines with tuning septums,
IEEE Trans. Microw. Theory Tech., MTT-26 (10): 820826, 1978.
3. D. Mirshekar-Syahkal, Spectral Domain Method for Microwave
Integrated Circuits, Somerset, UK: Research Studies Press, 1990.
4. T. Itoh, Analysis of microstrip resonators, IEEE Trans. Microw.
Theory Tech., MTT-22 (11): 946952, 1974.
5. Q. Zhang and T. Itoh, Spectral-domain analysis of scattering from
E-plane circuit elements, IEEE Trans. Microw. Theory Tech.,
MTT-35 (2): 138150, 1987.
6. C. Scott, The Spectral Domain Method in Electromagnetics, Norwood, MA: Artech House, 1989.
7. T. Kitazawa and T. Itoh, Asymmetrical coplanar waveguide with
finite metallization thickness containing anisotropic media, IEEE
MTT-S Int. Microwave Symp. Dig., 1990, pp. 673676.
8. K. M. Rahman and C. Nguyen, On the computation of complex
modes in lossless shielded asymmetric coplanar waveguides,
IEEE Trans. Microw. Theory Tech., MTT-43 (12): 27132716,
1995.
9. T. Itoh, Spectral-domain immittance approach for dispersion
characteristic of generalized printed transmission lines, IEEE
Trans. Microw. Theory Tech., MTT-28 (7): 733736, 1980.
10. T. Uwano and T. Itoh, Spectral-domain approach, in T. Itoh (ed.),
Numerical Techniques for Microwave and Millimeter-Wave Structures, Wiley, New York, 1989, pp. 334380.
CAM NGUYEN
Texas A&M University
SPECTROMETER. See MASS SPECTROMETERS; PARTICLE
SPECTROMETERS.
SPECTROMICROSCOPY. See X-RAY MICROSCOPY.
SPECTROSCOPY. See PULSE HEIGHT ANALYZERS.
111
260
SPIRAL ANTENNAS
SPIRAL ANTENNAS
Spiral antennas are so named because of their shape. Often
flat, antenna arms of thin conducting sheet material are defined by spiral curves. Two widely used shapes are related to
the spiral curves shown in Figs. 1 and 2. The spacing between
turns of an Archimedean spiral is constant, but in the logarithmic spiral the spacing grows with the distance from the
origin. An arm of an Archimedean spiral antenna has the spiral curve as centerline and the arm has constant width. An
arm of a logarithmic spiral (log-spiral) antenna has edges defined by two logarithmic spiral curves that differ by rotation
around the origin. Hence, the arm width of a log-spiral antenna also increases with distance from the origin. Usually,
two or four arms comprise an antenna.
The antenna arms cannot follow the spiral curves in the
regions close to or far from the origin. The truncation of the
conductors near the origin provides terminals for connection
to transmission lines or cables. Truncation at the other end,
which defines the length of the arms, is dictated by practical
matters, such as cost, weight, and the available space. The
positions of two arms to form a planar Archimedean spiral
antenna are shown in Fig. 3. The second arm is identical to
the first, but rotated by 180. Figure 4 shows a two-arm logspiral antenna. Note that the small and large limits of this
structure are defined by lines, but could also be defined by
concentric circles of small and large radii centered at the
origin.
Planar spirals can be made by cutting the desired shapes
from thin sheets of good-conducting metal, such as copper or
aluminum. However, a better way is to use photolithographic
techniques like those used to make printed circuits. Thin
sheets of metal-clad microwave substrate (dielectric) material
are commercially available in rather large sizes. The shapes
of the spiral antenna arms are drawn on paper using computer software like that used in preparing Figs. 3 and 4. The
artwork is used to make a photographic negative. The image
Figure 1. Segment of an Archimedean spiral curve. The dot in the
center is the origin.
J. Webster (ed.), Wiley Encyclopedia of Electrical and Electronics Engineering. Copyright # 1999 John Wiley & Sons, Inc.
SPIRAL ANTENNAS
261
Figure 2. Segment of a logarithmic spiral curve. The dot in the center is the origin.
of the drawing is then used to expose photo-resist on the
metal-clad substrate. The excess metal is etched away, leaving the antenna arms in the desired shape, supported by the
thin layer of dielectric.
Many spiral antennas conform to the surface of a cone
rather than a plane. Infrequently, spiral antennas conform to
some surface other than a plane or a cone. The construction
of nonplanar antennas is somewhat more complicated but can
be accomplished with proper modification of the photolithographic method. In the case of conical spirals flat artwork is
prepared and used to fabricate a planar set of conductors on
flat, but flexible, substrate. The substrate is then formed into
a conical shape and the conductors are soldered together
along the seam.
Most spiral antennas can radiate or receive at any frequency within a very wide band, so wide that it is characterized by the ratio of the upper to the lower limit (e.g. 20:1).
The lowest frequency of the operating band is determined by
the distance from the origin to the far truncation, the maximum extent of the arms. The highest frequency is determined
by the distance from the origin to the near truncation, the
size of the terminal region.
Transmitting planar spirals in free space radiate the same
amount of power on each side of the plane containing the spiral arms. Receiving planar spirals in free space receive
equally well on both sides of the plane. The radiation pattern
of a planar spiral is generally rather broad, encompassing
much of the angular space above and below the plane of the
Figure 3. A two-arm, balanced Archimedean spiral antenna. The inner ends are the input terminals.
Figure 4. A two-arm, balanced logarithmic spiral antenna. The outlined edges could be the boundaries of thin metal arms, or of a slot
in a thin metal sheet.
spiral. However, conical spirals are usually designed to radiate mostly in one hemisphere. Depending upon the choice of
the parameters of the spiral curve, the radiation beam of a
conical spiral antenna can be made quite broad or rather
narrow.
In most of the radiation beam of spiral antennas the polarization is circular, or nearly so. The sense of the polarization
is determined by the direction of rotation in which the spiral
expands. Pointing the thumb in the direction of propagation
of the wave radiated by the spiral with the fingers parallel to
the spiral (in the direction of the spiral wrap), the hand used
indicates the sense of polarization following the standards of
the Institute of Electrical and Electronics Engineers. (Physicists often use the opposite designation.) Planar spirals radiate one sense in one hemisphere and the opposite sense in the
other hemisphere. The spiral of Fig. 3 radiates right-hand
sense in the space above the antenna; left-hand sense in the
space below. Spirals with more than two arms must be excited with an arm-to-arm phase progression that agrees with
the sense of the polarization as determined by the direction
of wrap of the spiral. Spiral antennas are sometimes used
even in narrowband applications because of their polarization properties.
One of the first applications of a spiral antenna was on
a navigation satellite. Many other satellite applications have
occurred since then. Circular polarization is preferred over
linear polarization for waves that penetrate the ionosphere
where the plane of linear polarization undergoes rotation.
Conical spiral antennas are used as feeds for reflector antennas. The University of Illinois radio telescope used a linear
array of conical spirals as a slow-speed scanning feed for a
parabolic cylinder reflector (1). The direction of the beam of
such an array can be changed by rotating the antenna elements. This follows from the circular polarization of the spiral
antennas, because the phase of the radiated field is directly
related to the angle of rotation. The broadband capability of
spiral antennas is utilized in direction finding, surveillance,
and electronic warfare systems (25).
262
SPIRAL ANTENNAS
SPIRAL ANTENNA GEOMETRY
lim
The defining equation for a planar Archimedean spiral is
= k
(1)
where is the distance from the origin to a point on the spiral, is the angle measured from a reference axis in the counterclockwise direction, and k is a parameter that determines
how tightly the spiral is wrapped. The sign of k determines
the direction of wrap. The spacing between the turns of an
Archimedean spiral is given by 2k. If Eq. (1) is used to define the centerline of one arm of an antenna, then the centerline of the other arm of a balanced, symmetric antenna would
be given by
= k( )
(2)
Input terminals are provided by starting the spiral, not at
0 or 0, but at some other value of either or . The
structure is truncated at the large end by defining a maximum value of either or . The antenna of Fig. 3 was drawn
by using Eq. (1) with k 0.01, min /2 and max 10 and
Eq. (2) with k 0.01, min 3/2 and max 11.
The equation for the logarithmic spiral curve of Fig. 2 is
= ea
(4)
the parameter a is given by
a=
1 d
d
(5)
which shows a to govern the rate at which the distance from
the origin increases as increases. Hence, a is called the spiral-rate constant. Negative values of a give spirals that approach the origin as increases, said to be wrapped in the
direction opposite to those with positive a.
An incremental change in produces an incremental
change in as shown in Fig. 5. The hypotenuse of the differential triangle becomes tangent to the curve in the limit as
0.
(6)
Hence, the angle between the radial line and the tangent
is another measure of the spiral rate. Spirals with near 90
are said to be tightly wrapped, but those with small values of
are loosely wrapped. For a given logarithmic spiral, the
angle is fixed, leading to the alternative name, equiangular spiral.
Suppose that the scale is changed on a log-spiral curve, say
by multiplying the radial coordinate by a constant, K 1.
2 = Kea
(7)
The same spiral would be given by
1 ln K )
2 = ea(+ a
(8)
which is the equation for the original spiral rotated through
the angle
=
(3)
where and are as defined above, e is the base of Napierian
logarithms, and a is a parameter. Since
d
= aea = a
d
1 d
=
= a = cot
d
1
ln K
a
(9)
clockwise being considered the positive direction of rotation.
Hence, a change in the scale of a log-spiral curve is equivalent
to a rotation.
Two log-spiral curves separated by rotation by some angle
form the edges of one arm of a log-spiral antenna. Either
or K can be used as the arm-width parameter, because
they are related by Eq. (9). A third edge for the arm can
be defined either by d/2 or by 1; a fourth edge,
either by D/2 or by 2. For a symmetrical, balanced antenna the second arm is obtained by rotating the
first by 180. One of the arms of the log-spiral antenna of
Fig. 4 was defined by
1 = e0.25
2 = e0.25( /2)
1 = 3
2 = 0.5
and the other one, by
3 = e0.25( )
A
B
4 = e0.25(3 /2)
3 = 2
4 =
Figure 5. Construction showing the angle, A, between an extended
radius and the tangent to a logarithmic spiral curve. A , B
, C .
3
2
Frequently, nonplanar spiral antennas are needed. Conical
spirals are widely used because of their unique performance.
Antennas for fast moving vehicles are often required to be
conformal with the surrounding surface. The geometry of a
nonplanar spiral antenna is usually derived from that of a
SPIRAL ANTENNAS
= e a
(a)
= k
5
4
3
2
1
(b)
Figure 6. Projection of log-spiral and Archimedean spiral curves in
a plane onto the surfaces of cones. Note that the angle, , between a
radius and the tangent to the spiral curves is constant for the logspiral, but is not constant for the Archimedean spiral. After J. D.
Dyson and P. E. Mayes, New circularly polarized frequency-independent antennas with conical beams or omnidirectional patterns. IRE
Trans. Antennas Propag., AP-9: 4C, 1961.
planar spiral antenna by orthogonal projection. Figure 6
shows how the procedure would work graphically, however,
analytic expressions can be used to compute points on the
space curves needed to outline the antenna. If the cone is defined by 0, then the equations for the edges of one
conical log-spiral arm are
1 = e(a sin 0 )
2 = e(a sin 0 )( ) = K1
As with the planar log-spiral antennas, rotation of both edges
of one arm will define the edges of a second arm of a balanced
antenna. The angle between the radius vector and the tangent to any edge of the conical log-spiral antenna remains
constant for any cone angle 0.
HISTORY OF DEVELOPMENT
Experiments with Archimedean spiral antennas were begun
in 1953 by E. M. Turner at Wright Air Development Center,
Dayton, Ohio (6). The problem of limited antenna bandwidth
was hampering the development of broadband homing, direction-finding and other electronic systems of interest to the
U.S. Air Force. The Research Laboratory of Electronics at
Massachusetts Institute of Technology (MIT) was awarded a
263
contract to develop a theory for the Archimedean spiral antenna and much development work was done at the Naval
Research Laboratory, Diamond Ordnance Fuze Laboratories,
and elsewhere. Other contracts went to the Antenna Laboratory of the University of Illinois at Urbana-Champaign
(UIUC) to study the general problem of greatly increasing antenna bandwidth.
In September of 1954, V. H. Rumsey came to the UIUC
Antenna Laboratory from Ohio State to assume the position
of director. It was well known that the performance of an antenna was determined by its size in wavelengths. Rumsey had
already tested some preliminary ideas about antennas that
were described, insofar as possible, by angles rather than
lengths. He also recalled that Mushiake had pointed out that
a self-complementary antenna would have a constant input
impedance, independent of frequency (7). Little practical use
had been made of this principle because all self-complementary shapes extend to infinity. It occurred to Rumsey that an
unbounded logarithmic (equiangular) spiral could be specified
completely by a single angle and it could be used to define a
self-complementary antenna (8). A two-arm, log-spiral antenna is said to be self-complementary when the arm edges
are spirals that are 90 apart (as are those in Fig. 4). The
space between arms is then congruent to the arms, the condition to be self-complementary.
But could a physical antenna, necessarily a truncated version of the ideal, be made to operate like the infinite one, even
over a finite band? He suggested to John Dyson, a graduate
student in the Antenna Laboratory, that self-complementary
antennas using log-spirals as boundaries should be investigated as potential broadband antennas.
It was not readily apparent how to isolate the performance
of the spiral-shaped antenna from the effects of the cable that
must be attached to connect the antenna to transmitters or
receivers for testing. By making the spiral arms as a slots in
a large ground plane, rather than a thin conducting sheet, the
cable could be affixed to the ground plane where it would
cause minimum perturbation of the field. So the terminals of
the spiral slots were extended directly toward one another to
meet at the origin, the shield of the feed cable was soldered
to the area between the spiral slots up to one side of the slot
near the origin, and the center conductor was attached on the
other side of the slot. In some instances, a dummy cable was
soldered in a symmetrical location to improve the balance of
the feed region and the symmetry of the radiation patterns.
This feed provides a simple conversion from the unbalanced
cable to the balanced antenna, the function of devices called
baluns.
Dyson demonstrated that planar log-spiral slot antennas
in the shape shown in Fig. 4 (with the terminals connected
by continuation of the slot) could indeed be made to have impedance and pattern characteristics that changed very little
over bandwidths greater than 10:1. Dyson later studied planar sheet log-spirals and conical log-spirals and gave design
and application information (9).
THEORY OF OPERATION
As with other antennas, the distribution of current produced
by applying a source of radio frequency power to the terminals is the key to understanding the performance of spiral
264
SPIRAL ANTENNAS
antennas. Although powerful computer methods are available
and have been used to find the current distributions on spiral
antennas with conducting arms (10,11), the computed results
serve primarily to reinforce intuitive concepts and measurements that were used in the original development of spiral
antennas.
Consider the Archimedean spiral shown in Fig. 3. A balanced source at the input would produce currents at the terminals which differ in phase by 180. Because currents tend
to travel along thin wires with a constant speed determined
by the surrounding medium, the phase difference at any two
points that are diametrically opposite one another will remain
at 180. However, it is the relative phase of the currents in
adjacent arms of the spiral that is important to understanding the radiation phenomenon. Near the input terminals the
distance the current must travel from one terminal until it
draws near the current from the out-of-phase terminal is
small compared to the wavelength. So the adjacent currents
near the input are approximately out of phase. Out-of-phase
currents act in a manner similar to currents in a two-wire
transmission line, producing little radiation.
At some distance away from the input of an Archimedean
spiral antenna there will be a half-turn that is approximately
one-half wavelength in length. In this region consider the currents at points A and B that are diametrically opposed on the
spiral. As always, these currents will be out of phase. But
between point A and a point C on the same spiral arm, where
points B and C are located side-by-side on adjacent turns, the
current will undergo 180 shift in phase. The condition of zero
phase difference at points B and C is conducive to radiation.
It is satisfied exactly only at points B and C, but it is satisfied
approximately over a band that may constitute several turns
of the spiral antenna. The input power is thus guided by the
spiral arms from the input terminals to a radiation band that
has a circumference approximately equal to the wavelength
of the input signal. In the radiation band the currents change
in phase by approximately 360 each time they propagate
around the quasicircular path. The analysis of such a current
distribution leads to a rather broad beam of circularly polarized fields in both the upper and lower hemispheres (12).
The power that is delivered to the input terminals of an
Archimedean spiral antenna begins to propagate along the
spiral arms as though they were conductors of a two-wire
transmission line. As the relative phase of the currents in adjacent conductors begins to change from 180 toward zero degrees, the spiral begins to radiate. The radiation per unit
length of the spiral arms will become maximum at the region
where the phase difference goes to zero. However, radiation
will continue beyond the maximum point. For best performance, it is desirable for almost all of the input power to be
radiated before the relative phase of the currents in adjacent
conductors approaches 180 once more. If the maximum circumference of the spiral is no greater than one wavelength at
the lowest frequency of operation, leftover power will be refected from the truncated ends of the spiral arms. Because
the reflected power travels through the radiation zone in the
opposite direction to the incident power, the radiation will be
polarized in the opposite sense. This will be detrimental to
the axial ratio of the antenna. Furthermore, any reflected
power that reaches the input terminals will cause a variation
in the input impedance as frequency changes.
If the antenna circumference is large enough, additional
radiation zones will be present. The phenomenon described
above will occur again near any region where the circumference is an odd integer multiple of the wavelength. Because
the amplitude and phase of the radiation fields depend upon
the order of the radiation zone, simultaneous radiation from
more than one zone is generally detrimental to the radiation
pattern.
The current distribution on log-spiral antennas is not so
easily determined or understood, particularly for the planar
case. A combination of numerical techniques, approximate analytical techniques, and experimental measurements have
been used. Some of the experimental results are presented in
the next section and the numerical methods are discussed
briefly in the last section.
EXPERIMENTAL DATA
Planar Log-Spiral Antennas
In the early investigation by J. D. Dyson, the radiation patterns of more than forty log-spiral slot antennas with 0.2
a 1.2 and 0.375 K 0.97 were measured (13). Each antenna shape was cut from a 35.5-cm square metal plate and
a feed cable was affixed in the area between the slots. The
metal plate was then inserted into a like-sized hole near the
center of a ground plane 3.66-m square. A movable boom
could be swung across the space above the ground plane to
measure the variable cut in a hemispherical coordinate system with the polar axis normal to the ground plane. The plate
containing the slot antenna could also be rotated about the
normal so that the elevation cuts could be made at any value
of . Or a variable cut could be measured, while the boom
was held fixed at a given location. Typical measured patterns of both the and components of the electric field are
shown in Figs. 7 and 8. The patterns at 595 and 12,000 MHz
show the effects of truncation at the large and small ends,
respectively. Across the band between these frequency limits,
a bandwidth of 20:1, the patterns are very much the same.
On loosely wrapped log-spirals, the beamwidths in orthogonal planes are not equal. Then it can be observed that the
pattern rotates with changes in frequency. Thus, the pattern
repeats periodically with the logarithm of the frequency. Logspiral antennas are special cases of logarithmically periodic
(log-periodic) antennas. Antennas with tightly wound arms
and those with wide arms have more uniform patterns that
suffer smaller variations in beamwidth. Antennas with negligible variation in patterns and impedance over bandwidths
determined solely by the truncation dimensions are called frequency-independent antennas.
Measurements of fields near the antenna arms indicate a
rapid decay (up to 20 dB per wavelength) with displacement
from the feedpoint. As frequency increases, so does the decay.
The portion of the antenna having appreciable near field is
almost constant when measured in wavelengths. The decay
in near fields makes possible the truncation at a finite distance from the origin while retaining the radiation characteristics of a structure of infinite size. The large-end truncation
effect is negligible as long as the arm length is equal to or
greater than approximately one wavelength. Because of the
spiral shape, this arm length can be contained within a circular area of diameter equal to or less than one-half wave-
SPIRAL ANTENNAS
=0
f = 595 MHz
r = 1.94
= 90
265
length. Planar log-spirals that are truncated at the large end
in a circle operate to a lower frequency for a given maximum
diameter than those that are truncated in lines as in Fig. 4.
Once the operating frequency is higher than the required
minimum, the input impedance is observed to change little
until the frequency is high enough that the feed region becomes an appreciable part of the operating wavelength. The
average value of the impedance of a balanced log-spiral antenna is predominantly resistive. For a slot version, the average input impedance varies, depending primarily upon the
width of the slot, from 180 for wide slots (K 0.5) to 60
for narrow ones (K 0.9). The efficiency of planar log-spiral
antennas, whether slots in a metal plane or metal arms in
free space, was measured to be approximately 98 %, as long
as the arm length was equal to or greater than one wavelength.
E
E
f = 800
r = 1.39
f = 1000
r = 1.19
f = 1400
r = 1.04
Conical Log-Spiral Antennas
f = 2000
r = 1.04
E
E
Figure 7. Measured radiation patterns of a two-arm, balanced, logspiral antenna, r axial ratio. (After J. D. Dyson, The equiangular
spiral antenna. IRE Trans. Antennas Propag., AP-7: 2C, 1959.
=0
f = 2800 MHz
r = 1.05
A very interesting and useful phenomenon occurs when the
arms of a log-spiral antenna are developed on the surface of
a cone, as shown in Fig. 9, particularly a cone with small apex
angle (20). As the apex angle decreases, the radiation becomes confined more and more to the half-space in the direction the tip of the cone is pointing. The measured patterns
displayed in Fig. 10 show that, when 0 15, the front-toback ratio is very high. This unidirectional pattern is needed
for many applications and is the principal contributing factor
to the popularity of conical log-spiral antennas.
The conical log-spiral can be fed by bonding the feed cable
to one arm in a manner similar to the feed of the planar logE
= 90
;;;;
Z
f = 3950
r = 1.07
d
2
f = 3600
r = 1.07
f = 9027
r = 1.07
f = 12000
r = 1.24
H
h
d
2
E
E
Figure 8. Measured radiation patterns of a two-arm, balanced, logspiral antenna, r axial ratio. After J. D. Dyson, The equiangular
spiral antenna. IRE Trans. Antennas Propag., AP-7; 2C, 1959.
Figure 9. Parameters of a conical log-spiral antenna and the coordinates used to describe radiation patterns. After J. D. Dyson, The
characteristics and design of the conical log-spiral antenna, IEEE
Trans. Antennas Propag., AP-13: 7C, 1965.
SPIRAL ANTENNAS
E
E
= 90
o = 10
o = 15
o = 30
o = 90
Figure 10. Radiation patterns of a conical log-spiral antenna showing increasing front-to-back ratio with decreasing cone angle, 0. After
J. D. Dyson, The unidirectional equiangular spiral antenna. IRE
Trans. Antennas Propag., AP-7: 4C, 1959.
spirals. However, because the cable so used may be much
longer than for the planar antennas, cable loss may be unacceptably high, particularly at frequencies above 1 GHz. It is
possible to bring a feed line along the interior axis of the cone
with minor effect on the performance of the antenna. A singlefeed cable can be used with a broadband balun at the feed
point, or twin cables can be excited with 180 phase difference
by a hybrid network located at the base of the cone.
Since the arms of a conical log-spiral appear to be wrapped
more tightly than those of the corresponding planar antenna,
the radiation patterns are more nearly rotationally symmetric. Even though the pattern rotates with frequency, the variation in beamwidth is hardly noticeable. A conical log-spiral
with 0 10 and 73 was observed to have half-power
beamwidths of 70 for electric field polarized in the direction
and 90 for electric field polarized in the direction. The result is radiation that is very nearly circularly polarized over
much of the beam.
Much of the behavior of the radiation pattern can be understood in light of knowledge of the near field. Figure 11
shows measured amplitude and phase of the magnetic field
close to one arm of conical log-spirals that differ only in angle
of wrap, . For the tightly wrapped case ( 80) the near
field displays a single dominant peak beyond which occurs a
precipitous drop. The measured phase is smooth and indicative of the phase progression of a wave traveling toward the
feedpoint (i.e., radiation in the backfire direction). Most of the
radiation takes place in the vicinity of the peak of the near
field, the so-called active region. The radiation in the backfire
direction results in the rapid decay in the near field. When
the active region is relatively small, encompassing only a few
turns of the spiral, most of the radiation is phased for back-
fire. As frequency changes, the active region moves, maintaining constant size in wavelength measure.
When the antenna is wrapped more loosely ( 60), the
active region is broader and includes turns that are phased
away from backfire toward broadside. This accounts for the
broader beams. This effect becomes even more pronounced as
the wrap is loosened further ( 45). Fluctuation in the
near-field amplitude in Fig. 11 is caused by the probe passing
close to conductors of the spiral. The dashed lines show
smoothed data more indicative of the behavior of the near
field along the arm rather than that along a radial line.
The plots of phase versus displacement in Fig. 11 illustrate
how the phasing is affected by the increasing length of conductor in each successive turn of the log-spiral. Appreciable
slope is seen in the active region for 80. As the slope in
the active region decreases, more turns are included in the
active region and some of these turns are phased in directions
slightly removed from backfire. The bandwidths of antennas
with wide active regions are smaller for a given physical size
than those of antennas with more limited active regions.
Apparently because of the effective small wrap angle of
conical log-spiral antennas, the pattern performance does not
depend greatly upon the width of the arms. In fact, for moderate bandwidths it is possible to eliminate the flat conductors
altogether if a dummy cable is used on the second arm as
recommended to preserve the symmetry.
Rel phase adance
=0
Active region
Phase
1000
0
3
900
= 80
800
700
10
15
Relative amplitude in dB
266
0
500
0
.5
1.0
1.5
2.0
2.5
2.0
2.5
400
= 60
300
3
10
15
0
200
.5
1.0
1.5
100
0
= 45
3
10
15
30
1.5
.5
1.0
2.0
Distance along conical surface ( / )
2.5
Figure 11. Relative amplitude and phase of magnetic fields (currents) measured along the surface of conical log-spiral antennas
(20 15, 90) and corresponding radiation patterns. After J. D.
Dyson, The characteristics and design of the conical log-spiral antenna. IEEE Trans. Antennas Propag., AP-13: 7C, 1965.
SPIRAL ANTENNAS
Several parameters are required to describe a single conical log-spiral antenna. The pattern shape is most dependent
upon the apex angle, 20, of the cone. However, the arm
width, , and the wrap angle, , can also have an effect. Radiation patterns for several values of 20, , and are shown in
Fig. 12. As to be expected, tightly wrapped antennas have the
smoothest, most symmetrical patterns for a wider range of
20 and . However, the front-to-back ratio is best for the
smallest apex angle, 20 15. Antennas with arm widths
given by 90 have patterns superior to those with either
wider or narrower arms. Infinite planar spiral structures with
90 are self-complementary. Although the argument for
constancy of impedance applies only to such planar cases, evidently there are other advantages to be gained by applying
the self-complementary condition.
The measured input impedance does not change much over
the frequency band of good patterns, a span of 20 : 1 or more
in frequency is possible. Table 1 gives data for the mean impedance and SWR as a function of the cone angle, 0.
Multiarm Log-Spiral Antennas
Although most log-spiral antennas in use today have two
arms, that is by no means a requirement. The needs of certain
applications can be better met with an antenna having more
than two arms. Log-spiral antennas with only one arm can be
fed against ground, but the performance is unsuitable for
most applications (14,15). Two-arm log-spirals are conventionally fed with 180 degree phase difference between the
arms at the input. As the number of arms increases, other
= 60
= 70
= 80
= 0 = 90 = 0 = 90 = 0 = 90
267
Table 1. Measured Input Impedance of Balanced Conical
Log-Spiral Antennas (K 0.925, L 150 cm, a 0.303 sin 0 ,
73)
Approx. mean impedance ()
Max. SWRa
10
15
30
90
129
147
153
164
1.9
1.9
1.95
2.1
a
Referred to the mean impedance.
Source: J. D. Dyson, The unidirectional spiral antenna, IRE Trans. Antennas
Propag, AP-7: 4C, 1959.
possibilities arise, as shown in Fig. 13. Some of these are simply related to the azimuthal ( dependent) solutions of Maxwells equations, exp(jm), where m is an integer to ensure
single-valued fields. The conventional feed for two-arm spirals
can produce fields having only odd integer values of m, because a rotation of 180 is in agreement with a phase change
of 180. However, measurements of the fields radiated from
conventionally fed two-arm spirals find that contributions
from values of m greater than one are negligibly small.
Hence, it might be expected that a field which varies as
exp(j2) could be produced by feeding a four-arm spiral as
shown in Fig. 13(c). Recalling that the sense of polarization is
determined by the direction of wrap, trying to excite a
multiarm spiral in the wrong sense leads to poor performance. The excitation of Fig. 13(c) applies equally well to
m 2, but the direction of wrap determines the sense of
the polarization. The purity of the phase law as given by
exp(jm) is demonstrated by the measured phase of the
= 16
= 90 2 o = 15
+
(a)
= 164
= 16
+1
1
= 90 2 o = 20
+1
1
= 164
(b)
= 16
1
= 90 2 o = 30
+1
+1
1
= 164
(c)
Figure 12. Typical radiation patterns indicating changes in shape
with cone angle, 0, spiral angle, , and angular arm width, . After
J. D. Dyson, The characteristics and design of the conical log-spiral
antenna. IEEE Trans. Antennas Propag., AP-13: 7C, 1965.
Figure 13. Possible simple excitations of multiarm antennas. After
J. D. Dyson and P. E. Mayes, New circularly polarized frequencyindependent antennas with conical beams or omnidirectional patterns. IRE Trans. Antennas Propag., AP-9: 4C, 1961.
SPIRAL ANTENNAS
fields radiated by two-arm and four-arm spirals shown in
Fig. 14.
Maxwells equations show that any solution of the form
exp(jm), with integer m differing from unity, has a null
along the polar axis (a conical beam). Figure 15 shows elevation-plane patterns for four-arm conical log-spirals fed in the
manner of Fig. 13(c). Note that the angle of the peak of the
conical beam can be controlled by changing the angle of wrap
parameter, . Using 45 produces a circularly polarized
field that has an omnidirectional pattern in the azimuthal (
90) plane. Recalling that the angle of wrap of the logarithmic spiral is constant, whereas that of the Archimedean spiral is not, it can be expected that consequences of this difference in geometry would be displayed in the radiation
characteristics of these two types of spirals. A widening of
the beamwidth of the two-arm Archimedean spiral is hard to
observe for planar antennas, but becomes more apparent for
conical versions. For four-arm spirals excited to produce conical beams, the difference in radiation patterns is evident, as
shown in Fig. 16. Thus, it is possible by using a four-arm
conical Archimedean spiral to obtain a beam that scans toward the horizon as frequency increases. If this frequencyscanning behavior is not desirable, then a logarithmic spiral
should be used.
E
E
= 73
= 60
= 45
40
50
in degrees
268
60
70
80
90
40
60
70
50
in degrees
80
90
Figure 15. Typical radiation patterns and orientation of the conical
beam as a function of the spiral angle, (7.5 0 10). After J. D.
Dyson and P. E. Mayes, New circularly polarized frequency-independent antennas with conical beams or omnidirectional patterns. IRE
Trans. Antennas Propag., AP-9: 4C, 1961.
720
680
+1
640
1
RECENT AND FUTURE WORK
600
Relative phase of far field in degrees
Planar spiral antennas in free space find little application because of the equal radiation on both sides of the plane. If the
application will allow 3 to 4 dB loss, the power radiated from
+1
560
520
480
440
Archimedes spiral
Equiangular spiral
400
sin 10
(
)
= e tan 45
= 1.026
360
320
f = 1000 MHz
E
E
280
240
200
f = 1400 MHz
160
120
+1
80
f = 2000 MHz
40
0
40
80 120 160 200 240 280 320 360
Azimuthal angle in degrees
Figure 14. Phase of radiated field measured in the 90 plane as
a function of the azimuth angle, , for two antennas with 45.
After J. D. Dyson and P. E. Mayes, New circularly polarized frequency-independent antennas with conical beams or omnidirectional
patterns. IRE Trans. Antennas Propag., AP-9: 4C, 1961.
(a)
(b)
Figure 16. Radiation patterns of symmetrical, four-arm (a) Archimedean and (b) equiangular spiral antennas; 0 10, D 29.5 cm,
d 4.5 cm, ( 90, variable). After J. D. Dyson and P. E. Mayes,
New circularly polarized frequency-independent antennas with conical beams or omnidirectional patterns. IRE Trans. Antennas Propag.,
AP-9: 4C, 1961.
SPIRAL ANTENNAS
one side can be absorbed. This is often accomplished by placing a conducting cavity on one side of a planar spiral and
introducing resistance cards, lossy foam, or some other absorbing material into the cavity. Sometimes the losses can be
avoided if the band to be covered is sufficiently narrow and
some degradation in performance can be permitted.
Some recent publications have indicated that spiral antennas can operate well when placed parallel to a closely spaced
conducting ground plane (16). The effect of the close ground
is to reduce the rate of decay of the currents on metal spiral
arms. Hence, it may be necessary to absorb the power that is
left at the truncation to prevent large reflected power from
degrading the pattern and impedance. A ring of absorber
around the outer rim of a spiral antenna may dissipate less
power than an absorbing cavity and provide thereby increased gain.
An alternative to the peripheral absorber ring is a vertical
resistance card that spans the space between the spiral arms
and the ground plane and follows along an edge or the centerline of the spiral. This method has the potential to control the
rate of decay of the currents so that negligible current remains to be reflected at the truncation. Although computer
simulations have shown the card-loaded spirals over close
ground to work well, the construction of such an antenna is
difficult and the measured performance is not as good as predicted (15).
The electromagnetic boundary-value problem presented by
a log-spiral antenna has, so far, proved intractable. Analytic
results have been obtained for anisotropic conducting sheets
with spiral lines of conductivity (17,18), but this model is
somewhat divergent from the practical case. The availability
of high-speed digital computers has made it possible to solve
discretized versions of integral equations for practical log-spiral geometries. The conducting arms of a spiral antenna are
divided into many small areas (patches). The current on each
patch is represented by some (basis) function with an unknown complex-valued coefficient. Expressing the field in
terms of these coefficients enables one to convert the integral
equation into a set of linear algebraic equations that can be
solved with standard techniques. The digital computer is capable of solving such systems of equations that involve thousands of unknowns. The coefficients so determined can be
used with the basis functions to calculate an approximate current distribution on the antenna. The calculation of the radiated fields, polarization and patterns, from the approximate
currents is straightforward. An approximate value for the input impedance can be determined from the value of the current at the input terminals. Additional information about
applying these method of moments techniques to spiral antennas can be found elsewhere (19,20).
Finding the currents on multiarm spirals, particularly
those used with closely spaced ground planes or cavities (with
or without absorber) and intended for use over very wide
bands, can tax the capability of even the fastest supercomputer. One way to reduce the number of unknowns on many
spiral antennas is to make use of the discrete rotational symmetry, the fact that each arm is an exact duplicate of the
others and the separations between adjacent arms are the
same. Further work in numerical analysis will likely yield
faster algorithms for computing the electromagnetic fields
produced by various distributions of currents and this will
reduce the computer time needed to solve problems with large
numbers of unknowns. Alternative methods for evaluating
269
the input impedance may prove to be more accurate than the
value obtained directly from just the approximate current at
the input terminals.
BIBLIOGRAPHY
1. G. W. Swenson and Y. T. Lo, The University of Illinois radio telescope, IRE Trans. Antennas Propag., AP-9: 916, 1961.
2. J. D. Dyson, Multimode logarithmic-spiral antennaspossible
applications, Proc. Natl. Electron. Conf. 206213, 1961.
3. G. Hahn and R. Honda, Conical spiral arrays for passive direction
finding, Proc. USAF Res. Develop. Program, Urbana: University
of Illinois, 1968.
4. H. Montague, M. J. Horrocks, U. W. Margosian, and J. D. Dyson,
The dual-aperture counter-wound log-spiral antenna directionfinder system, IEEE Trans. Antennas Propag., AP-21: 224226,
1973.
5. D. D. Cornell and B. J. Lamberty, Multimode planar spiral for
DF applications, Proc. Antenna Appl. Symp., Urbana: University
of Illinois, 1981.
6. E. M. Turner, Spiral slot antenna, Wright-Patterson AFB, Ohio,
Tech. Note WCLR-55-8 WADC, June 1955.
7. Y. Mushiake, J. Inst. Electron. Eng., Japan, 69: 8688 (in Japanese), 1949.
8. V. H. Rumsey, Frequency-independent antennas, IRE Natl. Conv.
Rec., I: 114118, 1957.
9. J. D. Dyson, The characteristics and design of the conical logspiral antenna, IEEE Trans. Antennas Propag., AP-13: 488
499, 1965.
10. Y. S. Yeh and K. K. Mei, Theory of conical equiangular-spiral
antennas, part I: Numerical technique, IEEE Trans. Antennas
Propag., AP-15: 634639, 1967.
11. A. E. Atia and K. K. Mei, Analysis of multiple-arm conical logspiral antennas, IEEE Trans. Antennas Propag., AP-19: 320
331, 1971.
12. B. H. Burdine, Radiation from the double-spiral antenna, MIT
Res. Laboratory Electron. Rep., 1955.
13. J. D. Dyson, The equiangular spiral antenna, IRE Trans. Antennas Propag., AP-7: 181187, 1959.
14. R. L. Carrel, Experimental investigation of the conical spiral antenna, Tech. Rep. 22, Contract AF33(616)-3220, Urbana: Antenna
Laboratory, University of Illinois, 1957.
15. P. E. Mayes, Planar and other wide-angle logarithmic spirals
over ground, Electromagnetics, 14: 329362, 1994.
16. J. J. H. Wang and V. K. Tripp, Design of multioctave spiral-mode
microstrip antennas, IEEE Trans. Antennas Propag., AP-39: 332
335, 1991.
17. B. R-S. Cheo, V. H. Rumsey, and W. J. Welch, A solution of the
frequency-independent antenna problem, IRE Trans. Antennas
Propag., AP-9: 527534, 1961.
18. S. R. Laxpati and R. Mittra, Boundary-value problems associated
with source-excited, planar equiangular-spiral antennas, Proc.
IEE, 114: 352358, 1967.
19. D. W. Smith, Numerical and experimental analysis of circularly
polarized radiating line antennas, Ph.D. Thesis, Electromagnetics Laboratory, University of Illinois, Urbana, 1990.
20. N. J. Champagne II, J. T. Williams, and D. R. Wilton, Resistively
loaded printed spiral antennas, URSI Radio Science Meeting Digest, Chicago, 1992.
Reading List
E. C. Jordan, G. A. Deschamps, J. D. Dyson, and P. E. Mayes, Developments in broadband antennas, IEEE Spectrum, 1: 5871, 1964.
270
SPONTANEOUS EMISSION
V. H. Rumsey, Frequency-Independent Antennas, New York: Academic Press, 1966.
P. E. Mayes, Frequency-independent antennas. In Y. T. Lo and S. W.
Lee (eds.), Antenna Handbook, New York: VanNostrand Reinhold, 1988.
P. E. Mayes, Frequency-independent antennas and broadband derivatives thereof, Proc. IEEE, 80: 103112, 1992.
PAUL E. MAYES
University of Illinois at UrbanaChampaign
562
TELEVISION ANTENNAS
TELEVISION ANTENNAS
At the present time, the superturnstile antenna (of which the
batwing antenna is a radiating element) (1), the supergain
antenna (dipole antennas with a reflector plate), invented in
the United States, and the Vierergruppe antenna, invented
in Germany and sometimes called the two-dipole antenna in
Japan, are widely used for very high frequency television
(VHF-TV) broadcasting around the world. The superturnstile
antenna (2) is used in Japan, as well as in the United States.
Figure 1 shows the appearance of the original batwing antenna, when it was first made public by Masters (1). The secret lies in its complex shape. It is called a batwing antenna
in the United States and Schmetterlings Antenne (butterfly
antenna) in Germany, in view of this shape.
The characteristics of the batwing antenna were calculated
using the moment method proposed by Harrington (3) and
experiments were conducted on this antenna. The model antenna was approximately two-fifths the size of a full-scale batwing antenna, with its design center frequency at 500 MHz.
It has been reported that for thick cylindrical antennas, a
substantial effect on the current distribution appears due to
nonzero current on the flat end faces. The assumption of zero
current at the flat end face is appropriate for a thin cylindrical antenna; however, in the case of a thick cylindrical antenna, this assumption is not valid.
Specifically, this article applies the moment method to a
full-wave dipole antenna, with a reflector plate supported by
a metal bar, such as those widely used for TV and frequency
modulated (FM) broadcasting (4). In the present study, the
Figure 1. Historical shape of the batwing radiator.
analysis is made by including the flat end-face currents. As a
result, it is found that the calculated and measured values
agree well, and satisfactory wideband characteristics are obtained.
Next, the twin-loop antennas, most widely used for ultra
high frequency (UHF)-TV broadcasting, are considered. Previous researchers analyzed them by assuming a sinusoidal current distribution. Others adopted the higher order expansions
(Fourier series) of the current distribution; however, their
analyses did not sufficiently explain the wideband characteristics of this antenna.
This article applies the moment method to a twin-loop antenna with a reflector plate or a wire screen-type reflector
plate. As for the input impedance, 2-L-type twin-loop antennas have reactance near zero [in the case where l1 0.15 0,
that is, where the voltage standing-wave ratio (VSWR) is
nearly equal to unity]. Also, satisfactory wideband characteristics are obtained. The agreement between the measurement
and the theory is quite good. Thus it may become possible in
the future to improve practical antenna characteristics, based
on the results obtained. Several very popular television
(transmitting) antennas (5,6) were mentioned, such as slot
array, normal mode helix, and V-dipole antennas.
THE BATWING ANTENNA ELEMENT
The antenna is installed around a support mast, as shown in
Fig. 2, and fed from points f and f, through a jumper from a
branch cable with a characteristic impedance of 72 . The
conducting support mast is idealized by an infinite, thin mast.
The batwing antenna element is divided into 397 segments
for the original type, with triangular functions as the
weighting and expansion functions, and the analysis of the
batwing antenna elements carried out using the Galerkins
method. The batwing antenna is fed with unit voltage. The
currents flowing in each antenna conductor are calculated
over a frequency range of 300 to 700 MHz.
Figure 3 illustrates these current distributions Ii(i 1
12) on the conductors at frequencies of 300, 500, and 700
MHz. Since the distribution of currents along each conductor
is calculated, this allows calculation of the radiation characteristics. Figure 4 illustrates the amplitude and phase characteristics of radiation patterns in the horizontal and vertical
planes. It is seen from this figure that the theoretical values
agree well with the measurements.
Figure 5 illustrates the theoretical and measured input impedance of a batwing antenna mounted on an aluminum
plate, 3 m 3 m. Both curves coincide closely with each
other, with the input impedance having a value close to 72 ,
which is the proper match to the characteristic impedance of
the branch cable. Vernier impedance matching is carried out
in practice by connecting a metal jumper between the end of
the branch cable and the feed point of the antenna element
or the support mast. The feed straps length, width, or form
is varied to derive VSWR values below 1.10.
The power gain of the antenna at 500 MHz is calculated to
be 3.3 dB. Figure 6 shows the gain of the antenna in the
0 direction as a function of the frequency, referenced to a
half-wavelength dipole. Figure 7 illustrates three-dimensional amplitude characteristics of radiation patterns in the
horizontal and vertical planes at each frequency.
J. Webster (ed.), Wiley Encyclopedia of Electrical and Electronics Engineering. Copyright # 1999 John Wiley & Sons, Inc.
TELEVISION ANTENNAS
conductors. We assume the supporting bar to be separated
from the feed point by a distance l1. Also, the radius of the
supporting bar is fixed at a fourth of the radius of the antenna
element (i.e., at 0 /100), and then is varied to be 0.2 0, 0.25
0, and 0.3 0. Figures 9 to 12 indicate various calculated performance characteristics. Note that the leakage current of the
supporting bar is minimized for f/f 0 0.7, and that this current is substantial at other frequencies. The current distribution is shown only for the case of l1 0.25 0.
P( . )
E ( )
563
CHARACTERISTICS OF 2L TWIN-LOOP ANTENNAS WITH
INFINITE REFLECTOR
As shown in Fig. 13, a twin-loop antenna has the loops connected by a parallel line: The 2L-type, 4L-type, and 6L-type
E ( )
3.2
12.2
3.1
A
I8
H
I7
B
I6
G
37.8
I5
D
I4
I3
E
I11
I10
I2
I1
I9
300 MHz
(a)
Jumper
72 Branch cable
l
(Unit: cm)
Figure 2. Construction of the model batwing antenna and its coordinate system.
500 MHz
(b)
THEORETICAL ANALYSIS OF METAL-BAR-SUPPORTED
WIDEBAND FULL-WAVE DIPOLE ANTENNAS WITH A
REFLECTOR PLATE
Wideband full-wave dipole antennas with a reflector plate
supported by metal bars were invented in Germany. The construction is shown in Fig. 8. A full-wave dipole antenna is
located in front of a reflector, and supported directly by a
metal bar attached to a reflector. This antenna was also analyzed by the moment method described previously.
Because the supporting bar (see Fig. 8) is metallic, leakage
currents may cause degradation of the radiation characteristics. To calculate these effects, the radial component of field,
E must be taken into consideration. In other words, E is
needed for the calculation of Zm,n, as defined by inner products
of the expansion functions on the supporting bar and
weighting functions on the antenna element or on the parallel
700 MHz
(c)
Figure 3. Amplitude characteristics of current distribution for frequency range from 300, 500, 700 MHz of shaded areas.
564
TELEVISION ANTENNAS
1.0 0
1.0
1.0
E ( )
Amp.
0.5
E ( )
0
200
90
200
0
phase (deg.)
180
300 MHz
0.5
0.5
Horizontal pattern
700 MHz
500 MHz
(a)
(b)
(c)
1.0 0
1.0
1.0
Theoretical values
Measured values
(amplitude pattern)
Theoretical values
Measured values
(phase pattern)
E ( )
0.5
E ( )
0
200
90
0
200
phase (deg.)
180
300 MHz
0.5
0.5
Vertical pattern
700 MHz
500 MHz
(a)
(b)
(c)
Figure 4. Amplitude and phase characteristics of radiation patterns at 300, 500, 700 MHz (with
support mast).
Normalized impedance 72
Theoretical
value
Input resistance
100
4
Measured
value
50
Input reactance
0
300
400
500
3
Gain (dB)
Input impedance ()
150
600
700
Frequency (MHz)
50
100
Figure 5. Theoretical and measured values of the input impedance
as function of frequency (mast is infinite thin, 0).
0
300
400
500
Frequency (MHz)
600
Figure 6. Batwing antenna gain with /2 dipole.
700
TELEVISION ANTENNAS
565
Figure 14 shows l1 0.15 0 and l1 0.25 0 characteristics
of the radiation pattern in a polar display. The antenna gain
for both lengths (l1 0.15 0 and l1 0.25 0) shows a small
change of approximately 9.58.5 dB.
The input impedance has a value very close to 50 , essentially the same as the characteristic impedance of the feed
cable. As for the input impedance, the 2L twin-loop antenna
has reactance nearest zero (for the case where l1 0.15 0),
that is, the VSWR is nearly equal to unity.
In the above calculation, the reflector was considered to be
an infinite reflector, and the effect of the reflector on the antenna elements was treated by the image method. In the case
of practical antennas, however, it is the usual practice to
make the reflector finite, or consisting of several parallel conductors. Therefore, a calculation was executed for a reflector
in which 21 linear conductors replaced the infinite reflector,
as shown in Fig. 15.
The results are shown along with those for the infinitereflector case. Based on these results, it was concluded that
no significant difference was observed in input impedance and
gain between the infinite reflector case and the case where
the reflector consisted of parallel conductors.
The wire screen-type of reflector plate had a height of 3 0
(120 cm), a width of 0 (40 cm), and a wire interval of 0.15 0
(6 cm). The radiation pattern is shown in Fig. 16. With regard
to the pattern in the horizontal plane, no difference was found
in comparison with an infinite reflector, but a backlobe of approximately 16 dB exists to the rear of the reflector. The
same figure also shows the phase characteristics. With regard
to the pattern in the vertical plane, the phase shows a large
change where the pattern shows a cut.
SEVERAL POPULAR TRANSMITTING ANTENNAS
Slot Antennas
Both resonant and nonresonant end-fed arrays of slots are
used for TV broadcasting. The resonant arrays are restricted
2h
Figure 7. Three-dimension amplitude characteristics of radiation
patterns at 300, 500, 700 MHz.
2as
h/2
Metal-bar
supported
are used, according to the number of loops. For actual use, a
reactive load is provided by the trap at the top end, which
also serves as the antenna support. The dimensions used for
this article are as follows: center frequency f 0 750 MHz
(wavelength 0 40 cm), length of the parallel line part
2l1 0 /2 (l1 10 cm), 12 0 /2 (l2 20 cm), interval of
the parallel line part, d 0 /20 (d 2 cm), loop radius
b 0 /2 (b 6.366 cm), distance from the reflector to the
antenna l3 0 /4 (l3 10 cm), conductor diameter 10
mm, and top-end trap lt 0 to 0 /4, changed in intervals of
0 /16. 2L twin-loop antennas were arranged in front of an infinite reflector, and calculations were executed in regard to
the frequency characteristics of the trap length.
The radiation pattern of the type 2L antenna is shown in
Fig. 14. Up to lt 0 /8, the main beam gradually becomes
sharper with increasing frequency, and it can be seen that
the sidelobes increase. When lt increases in this way to 0 /8
and 0 /4, the directivity becomes disturbed.
lr
2a
Dipole
antenna
#5
#6
#4
#2
#0
Reflector #1
plate
=
=
=
=
=
=
=
=
#3
2as
(a)
a/ o
a/ o
s/ o
d/ o
d/ o
lr/ o
lt/ o
as/ o
0.04
0.01
dr
0.4
0.8
0.15
1.2
0.2 or 0.25 or 0.3
0.01
lt
2a
2ar
(b)
Figure 8. Metal-bar supported wideband full-wave dipole antennas
with a reflector plate.
TELEVISION ANTENNAS
Current l1/ 0 = 0.25
f/f0
0.6
0.6
#3
0.4
0.6
#4
0.4
0.2
#5
0.4
#6
0.35
0.3
0.2
0.2
0
180
0
180
0.6
#2
0.5
180
0
180
(mA/V)
0.5
#0
2
#1
0.6
0.4
0.4
0.4
0.2
0.2
0.2
0
180
0.6
180 3
0
180
0.6
#3
0.4
0
180
0.6
#4
0.4
0.2
0.1
0.7
0
180
0.35
0.3
0.2
0
180
0
180
0.6
#2
#1
0.6
#6
0.4
0.2
#0
180
#5
180
0.1
180
0.5
0.4
0.2
0
180
0.6
#3
0.4
0.4
0.2
0.2
0
180
0.2
2
#4
0.6
0.4
0
180
0.6
#5
0.4
0.4
0.2
0.2
#6
0.35
0.3
0.1
180
1.0
566
0
180
0
180
2
#0
0
180
2
0.6
#1
0.6
180
#2
0.5
0.4
0.4
0.4
0.2
0.2
0.2
0
180
0
180
0
180
Figure 9. Current distribution of metal-bar supported full-wave dipole antennas (two-bay) with
a wire screen-type reflector plate.
TELEVISION ANTENNAS
1.0
0.9
0.8
0.7
0.6
0.5
f/f0
l1/ 0 = 0.2
l1/ 0 = 0.25
l1/ 0 = 0.3
1.0
1.0
1.0
.5
.5
.5
180 0
0 180
180 0
0 180
180 0
0 180
180
180
180
1.0
1.0
1.0
.5
.5
.5
180 0
0 180
180 0
0 180
180 0
0 180
180
180
180
1.0
1.0
1.0
.5
.5
.5
180 0
0 180
180 0
0 180
180 0
0 180
180
180
180
1.0
1.0
1.0
.5
.5
.5
180 0
0 180
180 0
0 180
180 0
0 180
180
180
180
1.0
1.0
1.0
.5
.5
.5
180 0
0 180
180 0
0 180
180 0
0 180
180
180
180
1.0
1.0
1.0
.5
.5
.5
180 0
0 180
180 0
0 180
180
180
(
Amplitude
567
180 0
0 180
180
Phase)
Figure 10. Radiation pattern of metal-bar supported full-wave dipole antennas (two-bay) with
a wire screen-type reflector plate.
to UHF applications because of their limited bandwidth. The
traveling-wave slot antenna illustrated in Fig. 17 is a large
end-fed coaxial transmission line with a slotted outer conductor. The slots are arranged in pairs at each layer, with the
pairs separated by a quarter wavelength along the length of
the antenna. Adjacent pairs occupy planes at right angles to
each other. The slot pairs, which are approximately one-half
wavelength long, are fed out of phase by the coaxial line by
capacitive probes projecting radially inward from one side of
each slot so as to produce a figure-eight pattern. The probes
568
TELEVISION ANTENNAS
300
Re
200
lt
Zin ()
100
Im
Feed point large scale
Parallel line
0.6
0
0.7
f/f0
0.5
0.8
0.9
Loop
Trap
0.2
0.25
0.3
An infinite reflector
200
Figure 11. Input impedance characteristics of metal-bar supported
full-wave dipole antennas (two-bay) with a wire screen-type reflector
plate.
are placed on opposite sides of adjacent in-line slots which are
spaced one-half wavelength to provide in-phase excitation.
The quarter-wavelength separation of layers in conjunction
with the space-quadrature arrangement of successive layers
of slots effects a turnstile-type feed which produces a horizontally polarized azimuth pattern with a circularity of 1 dB
for VHF applications. An equal percentage of the power in the
coaxial line is fed to each layer of slots, which results in an
exponential aperture distribution that provides null fill.
Reflections from adjacent layers tend to cancel, which
allows the traveling-wave operation. The top slots are
strongly coupled to the line to reduce reflections. For highgain applications, one-half of the slots may be eliminated, resulting in a one-wavelength spacing of in-line slots. The
standing-wave antenna consists of layers or bays of one or
more axial slots spaced by one wavelength and fed by a coaxial line with the slotted pipe forming the outer conductor. Azi-
10
Gain (dB)
Figure 13. Structure of 2L-type twin-loop antenna and its coordinate
system for analysis.
muth patterns are controlled by the number of slots per bay.
One slot per bay produces a skull-shaped pattern, two slots a
peanut-shaped pattern, and three slots a trilobe pattern. Four
or more slots per bay are usually required for an ominidirectional pattern with a circularity of 1 dB. A typical omnidirectional pattern is shown in Fig. 18.
Helix Antennas
Figure 19 shows a single bay of a single-arm right-hand and
left-hand helix fed in phase at the center so that the vertically
polarized components of the two helices cancel in the broadside direction. The pitch angle is about 12 so that the vertically polarized radiation from each helix is about 10 dB down
from the horizontally polarized radiation, which produces
about 0.5 dB loss in gain due to cross-polarization radiation.
Since the beam of each helix scans about 2.7 per 1% change
in frequency, the bay length is limited to about six wavelengths. Sidefire helical antenna pattern is shown in Fig. 20.
V-Shaped Antennas
l1/ 0
0.2
0.25
0.3
4
0
2l1
l1/ 0
100
1.0
0.5
0.6
0.7
f/f0
0.8
0.9
1.0
Figure 12. Gain of metal-bar supported full-wave dipole antennas
(two-bay) with a wire screen-type reflector plate.
The multi-V antenna has been designed especially to mount
on the side of existing towers which are used for standard
broadcast radiators or TV supporting towers. The multi-V antenna array (Fig. 21) consists of a number of V elements
stacked vertically to provide power gain in the horizontal
plane and vertical directivity. The resulting horizontal radiation pattern is essentially omnidirectional. The pattern remains approximately circular when side-mounted on a tower
having a uniform cross section with 2 ft on a side.
The bays, or Vs, are so designed that the input impedance
of each pair is 50 , enabling them to be fed by standard
coaxial transmission lines. The feed system enables an even
number of bays from two to eight to be employed to obtain
power gains ranging from 1.6 to 7.3. The design is such that it
is necessary to tune the antenna at the factory for the desired
TELEVISION ANTENNAS
l1 = 0.15 0
0.4
600 MHz
l1 = 0.25 0
0.4
0.2
0.2 0.4 0.6 0.8
=
90
1.0
0.2
0.4
0.4
0.2 0.4 0.6 0.8
=
90
1.0
0.2 0.4 0.6 0.8
=
90
1.0
0.2 0.4 0.6 0.8
=
90
1.0
0.2 0.4 0.6 0.8
=
90
1.0
180
0
0.2
0
0.2 0.4 0.6 0.8
=
90
1.0
180
0
0.4
0.2 0.4 0.6 0.8
=
90
1.0
0.2
0
0.2
0.2
0.4
0.4
0.6
180
0
0.2
0.4
0.6
180
0
0.2
0
0.2 0.4 0.6 0.8
=
90
1.0
0.2
0.4
900 MHz
0.4
0.4
180
0
0.2
0.4
0.6
180
0
0.2
0.2
825 MHz
0.2 0.4 0.6 0.8
0.2
0.4
0.4
180
0
0.2
0.4
0.4
=
90
1.0
0.2
0.2
750 MHz
0.2
0.4
0.4
180
0
0.2
0.4
0.4
0.2
0.2
675 MHz
569
180
0.6
Theoretical
0.2 0.4 0.6 0.8
=
90
1.0
180
Measured
frequency. It is very simple, with one feed point per bay and
a maximum of four power-dividing elements for an eight-bay
array. The radiating elements are grounded for maximum
lightning protection. Figure 22 shows the four bay V antenna
pattern for various values of space.
CONCLUSION
Previous researchers (1,2) have analyzed batwing antennas
by approximating the current distribution as a sinusoidal distribution. Wideband characteristics are not obtained with a
sinusoidal current distribution. In this article, various types
of modified batwing antennas, as the central form of the superturnstile antenna system, were analyzed theoretically
with the aid of the moment method. The results were compared with measurements, in order to examine the performance of the antenna elements in detail.
Figure 14. Vertical radiation pattern of
2L-type twin-loop antenna.
It is also evident from this research that the shape of the
jumper has a remarkable effect on the reactance of the input
impedance, and that the distance between the support mast
and the antenna element also markedly influences the resistance of this impedance. Thus, a satisfactory explanation is
given with regard to the matching conditions. As a result, it
was found that the calculated and measured values agree
well, and satisfactory wideband characteristics are obtained.
Next an analytic method and calculated results for the performance characteristics of a thick cylindrical antenna were
presented. The analysis used the moment method and takes
the end face currents into account. The calculated results
were compared with measured values, demonstrating the accuracy of the analytic method.
Using this method, a full-wave dipole antenna with a reflector supported by a metal bar was analyzed. The input impedance was measured for particular cases, thus obtaining
the antenna dimensions for which the antenna input imped-
570
TELEVISION ANTENNAS
600 MHz
= 90
0
0.4
#1
#2
#3
#4
#5
0.15 0
0.8
= 90
1.0
0.4
0.8
1.0
675 MHz
= 90
= 90
S
0.4
0.8
1.0
0.8
0.4
1.0
3 0
l2
750 MHz
l1
= 90
0.4
0.8
= 90
1.0
0.4
0.8
1.0
825 MHz
#17
#18
#19
#20
#21
= 90
0.4
0.8
= 90
1.0
10
0.4
0.8
1.0
900 MHz
Figure 15. Structure of 2L-type twin-loop antenna with a wire
screen-type reflector plate.
= 90
0.4
Slot
Slot
0.8
= 90
1.0
0.4
0.8
1.0
Wire screen type reflector
An infinite reflector
(a)
(b)
Figure 16. Comparison between characteristics of 2L-type twin-loop
antenna with a wire screen-type reflector and with an infinite reflector.
Probe
0
30
330
Probe position
60
300
Slot
Probe
Relative field
20 40 60 80
270
90
Excitation of slot
(b)
240
120
Construction
(a)
Figure 17. The traveling-wave slot antenna.
210
150
180
Figure 18. A typical omnidirectional pattern.
TELEVISION ANTENNAS
571
0.235
A triangular
prism
0.340
Diameter
/100
Pitch
p = /2
Figure 21. The construction of V-shaped antenna.
Feed point
A cylindrical conductor
Short
(a)
(b)
Figure 19. A sidefire helical antenna.
Freq.
Pattern
470 MHz
480 MHz
Horizontal
pattern
180
90
0 180
270
90
Vertical
pattern
2nd mode
90
180
270
90
270
Horizontal
pattern
Vertical
pattern
3rd mode
0 180
270
90
180
0 180
0 180
270
270
90
0 180
270
90
0 180
270
730 MHz
90
0 180
0 180
270
0 180
270
270
90
270
90
720 MHz
90
270
90
0 180
0 180
710 MHz
90
90
270
90
270
700 MHz
90
0 180
0 180
510 MHz
90
270
90
270
690 MHz
500 MHz
90
0 180
0 180
180
490 MHz
ance permits broadband operation. In conclusion, wideband
characteristics are not obtained with a one-bay antenna. The
wideband characteristic is obtained by means of the mutual
impedance of the two-bay arrangement. In the frequency region of f/f 0 0.7, the resistance of the input impedance is
considered to be constant. In this case, the leakage current to
the support bar is small. With regard to the radiation pattern,
it was seen that a degradation of characteristics was caused
by the metal support bar.
It is noted that the present method should be similarly
useful for analyzing antennas of other forms where the end
face effect is not negligible.
Next, the twin-loop antennas were considered for use as
wideband antennas. The analysis results for 2L type showed
270
90
0 180
270
270
Figure 20. The horizontal and vertical
pattern of sidefire helical antenna.
572
TELEVISION BROADCAST TRANSMISSION STANDARDS
9 10
reported here that a rigorous theoretical analysis has been
achieved almost 30 years after the invention of these VHF
UHF antennas.
0
/
0=180
120
20
10
BIBLIOGRAPHY
1. R. W. Masters, The super turnstile. Broadcast News, 42: 1946.
30
2. Y. Mushiake, (ed.), Antenna Engineering Handbook. The OHMSha, Ltd., 1980 (in Japanese).
3. R. F. Harrington, Field Computation By Moment Method, New
York: Macmillan, 1968.
4. W. Berndt, Kombinierte Sendeantennen Fur Fernseh-und UKWrunfunk (Teil II), Telefunken-Zeitung, Jahrgang, Heft. 26, 101:
268, 1953.
60
90
5. Y. T. Lo and S. W. Lee, Antenna Handbook, New York: Van Nostrand Reinhold, 1988.
0
0.5
270
90
0.5
6. R. C. Johnson and H. Jasik, Antenna Engineering Handbook,
New York: McGraw Hill, 1984.
HARUO KAWAKAMI
Antenna Giken Corporation
0.5
180
0
0.75
270
90
0.75
0.75
180
0
1.0
270
90
1.0
1.0
180
Figure 22. Vertical and horizontal pattern of V-shaped antenna.
that the change in the characteristics with a change in frequency becomes more severe with increasing trap length lt,
and the bandwidth becomes small, while a short trap length
lt shows a small change and a tendency for the bandwidth to
become wide. For lt 0, a wide bandwidth for pattern and
gain was obtained for the 2L type. The input impedance has
a value very close to 50 , essentially the same as the characteristic impedance of the feed cable over a very wide frequency range. Thus, a satisfactory explanation was given
with regard to the matching conditions. Popular television antennas cover the properties of many basic types of antennas
which are the mainstream of antenna technology. It has been
TELEVISION BROADCASTING BY DIRECT SATELLITE. See DIRECT SATELLITE TELEVISION BROADCASTING.
492
WAVEGUIDE ANTENNAS
Figure 1. Waveguide slot array for an airborne radar application.
(Courtesy of Raytheon Systems Co.)
WAVEGUIDE ANTENNAS
Waveguide antennas are useful for a variety of airborne,
ground based, space, and missile applications in the microwave frequency band because they offer high radiating efficiency, a high degree of control over the radiation pattern,
compact architecture, and low weight. The high radiating efficiency and compact architecture, namely, the small depth
compared to a parabolic reflector antenna with equivalent
gain, combine to optimize the use of the available aperture
area. Accurate control of the radiation pattern allows designing arrays with a radiation pattern envelope tailored to the
specific application. The low weight minimizes the impact on
the installation and allows mounting the antenna on a mechanical gimbal assembly to scan the radiation pattern. A
waveguide slot array designed to mount on a gimbal for an
airborne radar application is shown in Fig. 1.
In waveguide slot arrays the radiating and transmission
structures are inexorably intertwined, thus it is appropriate
to treat the history of the waveguides together with that of
slot radiators. In the 1930s, G. C. Southworth (1) at Bell Telephone Laboratories investigated electromagnetic propagation
in hollow waveguides. Carson, Mead, and Schelkunoff performed the accompanying theoretical analysis (2). Independent work by Barrow of MIT (3) was published at about the
same time (4). The analysis of waveguides and slot radiators
in terms of transmission lines is also due to Schelkunoff (5).
Another aspect of array work is the need for pattern synthesis. S. A. Schelkunoff s classic 1943 paper (6) and the subsequent C. L. Dolph (7) paper on the synthesis of equal sidelobe
levels were used by T. T. Taylor (8) to achieve practical array
designs. The major work on waveguide slot arrays during
World War II was carried out in Canada at McGill University
by W. H. Watson and his colleagues (9). The basis of their
work included A. F. Stevensons (10) analysis of the behavior
of radiating slots in waveguides. Significant advances were
made at the MIT Radiation Lab and are summarized in S.
Silvers book (11). Further work was carried out by members
of this group and others at Hughes Aircraft Co. in Culver
City, CA (12). A. A. Oliner (13,14) and others refined Stevensons theory at Polytechnic Institute of Brooklyn. R. S. Elliott
and his students at UCLA, and numerous other researchers
worked to further refine the design process.
The rectangular waveguide is the guide geometry used for
most waveguide slot arrays because it provides a compact,
low-loss transmission line which contributes to the high efficiencies achievable with these arrays. The basic radiating element, a single rectangular slot located in an infinite ground
plane, is the complement to a dipole radiator (15) and, like
the dipole, produces a broad radiation pattern. The radiating
slot in the rectangular waveguide is realized in a variety of
configurations, thus providing flexibility for the designer. The
slot radiation characteristics are controlled by adjusting the
slot location on the waveguide and the slot length and width.
An array of slots with a specified aperture distribution
generates a highly directive or specially shaped radiation pattern. Arrays of slots are created by placing slot radiators in a
waveguide to create a linear array and by joining a number
of these waveguides with a power-divider network to create a
planar array. Slots are used to couple energy between waveguides, and these coupling slots provide a compact way to realize the power-divider network. The closed-feed network
eliminates the spillover loss suffered in horn-fed reflectors
and other space-fed antennas. The aperture distribution is determined by the radiating slot characteristics and the power
dividers in the feed network. The designer adjusts these pa-
J. Webster (ed.), Wiley Encyclopedia of Electrical and Electronics Engineering. Copyright # 1999 John Wiley & Sons, Inc.
WAVEGUIDE ANTENNAS
rameters to achieve a high degree of control of the aperture
distribution and thereby achieves a low-sidelobe or shapedbeam radiation pattern. The design process for these arrays
is complex because of the need to accurately characterize the
relationship between the slot geometry and performance and
also because of interactions between the various elements of
the array. The characteristics of the individual elements are
derived from the interaction of the slot with the modes in the
waveguide, and a simple lumped-circuit model is used to approximate the slot characteristics. The initial array design approach uses these lumped circuit elements to approximate
array performance. The lumped-circuit model neglects interactions between the elements created by radiated and internal electromagnetic coupling. Refinement of the design to
compensate for the interactions is done with experimental
hardware or sophisticated electromagnetic simulation software.
WAVEGUIDE RADIATING SLOTS
A rectangular slot cut into a metallic ground plane radiates
when an electric field is excited across the slot. The electric
field distribution in the slot must meet the boundary condition that the tangential electric field at the slot boundary is
zero. The lowest mode that meets the boundary conditions
has a constant electric field directed across the width of the
slot with a cosinusoidal distribution along the length of the
slot:
E s (x, z) = x
Vs
cos(z/l) for w/2 < x < w/2, l/2 < z < l/2
w
(1)
where Vs is the peak voltage across the slot. The far-field radiation pattern of the slot located in an infinite ground plane is
calculated by using Babinets principle (15) (or image theory).
For a slot located in the xz plane, this results in a radiation
pattern given by
e jkR
R
jV s k sin cos(kl/2 cos )
E (, ) =
l
(/l)2 k2 cos2
E = E (, )
(2)
If the slot size is adjusted to the lowest resonance, the loss
due to reactive energy is minimized, and the slot coupling is
enhanced. At resonance, when the slot is assumed to be equal
to one-half the free space wavelength, the radiation pattern
becomes
E (, ) =
jV s cos(/2 cos )
sin
(3)
The directivity for a radiator measures the focusing effect of
the radiator and is defined as the directivity relative to that
for an isotropic radiator:
D(, ) =
2
0
4P(, )
P(, ) sin d d
(4)
and the power is given by
P(, ) =
1
2
r
|E(, )|2
493
(5)
The gain of the element is given by the directivity minus any
losses in the antenna. The directivity and gain are typically
quoted in decibels. For a resonant slot, assuming that radiation occurs only above the ground plane, the peak directivity
is 5.6 dB above that of an isotropic radiator.
Rectangular Waveguide Modes
The field in a waveguide slot is created by a displacement
current caused by disruption of the currents generated on the
waveguide walls by the energy in the waveguide modes. A
metal-walled rectangular waveguide has a set of waveguide
modes that satisfy the boundary conditions imposed on the
fields by the metallic walls. These modes are classified into
two types: the transverse electric (TE) which has only a magnetic field component in the direction of propagation and the
transverse magnetic (TM) which has an electric field component only in the direction of propagation. The complete sets
of field components for these modes for a lossless waveguide
are as follows (16): for the TEmn modes,
Hz = cos
mx
a
cos
ny
b
e mn z
mx ny
mn
mmn
Ey = 2
sin
cos
e mn z
j
mn a
a
b
mx ny
mn
nmn
Ex = 2
cos
Hy =
sin
e mn z
j
mn b
a
b
m 2 n 2
2 =
+
a
b
Hx =
(6)
and for the TMmn modes,
mx
ny
e mn z
b
mx ny
mn
mmn
Hy = 2
cos
Ex =
sin
e mn z (7)
j
mn a
a
b
mx ny
mn
nmn
Hx = 2
sin
Ey =
cos
e mn z
j
mn b
a
b
Ez = sin
sin
The waveguide mode propagates unattenuated, except for
ohmic losses, for frequencies where the propagation constant
mn =
r m
n 2
b
k2
(8)
is imaginary. The wavelength at which this first occurs, the
cutoff wavelength, is given by
= 2/
(m/a)2 + (n/b)2
(9)
For almost all applications it is desirable to operate with only
a single propagating mode. The mode with the lowest cutoff
frequency, the TE10 mode, is used. The guide dimensions are
chosen to allow only this mode to propagate over the desired
frequency range. The ohmic losses increase as the frequency
approaches the cutoff frequency, therefore it is generally de-
494
WAVEGUIDE ANTENNAS
sirable to operate at least 20% above the waveguide cutoff
frequency. It is also preferable to operate a similar amount
below the onset of the next propagating mode. The waveguide
fields for the dominant TE10 mode are given by
Hz = j cos
x
a
e j 10 z
x
10
sin
e j 10 z
/a
a
x
0
sin
Ey =
e j 10 z
/a
a
Hx =
(10)
Ys
where the upper sign denotes a wave traveling in the zdirection and the lower sign denotes a wave traveling in the
z-direction. The propagation constant for this mode is given
by
10 = j10 = j
k2
2
a
(11)
and the wavelength in the guide is given by
g =
(12)
2a
The relationship J 1n H gives the electric currents flowing on the broadwall surface of the waveguide for this mode
as
Jx = j cos
Jz =
x
a
10
sin
/a
e j 10 z
x
a
e j 10 z
(13)
Waveguide Slot Radiation Mechanism
If slots are cut into the waveguide in locations where they
interrupt the current flow, shown in Fig. 2, a displacement
current is set up across the slot and radiation occurs. The
amount of radiation is determined by the amount of current
intercepted and is controlled by adjusting the slot position on
the waveguide and its length and width. A variety of locations
create slots with significant amounts of radiation. Slots are
also made to radiate by introducing discontinuities into the
waveguide, such as probes, wires, posts, and irises.
Figure 3. Waveguide radiating slots and their lumped-element circuit models: offset longitudinal shunt slot, and angled and transverse
series slot.
The slot scatters energy in the waveguide in addition to
radiating energy. The slot acts as an obstacle in the equivalent transmission line, and it is modeled as a lumped-circuit
element in a two-wire transmission line (15). The lumped-circuit model has a resistive component representing the power
lost to the radiated fields and a reactive component representing the stored energy. The backward and forward scattered
waves for a slot in the waveguide wall are calculated from
(10)
1
B10 =
( jEt Kt + Ez Kz )e j 10 z dS
2Sa Slot
(14)
1
A10 =
( jEt Kt Ez Kz )e+ j 10 z dS
2Sa Slot
where Sa is twice the Poynting energy flux for the dominant
mode, E is the electric field in the slot, and K is the surface
current density on the waveguide wall if the slot were not
present. The scattering characteristics are used to compute
the equivalent circuit characteristics and their dependence on
the slot location and geometry. The slots are characterized by
their location and orientation in the waveguide and by their
equivalent circuit characteristics. Several slot geometries and
their equivalent circuit models are shown in Fig. 3.
Waveguide Radiating-Slot Geometries
Offset Longitudinal Shunt Slot. A narrow slot located along
the waveguide broadwall centerline does not radiate a significant amount of energy because it causes minimal disruption to the current flow on the waveguide walls. As the slot is
offset from the waveguide centerline, it intercepts more of the
waveguide currents, and the amount of power radiated becomes significant. The phase of the radiated field is reversed
if the slot is located on the opposite side of the waveguide
centerline. For an offset slot, it is a reasonable approximation
to assume that the fields in the slot are symmetrical, and it
can be shown that the scattering from the slot in the waveguide is also symmetrical. This symmetrical scattering corresponds to a shunt obstacle for the lumped-circuit model. The
scattering for the slot is given by (17).
B10 = A10 =
Figure 2. Rectangular waveguide current distribution for the TE10
mode.
Zs
j2V s (/a)2 (/l)
10 l
cos
2
2
o ab10 [(/l) 10 ]
2
sin
x
a
(15)
This assumes a narrow offset slot with only a Ex component
that has the cosinusoidal distribution given in Eq. (1). This
WAVEGUIDE ANTENNAS
Vs
Y
= K1 f (x, l)
G0
V
2
kG0 10 ab
(16)
and
f (x, l) =
(/l) cos(10 l/2)
sin(x/a)
2
(/l)2 10
where x is the slot offset from the centerline and G0 is the
characteristic conductance of the guide. At resonance, when
the slot length is one-half a free-space wavelength, the scattering coefficient is given by
B10 =
j2V
cos
0 (10 /k)ab
10
2k
sin
x
a
Offset=0.5 cm
0.6
Length=1.6 cm
Length =
1.5 cm
0.5
Offset=0.38 cm
0.4
0.3
Offset=0.25 cm
0.2
0.1
0
8.25
Offset=
0.13 cm
8.50
8.75
9.00
9.25
9.50
9.75
Slot resonant frequency (GHz)
Figure 4. Offset shunt-slot resonant frequency and conductance for
various slot lengths and offsets.
(17)
and the slot is represented by a shunt conductance. The conductance value is a function of the slot offset and the waveguide dimensions. The dependence of the conductance on the
slot offset is given by (10,18)
G
x
= g = g1 sin2
G0
a
g a
g1 = 2.09
cos2
b
2g
Length=1.7 cm
0.7
(18)
These equations are useful for understanding the slot characteristics but are not accurate enough for many design applications. The amount of radiation from the slot is controlled primarily by the offset and length as expressed previously but
also depends on the slot width and the thickness of the wall
in which the slot is cut. The slot resonant frequency must also
be determined accurately. The slot is defined to be resonant
when the phase of the scattered field is 180 out of phase with
the incident wave. The resonant frequency occurs when the
slot length is approximately one-half of a free-space wavelength, but the exact frequency is also a function of the slot
offset, width, and thickness. The characteristics of the slot
are determined more accurately by using an electromagnetic
simulation technique, such as the method of moments (19), or
by careful measurements. Figure 4 shows the variation of the
slot conductance and resonant frequency as the offset and
length are varied. The slot characteristics were calculated by
using a computer simulation based on the method of moments
formulation described in Khac (20), with slot thickness effects
were added. The variation of the slot admittance normalized
to the peak conductance values versus frequency relative to
the resonant frequency for a slot in full height (2.3 cm 1.0
cm) X-band (8 GHz to 12 GHz) waveguide is shown in Fig. 5.
Away from the resonant frequency, the slot becomes reactive,
and the radiation is reduced. The variation with frequency is
relatively insensitive to the slot offset and length, so the
curves are applicable over a wide range of offset and length
values. If the slot offset is large or the waveguide height is
reduced considerably, b a/4, the fields in the slot are less
symmetrical, and the slot behaves less like a pure shunt element (21). For these cases a more complicated circuit model is
required, or the scattering matrix description should be used
instead of a lumped-circuit model.
Broadwall Series Slots. A slot cut transversely across the
waveguide broadwall radiates when centered in the waveguide, and the amount of radiation is controlled by adjusting
the offset. For this element the forward scattered field is opposite in sign to that of the backscattered field, so this slot is
0.8
1
0.9
g/g0
b/g0
0.8
0.7
0.6
0.4
0.6
0.2
0.5
0.4
0.3
0.2
Normalized susceptance b/g0
2
K1 =
ja
Normalized conductance g/g0
where
0.8
Normalized slot conductance at resonance
can be related to the modes in a two-wire transmission line
to calculate the conductance of the lumped circuit element:
495
0.2
0.1
0.4
0
0.90 0.92 0.94 0.96 0.98 1.00 1.02 1.04 1.06 1.08 1.10
Normalized frequency f/f0
Figure 5. Variation of normalized conductance and admittance for
an offset-shunt slot relative to the resonant frequency.
496
WAVEGUIDE ANTENNAS
modeled as a series element. The normalized series resistance
for this element is given by
x
R
1
= r = r1 cos2
R0
2
r1 = 0.523
3
2
cos2
ab
4a
(19)
The maximum coupling for this element occurs when the slot
is centered. This slot has a high coupling value, but the range
of coupling attainable is limited because the slot length restricts the range of offsets, and the slot is confined to the
broadwall.
A slot located along the waveguide centerline also radiates
when rotated at an angle relative to the centerline. The forward scattering from this element is opposite in phase to that
of the backscattered field, and hence it is modeled as a series
lumped element. The amount of power coupled is controlled
primarily by the slot angle and is adjustable over a wide
range of coupling values.
2
R
= r = 0.131
R0
cos
I( )
2
=
1 2
J( )
=
cos
g
I( ) sin +
ab
cos
2
1 2
g
J( ) cos
2a
2
(20)
sin
2a
The resonant frequency for this slot is relatively insensitive
to the slot angle. The polarization characteristics of the slot
rotate with the slot orientation and generate a cross-polarization component which limits its usefulness in systems where
polarization purity is required.
Narrow-Wall and End-Wall Slots. Transverse angle slots located in the narrow wall of the waveguide are also used as
radiating elements. The height of the narrow wall is typically
less than the half free-space wavelength required for a resonant slot. So to achieve resonance, the ends of the slot are
extended into the broadwall. The coupling is controlled by the
slot angle and the depth to which the slot extends into the
broadwall. This slot exhibits a cross-polarization component
that varies with the slot angle. This geometry is difficult to
model accurately.
Slots located in the shorted end of a waveguide are used
for phased-array applications. The slot radiation is created by
the disruption of the currents in the terminating short of the
waveguide. For most phased-array applications, the slot characteristics are chosen to provide the best match in the array
environment, and the power to the slot is controlled by the
feed network. Open-ended waveguides are frequently used for
scanned-array applications.
linear radiating slot arrays, the coupling slots provide the
building blocks for the planar slot array. The slot configurations described for the radiating elements are also applied as
coupling elements. Coupling between the broadwalls is useful
to create arrays with minimal depth, and coupling is achieved
with the waveguides parallel or perpendicular to each other.
The coupling slots can be modeled by using lumped-circuit
elements. For these elements a transformer between two sets
of two-wire lines models the coupling characteristics, and a
reactive element models the stored energy in the slot. The
slot-coupling ratio determines the turns ratio for the transformer model. The slot is described in terms of its lumpedelement characteristics in each of the lines. The element may
be a seriesseries, shuntshunt, or shuntseries element depending on its orientation in the two waveguides.
Coupling-Slot Characterization
The characteristics of these elements are determined by a
method similar to that for the radiating elements. The forward and backward scattering components are the same as
those for the radiating slot given in Eq. (14). Now they are
applied in both the input and coupled guides to give the backscattered, forward-scattered and coupled fields. The coupling
coefficient between the waveguides is a function of the slot
characteristics in the two waveguides and the characteristic
impedances of the two guides. The power balance requirement is applied to the junction to show that the coupling coefficient can be calculated from the ratio of the scattered fields
and characteristic impedances in the two guides. The scattered-field terms are the same as those derived for the radiating slot except for the value of the coefficient of the field in
the slot. For the coupling slot, the ratio of the scattered fields
in the two guides is independent of the value of the coefficient,
so it need not be determined. The coupling ratio for the slot
is given by (22)
M=
(21)
where Y1 and Y2 are the characteristic impedances in the input and coupling guides, respectively, and B1 and B2 are the
backscattered fields in the two guides. The value of 1/ M
gives the turns ratio for the transformer in the lumped-circuit model.
Coupling-Slot Geometries
The seriesseries angle slot is a convenient element for
broadwall coupling between transverse waveguides because
the coupling is controlled by adjusting the slot angle without
changing the relative locations of the waveguides and a wide
range of coupling values is attainable. This slot and its equivalent circuit model are shown in Fig. 6. The coupling value
for this element at resonance, assuming a narrow rectangular
slot with zero thickness, is given by
WAVEGUIDE COUPLING SLOTS
In addition to their function as radiating elements, slots are
also used to couple energy between waveguides. The electric
field excited in the slot by the incident mode at the input port
excites fields in the coupled waveguide. In conjunction with
a1 b1Y1 |B1 |2
a2 b2Y2 |B2 |2
M=
a2 b 2 g 2
a b g
1 1
I1 ( ) sin +
I2 ( ) sin +
g 1
2a1
g 2
2a2
J1 ( ) cos
J2 ( ) cos
(22)
WAVEGUIDE ANTENNAS
497
tern for a linear array of N elements located along the z-axis
is given by
S(, ) = A(, )
N
Vn e jkz n cos
(24)
n=1
(a)
1
..
M1/2
Bs
(b)
Figure 6. Seriesseries angle slot coupling element and its lumpedelement circuit model.
where the expressions in the equation are the same as those
defined in Eq. (20) for the angled radiating slot. The transverse seriesseries slot, which is the limiting case for the
angled seriesseries slot, provides large coupling values between the guides.
The offset shuntseries coupling slot provides coupling between transverse waveguides. The slot is oriented as an offset
shunt slot in one waveguide and a transverse series slot in
the other. The coupling values are adjusted by changing the
slot offset which is equivalent to moving the location of the
shunt waveguide relative to the slot. The coupling value for
this slot at resonance is given by
M=
1 a2 b 2 g 1 g 2
4
a31 b1
1 (l/a2 )2
1 (2l/g 1 )2
2
cos2 (1 l/2) sin2 (x1 /a1 )
cos2 (l/2a2 ) cos2 (x2 /a2 )
(23)
where waveguide 1 contains the shunt-slot orientation.
These approximate values for the slot coupling depend on
the accuracy of the assumption made for the field in the slot
and neglect effects, such as slot thickness and the reactive
component of the slot, as the frequency departs form the resonant frequency. More accurate models for the slot are derived
by the method of moments (23). Computer simulation tools
based on the finite-element method (24) are also suitable for
modeling coupling slots because they operate in a closed
structure. This differs from the radiating slot case which does
not lend itself well to this type of model because the problem
is unbounded.
LINEAR SLOT ARRAYS
A linear array of slots serves alone as an antenna or as a
building block for a planar array. The far-field radiation pat-
The A(, ) term is the radiation pattern of a single radiating
element, which is assumed to be the same for all elements in
the array. The element pattern is similar to that given in Eq.
(2), but it is modified by the array environment. The summation term represents the array factor where Vn are the excitation values and zn are the element locations. The excitation
values of the elements are chosen to generate a specified
array radiation pattern. Slot-radiating elements allow controlling the excitations to achieve the desired pattern characteristics.
Linear arrays are divided into two categories: the standing-wave (or resonant array) and the traveling-wave array (or
nonresonant array). The antenna and system performance requirements dictate the choice between the two configurations.
In the standing-wave array, the slot locations are chosen to
maximize the coupling at the design frequency, and the waveguide is terminated with a short circuit. This creates an efficient antenna, but the bandwidth is limited. In the travelingwave array, the slot spacings are nonresonant, and the
waveguide is terminated with an absorptive load. The traveling-wave array operates over a larger frequency bandwidth
than the standing-wave array, but the power absorbed by the
load reduces the radiating efficiency. It is difficult to generate
a beam perpendicular to the guide with the traveling-wave
array, and it also exhibits beam scan with frequency.
Standing-Wave Arrays
For the resonant array, standing-wave fields are created in
the waveguide by terminating the waveguide with a metal
wall to create a short circuit. The slot locations relative to
the short circuit are chosen to maximize the array radiation
coupling. For series elements, such as the angled series slot,
the slots are located multiples of a half guide wavelength
from the short circuit. For shunt elements, such as the offsetshunt slot, the maximum coupling locations are multiples of
a half guide wavelength from an open circuit which is
achieved by locating a short circuit at one-quarter guide
wavelength from the nearest slot. For broadside operation the
slots are located in the guide to achieve a uniform phase distribution. A uniform phase distribution is achieved by locating the slots at one guide-wavelength intervals. However this
results in an array spacing greater than one free-space wavelength which creates multiple main beams and reduces the
antenna gain and performance. The spacing can be reduced
to one-half guide-wavelength intervals by alternating the slot
offsets to compensate for the 180 phase reversal in the guide.
Circuit Model for Resonant Linear Array. A circuit model of
the array provides a good starting point for predicting the
array performance and for determining the slot dimensions.
The array is modeled by using the lumped-circuit elements
for the individual slots, transmission line sections representing the waveguide sections, and a short circuit termination.
An offset-shunt-slot linear array and its circuit model are
498
WAVEGUIDE ANTENNAS
Y1
To achieve an input match for an end-fed standing-wave
array, the slot parameters should be chosen so that the slot
conductances or impedances sum to one. The slot conductances must also be selected to obtain the amplitude distribution required to achieve the desired radiation pattern. If the
slots are all assumed to be resonant and the desired distribution is all equiphase, the ratios of the slot conductances are
proportional to the relative slot radiated powers. This gives a
set of N equations that can be solved to give the required slot
conductances for a shunt-slot array:
Yn
N
Gi
P rn
G
= 1
G0
G0 n=1 P r1
d
Figure 7. Waveguide linear array of offset longitudinal shunt-slot
radiators and its transmission line circuit model.
shown in Fig. 7. The transmission matrix is defined as (25)
Vi
A
=
C
Ii
B Vi+1
D Ii+1
(25)
A section of transmission line of length d with a propagation
constant and a characteristic impedance Z0 has a transmission matrix
cos(d)
j
sin(d)
Z0
jZ0 sin(d)
cos(d)
(26)
The transmission matrices for the shunt element with admittance Y and series element with impedance Z are
1
Y
0
1
and
1
0
Z
1
(27)
To achieve a satisfactory design, the coupling values for the
individual slots must be chosen so that the reflection coefficient at the array input achieves the required value and so
that the relative coupling from the slots generates the desired
radiated phase and amplitude distribution for the array. The
input match to the array and its variation with frequency is
modeled by cascading the transmission matrices for the slots
and the waveguide sections. At the array resonant frequency
the normalized input conductance or resistance for an array
of N shunt or series slots reduces to
Array Mutual-Coupling Effects. The simple circuit model design is useful for understanding the slot array or for determining a set of initial design values, but it is not accurate
enough for most applications. The assumption was made that
the slots in the array have the characteristics of an isolated
slot. In the array environment the slots are affected by the
presence of the other slots, and slots at the edge of the array
exhibit behavior that differs significantly from that of slots in
the center of the array. If neglected, these effects result in
an array where the input reflection coefficient and radiation
pattern depart significantly from the desired values. The
characteristics of the isolated slot were determined by using a
single source of excitation for the slot: the incident waveguide
mode. Multiple sources in the array environment excite the
slot: the incident field, the fields scattered in the waveguide
by the other slots in the guide, and the radiated fields from
all the other slots. The characteristics for each slot are influenced by all the other slots in the array to some extent, so
a slots configuration cannot be determined independently. A
series of simultaneous equations is required to perform the
array design. Elliott (17) has developed a series of design
equations that account for the internal and external mutual
coupling effects:
Ypa /G0
f p Vps Vn
f n Vns Vp
(30)
2 f n2
Yna
=
N
G0
2 f n2
Vms
+ j(10 /k)(k0b)(a/)3
gmn
Y /G0
Vs
m=1 n
or
(28)
f n (xn , ln ) =
xn
(/l) cos(10 ln )
sin
2
2
(/ln ) 10
a
gmn (xn , ln , xm , lm ) =
The input reflection coefficient is given by
1 Yi /Y0
Z /Z 1
i =
= i 0
1 + Yi /Y0
Zi /Z0 + 1
The offset-shunt slot is a commonly used element for linear
arrays. It offers a wide range of conductance values and linear
polarization. The slot offsets and length can be computed using the equations for the offset shunt slot given earlier, or
interpolated from a graph, such as Fig. 4.
Yna /G0
N
Yi
Yn
=
Y0
Y
n=1 0
N
Zi
Zn
=
Z0
Z
n=1 0
Gn
Gm P rn
=
G0
G0 P rm
(29)
l m /2
l m /2
cos(m /lm )
cos(n /ln )
l n /2
l n /2
jk R
0
2 e
dn m (31)
+
k
0
2
n
R
2
WAVEGUIDE ANTENNAS
499
The Yna /G0 are the active slot admittances in the array environment, the Vsn, Vn are the slot and the waveguide mode voltages, and the gmn terms represent the external mutual-coupling effects between the slots. The set of equations is applied
iteratively along with the equation for the required input impedance to determine the slot offsets and lengths. These equations assume an ideal cosinusoidal distribution in a narrow
slot, and may not be accurate enough for some applications.
A more thorough analysis, where the field distribution in the
slots are unknowns which must be determined, provides a
more accurate model of the impact of element mutual coupling on array performance (26). The computational requirements for a rigorous analysis may limit the size of the array
for which it is applicable. A good approximation to the final
design should be found first by using an approximate method
to reduce the number of iterations required.
because the summation of the individual reflections creates a
large reflection. The traveling wave generates a linear phase
slope across the aperture which corresponds to a scanned
beam with the beam pointing direction given by (27)
Variation of Performance with Frequency. The design
achieves the desired input match at the design frequency, but
the input match of the linear array degrades as the frequency
moves away from resonance. This effect is modeled using the
transmission matrices by adjusting the propagation constant
in the transmission line sections and including the variation
of the slot admittance with frequency. The bandwidth is generally defined as the frequency range over which the input
reflection coefficient remains below a specified level. The
bandwidth is determined primarily by the number of elements in the array, and it degrades as the number of elements in the array increases. Watson (9) derived an approximate expression for the variation of the input reflection
coefficient which neglects the multiple reflections:
Circuit Model for Nonresonant Arrays. The circuit model for
the traveling-wave array is similar to that used for the resonant array, but a matched-load termination replaces the terminating short, and the line lengths are adjusted to reflect
the nonresonant spacing. The initial design is simplified by
assuming that the scattering from each slot is small, and
hence multiple reflections are neglected. With this assumption and a perfect load termination, the reflection coefficient
at the waveguide input is given by
=
N
Yn /Y0 e j2N d
n=1
2+
N
Yn /Y0
N
Yn /Y0 e j2n d
n=1
e j2N d
n=1
N
j2n d
n=1
The array excitations also change away from the center frequency because the slot locations relative to the standing
wave in the guide are no longer optimum and the slot radiation characteristics also exhibit frequency variation. This
causes pattern degradation and main-beam distortion. The effects are modeled by using the circuit model to determine the
slot excitations and computing the far-field pattern generated
by these excitations.
Traveling-Wave (Nonresonant) Arrays
For the nonresonant array, the array distribution is generated by progressively coupling energy from the waveguide
mode as it travels down the guide. The element spacings are
chosen to be nonresonant, and the waveguide is terminated
with a matched load. The design process differs from that of
the standing-wave array because the reflections from the slots
do not add coherently at the input. The input reflection is
controlled by the destructive interference due to the nonresonant spacing, and the reflection coefficients of the individual
elements are kept small to minimize the reflection. The traveling-array input-match bandwidth does not decrease as the
array size increases, unlike that of the resonant array. The
resonant slot spacing case for a broadside beam is avoided
(33)
for an array with alternating slot offsets. The beam scans
with frequency due to the change in the guide wavelength.
This frequency-scanning effect is often exploited by making
the waveguide path length between the slots greater than the
array spacing to enhance the amount of scan for a given frequency change (28). The beam position for this case is given
by
cos 0 = dg /dg /2d
i =
N
n e j2n d
(34)
(35)
n=1
where the reflection coefficient for a resonant slot
n =
(32)
Yn /Y0 e
cos 0 = /g /2d
2G/G0
2 + G/G0
(36)
Neglecting the slot reflections, the ratios of the slot radiated
powers for a traveling-wave array of resonant shunt elements
are given by
P rn
Gn /G0 n1
Gi
=
1
P1r
G1 /G0 i=1
G0
(37)
where the product term gives the decrease in the power along
the guide caused by radiated energy. The ratio of the power
remaining after the final element is absorbed in the load termination relative to the input power is given by
N
Pl
Gn
=
1
Pi
Go
n=1
(38)
If the waveguide loss is significant, the attenuation factor
ez should be included in the power calculation. The initial
design process requires solving these equations to achieve
both the desired amplitude distribution and an acceptable
value for the power lost to the load. For the traveling-wave
array it is advantageous to have a longer array because this
reduces the amount of power which must be absorbed in the
load termination. The range of coupling values available for
the radiating elements also affects the amount of power which
will be lost in the load. The circuit models can be cascaded to
account for the multiple reflections and frequency effects. The
500
WAVEGUIDE ANTENNAS
design process for the array including mutual coupling effects
is similar to that for the resonant array. However, the variation of the mode voltage along the array must be included in
the model (29).
WAVEGUIDE PLANAR ARRAY
The waveguide planar array is constructed by joining a number of linear arrays via a feed network. The planar aperture
provides the area required to achieve high gain which is a
critical parameter for most designs as it determines the effective range of system operations. The rectangular waveguide
exhibits low loss values over a wide range of frequencies (up
to the millimeter wave band) which makes the waveguide slot
array desirable for high-gain applications. The sidelobe distribution is also important for many applications as high sidelobes may result in unwanted interference between adjacent
installations or false returns off scatterers located in the sidelobe region. For the planar-slot array, the aperture distribution is achieved by controlling the characteristics of the individual radiating elements and the coupling characteristics of
the feed network. The ability to accurately control the excitation across the aperture allows the designer to set the radiation pattern sidelobes.
Planar-Array Far-Field Patterns
The equation to compute the far-field pattern of a planar
array located in the xy plane is
S(, ) = A(, )
N
M
Vmn e jk(xmn sin cos +ymn sin sin )
(39)
m=1 n=1
The array elements are assumed to have identical element
patterns which are similar to Eq. (2) with the appropriate
coordinate transformation to match the element orientation.
The array factor is given by the summation term where Vmn
is the excitation of each element. If the array elements are on
a rectangular grid, the equation simplifies to
N
M
S(, ) = A(, )
Vmn e jk(md x sin cos +nd y sin cos )
(40)
cardinal plane patterns of the array match the linear-array
patterns, and the intercardinal planes have sidelobes which
are the product of the two linear-array distributions. For circular apertures, there are a variety of circular distributions
available which are sampled to generate the array distribution (31). More complex distributions are synthesized for specific applications. The planar-array directivity is computed by
using the array far-field pattern in the directivity expression,
Eq. (4). The directivity is reduced by the losses in the antenna
system to determine the array gain.
Resonant Planar-Array Design
A common architecture for a resonant planar array, shown
in Fig. 8, uses broadwall offset-shunt slots as the radiating
elements. The linear arrays are attached along the sidewalls,
and common sidewalls are used to reduce the weight. Different lengths are used for the linear arrays to fit the array
within a specified area. Power is fed to the individual linear
arrays via a waveguide network beneath the radiating aperture. A compact feed network is formed with a linear array of
broadwall coupling slots to distribute power to the radiating
waveguides. Angled seriesseries slots provide a convenient
coupling mechanism because the power to each linear array
is controlled by adjusting the slot angle. Offset shuntseries
slots are also used as coupling elements, but they have the
disadvantage that the linear arrays must be offset to maintain the proper feed point and compensate for the staggered
offsets of the coupling slot.
The radiating slots and feed network are designed together
to provide the desired input match and to achieve the aperture radiating distribution. The design values for the radiating slots and the coupling slots must be chosen together to
meet the performance goals as the input impedance and the
slot radiation are functions of both the radiating slot values
and the coupling values of slots in the feed waveguide. The
array spacings to achieve resonance in both the radiating and
feed planes are related because the radiating guide width determines the spacing in the feed plane. The spacings and the
waveguide dimensions must be chosen as a set to achieve the
correct resonant frequency in each plane of the array.
m=1 n=1
This formulation is in the form of a Fourier transform, and
with appropriate manipulations fast Fourier transform (30)
computer routines are used to reduce computation time
greatly. For a rectangular, separable array, the distribution
is chosen to be of the form
Vmn = VmVn
Now the resulting planar-array equation is the product of the
two linear-array distributions. This results in a dramatic reduction in computation time:
S(, ) = A(, )
M
m=1
Vm e jkmd x sin
N
Vn e jknd y sin sin
(41)
n=1
For arrays with separable distributions, the desired lineararray distributions are selected, and then the products are
used to determine the individual element distributions. The
Figure 8. Waveguide planar array with offset longitudinal shuntslot radiators and seriesseries angle-slot coupling elements.
WAVEGUIDE ANTENNAS
Planar-Array Circuit Models. The lumped-circuit models
used for the linear-array design combine with the couplingslot models to provide a simple model for the planar array.
The impedance at resonance of a seriesseries slot coupling
to a linear array of M radiating slots is given by
Rn /R0 =
1 M(n)
Gmn /G0
Mn m=1
(42)
The coupling elements form a linear array of N slots in the
feed waveguide with an input impedance at resonance of
Zi /Z0 =
N
1 M(n)
Gmn /G0
Mn m=1
n=1
(43)
The ratios of the slot conductance and coupling values must
be chosen to provide the correct ratios for the radiating-slot
voltages. The slot conductance values within each radiating
guide are given by
Gmn = G1n
r
Pmn
r
P1n
(44)
and the coupling ratio for each seriesseries slot is given by
M(n)
Mn = M1
Gmn /G0
M(1)
m=1
m=1
M(1)
M(n)
Gm1 /G0
m=1
P rm1
(45)
P rmn
m=1
The values for the G1n and M1 must be chosen to create the
desired input impedance value. For compactness and improved bandwidth characteristics, it may be desirable to feed
the array at the center of the feed waveguide using an additional feed layer with a coupling slot. For a shuntseries coupling element where the waveguide terminated by a short at
one-quarter guide wavelength, the input admittance to the
array is given by
Yi =
Z
M0
(46)
where Z is the impedance in the feed waveguide and M0 is
the input slot coupling ratio. The values for the seriesseries
coupling slots must be chosen to account for the power division between the elements to each side of the input slot.
Mutual-Coupling Effects. As in the linear array, the simple
circuit model neglects effects, such as mutual coupling between the slots, that perturb the performance. The mutualcoupling effects between the radiating elements are handled
as described for the linear array with a separate input admittance requirement for each radiating waveguide. The external
mutual-coupling effects are generally more significant in the
E-plane of the slots than in the H-plane. Additional coupling
mechanisms which impact the array performance include coupling-slot interaction with the radiating slots and with the
other coupling slots in the feed guide. The radiating slots closest to the coupling slots in each guide are significantly affected by these interactions. These interactions are modeled
by the method of moments (32).
501
Large Resonant Arrays. The array bandwidth decreases as
the size increases because of the increased number of elements in the radiating and feed linear arrays. To avoid the
bandwidth reduction for a large array, the array is subdivided
into a number of subarrays or modules with shorter radiating
and feed arrays. The array power is distributed to the individual modules via a feed network made up of a series of waveguide power dividers. The configurations of the modules are
dictated by the required array bandwidth, the available aperture area, and the packaging limitations for the feed network.
It is desirable to minimize the modular size to improve the
array bandwidth, but this requires more modules and complicates the feed network. Dividing the array into subarrays is
fairly straightforward for rectangular arrays. The choice of
modular layouts is more complex for arrays with elliptical or
irregular boundaries. The available range of coupling values
for the slots and the feed network limits the choice of modular
configurations. Systems requirements, such as the need to
provide both sum and difference patterns which necessitates
dividing the array into equal quadrants, also determine the
available configurations.
A typical configuration for a resonant planar array is built
up of several modules with offset-shunt radiating slots and
angled seriesseries coupling slots. Offset-shunt series slots
are used to couple power from a waveguide corporate feed
network made up of reactive H-plane waveguide three-port
power dividers. For an array with difference pattern requirements, the power is distributed to the quadrants of the array
by a network of waveguide magic-tee four-port power dividers. This architecture provides some inherent mechanical advantages due to the boxlike structure of the radiating aperture, and, depending on the array size, is packaged so that
the depth of the array is as small as three times the waveguide b dimensions. Numerous other variations of array architecture are also used.
To begin the design of an array of this type, the desired
gain and sidelobe performance and frequency bandwidth
must be specified. The size of the array is driven by the array
gain. The area is chosen so that the area directivity minus
the taper loss, all losses in the aperture and feed networks,
and any of the return loss at the array input meets the required gain value with some margin. Because of variations
the required aperture distribution is selected to meet the sidelobe requirements with some margin. The array slot spacings
are chosen to meet the frequency requirements and make the
most effective use of the available area. The aperture distribution is discretized to give the required value at each slot.
Then the array is subdivided into modules whose size is based
on the required frequency bandwidth. The design of individual modules is similar to that described previously with the
additional complication that now the external mutual coupling includes both the slots within the module and all slots
in the other modules.
The feed network is a multiport power-dividing network
with an output port for each module. The relative power to
each port is calculated by summing the specified powers of
the slots in each module. For most applications a corporate
feed structure with equal path lengths to each module is required to meet the bandwidth requirements. A reactive Hplane tee in a rectangular waveguide provides a low-loss, lowprofile power-dividing network. The individual tees are
matched, and the power splits are controlled by adjusting in-
502
WAVEGUIDE ANTENNAS
ductive-tuning irises at the tee input and the length and position of a septum on the back wall of the tee. The performance
of the reactive-feed network is sensitive to the loads placed
on the feed by the modules. To reduce this sensitivity and
provide broader band performance, matched power dividers
are used though they have higher loss and larger sizes.
Frequency and Power-Handling Effects. As the operating frequency of a resonant array moves away from the design frequency, the input match degrades and the pattern performance suffers. For arrays whose radiating guide and feed
waveguides are fed near or at their centers, the amplitude
and phase of elements at the ends of the guide diverge from
those near the center. For a single module this broadens the
main beam and increases the sidelobe level. For an array of
modules, this creates an error across the array with a period
equal to the modular spacing which creates error lobes in the
far-field pattern and reduces the array directivity. These effects are modeled by using a circuit model for the modules
and feed network.
The amount of power that the array is required to handle
also influences the design. The waveguide power-handling capability is reduced as the waveguide b dimension is decreased, so for high-power applications full height waveguide
should be used. For arrays which must operate at high altitudes, the waveguides are pressurized. Care should be taken
to avoid any obstacles which protrude into the guide in the b
dimension and mismatches should be minimized to reduce the
standing-wave effect. The amount of power is reduced as it
divides through the feed network and subarrays, so the
power-handing problem is less severe at the aperture.
Nonresonant Planar Arrays
The traveling-wave planar array is created by joining a set a
linear, traveling-wave arrays with a feed network. The sidewall edge-slot array is frequently chosen, particularly for applications where the array is scanned in the feed plane, because the linear arrays can be packed closely together. The
radiating arrays are designed similarly to the linear array,
but mutual coupling between the linear arrays must be taken
into account. It is generally preferable not to subdivide the
array and to end feed the radiating aperture to minimize the
power lost to the loads. The feed network can be traveling
wave or standing wave depending on the application. A linear
array of coupling slots provides a compact feed network for
many applications. Where beam squint in the feed plane is
not desired, a corporate feed network of waveguide power dividers is used.
the features and the geometry of the individual features must
be controlled. The potential cost impacts of overly tight tolerances must be weighed against the performance impacts. The
operating frequency of the array is a major factor in the allowable tolerances and hence the preferred manufacturing technique. Typical manufacturing processes for slot arrays include machined metal assemblies joined either by brazing or
bonding, stamped or punched slots, etched slots on metal clad
dielectric, molded metallized plastic assemblies, or cast parts.
For slots manufactured by conventional machining techniques, it is more convenient and cost effective to machine the
slots with a rounded end that matches the diameter of the
endmill rather than create a true rectangular slot. For narrow
slots, the main effect of the rounded end is a change in the
slot resonant frequency. The effect is modeled and compensated for by assuming that a rounded end slot has the same
resonant frequency as a square-ended slot of equal area (14).
For large arrays, errors that are randomly distributed
across the array face result primarily in increased average
sidelobe level and a decreased array gain. The effects of these
errors are estimated by statistical techniques (33,34) or by a
Monte Carlo simulation of the array pattern. The variation in
the element excitation must be estimated from the tolerances
on the element features, and then these values are used to
estimate the change in array performance. The sensitivity of
the characteristics of the individual slots to a given variation
in slot dimensions is determined by using the slot design
equations or design curves such as those shown in Fig. 4.
Systematic errors have more significant impacts. A consistent change in the waveguide a dimension shifts the resonant
frequency of the standing-wave array and alters the beampointing angle in the traveling-wave array. A systematic error
in the slot locations creates undesirable sidelobes. In the traveling-wave array, if the slot coupling or the waveguide losses
are consistently higher or lower than the design value, the
error in the array excitation accumulates along the array because the mode amplitude is increasingly perturbed. If the
coupling is significantly lower than the design value, excess
power is lost in the load. Periodic errors raise the sidelobe at
the position where a grating lobe would occur for an array
whose spacing is equal to the spacing of the periodic error.
Feed-network or other errors that affect large portions of the
array have significant impacts on the array sidelobe and gain
performance. An array with a difference pattern requirement
is sensitive to asymmetrical errors across the array because
any imbalance affects the cancellation which occurs at the difference pattern null.
ACKNOWLEDGMENTS
Effects of Manufacturing Variations
Array performance is influenced by how accurately the design
is realized in manufacturing. For arrays with stringent sidelobe requirements, the effects of manufacturing deviations
from the design values significantly alter the performance,
and the deviations must be carefully controlled. The potential
impacts of the manufacturing process should be accounted for
in the design. Where possible, the design values should be
chosen to minimize sensitivity to the variations. The acceptable limits for process variations should be set on the basis of
acceptable performance limits; the processes should be chosen
for compatibility with these limits. The relative locations of
The author wishes to thank W. H. Kummer for his assistance
with the manuscript and historical background, R. C. Hansen
for providing historical material, G. R. Kulakowski for his assistance with the figures and the reviewers for their helpful
comments.
BIBLIOGRAPHY
1. G. C. Southworth, Hyper-frequency wave-guidesGeneral considerations and experimental results, Bell Syst. Tech J., 15: 284
309, 1936.
WAVEGUIDES
503
2. J. R. Carson, S. M. Mead, and S. A. Schelkunoff, Hyper-frequency
wave guidesMathematical theory, Bell Syst. Tech. J., 15: 310
333, 1936.
28. N. A. Begovich, Frequency scanning, in R. C. Hansen (ed.), Microwave Scanning Antennas, Vol. 3, New York: Academic Press,
1964, Chap. 2.
3. W. L. Barrow, Transmission of electromagnetic waves in hollow
tubes of metal, Proc. IRE, 24: 12981398, 1936.
29. R. S. Elliott, On the design of traveling-wave-fed longitudinal
shunt slot arrays, IEEE Trans. Antennas Propag., AP-27: 717
720, 1979.
30. E. O. Brigham, The Fast Fourier Transform, Englewood Cliffs,
NJ: Prentice-Hall, 1974.
31. T. T. Taylor, Design of circular apertures for narrow beamwidth
and low sidelobes, IRE Trans. Antennas Propag., 8: 1722, 1960.
32. D. C. Senior, Higher order mode coupling effects in a shunt-series
coulping junction in a planar slot array antenna, Ph.D. Dissertation, Univ. California at Los Angeles, 1986.
33. J. Ruze, The effects of aperture errors on the radiation pattern,
Nouvo Cimento, 9 (Suppl. No. 3): 364380, 1952.
34. R. S. Elliott, Mechanical and electrical tolerances for two-dimensional scanning antenna arrays, IRE Trans. Antennas Propag.,
10: 114120, 1958.
4. K. S. Packard, The origin of waveguides: A case of multiple rediscovery, IEEE Trans. Microw. Theory Tech., MTT-32: 961969,
1984.
5. S. A. Schelkunoff, The impedance concept and its application to
problems of reflection, refraction, shielding, and power absorption, Bell Syst. Tech. J., 17: 1748, 1938.
6. S. A. Schelkunoff, A mathematical theory of linear arrays, Bell
Syst. Tech. J., 22: 80107, 1943.
7. C. L. Dolph, A current distribution for broadside arrays which
optimizes the relation between beamwidth and side lobe levels,
Proc. IRE, 34: 335348, 1946.
8. T. T. Taylor, Design of line source antennas for narrow beamwidth and low sidelobe, IRE Trans. Antennas Propag., AP-7: 15
28, 1955.
WINIFRED E. KUMMER
9. W. H. Watson, The Physical Principles of Wave Guide Transmission and Antenna Systems, Oxford, UK: Clarendon Press, 1947.
Raytheon Systems Company
10. A. F. Stevenson, Theory of slots in rectangular waveguides, J.
Appl. Phys., 18: 2438, 1948.
11. S. Silver, Microwave Antenna Theory and Design, Vol. 12, MIT
Rad Lab Series, New York: McGraw-Hill, 1949.
12. I. P. Kaminow and R. J. Stegen, Waveguide slot array design,
Tech. Memo. No. 348, Hughes Aircraft Co. July 1954.
13. A. A. Oliner, The impedance properties of narrow radiating slots
in the broad face of rectangular waveguidePart Itheory, Part
IIcomparison with measurement, IEEE Trans. Antennas Propag., AP-5: 420, 1957.
14. A. A. Oliner, Historical perspectives on microwave field theory,
IEEE Trans. Microw. Theory Tech., MTT-32: 10221045, 1984.
15. H. G. Booker, Slot aerials and their relations to complementary
wire aerials (Babinets principle), J.I.E.E. (London), 93, part
IIIA: 620626, 1946.
16. Ref. 11, pp. 226229.
17. R. S. Elliott, An improved design procedure for small arrays of
shunt slots, IEEE Trans. Antennas Propag., AP-31: 4853, 1983.
18. Ref. 11, pp. 286299.
19. R. F. Harrington, Field Computation by Moment Methods, New
York: Macmillan, 1968.
20. V. T. Khac, A study of some slot discontinuities in rectangular
waveguides, Ph.D. Dissertation, Monash Univ., Australia, Nov.
1974.
21. G. J. Stern and R. S. Elliott, Resonant length of longitudinal slots
and validity of circuit representation: Theory and experiment,
IEEE Trans. Antennas Propag., AP-33: 12641271, 1985.
22. D. J. Lewis, Waveguide Coulping Slots, Internal memo., Hughes
Aircraft Co.
23. S. R. Rengarajan, Analysis of a centered-inclined waveguide slot
coupler, IEEE Trans. Microw. Theory Tech., 37: 884889, 1989.
24. P. P. Silvester, Finite Elements in Electrical and Magnetic Field
Problems, New York: Wiley, 1980.
25. J. L. Altman, Microwave Circuits, New York: Van Nostrand,
1964, pp. 399406.
26. J. J. Gulick and R. S. Elliott, The design of linear and planar
arrays of waveguide-fed longitudinal slots, Electromagnetics, 10
(4): 327347, 1990.
27. R. S. Elliott, Microwave Antenna Theory and Design, Englewood
Cliffs, NJ: Prentice-Hall, 1981.
WAVEGUIDE ATTENUATOR. See ATTENUATION MEASUREMENT;
ATTENUATORS.
WAVEGUIDE CIRCUITS. See MICROWAVE CIRCUITS.
WAVEGUIDE FINLINES. See FINLINES.
MODELING WIRE ANTENNAS
417
cluded for reference (1). The numerical approaches described
are rather general, and specific details are summarized as implemented in two computer codes. One is Numerical Electromagnetics Code (NEC) (2) and the other is Thin-Wire Time
Domain (TWTD) (3). The former is a widely used frequencydomain model, and the latter is its time-domain counterpart.
Both were developed at Lawrence Livermore National Laboratory, and versions of both are in the public domain.
PRELIMINARIES
Integral Equations and Wire Modeling
MODELING WIRE ANTENNAS
Wire antennas represent the oldest category of antenna types,
dating back to the successful transatlantic transmissions by
Marconi and even before that to the theoretical and experimental efforts of Hertz, Maxwell, and others. In 1898,
Pocklington developed an integral equation for a dipole antenna whose numerical solution remained relatively intractable until the digital computer made its numerical solution
possible. This integral equation, and its various generalizations, provide the starting point for essentially all wire-antenna numerical modeling now routinely done and on which
the discussion below focuses. The emphasis here is on describing and demonstrating the capabilities of wire-antenna models rather than on cataloging a large number of antenna types
and characteristics, because the extremely large variety of the
latter is incompatible with the space available and the scope
of the discussion.
Because our primary goal here is to summarize the modeling and use of wire antennas rather than to concentrate on
their numerical modeling alone, we consider only the Pocklington form of the types of integral equations available for wireantenna analysis and design, although several others are in-
The derivation of an integral equation for a perfect electric
conductor (PEC) is approached in various ways, but perhaps
the most physically appealing is to begin with Maxwells
equations written in their source-integral form whose kernel
is a Greens function for an electric source in an infinite medium. The source integral gives the secondary field (also
called induced, scattered, or radiated field) caused by a current on the body flowing in response to a primary field (also
called incident, applied, or exciting field). By expressing the
secondary field over loci of points where the behavior of the
total field (applied plus radiated) is known via boundary or
continuity conditions, an integral equation for the induced
source is obtained.
Two broad general classes of integral equations are encountered, depending on whether the unknown source occurs
only under the integral (a first-kind Fredholm integral equation) or also outside of it (a second-kind integral equation) (4).
In electromagnetics, a first-kind integral equation arises
when the forcing function (primary field) is an electric field,
and a second-kind integral equation arises when the forcing
function is a magnetic field. Although it is usual for derivatives of the unknown also to occur, the resulting equation is
commonly called an integral equation (IE), rather than an integro-differential equation as would be strictly correct.
Generally speaking, a magnetic-field IE (MFIE) is best
suited for smooth, closed objects, and it is analytically inapplicable to objects thin in one dimension, such as plates, shelllike structures, and wires. The electric-field IE (EFIE) is not
limited by these constraints, and so becomes the only practical type for modeling the wires of interest here. Because of
anomalous internal resonances, both the MFIE and the EFIE
fail numerically near certain discrete frequencies, and a combined-field IE (CFIE) comprised of their sum is needed for
modeling smooth, closed objects.
At its simplest, a wire is a PEC of constant, circular cross
section whose circumference in wavelengths C is no greater
than 1 so that its radius in wavelengths, a C/2, although
it is usual to employ a maximum radius only a tenth or so as
large. By definition, to be an electromagnetic wire, it must
satisfy at least two conditions: (1) the effect of any circumferential or azimuthal current (one that flows on the wire surface in a direction normal to its axis) is negligible; and (2) the
longitudinal current (the surface current flowing along its
axis) is independent of the azimuth angle, that is, it is uniform around the circumference.
A further requirement also usually imposed for analytical
and numerical convenience is that the thin-wire (or reducedkernel) approximation is employed, so that the two-dimensional surface integration required to evaluate the fields of
J. Webster (ed.), Wiley Encyclopedia of Electrical and Electronics Engineering. Copyright # 1999 John Wiley & Sons, Inc.
MODELING WIRE ANTENNAS
sources on a general surface is approximated by a line integral along the wires axis. This is discussed later.
Most wire modeling, even when approximating a closed
surface by a wire mesh or grid, is done with piecewise linear
(straight) segments. Because meshes used as approximations
to curvilinear surfaces cannot be made of planar square or
rectangular elements, this means that junctions of wires
meeting at variable angles are encountered. Although requirement (2) above is then almost certainly violated, the
thin-wire approximation remains accurate enough to produce
useful results. The added difficulty of including such higherorder effects as a circumferentially nonuniform longitudinal
current is evidently not commensurate with the improvement
that might be realized. Use of straight segments is also common in modeling such simple objects as circular loops, helices,
conical spirals, etc.
Guidelines have been developed over the years to help users of wire codes choose modeling parameters more likely to
lead to acceptable results. For example, when modeling circular loops, the centerline circumference of the polygonal approximation should equal the circumference of the actual loop
and a minimum of six wire segments must be used to model
the loop in the vicinity of its first resonance, that is, where
C 1. This relates to the fact that experience has shown that
nominally six wire segments per wavelength are needed to
achieve acceptable accuracy, when the sampling density of
the unknown current is wavelength-driven. When more segments are needed to properly represent the geometry of the
object being modeled, the required segment lengths are required to be much smaller, and the sampling density is geometry-driven. When approximating a solid surface with a wire
mesh as shown in Fig. 1, the maximum mesh openings are no
greater than 0.1 wavelength on a side. The radius of the mesh
wires also needs to be such that their total area is twice that
of the surface they represent, so that the wire area in each
direction of a local, orthogonal coordinate system approximates that of the surface being modeled (6), the equal-area
rule. A computed result showing that the front-to-back ratio
of a backscreen of parallel wires is maximized when the
Back
Front
2a
xe
103
0.5 f0
Front-to-back ratio
418
102
101
0.75 f0
f0
1.25 f0
1.5 f0
1
0.05 0.15
a/D
0.25
Figure 2. This two-dimensional backscreen model, consisting of
equally spaced, infinite wires with an excited element in front, was
studied parametrically to determine the wire radius with the largest
front-to-back ratio (FBR) to gain insight about modeling a continuous
surface with wires (5). For a 50-wire backscreen, a 2D provides
the maximum FBR, where the wire and a continuous backscreen have
equal areas over a 3 : 1 frequency range.
equal-area rule is satisfied is shown in Fig. 2 (5). Modeling
guidelines for wire antennas are discussed in more detail
elsewhere (6).
Finally, it should be recognized that a wire model is not
limited to circular PECs. A thin-wire model can be used for
literally any object that satisfies condition (1) previously
stated. Its cross section can vary from a flat strip to an arbitrarily irregular shape. For a flat strip of width w, an equivalent radius is found to be a w/2. Other simple cross sections, such as triangular, are also amenable to thin-wire
modeling, where the radius of the equivalent circular wire is
established from simple formulas by quasi-static analysis (8).
IMPORTANT PROPERTIES OF WIRE ANTENNAS
Figure 1. This wire-grid model of a ship (7) is intended for the HF
band and lower, where the length of the grid wires is less than 0.1
wavelength and has more than 829 segments. Because the ship is in
salt water, which becomes a perfectly conducting half-space to which
all of the wires forming its lower sides are connected, the model is
not closed on the bottom. Wire-grid models with thousands of segments have been used.
The purpose of the analytical treatment and its implementation to develop a numerical model, described later, is to assess
the performance characteristics of antennas of interest, of
which some of the more important are described here (1).
Input Impedance and Admittance. The input impedance
Zin, of an antenna is needed to match an antenna to its source
MODELING WIRE ANTENNAS
and is also an important quantity in determining its efficiency, defined as
= Prad/(Prad + Ploss ) = Prad/Pin
Yin = I(sin )/V (sin ) = Iin /Vin = 1/Zin
(2)
where sin is the place at which the driving voltage Vin is applied and Iin is the current there. Determining a realistic
value of Vin from a numerical model presents a problem because it is sensitive to the way the feed region is described
and it is difficult to replicate the physical arrangement with
good fidelity. Feedpoint-modeling errors affect both Gin and
Bin though uncertainty in Vin is sometimes resolved, as far as
its effect on Gin is concerned, by far-field integration. In a computer code like NEC that uses point matching where it is assumed that Vin Ein, with the length of the feed segment
(see section later on numerical implementation), this uncertainty is usually resolved by integrating Etan in the vicinity of
the feedpoint.
The Far-Field and Radiated Power. The power radiated by
an antenna is usually determined by integrating the far-field
power flow over a closed surface containing the antenna as
given by
Prad =
1
Re
2
1 2
E H ds Iin
Rrad
2
S
the far field in the direction (, ) to its average value over
S, so that
Re[(E H ) r ]
1
lim
(E
H
)
r
ds
Re
r
4r2
S
Re[(E
H
)
r
]
= 2r2
Prad
(1)
where Prad and Ploss are the radiated powers and Ploss arises
from the losses to which the antenna is subject. From a numerical and analytical perspective, the antennas admittance
Yin is normally the more accurately obtainable quantity, because the real conductance Gin is determined wholly by Pin,
whereas the reactive susceptance Bin is very sensitive to feedregion geometry (discussed later). The real and reactive components of the impedance, on the other hand, which is simply
the inverse of the admittance, is affected by whatever errors
arise in the susceptance. Furthermore, when there are no
losses, it is straightforward to obtain the conductance by simply integrating the far-field power flow, a quantity that is also
relatively insensitive to errors in the computed current distribution. By contrast, the susceptance requires an accurate solution of the antennas near fields, because it is related to the
stored power that they represent.
The admittance is defined as
GD (, ) =
where S is the enclosing surface, usually a sphere centered
on a convenient part of the antenna, Re denotes the real part,
ds has a unit normal in the outward direction, E and H are
the peak values of the fields and Rrad is the radiation resistance. The input power equals Prad unless there are losses, in
which case it is found from
(4a)
where Prad /(4r2) is equivalent to the power density of an isotropic antenna radiating the power Prad. The directivity D of
an antenna is defined as the maximum value of the directive
gain, or
D = max{GD (, )}
(4b)
and finally the gain is defined as in Eq. (4a) but including the
losses so that
G(, ) = 2r2
Re[(E H ) r ]
= GD (, )
Pin
(4c)
A Summary of Analytical Results for Some Common Antennas
Although the complexity of real antennas located in their actual environments precludes analytical solutions, and necessitates numerical models like those previously discussed, a
wide variety of simple antennas have been studied over the
years. Some important properties of a large number of these
antennas are summarized for ready reference in Table 1 (1,9).
NUMERICAL MODELING OF ANTENNAS
Although attention here addresses specifically only the numerical treatment of the Pocklington-type IE, for completeness, several different types of thin-wire IEs are included in
Table 2 (1). These are all frequency-domain IEs, because most
wire modeling uses that approach, but a brief review of wire
time-domain modeling concludes this section.
As a starting point, we include the EFIE and MFIE for
smooth surfaces from which such equations can be derived
(1):
1
E (r) = n
n
4 j
(3a)
419
2
JS (r) G (r, r )d r
(5)
and
Hi (r) +
JS (r) = 2n
1
2
JS (r ) (r, r )d 2 r
(6)
where
Pin = Re(Vin Iin
)
2
(3b)
Directivity and Gain. The directive gain of an antenna
GD(, ) is defined as the ratio of real-power flux density in
G (r, r ) = ( + k2 )g (R)
(r, r ) = I g (R)
(7)
Equation (1) is the EFIE and Eq. (2) is the MFIE, where the
is an outwardunknown surface-current density is Js(r), n
420
MODELING WIRE ANTENNAS
Table 1. Properties of Some Simple Antennas
Type
3-dB
Impedance:
Resistive at Bandwidth
Percent
fr9 R()
Configuration
Isotropic
radiator
(theoretical)
Small
dipole
Isotropic
Dipole
Polarization
Pattern
number
2.14
none
1.74
0.4
60
34
2.14
49
55
2.14
37
100
2.14
150
130
3.64
1.5
L
L > /2
Gain:
dB above
Thin
dipole
L
L = /2
L/D = 276
Thick
dipole
L
L = /2
L/D = 51
Cylindrical
dipole
L
L = /2
L/D = 10
Cylidrical
dipole
L
L =
L/D = 9.6
90
90
90
0 z 180
180
0 x
270
270
270
90
90
90
0 z 180
0 z 180
180
270
0 z 180
270
pointing, surface-normal, unit vector andthe other quantities
are defined in Table 2. The quantities G(r, r) and (r, r)
are known as Greens dyads for the electric and magnetic
fields of electric current sources.
It should be understood that C(r) in the equations of Table
2 represents the geometrical configuration of the wire or collection of wires to be modeled and, therefore, does not need to
be spatially continuous. The subscript is used on various
quantities in the table to emphasize their association with an
infinite medium, as we next consider the half-space problem.
0 x
270
Extension of the previous infinite-medium IEs is straightforward for a perfectly conducting half-space, a common starting point for treating imperfectly conducting or electromagnetically penetrable grounds, an especially important
application of wire-antenna modeling. By convention, the
half-space occupies the region z 0. It can be deduced that
(1), because an object located above a PEC half-space is electromagnetically imaged or mirrored in it, the vertical components of the image electric currents flow in the same direction as the actual currents and the horizontal components are
MODELING WIRE ANTENNAS
421
Table 1. (Continued)
Type
3-dB
Impedance:
Resistive at Bandwidth
Percent
fr9 R()
Configuration
Folded
dipole
Gain:
dB above
Isotropic
Dipole
Polarization
Pattern
number
L
d
L = /4
L/d = 13
6000
1.64
0.5
300
45
2.14
72
100
2.14
350
200
2.14
150
50
0.86
150
20
7.14
R
Folded
dipole
L
d
L = /2
L/d = 25.5
Biconical
x
40
L = /2
R
Biconical
x 60
L
L =
R
Turnstile
L
L = /2
L/d = 25.5
d
R
Folded
dipole over
reflecting
sheet
L = /2
L/d = 25.5
/8 above
sheet
y
L
d
R
90
90
90
0 z 180
180
270
270
90
90
z 180
0 x180
270
reversed. Thus, the electric-field Greens dyad for the PEC
becomes
GPG (r, r ) = G (r, r ) + GI (r, r )
(8a)
where the image term, denoted by subscript I is given by
GI (r, r ) = I R G (r, I R r )
I R = x x + y y z z
0 x
270
No signal
response
in this
plane
0 x180
(8b)
0 x
270
with a similar expression for the magnetic-field form
PG (r, r ) = (r, r ) + I (r, r )
I (r, r ) = I R (r, I R r )
(8c)
(8d)
Then the Pocklington IE is written without any additional
approximation as
1
s Ei (s) =
I(s )[G (s, s ) + GI (s, s )]ds
4 j C(r )
422
MODELING WIRE ANTENNAS
Table 1. (Continued)
Type
3-dB
Impedance:
Resistive at Bandwidth
Percent
fr9 R()
Configuration
Dipole
over small
ground plane
Isotropic
Dipole
Polarization
Pattern
number
z
D
L = /4
L/D = 53
l = 2
Gain:
dB above
28
40
2.14
150
45
2.14
50
16
2.14
72
200
2.14
50
300
2.14
20
25
14.14
12
l
R
Folded
unipole
over small
ground plane
L = /4
L/D = 53
l=2
L/d = 13
L
l
Coaxial
dipole
L
L = /4
L/D = 40
R
Biconical
coaxial
dipole
d
L
L = /2
d = /8
D = 3 /8
Disc-cone
or rod
disc-cone
L = /4
l=
D
R
z
L
l
R
Biconical
horn
L = 9 /2
D = 14
D
L
R
90
90
90
0 z 180
180
270
270
90
90
0 z 180
180
C(r )
s =
|C(r )|
0 x
270
with the effect of the PEC ground included in the GI (image)
term of the modified IE and where s* is the axial coordinate
of the image current. Note that image theory is useful in numerous other ways in electromagnetics, but space precludes
discussing details here.
gI (R ) =
r (x, y, z) = r (x, y, z)
0 x 180
270
GI (s, s ) = [k2 s s + (s )(s )]gI (R )
270
with
0 x
90
270
e jk R
R
R = |r r |
0 x 180
(9)
Imperfectly Conducting Ground
Modified Image Theory or Reflection-Coefficient Approximation. When the half-space of interest is finitely conducting or
MODELING WIRE ANTENNAS
423
Table 1. (Continued)
Type
Configuration
Slot in large
ground
plane
L
L = /2
l/d = 29
Isotropic
Dipole
Polarization
Pattern
number
350
70
2.14
45
13
3.14
130
200
10.14
circ.
600
100
16.74
14.5
300
30
14.74
12.5
50
35
15.14
13
D = /
D/d = 36
Helical over
reflector
screen,
tube 6 long
coiled into
6 turns
/4 apart
Rhombic
y
600
z
L
Parabolic
with folded
dipole feed
( /2)
Gain:
dB above
Vertical
full-wave
loop
L = 9
l = 9 /2
3-dB
Impedance:
Resistive at Bandwidth
Percent
fr9 R()
z
R
D = 5 /2
Horn,
coaxial feed
y
l
L = 3
l = 3
z L
y
No signal
response
in this
plane
z
90
z 180
0x180
0 x
270
270
270
90
90
90
y
90
0 z 180
180
270
0 x180
270
a simple dielectric, then the image-theory approach for the
PEC half-space is no longer exact. Nevertheless, as a means of
approximating the effect of the half-space, and thereby
avoiding the considerable complexity that otherwise results
from the rigorous Sommerfeld approach, a modified image theory is useful. This leads to the reflection-coefficient approximation (RCA) (1,10) and yields a new form for the Pocklington-
0 x
270
type IE which follows from the following RCA Greens dyads
GIG (r, r ) = G (r, r ) + RM GI (r, r ) + (RE RM )[GI (r,r ) p]
IG (r, r ) = (r, r ) + RE I (r, r ) + (RM RE )[ I (r, r ) p]
=
p
(r r ) z
|(r r ) z|
(10)
424
MODELING WIRE ANTENNAS
Table 2. Different Forms of Thin-Wire-Approximation Integral Equationsa
s Ei(s)
1
k2s
4j s
s Ei(s)
1
4
s Ei(s)
1
4
s sjI(s)g(R)ds
C(r)
C(r)
C(r)
sI(s)g(R)ds
s sjI(s)g(R)
1 I(s)
s g(R) ds
j s
1
4j S
C(r)
I(s)
g(R)ds
s
Pocklingtons Integral Equation
1
4j
s Ei(s)
C(r)
I(s)G(s, s)ds, with G(s, s) [k2s s (s )(s )]g(R)
Magnetic Vector Potential Integral Equation for Arbitrarily Curved Wires
C(r)
I(s) g(R)s s
1
2
C(r)
Aejks Bejks
d( s)ejks
1
2( /)
g(R)( s) [( s)g(R)]
s
ds
s Ei(s)ejkss ds
C(r)
(u) 1, u 0; (u) 0, u 0
Magnetic Vector Potential Integral Equation for Straight Wires
(Hallens Integral Equation)
C(r)
I(s)g(R)ds Ajks Bjks
1
2( /)
C(r)
s Ei(s)ejkss ds
In the previous equations:
s and s denote the axial coordinates at the observation and source points, respectively;
r and r are the vector coordinates at the observation and source points;
C(r)
C(r)
and s
are unit tangent vectors at r and r;
s
C(r)
C(r)
C(r) is the range of integration over the wire;
a(r) is the wire radius at r;
s C(r) a(r) so that R a(r) as required by the thin-wire approximation;
g(R)
ejkR
;
R
R [a(r)2 r r2]1/2;
k /( )1/2 is the wave number of the infinite medium in which the wire is located; the superscript i denotes an incident-field quantity;
and the denotes a unit-length vector.
a
After (1).
as
s Ei (s) =
and
1
4 j
C(r )
I(s )[G (s, s ) + RM GI (s, s )
(r) = tan1
+(RE RM ) sin sin sin( ) sin( )gI (R )]ds
(11a)
where
RE =
RM =
( E sin2 )1/2 cos
( E sin2 )1/2 + cos
E cos ( E sin2 )1/2
E cos + ( E sin2 )1/2
(11b)
(y y )
(x x )
(11c)
Here RE and RM are the Fresnel plane-wave reflection coefficients for transverse-electric (TE) and transverse-magnetic
(TM) polarizations, respectively and the subscript IG signifies an imperfect ground. Also, (r, r) is the angle with respect to a normal to the interface and a straight line in the
vertical plane that joins s and s (known as the specular reflection point from optics), and (r) and (r) are the directional angles of the wire at r with respect to the x and z axes,
but these explicit dependencies are omitted for clarity. Note
MODELING WIRE ANTENNAS
90
45
terms in the RCA. Thus, one form of the Pocklington-type
EFIE for the imperfect ground is written as follows
(1,9,11,12):
1
s Ei (s) =
I(s ) G (s, s ) + GI (s, s )
4 j C(r )
2
(13)
[sin gHz cos gVz ]
+ k2 cos +
sz
2
g
ds
+ sin k2 sin cos( ) +
Ht
st
00
90
(a)
00
45
where the partial derivative with respect to t is in the direction of the horizontal projection of the wire at s. Also
(b)
Figure 3. The real component of the radial electric field of a horizontal, delta-functional current source in free space is shown as a
function of distance R and elevation angle near a half-space for
IG / 4, IG 0.001 S/m, and (b) IG / 16, IG 0 (17). The latter
exhibits the beating effect of waves in the upper and lower half-spaces
with two different wavelengths which the loss in the former eliminates.
IG
j IG = Er j Ei
J ()e (z+z ) d
+ E 0
0
cos( ) E
gHz =
J ()e (z+z ) 2 d
k2
E + E 1
0
E
gVz = 2
J0 ()e (z+z )
d
(14)
0
E
E
= (x x )2 + (y y )2 + a2
= 2 k2
E = 2 E k2
gHt = 2
that E is the permittivity of the lower half-space relative to
the upper, that is,
E =
425
(12)
where the integrals are the field expansions mentioned previously, first derived by Sommerfeld (11,13). The quantity
Jn(x) is a Bessel function of order n and argument x, z and z
are the source and observation heights above the interface,
and the a in the expression for radial separation imposes
the minimum separation required by the thin-wire approximation. Note that these integrals are functions of only two
variables and z z.
The treatment thus far is limited to the one-sided problem, where the object being modeled is entirely contained on
one side of the interface, so that the source and observation
points are always in the same medium. Similar, but still more
complicated expressions result when the object(s) being modeled occupy both half-spaces, that is, the two-sided problem.
It is worth noting that the RCA has been extended to the twosided problem (14), but the latter primarily has been approached with the Sommerfeld theory or its equivalent. Ana-
where IG and IG are the permittivity and conductivity of the
lower half-space, the subscripts r and i denote real and
imaginary components, and it should be understood that
itself can be complex, should the above-ground medium be
conducting.
The RCA provides reasonably good results for input impedance and radiation patterns, with errors of 10% or less in the
former, for wires no closer than 0.1 wavelength of the interface. For situations requiring greater accuracy or involving
wires closer to the interface, the Sommerfeld approach, or its
equivalent, is needed.
The Sommerfeld Approach. The Sommerfeld approach expands the fields of point sources in a continuous-wave expansion along the interface, which leads to infinite-range integrals for the reflected fields that appear in place of the image
30
20
B (mS)
G (mS)
20
10
10
0
0
4
k Lh
4
k Lh
Figure 4. These three results for the input admittance of a monopole antenna of
length h with air insulation b/a 2.5 in
water having IG / 80 j0.197 as a
function of the composite wave number
demonstrate the general validity of the
sheath model described above. The xs are
from NEC, the solid lines are independent
computations (21), and the solid circles
are measurements (22).
426
;;;
MODELING WIRE ANTENNAS
/4
/4
1.40
1.<180
replaces the Sommerfeld fields with a series of images in complex space that provide the interface-reflected fields. Another
alternative is to generalize the problem to include the tangential electric and magnetic fields on the interface as additional
unknowns in the model (16). Because the interface extends to
infinity, this is impractical because the number of unknowns
is commensurate with the area being modeled. However, in
terms of accounting for the interaction of an object with the
interface, only a limited area of the interface under the objects projection onto it measurably affects the objects currents. Thus, the number of interface unknowns is limited.
Furthermore, for a particular choice of ground parameters
and frequency, the interface part of the problem needs to be
solved only once and then continues to be reused for whatever
object is to be modeled for those specific conditions.
Approaches that fall in category (3) are of a generally different character and are described as employing a signal-processing philosophy. The basic motivation here is to compute
a minimum number of Sommerfeld-integral values, or their
equivalent, beyond which the field values needed in the IE
Figure 5. The two-element monopole array shown produces the
ground-plane currents presented as a vector plot (24). By revealing
the current-flow pattern in a perfect ground, such plots indicate the
most effective ground-screen geometry. The arrows represent the
semimajor and semiminor axes of the ground-current polarization ellipse, respectively.
12.
10.
lytical treatment of the two-sided problem is beyond the scope
of this discussion. The details are found elsewhere (e.g.,
(12,13), but it is relevant to mention that the fields for the
two-sided case are functions of the three variables , z and z.
R ()
8.
N = 20
6.
N = 50
4.
N = 100
Simplifying the Sommerfeld Treatment. The appearance of
the Sommerfeld integrals as part of its kernel function adds
significant further computational complexity to solving the IE
in Eq. (6). There are at least three ways of reducing the computer cost of solving interfacial problems while retaining the
rigor of the Sommerfeld-type approach:
2.
g = 10 j100
0.
0.0
0.1
0.2
0.3
Screen radius/ 0
0.4
0.5
10.
Category (1) has long been pursued, and a variety of analytical approximations have been derived for various regions of
the parameter space for which Sommerfeld-integral values
are needed, as have been numerous numerical approaches.
Unfortunately, although some very useful results have been
obtained, individual approaches generally do not cover the parameter space encountered in many practical problems, so
this possibility has had only limited success. The Sommerfeld
integrals are one of the wave-type, or Fourier-type expansions, developed for the interface problem, of which a summary is found in (13).
As an alternative to wave expansions that lead to Sommerfeld-type integrals, one example of category (2) is the development of complex-image theory (15). This particular approach
8.
6.
X ()
1. develop more efficient ways of evaluating the Sommerfeld integrals;
2. develop an alternative, more efficient formulation in
their place;
3. minimize the need for evaluating these integrals in the
numerical model.
N = 20
4.
N = 50
2.
N = 100
0.
2.
0.0
0.1
0.2
0.3
Screen radius/ 0
0.4
0.5
Figure 6. Input impedance of a quarter-wave monopole antenna
driven against a buried ground screen with N wires subtracted from
its impedance for a perfect ground with E 10 j100 (25). The NEC
results, which represent a numerically rigorous solution of the problem are compared with approximate results from the compensation
theorem (26) discussed in the next section.
MODELING WIRE ANTENNAS
130
427
180
h/
120
160
0.25
100
Input resistance ()
Input resistance ()
110
0.30
90
0.35
80
0.45
70
140
h/
120
0.3
100
60
60
40
50
20
40 5
10
104 103 102 101
Conductivity, (mhos/m)
0 5
10
100
0.5
80
0.1
104 103 102 101
Conductivity, (mhos/m)
(a)
(b)
model are obtained from a reduced-order analytical approximation based on these values. A motivation for doing this is
demonstrated in Fig. 3 (17), where the real component in the
radial direction of the interfacially reflected field for a horizontal delta-function electric-current source is shown graphically as a function of the two applicable variables R
[2 (z z)2] and tan1[(z z)/ ]. The spatial variation of the fields is quite smooth, certainly less complex than
the analytical descriptions of Eq. (14) imply. Furthermore the
possibility is suggested that, if the fields are sampled over
a mesh, simple interpolation might provide accurate enough
values between these samples, so that no further rigorous
field computations are required in numerically modeling objects near an interface, as was shown feasible (18).
7
Ground screen
Re(Zin)()
d-layer T = 15,
= 0.01
Sea water T = 81,
=4
120 radials
240 radials
4
100
Figure 7. These plots of the input resistance of (a) a half-wave vertical dipole and
(b) a half-wave horizontal dipole as a
function of ground conductivity for various antenna heights compare the RCA
(solid line), the Sommerfeld theory (os),
and the compensation theorem (xs) for
the vertical antenna (10). The frequency
is 3.0 MHz, E 10, and the antenna radius is 5 104 wavelengths. The RCA
becomes exact for PEC and so is expected
to be most reliable for reasonably conductive grounds.
However, the fields for the two-sided problem are functions
of three variables and are less amenable to interpolation. But
they have been similarly treated using a procedure called
model-based parameter estimation (MBPE) (12). This involves using approximate, asymptotic formulas for the fields,
derived from the Sommerfeld integrals as the interpolating,
or fitting, functions. Matching the sum of the fitting models
to rigorously computed field values provides multiplying coefficients (the parameters) for the fitting models (the parameter-estimation step). As in the simple interpolative approach,
the fitting models are subsequently used in place of the Sommerfeld integrals in the numerical solution, where, for the
interface problem, the computetime saving is a factor of 100
1000.
A principle that needs to be kept in mind whenever comparing alternate approaches to a given problem is that of the
conservation of difficulty. Alternative formulations that lead
to integrals different from, but comparable to, the Sommerfeld integrals, may offer no computational advantage. Another principle is that of the information content of the phenomenon of interest. From an engineering viewpoint, only
some minimum of detail is needed to acceptably represent
that phenomenon, so that when there is the possibility of
choosing between mathematical rigor and computational
practicability, in this case between the Sommerfeld integrals
and a model-based procedure, the latter offers an attractive
option.
360 radials
3
0.1
10
100
d (meters)
Figure 8. These results for the input resistance of the SLT as a function of depth to a salt-water half-space show that between 240 and
360 radials are needed to limit the input-resistance change to less
than 10% (28). They are based on combining the RCA with Eqs. (19),
(21), and (24). Also, E,1r 15, 1 0.01, and E,2r 81, 1 4.
Modeling Finitely Conducting, Loaded, and Sheathed Wires
When modeling wire antennas, it is often desirable or necessary to include such effects as the distributed impedance due
to a wires finite conductivity, lumped loads when used as terminations or for matching, and dielectric sheaths used for insulation or for other purposes. The original integral equation
is simply modified by including a voltage-drop term, where
MODELING WIRE ANTENNAS
Figure 9. A hybrid of an integral equation and GTD model is used here to obtain
the input impedance of a monopole antenna at the center of an octagonal plate
(29). Agreement between the experimental results (open circles) and the computations (solid line) is within 10%.
64
36
60
32
56
28
52
24
Input resistance ()
Input resistance ()
428
48
44
40
34
20
16
12
8
32
28
24
0.2
0.4
0.6
d/
ES(s) represents the field scattered from the wire by the incident field, as follows (1,2):
1
s [Ei (s) + Es (s)] = s Ei (s)
I(s )G (s, s )ds
4 j C(r )
(15)
= ZL (s, s )I(s )
0.8
1.0
1.2
0.2
0.4
0.6
d/
0.8
1.0
1.2
tions, with the requirement that their effect is truly local and
that the load is itself nonradiating except for the current
flowing through it at its connection point si (2).
Loading Due to a Sheath. The effect of a sheathed wire is
handled by including an equivalent electric current Js radiating in an infinite medium as given by (2,20)
where ZL(s, s) is the load value affecting the field at s because
of a current at s, to allow for mutual and self-loading and
whose specific value is determined as described later. For selfloading, where ZL(s, s) ZL(s)(s s), the effect of loading
is simply to modify the system matrix (see below) along its
diagonal, whereas mutual loading more generally affects any
of the system-matrix coefficients. Mathematically, the effect
of a load is the same as that of the incident-field term, except
it is of opposite sign (unless it is an active load); the former
represents a voltage rise along the wire and the latter a
voltage drop.
where s is the permittivity of the sheath relative to the infinite medium where the wire is located. So as not to destroy
the one-dimensional nature of the wire integral equation, the
sheath is assumed to be electrically thin, and the total sheath
field is assumed to be dominated by its radial component,
which is due to the wire charge, so that
Distributed Loading Due to Finite Conductivity. For a wire of
radius ai, conductivity i, permeability i, and length i, the
impedance per segment i is approximately given by (2,19)
Ber(qi ) + jBei(qi )
ji i
ZL (si ) =
ai
2i Ber (qi ) + jBei (qi )
where I(s) is the spatial derivative of the current at s. Because the incident electric field in the radial direction can be
neglected relative to that due to I(s), the equivalent sheath
current becomes
(16)
where
qi =
i i ai
and Ber and Bei are Kelvin functions.
Lumped-Impedance Loading. A lumped impedance is generally described as a complex quantity given by
1
ZL (si ) = Ri + j Li
(17)
Ci
for a series R, L, C load located at observation point si. Other
combinations are readily introduced for more general situa-
Js (, s, ) = j (s 1)[Es (, s, ) + Ei (, s, )]
Es (s)
Js (, s, )
jI (s)
2 s
(s 1)
I (s)
2s
(18a)
(18b)
(18c)
for a(s) b(s), where b is the sheath radius and a is the
wire radius. As shown elsewhere (2), with some additional
approximations, the sheath is predominantly a self-term effect, whose axial field is given by
s i Ess (si )
jk2s (s 1)
b(si )
Ci Ln
2 s
a(si )
(18d)
where ks is the sheath wave number, to be included in the
thin-wire integral equation. An example of modeling the
sheath effect is shown in Fig. 4.
MODELING WIRE ANTENNAS
6.00e4
G, B(mhos)
4.00e4
2.00e4
G One-segment
B Source
0.00e+0
and
0
20
40
60
80
100
Number of segments (unknowns)
(a)
10000
Resistance, reactance ()
Ex,IG = ZsurfHy,IG
G Constant-width
B source
4.00e4
5000
Resistance
Reactance
0
5000
the angle of incidence at the specular reflection point and also
by the effective value of the surface impedance there. To begin with, note that the surface impedance of the imperfect
ground itself is given by
Zsurf = IG /IG = / E
(19a)
where is the wave impedance of the upper medium. Then
the SIA is expressed as
2.00e4
6.00e4
429
(19b)
Ey,IG = ZsurfHx,IG
which is valid when sin2 / E 1 is satisfied.
For a wire ground screen located at the interface and in
good electrical contact with the lower half-space so that it is
considered electrically in parallel with the ground, the modified surface impedance for the imperfect ground is given by
(23)
Zsurf =
Zsurf Zscreen
Zsurf + Zscreen
(20)
where Zscreen is the screen impedance. The screen impedance
for a radial screen at a distance from its center and comprised of N wires of radius a is given by
0
100
200
Number of unknowns
(b)
Figure 10. The input impedance (b) and admittance (a) are shown
for an L 2, center-fed dipole antenna as a function of the number
of unknowns, where only the center segment is excited. The results
imply that the impedance is not well-converged whereas the admittance behavior suggests otherwise. When multiple segments are excited to keep the physical width of the source region a constant value
of L/11, the additional results of (b) are obtained. Its clear that physical and numerical effects influence the apparent convergence rate of
the numerical solution.
Modeling Ground Screens
When used via MBPE as outlined previously, the Sommerfeld
treatment can be regarded as a numerically rigorous approach for modeling antenna-ground-screen combinations for
not too great an increase in computational cost compared with
modeling the same configuration in an infinite medium. However, the size in wavelengths of complex screen geometries
can be so large itself to make a more efficient model desirable
so long as it offers acceptable accuracy. Two approaches to
ground-screen modeling are outlined here, one using a surface-impedance approximation together with the RCA and the
other a technique known as the compensation theorem.
Surface-Impedance Approximation for Ground Screens. The
surface-impedance approximation (SIA) represents the wires
in an actual ground screen by approximating the change they
cause in the reflecting properties of the imperfect ground.
Thus the reflection coefficients in the RCA are determined by
Zscreen,rad =
j
Ln
N
Na
(21a)
and a corresponding formula for a parallel grid of wires whose
center spacing is d is
d
j d
Ln
(21b)
Zscreen,par =
2
2a
thus providing a means of obtaining a value for the relative
permittivity of the ground using Eq. (20) and thereby the
Fresnel reflection coefficient for the RCA. It is worth observing that the latter becomes zero when d 2a, that is, when
the area of the screen wires equals that of the surface they
cover, as previously discussed in connection with using wire
meshes as approximations to continuous surfaces.
Unless made of a square mesh, the impedance of a ground
screen varies with direction in the xy plane, that is, is anisotropic. Thus, implementing the RCA requires a further decomposition of the interactive field between source and observation segments that takes this anisotropy into account. The
electric field is first decomposed into components along the
principal directions of the screen wires relative to the specular point. For example, in the case of the parallel screen of
Eq. (21b), the component parallel to the screen wires reflects
from a medium whose effective permittivity, then is, given by
Zsurf + Zscreen
E = /Zsurf =
Zsurf Zscreen
j d
d
(22)
+ E 2 Ln 2a
=
d
j d
Ln
2
2a
430
MODELING WIRE ANTENNAS
Current
Charge
Figure 11. Current and charge snapshots at several time steps for a Guassian-pulse-excited dipole exhibit a variety
of important physics (31). Two outwardpropagating same-sign current pulses,
and opposite-sign charge pulses originate
from the source. A slight diminution of
the pulse amplitudes occurs because of radiation damping, and a further diminution occurs upon end reflection, a strong
source of radiation. Plots like these are remarkably insightful in revealing why antennas behave the way they do.
(a)
(b)
(c)
(d)
(e)
(f)
(g)
(h)
(i)
whereas the orthogonal component reflects from a medium of
permittivity E. Then each of these two fields is further decomposed into TE and TM components relative to the specular
plane and for which the appropriate Fresnel plane-wave reflection coefficients are computed and applied to determine
the total reflected field at the observation segment.
4
3
2
1
Current amplitude (mA/V)
(a)
4
3
2
1
(b)
4
3
2
1
2
1.5
0.5
Generator
Load
Distance, wavelengths measured from load
(c)
Figure 12. Comparison of current on a two-wire transmission line
obtained from the analytical solution (solid line) and from a thin-wire
integral equation (xs) for an (a) open circuit, (b) short circuit, and (c)
matched load (16). Although these results do not check the radiativepredictive properties of the numerical model, they do provide some
assurance that the model can handle a non-radiating problem.
Application to Ground-Screen Design. A ground screen is
used to improve antenna performance by simulating a perfect-image plane to the degree necessary for the particular
problem requirements. Any overdesign of the ground screen
which provides an improvement greater than that sought is
regarded as an inefficient allocation of resources. A reliable
and credible method for optimizing the ground-screen design
has a significant impact on the installation and operating
costs of large antenna systems.
Observe that simulating a perfect-image plane with the
ground screen implies that the ground screen should simulate
current-conduction paths similar to those that occur when the
same antenna is located over a perfect ground. If the current
flow in the perfect ground is confined along particular paths,
for example, then a ground screen with relatively few wires
is still highly efficient if the wires, which it has, follow the
same paths. Hence, a radial-wire ground screen is very effective for monopole antennas located at its center, because the
monopole produces entirely radial ground currents.
A method for determining an optimum ground-screen geometry for a given antenna system can be developed based on
the RCA as described below. The antenna is first modeled using a PEC ground from which the current-flow pattern in the
ground plane under the antenna is obtained. Next, the initial
ground-screen design is achieved with a wire geometry that
replicates the current-flow pattern in the PEC ground with
a wire spacing inversely proportional to the surface-current
density and chosen to produce a specified power loss per unit
area. To reduce the screen complexity, some relaxation in the
actual screen geometry from the ideal pattern should be anticipated, possibly limiting the final form to a combination of
rectangular cells and radial lines. To illustrate this approach,
the perfect ground currents for a simple antenna are presented in Fig. 5.
MODELING WIRE ANTENNAS
103
431
B(plasma)
1000
X
+B(plasma)
100
B0
(free space)
Reactance ()
G1B(Mhos)
104
+X
105
G(plasma)
10
106
G0 103
(free space)
107
0 20 40 60 80 100 120 140 160 180 200
Frequency (kHz)
1.0
1.5
2.0
f(MHz)
2.5
3.0
3.5
Figure 13. Comparison of results from the analytical formula of (35)
(lines) with a thin-wire IE for a short antenna in free space and in a
lossy plasma (xs) provides mutual validation for both approaches
(16). This is an important extension of the IE model because handling
lossy media is necessary for modeling antennas buried in the ground.
Using Sommerfeld Theory to Model Ground Screens. Although intended primarily for modeling antennas near an interface, the Sommerfeld approach, as included in the NEC, is
also efficient enough to include the ground-screen wires for
some of the simpler designs. An example demonstrating this
application is presented in Fig. 6 (25).
Using the Compensation Theorem. An alternative approach
to including the effects of ground screens and the interface
itself on antenna impedance is a technique widely used before
computer solutions became practical, but still effective for
some applications. This is the compensation theorem (26),
whereby the input impedance in an antenna located near an
interface is estimated from a reference solution for a PEC
half-space, as
Zin,IG = Zin,PG +
Iin,PGIin,PG
A
HPG (x, y) z EIG (x, y)dxdy
(23a)
where the subscripts PG and IG, as before, refer to quantities
associated with a perfect ground and imperfect ground, respectively. Upon using the surface-impedance approximation
of Eq. (19b) and further assuming that
Hx,IG = Hx,PG and Hy,IG = Hy,PG
(23b)
Figure 14. A comparison of the input impedance of a Vee-dipole antenna obtained from several different wire IE models shows that they
agree to an order of 30% except for the resistance at lower frequencies when decreases as f 2 (16).
then the IG input impedance is obtained entirely in terms
of the PG result. Some results comparing the compensation
theorem, the RCA, and the Sommerfeld integrals for modeling
half-wave vertical and horizontal dipoles are included in
Fig. 7.
Modeling Layered Grounds
The SIA can also be used as an approximation for modeling
layered grounds in place of the even more complicated Sommerfeld approach. A particular example of interest is that of
200
Resonant frequency (MHz)
0.5
29.6 in.
190
180
1 in.
170
160
150
Experimental
Numerical
140
130
120
20
40
60 80 100 120 140 160 180
Zigzag angle (deg)
Figure 15. The computed and measured resonance frequency of a
29.6-in zigzag dipole comprised of 1-in pieces of wire agree to within
23%, demonstrating the capability of the wire IE model to handle a
slow-wave structure (16).
432
MODELING WIRE ANTENNAS
12
Below ground
10
Above ground
X 103 ()
R ()
Figure 16. Although the thin-wire approximation limits how close a wire can
approach an interface, by using a small
radius of 108 wavelengths in free space,
the 0.1 wavelength wire can be brought
much closer than its length to an interface
of relative permittivity 16 (17). The
above-ground results approach an asymptotic value, whereas those below ground
continue to change. The smoothness of the
behavior shown here does not prove the
validity of the model, but does show that
the model, based on the MBPE form of the
Sommerfeld approach, exhibits appropriate behavior.
6
4
3
Above ground
2
1
Below ground
0
101
102
103
h/ o
antennas located near a body of water, fresh or saline, where
tidal effects vary the ground-water depth beneath the antenna with time of day. When an antennas input impedance
must be limited between narrowly defined values to maintain
matched conditions with the generator, it is necessary to determine the ground-screen parameters that make this possible. It is straightforward to obtain an approximate value for
the effective surface impedance (27) for a layer of thickness h
over a lower half-space from the expression
+
Zsurf = Zsurf E,1
E,2 +
j E,2 tan[k h E,1 ]
j E,1 tan[k h E,1 ]
(24)
where the subscript 1 applies to the upper layer and 2 to
the lower half-space and, as before, x is the permittivity relative to that of the uppermedium . Including this SIA with a
ground-screen approximation and the RCA provides a means
for modeling the combined effect of a layered ground and
ground screen. Use of the SIA for the U.S. Coast Guard Sectionalized Loran Transmitting (SLT) antenna (28) (see Fig.
22 for a computer plot of the SLT geometry) antenna is illustrated in Fig. 8.
104
105
106
101 102 103 104 105
h/ o
106
Hybrid Models
Unfortunately, a numerical model based on a single formulation is not equally applicable to all problems. Even though a
wire code models surfaces approximated as grids, there are
more efficient and accurate ways to handle PEC objects. It is
equally true that not all parts of a given problem are suitable
for modeling with a single formulation. Wires attached to
solid PEC objects or located near inhomogeneous dielectric
bodies are simple examples. For this reason, it is desirable to
combine two (or more) approaches to develop what is called a
hybrid model to handle such problems.
Considering that there are four kinds of numerical models
presently used in electromagnetics (EM), based on integral
equations, differential equations, modal expansions and the
geometrical theory of diffraction (GTD), formulated in either
the frequency domain or time domain, hybrid models have
been developed using various combinations of these (see
(29,30) for two examples). An especially useful hybrid model
involving wires is one that combines a wire IE with the GTD,
an example of which is shown in Fig. 9. The benefit of this
kind of hybrid model is that only the unknowns for the cur-
6
5
4
3
2
1
0
1
2
3
4
5
6
0.8
E ( )/E(a)
Antenna current (mA/V)
1.0
0.6
0.4
0.2
0.5
1.5
2
2.5
3
3.5
4
4.5
Distance from source in wavelengths
5.5
Figure 17. The real (x) and imaginary (o) components of the current
on a conical spiral antenna obtained from a Pocklington IE compared
with a Hallen IE (solid and dashed lines) (36) show agreement to
within about 10% (16). The difficulty of generalizing the Hallen IE to
arbitrary geometries restricts its application.
11
/a
Figure 18. The normalized, near, radial electric field of a circular
loop antenna obtained from the electric-field IE (os) (16) agrees
within a few percent, on a normalized basis, with analytical results
(solid line) (37).
MODELING WIRE ANTENNAS
433
25
Frequency = 25 MHz
7.5 ft
Field strength (dB)
Experimental
15
10
20
E-plane pattern
70.8 ft
10
0
164
Numerical
166
168
170
Frequency (MHz)
172
Figure 19. This comparison between the electric-field IE (o) and
measured data (solid line) of the frequency response of a circular loop
continuously loaded with capacitors demonstrates the possibility of
modeling a distributed capacitive load (16).
rents on the wire need to solved for, because the currents on
the plate, against which the monopole is driven, are accounted for via GTD diffraction coefficients.
Modeling in the Time Domain
Figure 20. Lumped loading is employed here to foreshorten the four
longest elements in a 19-element log-periodic antenna, with
agreement to 1 dB relative to the pattern maximum obtained between experimental measurements (lines) and numerical results
(solid circles) (16).
Equations analogous to those in the frequency domain can be
derived for many situations directly in the time domain. For
a wire in free space, an IE comparable to the Pocklington IE
can be developed as follows (31):
14
12
(25)
G(millimhos)
s s I(s , t )
s Ei (s, t) =
4 C(r )
R
t
c s R
I(s , t ) c
Q(s
+
,
t
)
ds
R2
s
R
t
I(s , t )
Q(s , t ) =
dt
s
10
8
6
4
2
0
where Q(s, t) is the charge density at space location s and time
t and the retarded time t R/c, where cis the speed of
light in the medium. A time-domain IE like that in Eq. (25)
is readily solved by time stepping whereby the solution is
developed as a function of space and time, as described later.
The Frequency Domain
A generic form of the frequency-domain IEs thus far considered can be written as (1,32)
(26a)
where L(s, s) is the integral operator, f(s) is the unknown
current, and g(s) is the known source or forcing function,
which, for our application, is sampled values of a specified
tangential electric field. The method of moments (MOM) is
an intuitively logical way of solving this operator equation,
whereby the unknown is expressed (or sampled) in terms of a
8
6
B(millimhos)
NUMERICAL IMPLEMENTATION VIA THE MOMENT METHOD
L(s, s ) f (s ) = g(s)
10
20
H-plane pattern
174
Field strength (dB)
Input impendance ()
20
4
2
0
2
4
0
4
6
Frequency (kL)
10
Figure 21. These plots compare the input admittance of a dipole
antenna driven by a tangential electric field (dashed line) compared
with one attached to a two-wire transmission line and excited at the
far end (solid line) (16). The upward frequency shift caused by the
transmission line is evidently caused by the capacitive-loading effect
of the antennatransmissionline junction.
434
MODELING WIRE ANTENNAS
10
Resistance
600 ft
A
(Feed point)
14
70
ft
1470 ft
Figure 22. The computer model of the SLT antenna, containing a
total of 237 wire segments demarcated by the arrows (28). At the time
this work was done, a 237-segment model was state-of-the-art, but
now this problem is considered small for desk-top computers.
Resistance or reactance ()
0
Reactance
10
20
30
90
set of basis or expansion functions f i(s) and unknown coefficients ai:
f (s ) =
Ns
ai f i (s )
(27a)
i=1
so that the operator equation is then written
L(s, s )
Ns
ai f i (s ) =
Ns
i=1
ai L(s, s ) f i (s ) = g(s)
(26b)
i=1
95
100
Frequency (kHz)
105
110
Figure 24. The modeled (os) SLT resistance results compare within
a few percent of the scale-model measurements (solid lines) while the
reactance values are shifted by about 2% in frequency (16) for 120
radials in the ground screen and E 15 and IG 2 102 S. Such
reactance shifts are fairly common in numerical modeling, and can
have a number of causes (34). Of course, besides limitations in the
numerical model, incomplete knowledge of the experimental conditions can also be a source of such differences.
where Ns is the number of spatial unknowns. Then upon sampling the operator Eq. (26b) with a set of testing or weight
functions
{w j (s)}, m = 1, . . ., M
R
10
X
17
16
15
14
13
12
11
10
9
(27c)
100
200
50
11
250
Xin (ohms)
ai w j (s), L(s, s f i (s ) = w j (s), g(s), m = 1, . . ., M
i=1
Rin (ohms)
Input impedance ()
Ns
(27b)
150
100
50
Perfect ground
90
180
270
Number of 8 AWG radials
360
Figure 23. The influence of the number of wires in a radial-wire
ground screen on the input impedance of the SLT antenna as modeled
using the RCA and ground-screen approximation shows that the reactance is much less affected by a changing ground conductivity than
is the resistance (16). For the dashed curves, E 15 and IG 102
S and for the solid curves E 15 and IG 103 S.
50
100
0.1
0.2
d/ 0
0.3
150
0.1
0.2
0.3
d/ 0
Figure 25. A numerical model like NEC makes it possible to evaluate the influence of parameters such as the length of a ground stake
against which a monopole antenna is driven, as illustrated here (24),
computed using the rigorous Sommerfeld approach for a wire radius
of 2.5 106 (solid line), 2.5 105 (dotted line), and 2.5 104
(dashed line), respectively, for a ground of E 16 j16. Results like
this make it feasible to determine what kind of ground stake is
needed to stabilize the input impedance.
MODELING WIRE ANTENNAS
The admittance matrix is a numerical representation of the
complete EM properties of the object being modeled, within
the approximations made in its computation, and as such it
has some interesting properties. First, it provides a solution
for arbitrary excitation. Second, it can be stored for subsequent reuse. Perhaps most interesting, in a very real sense,
it possesses the properties of a hologram, as seen by writing
the field radiated by the object as
Monopole
Feedpoint
L
a
2 Wire
1 Wire
3 Wire
(a) Elevation pattern
Er =
Dipole
(b) Azimuth pattern
Ns
Yij E ij
(b)
Dipole
(c)
Figure 26. Sparse ground screens like that shown here (a) cannot be
modeled using the simple ground-screen formulas, but can be handled
using the interpolated Sommerfeld approach (38), to obtain the elevation (b) and azimuth (c) patterns, with the number of ground wires
as a parameter. The frequency is 10 MHz and the ground has E 4
and IG 103 S.
where the signifies what is called an inner product (the
inner product of two functions p(r) and q(r) over a surface S
defined as s p(r)q(r)d2s).
Now the original operator equation is written in a discretized, sampled approximation as
Zji ai = b j , j = 1, . . ., M
(30b)
where the exciting field Ei and the observation field Eo are
tangential projections onto the object of an arbitrary incident
field and the field from a point source located at the observation point.
Much work has been done to identify the best expansion
and testing functions. A variety of combinations are described
by Poggio and Miller (1). Those in NEC use what is called
sub-sectional collocation, wherein the wire is divided into
Ns segments with
f i (s ) = Ai + Bi sin[k(s si )] + Ci {cos[k(s si )] 1},
(31a)
si i /2 s si + i /2, i = 1, . . ., Ns
and
w j (s) = (s s j ), j = 1, . . ., Ns
(31b)
where i is the length of segment i and (s sj) is a deltafunction, thus producing point sampling of the tangential
electric field. Of the 3Ns unknowns in Eq. (31), 2Ns are eliminated by enforcing current and current-slope (charge) conti-
(28a)
4
where
Zji = w j (s), L(s, s ) f i (s )
(28b)
b j = w j (s), g(s)
IZ Z/Z %
i=1
and
(30a)
j=1
Eio = s i Eo (si ) and E ij = s j Ei (s j )
1 Wire
Ns
Eio
where the first sum yields a solution for the current distribution and the second sums the field at a specified observation
point caused by that current. Also
(a)
3 Wire
Ns
i=1
Continuous
wire
2 Wire
435
Finally, the coefficients that quantify the numerical solution
for the current distribution are obtained as
1
ai =
M
Yij b j , i = 1, . . ., Ns
(29)
j=1
where Zij is known as the impedance matrix and Yij is the
admittance matrix, its inverse. Equations (28a) and (29) are
written more compactly in symbolic form as
ZA =BA = YB
10
Separation (ft)
Figure 27. Comparison of the TCA (dashed line) with the Sommerfeld theory (solid line) demonstrates the formers potential applicability to subsurface probing (14) for a ground of E 9 and IG 103 S
and with both wires 0.5 wavelengths long and for a frequency of 10
MHz. The goal here is to detect a buried object from its effect on the
impedance of an above-ground probing antenna whose effect is plotted as a normalized difference between its value with and without
the object.
436
MODELING WIRE ANTENNAS
Z Percent change in real part of zin
Upon sampling the field with the testing functions
10
{Wkm (s, t)} = {wk (s)vm (t)}, k = 1, . . ., Ns , m = 1, . . ., Nt
9
8
(33b)
the operator equation is written in a discretized, sampled approximation as
7
6
Nt
Ns
Aij vm (t), wk (s), O(s, t ; s , t )Pij (s , t )
i=1 j=1
= vm (t), wk (s), G(s, t) (32c)
3
2
Crossed
Parallel
1
0
10
20
30
40
50
(deg)
60
70
80
90
Figure 28. Here the TCA is used to determine the variation in the
input impedance of an above-ground, horizontal antenna (L 0.457
wavelengths) as it rotates about a vertical axis above a horizontal,
buried wire (L 0.152 wavelengths) (14) whose center is on the same
vertical axis with a ground of E 9 and IG 0 for a frequency of 1
MHz. When the wires are orthogonal, there is no interaction while
the maximum effect is produced when they are parallel.
where the inner product now involves a space integration and
also a time integration.
Employing subsectional collocation in both space and time,
the unknown current is written as
Pij (s , t ) = A1,i j + A2,i j (s si ) + A3,i j (s si )2 + A4,i j (t t j )
+ A5,i j (t t j )2 + A6,i j (s si )(t t j )
+ A7,i j (s si )2 (t t j ) + A8,i j (s si )2 (t t j )2
+ A9,i j (s si )2 (t t j )2 ,
si i /2 s < si + i /2, i = 1, . . ., Ns ;
t j /2 t t j + /2, j = 1, . . ., Nt
nuity at the junctions between segments, so that the final
number is Ns.
The behavior of a numerical model, as Ns is increased, is
often examined in what is called a convergence test. Although a convergence test is reassuring if the numerical results approach an asymptote, it is not guaranteed that convergence takes place, or if it does, that the convergence is to
the correct answer. In addition, depending on how results of
the convergence test are examined, quite different conclusions
might be reached, as demonstrated in Fig. 10.
(34a)
and the constant time step is , whereas while the testing
function is given by
Wkm (s, t) = (s s j )(t t j )
(34b)
so that the tangential field is point-sampled in both space and
time, and the time solution is developed from time stepping.
There are nine unknown coefficients associated with each
space-time step, eight of which are eliminated in a fashion
similar to that described for NEC, by matching the current to
The Time Domain
= 0 (Horizontal)
;;;;
A similar approach is used to solve the time-domain version
of the EFIE, Eq. (25), which is first written in operator form
as (31,33)
O(s, t; s , t )F (s , t ) = G(s, t)
30
(32a)
60
where t is used here to denote the retarded time, that is t
t R/c. Then we might represent the space and time dependence of F(s, t) as
F(s , t ) =
Nt
Ns
Aij Pij (s , t )
i=1 j=1
(33a)
90
30 ft
where the Aij are space-time samples of the unknown (i is the
space index and j the time index) of which there are Ns and
Nt space and time samples, respectively, so that Eq. (27) becomes
Nt
Ns
i=1 j=1
Aij O(s, t; s , t )Pij (s , t ) = G(s, t)
(32b)
=0
r= 9
50 ft Rotated antenna
Figure 29. The TCA is used here to determine the above-ground radiation pattern of a dipole antenna as it rotates in a vertical plane
(14). When the antenna is vertical or horizontal, the above-ground
patterns are symmetric about a surface normal, but the patterns are
otherwise slightly skewed.
MODELING WIRE ANTENNAS
Radiated fields times distance (V)
Source current (mA)
3
2
1
0
1
2
0
10
20
Time (ns)
437
0.2
0.1
Figure 30. The source current (a) and
the broadside electric field (b) for a dipole
excited at its center by a Gaussian voltage
as obtained using TWTD are quite similar, with the latter demonstrating the radiation for these conditions is first produced as the voltage is applied and then
subsequently as the currentcharge wave
reflects from the wire ends (31).
0.1
30
10
(a)
20
Time (ns)
30
(b)
its neighbors in the adjacent eight space-time samples (31,33)
at i 1, j; i, j 1; i j, j i. An alternative approach is to
write the basis function as Pij(s, t) Ri(s)Sj(t), which then
involves five unknowns (by setting the constant term in Sj(t)
to unity) and four are eliminated by matching at i 1, j; i, j
1. Finally, the space-time current samples are written in
the form
i
s
Iij = I(si , t j ) = Yik Ekt j = Yik (Ekj
+ Ekj
)
Ns
t
+
Zii Ii ,m f (i,i )
= Yik Ekj
(35)
i =1
where Eijt total Eij, Esij scattered Eij.
For a straight wire, f(i,i) i i, but it is generally a
more complicated function of object geometry. Also, Yik represents the inverse of the time-independent matrix that accounts for field-current interactions in the time step j, and
Zii, accounts for the fields produced over the object from currents that occurred at earlier times. Developed in this way,
the time-stepping model is called implicit, whereas if no interactions are allowed within time-step j, the model is explicit. An example of a time-domain solution for a straight
wire excited by a Gaussian voltage pulse, V V0exp(t2a2), is
given in Fig. 11, where the spatial distribution of current and
charge are plotted at several instants of time (31).
Some Observations Concerning Numerical Modeling
Numerical models based on integral equations, like those presented in previously and solved by the moment method, as
described previously, yield the source distribution, typically
current for a wire antenna, for an arbitrary right-hand-side or
incident field. All other observables of interest to the antenna
designer or user are subsequently determined from this solution.
The code, or model, user usually devotes most attention to
input impedance, radiation efficiency, and the directive gain
or other antenna properties judged most important for the
intended application. The model developer, on the other hand,
is usually at least as interested in the current distribution,
the convergent behavior of the numerical result revealed by
its dependence on Ns, the near fields, and so on. For both the
code user and code developer, validation of the code and the
results obtained from it are crucially important. Validation
ultimately consumes more time in computational electromag-
1
0.4
Resistance (k)
0.8
Resistance (k)
Experiment
Calculated
Experiment
Calculated
0.6
0.4
0.2
0.2
0.2
2
1
Frequency (GHz)
(a)
0.4
1
2
Frequency (GHz)
(b)
Figure 31. This comparison of the measured and computed, using TWTD, input
impedance of a Gaussian-pulse excited
antenna shows the two results are graphically indistinguishable for about the first
8 resonances (31), providing mutual validation, as well as demonstrating the
broadband capability of a time-domain
computation and measurement.
438
MODELING WIRE ANTENNAS
netics than any other single activity, but a discussion of it
goes beyond the scope of this article. Some examples selected
to exhibit the relative validity of numerical results are presented later, and further information about validation is
found elsewhere (34).
The results presented later are selected from two perspectives. Aside from those chosen for explicit validation, one is to
demonstrate some of the kinds of capabilities that wire models provide, primarily through results generated from NEC
and TWTD. The other is to show the kinds of agreement expected between computation and measurement or other independent results.
SOME REPRESENTATIVE RESULTS
OF WIRE-ANTENNA MODELING
Validating the Numerical Model
There are few analytical solutions for wire objects that validate a computer model, one exception being the well-known
two-wire transmission line. A comparison of the results provided by an earlier version of NEC with those from analytical
formulas is shown in Fig. 12 (16). For small antennas, there
are also simple, analytical formulas for the input admittance
(35), one of which used for an antenna in a plasma to check
the results from NEC for a wire antenna in a lossy medium
is shown in Fig. 13. In both cases, the IE model agrees with
the analytical results to within a few percent.
Several different numerical models were used to determine
the frequency response of a Vee-dipole, an antenna with Vshaped end loads connected by a short, straight center section
where the antenna is excited (16). These model results, compared with each other over the frequency of the first resonance in Fig. 14, generally agree within 10%.
To be useful, wire codes must also model objects with fine
detail, such as the zigzag shown in Fig. 15 (16). The first resonance of a constant-length zigzag is shown as a function of
;;
Current (mA)
8
12
Time (ns)
16
20
Figure 32. This comparison of the measured center current on a
Vee-dipole due to a broadside-incident Gaussian plane wave with a
TWTD prediction provides a mutual validation for both (31). The time
displacement is evidently due to a slight time-scale difference.
the angle between the adjacent one-inch segments, for which
the measured and computed resonances agree within 5%.
A different kind of validation test is demonstrated in Fig.
16 where the input impedance of a dipole is presented as a
function of distance from the interface between free space and
a dielectric half-space (17).
Infinite-Medium Applications
Wire codes are typically used to determine input impedance,
current distribution and radiation patterns of candidate antennas, but are also used for a variety of other applications.
Some examples of these applications are shown here. In Fig.
17, the current distributions on a log-conical spiral antenna
obtained from two different codes are compared. They agree
within 10% relative to the peak values except right at the
source (16). The radial component of the near field of a loop
antenna is compared with independent data in Fig. 18. In Fig.
19, the input impedance of a loaded loop antenna is compared
with measurement (16). A comparison of the measured and
computed radiation patterns of a log-periodic dipole array antenna with foreshortened elements to reduce its physical size
is presented in Fig. 20 (16). Finally, the frequency-dependence of the input impedance of a dipole antenna is illustrated in Fig. 21 for what are termed implicit and explicit
source models (16). All of these results demonstrate the applicability of wire codes to various kinds of antennas for obtaining various observables.
Interface Applications
Because the segmentation or description of a continuous
wire object using a piecewise linear approximation is not easily visualizable, a computer model of the Sectionalized-Loran
Transmitting (SLT where LORAN stands for long range navigation) antenna is presented in Fig. 22 (28). The segments
that comprise the numerical model are shown by arrows,
where the reference direction for the current is in the direction to which the arrow points. This model contains 237 segments, or unknowns, a rather small number by todays standards where problems with thousands of unknowns are
routinely modeled. This particular antenna was modeled over
an imperfect ground with a radial-wire ground screen centered at the antenna feedpoint, using the combined RCAground-screen approach outlined previously.
Because a ground screen for an antenna this large requires
a large amount of copper wire, it is desirable to use no more
than necessary to achieve the desired performance characteristics. Therefore, parametric experiments were performed to
determine the SLTs impedance sensitivity to its ground
screen and the ground parameters, an example of which is
shown in Fig. 23 (28).
One difficulty with modeling antennas located on or near
the ground is validating the results of computer experiments,
such as illustrated in Fig. 23. Fortunately, the U.S. Coast
Guard, which operates the LORAN system obtained experimental measurements using a 1/25th scale-model antenna
with 120 radials and with IG 15, IG 2 102 S, for which
some results are compared with computer predictions in Fig.
24 (28).
Two rather different kinds of interface modeling are presented next in Figs. 25 and 26. The effect on the input impedance of a vertical monopole antenna excited just above the
MODELING WIRE ANTENNAS
439
0.12
Radiated field R V
Source current (mA)
1.8
1.2
0.6
0.06
0.0
0.06
0.0
0.2
4
Time (ns)
0.12
(a)
4
Time (ns)
(b)
E R magnitude
Vs
Hz
6.0
4.5
3.0
Figure 33. Nonlinear problems can be
modeled in the time domain, as this result
demonstrates. A dipole antenna continuously loaded with diodes that permit it to
conduct in only one direction has the feedpoint current in (a), and broadside radiated field in (b) when excited by a
Gaussian pulse (31). The spectrum of the
radiated field is as shown in (c).
1.5
0.0
3
1
2
Frequency (GHz)
(c)
15
Relative magnitude
Scattered field r (V)
10
2
0
0.4
Time ( s)
(a)
0.8
1.2
20
40
60
Frequency (MHz)
(b)
80
Figure 34. Time varying problems are
also well-suited to time-domain modeling
as demonstrated here (31). A 16 MHz,
broadside-incident, plane wave illuminates a half-wave diople having a center
load whose resistance varies sinusoidally
at 4 MHz, producing the broadside scattered field in (a) whose spectrum is shown
in (b). Dynamically varying the reflectivity of a scatterer can be used to change
the scattered-field spectrum from what it
otherwise would be.
440
MODELING WIRE ANTENNAS
interface on the length of the buried ground stake against
which it is driven is illustrated in Fig. 25 (24). The dependence of the radiation pattern of a monopole antenna driven
against a ground screen which consists of only a few radial
wires just above the surface of the ground is demonstrated in
Fig. 26. These kinds of problems are ideally suited for computer analysis and parametric optimization, because developing such data experimentally is generally impractical.
The modeling of buried objects and antennas is illustrated
in Figs. 2729. The normalized impedance change of a horizontal dipole in the upper air medium plotted versus its distance from a vertical plane containing a buried, horizontal
wire to which it is parallel is shown in Fig. 27. These results
were obtained by the Sommerfeld theory outlined previously
and the Transmission-Coefficient Approximation (TCA), an
extension of the RCA described previously to the computation
of fields on the other side of an interface from the source (13).
In Fig. 28, the effect on its input impedance of rotating an
upper-medium, horizontal-dipole antenna relative to a buried
wire is demonstrated, also obtained by the TCA. Finally, the
radiation pattern of a buried antenna is presented for several
different rotational angles relative to the interface normal in
Fig. 29.
Time-Domain Applications
There are several advantages in performing time-domain
modeling. One advantage is that wideband data are available
from one model computation as opposed to the many frequency samples required to obtain equivalent data from a frequency-domain code. Another advantage is a more straightforward approach to modeling impedance nonlinearities in
the time domain. A third advantage is that time-domain models also handle time variations in load impedances. An example of the first is illustrated in Fig. 30 (31), where the transient feedpoint current and broadside radiated field are
shown for a Gaussian-pulse-excited dipole. Upon Fourier
transforming the feedpoint current, the input impedance is
obtained across a wide bandwidth, as presented in Fig. 31
(31). The measured current at the center of a Vee-dipole antenna excited by a Gaussian plane-wave incident pulse is presented in Fig. 32 and compared with a TWTD computation,
where the amplitudes of the two results agree to within a few
percent but a slight difference in the time axis is seen.
Use of a time-domain model for nonlinear loading is demonstrated in Fig. 33 where the input current, broadside radiated field and the radiated-field spectrum are presented, as
computed from TWTD. The radiated field has two oppositesign pulses, caused by the initial turn on of the drive voltage
and the stopping of the current-charge pulses as their outward propagation is stopped. When a dipole is illuminated by
a time-harmonic plane wave incident from broadside while its
center load varies sinusoidally in value, the results of Fig. 34
are obtained. Interaction of the incident field and the timevarying load causes intermodulation products that produce
upper and lower sidebands in the scattered field.
BIBLIOGRAPHY
1. A. J. Poggio and E. K. Miller, Low-frequency analytical and numerical methods for antennas. In Y. T. Lo and S. W. Lee, eds.,
Antenna Handbook, New York: Van Nostrand Reinhold, 1988, pp.
3-1 to 3-98.
2. G. J. Burke, Numerical Electromagnetics CodeNEC-4 Method
of Moments, Part II: Program DescriptionTheory. Lawrence Livermore National Laboratory, Rept. USRL-MA-109338, January
1992.
3. J. A. Landt, E. K. Miller, and M. L. Van Blaricum, WTMBA,LL1B: A Computer Program for the Time Domain Analysis
of Thin-Wire Structures, Livermore, CA: Lawrence Livermore
Laboratory, UCRL-52172, 1976.
4. A. J. Poggio and E. K. Miller, Integral-equation solutions of
three-dimensional scattering problems. In Computer Techniques
for Electromagnetics, New York: Pergamon Press, 1973, pp.
159264.
5. R. M. Bevensee, Design Considerations for Parasitic Screen Antennas. Lawrence Livermore Laboratory Report, Rept. UCID16657, 1975.
6. E. K. Miller, A selective survey of computational electromagnetics, invited tutorial-review paper. IEEE Trans. Antennas Propag.
AP-36:, 12811305, 1988.
7. E. S. Selden and G. J. Burke, DLGN-38 Final Report. MBAssociates, San Ramon, CA, Tech. Rept. MB-R-75/16, 1975.
8. D. L. Jaggard, On bounding the equivalent radius. IEEE Trans.
Antennas Propag., AP-28: 384388, 1980.
9. O. M. Salati, Antenna chart for system designers. Electron. Eng.,
January 1968.
10. E. K. Miller et al., Analysis of wire antennas in the presence of a
conducting half-space. Part I: The vertical antenna in free space.
Canadian J. Physics, 50: 879888; Part II: The horizontal antenna in free space, Canadian J. Physics, 50: 26142627, 1971.
11. A. Sommerfeld, Partial Differential Equations in Physics. New
York: Academic Press, 1964.
12. G. J. Burke and E. K. Miller, Modeling antennas near to and
penetrating a lossy interface. IEEE Trans. Antennas Propag. AP32: 10401049, 1984.
13. A. Banos, Dipole Radiation in the Presence of a Conducting HalfSpace. New York: Pergamon Press, 1966.
14. E. K. Miller and F. J. Deadrick, Analysis of Wire Antennas in
the Presence of a Conducting Half Space: Part IIIThe Buried
Antenna. Lawrence Livermore Laboratory, Rept. UCRL-52228,
January 1977.
15. I. V. Lindell, E. Alanen, and K. Mannersalo, Exact image theory
for computation of antennas above the ground. IEEE Trans. Antennas Propag. AP-33: 937945, 1985.
16. E. K. Miller, Wires and wire-grid models for radiation and scattering including ground effects. In Theoretical Methods for Determining the Interaction of Electromagnetic Waves with Structures, Rockville, MD: Sijthoff and Noordhoff, 1981, pp. 5795.
17. G. J. Burke, E. K. Miller, J. N. Brittingham, D. L. Lager, and R.
J. Lytle, Computer modeling of antennas near the ground. Electromagnetics, 1: 2949, 1981.
18. E. K. Miller, J. N. Brittingham, and J. T. Okada, Bivariate-interpolation approach for efficiently and accurately modeling antennas near a half-space. Electronics Letters, 13: 690691, 1977.
19. E. S. Cassidy and J. Fainberg, Back scattering cross sections of
cylindrical wires of finite conductivity. IRE Trans. Antennas Propag., AP-8: 1, 1960.
20. J. H. Richmond, Radiation and Scattering by Thin-Wire Structures
in the Complex-Frequency Domain, Columbus, OH: Ohio State U.,
Electro Science Lab., Rept. 2902-10, 1973.
21. R. W. P. King and G. S. Smith, Antennas in Matter. Cambridge,
MA: The MIT Press, 1981.
MODEL REFERENCE ADAPTIVE CONTROL
22. R. W. P. King et al., The insulated monopole: Admittance and
junction effects. IEEE Trans. Antennas Propag., AP-23: pp. 172
177, 1975
23. J. R. Wait, Characteristics of antennas over lossy earth. In R. E.
Collin and F. J. Zucker, eds., Antenna Theory. New York:
McGrawHill, 1969, pp. 286437.
24. E. K. Miller, Numerical modeling techniques for antennas. NATO
AGARD Lecture Series No. 131, NASA, Langley Field, VA, 1983,
pp. 7-1 to 7-29.
25. G. J. Burke, Recent advances to NEC: Applications and validation. NATO AGARD Lecture Series No. 165, NASA, Langley Field,
VA, 1989, pp. 3-1 to 3-35.
26. J. R. Wait and W. A. Pope, The characteristics of a vertical antenna with a radial conductor ground system. Appl. Sci. Res., Sec.
B, 4: 177195, 1954.
27. J. R. Wait, Electromagnetic Waves in Stratified Media. New York:
Pergamon Press, MacMillan Co., 1962.
28. E. K. Miller and F. J. Deadrick, Computer Evaluation of LORANC Antennas, Livermore, CA: Lawrence Livermore National Laboratory, Rept. UCRL-51464, 1973.
29. G. A. Thiele and P. Tulyathan, A hybrid technique for combining
moment methods with the geometrical theory of diffraction. IEEE
Trans. Antennas Propag. AP-26: 587592, 1978.
30. X. Yuan, R. D. Lynch, and J. W. Strohbehn, Coupling of finite
element and moment methods for electromagnetic scattering
from inhomogeneous objects. IEEE Trans. Antennas Propag., AP38: 386393, 1990.
31. E. K. Miller and J. A. Landt, Direct time-domain techniques for
transient radiation and scattering from wires, invited paper. In
Proc. IEEE, 68: 13961423, 1980.
32. R. F. Harrington, Field Computation by Moment Methods. New
York: MacMillan, 1968.
33. R. G. Martin, A. Salinas, and A. R. Bretones, Time-domain integral equations methods for transient analysis. IEEE Antennas
Propag. Mag, 34(3): 1522, 1992.
34. E. K. Miller, Characterization, comparison, and validation of electromagnetic modeling software. ACES J., special issue on Electromagnetics Computer Code Validation, 4: 824, 1989.
35. K. G. Balmain, Dipole admittance for magnetoplasma diagnostics. IEEE Trans. Antennas Propag., AP-17: 389392, 1969.
36. Y. S. Yeh and K. K. Mei, Theory of conical equiangular spiral
antennas: Part I. Numerical techniques. IEEE Antennas Propag.,
AP-15: 634639, 1967.
37. R. L. Fante, J. J. Otazo, and J. T. Mayhan, The near field of the
loop antenna. Radio Sci., 4: 697, 1969.
38. E. K. Miller, J. N. Brittingham, and J. T. Okada, Explicit modeling of antennas with sparse ground screens. Electron. Lett., 14:
627629, 1978.
EDMUND K. MILLER
441
REFRACTION AND ATTENUATION IN THE TROPOSPHERE
379
REFRACTION AND ATTENUATION
IN THE TROPOSPHERE
Electromagnetic energy, often referred to in communications
as the signal and traced in diagrams as a ray, travels at the
speed of light, c, when in a vacuum, whether in the form of
X rays, visible light, or a radio wave. The value of c is 3
108 m/s. When the signal passes through a medium that is
not a vacuum, it travels more slowly. The ratio between c and
the speed in the medium, v, is the refractive index of the medium, denoted by n c/v. The refractive index of a medium
does not just affect the speed of electromagnetic energy flowing through it; it can also cause the magnitude of the signal
to fluctuate over time periods on the order of a second or
longer due to variations in the refractive index within the medium. These signal fluctuations, called scintillations, are usually small and approximately symmetrical about the average
level of the signal, although this is not always so in severe
scintillations. Where two dissimilar media form a common
boundary, the different refractive indices of the media will
cause the signal direction to diverge from its path if the incident direction is not normal to the boundary. This is called
refraction, and the incident and refracted directions either
side of the boundary are given by Snells law (e.g., Ref. 1).
The average refractive index of a medium is called the bulk
refractive index, and it affects the average direction of the
electromagnetic energy. Most media are not completely homogeneous, however: small volumes of different refractive indices will exist within the larger scale of the medium. These
variations, which can exist over dimensions as small as a few
centimeters in the troposphere, are termed the small-scale refractive index variations of the medium. They are like bubbles
of air in a glass of soda. In the troposphere, the mixing of the
small-scale bubbles within the larger scale of the medium
can be caused by convection currents, due to heating of the
air by the ground, or by turbulent effects in wind shear aloft.
These variations give rise to the scintillations noted above.
In addition to changing the direction of travel of an electromagnetic signal and/or causing scintillations, a medium can
both extract energy from a signal passing through it and alter
the orientation sense of the electric vector (the polarization)
of the signal as it passes through. The loss of energy is called
attenuation, and the change in polarization sense is generally
referred to as depolarization. In general, the longer the path
through a medium is, the greater will be its effect on the electromagnetic signal passing through it. For this reason, paths
through the atmosphere at low elevation angles will have
greater ray bending, scintillation, attenuation, and depolarization than paths at high elevation angles.
REFRACTIVE EFFECTS
Bulk Refractive Effects
In still air, the atmosphere assumes a spherically stratified
form with layers of generally decreasing refractive index lying
one above the other. The refractive index of a given portion of
the atmosphere will therefore vary far more quickly with
change in height than with change in horizontal position. For
J. Webster (ed.), Wiley Encyclopedia of Electrical and Electronics Engineering. Copyright # 1999 John Wiley & Sons, Inc.
380
REFRACTION AND ATTENUATION IN THE TROPOSPHERE
The minus sign indicates that the refractive index decreases
with increasing altitude. For terrestrial radio systems, the elevation angle will be almost zero, and, since the refractive
index is almost unity, Eq. (1) will reduce to the approximate
form
Apparent direction
to satellite
Real direction
to satellite
1
dn
=
r
dh
Horizontal
If the radius of the earth is a, the difference in curvature
between the ray path and the curvature of the earth along
the terrestrial link is given by
a (km)
(2)
1 1
1
dn
1
1
= +
=
=
a
r
a dh
ka
Re
Center of earth
(a)
(3)
Here Re is the effective radius of the earth given by the product ka, where k is the effective earth radius factor. The effective radius of the earth is often used in ray-tracing diagrams
of the vertical profiles of terrestrial radio systems, because it
allows for the average ray bending along the path [see Fig.
1(b,c)]. Ray bending on satellite-to-ground paths may cause
signal loss due to the spreading of the antenna beam. This
effect, and that of elevation-angle-of-arrival variations, are
negligible at elevation angles above about 3 and will only be
significant for large antennas (diameter 100 wavelengths).
The atmosphere has a dry and a wet refractive index component, the former due to the gases in the atmosphere (principally nitrogen and oxygen), and the latter due to the water
vapor that is taken up into the atmosphere. Combining the
two components together into a composite refractive index n
gives
a (km)
C
(b)
ka (km)
n1=
77.6
e
P + 4810
106
T
T
(4)
where
C
(c)
Figure 1. (a) Apparent and real directions to an earth satellite; illustration of methods of ray tracing: (b) real radius and a curved ray
path, and (c) effective radius and a geometric ray path.
this reason, an electromagnetic signal (a ray) leaving the surface of the earth at an elevation angle of 90 will bend away
from the normal to the surface [see Fig. 1(a)]. The curvature
will be contained in one plane, the projection of which onto
the surface of the earth is the azimuth direction. If r is the
radius of curvature of the ray (2), then
cos dn
1
=
r
n dh
P atmospheric pressure (hPa)
T absolute temperature (K)
e water vapor pressure (hPa)
The hectopascal (hPa) is the same as a millibar.
The water vapor pressure e can be related to the relative
humidity by (3)
e = Hes /100
and
es = a exp
(1)
where
angle of the ray path to the horizontal at the point
considered
n refractive index of the atmosphere
dn/dh vertical gradient of refractive index at a height h
above the earths surface
(5)
bt
t+c
where
H relative humidity (%)
t temperature (C)
es saturation vapor pressure (hPa) at the temperature t
and the coefficients a, b, and c are as follows:
(6)
REFRACTION AND ATTENUATION IN THE TROPOSPHERE
For water:
a = 6.1121
b = 17.502
c = 240.97
For ice:
a = 6.1115
b = 22.452
c = 272.55
The vapor pressure e is related to the water vapor density
(g/m3) by (3)
e=
T
216.7
(hPa)
(7)
The refractive index of air is so close to unity that Eq. (4) is
often rewritten as
n = 1 + 106N
(8)
where N is the refractivity of the atmosphere and is given by
(77.6/T)(P 4810e/T), with the same units as in Eq. (4).
Refractivity Profile. The refractivity profile is described by
a mean refractive index gradient, which generally decreases
exponentially with increasing altitude. A reference standard
atmosphere for calculating the refractive index as a function
of height, n(h), is given by (3)
n(h) = 1 + 106 N0 exp(h/h0 )
(9)
where
h height above mean sea level (km)
h0 reference altitude 7.35 km
N0 average value of atmospheric refractivity at mean sea
level 315
Rewriting Eq. (9) in terms of the refractivity, we have
N(h) = N0 exp(h/h0 )
The rate of change of the refractivity profile with height,
dN/dh, can be related to the parameter k used earlier. Sample values are given below (4):
dN/dh (N/km)
157
0
40
157
200
1
3.65
Refractive Ducts. Ducting is defined as occurring when a
radio ray originating close to the earths surface is sufficiently
refracted that it is either bent back towards the earths surface or travels along a path that is parallel to the earths surface. Ducts are formed when layers of air having a relatively
high refractive index are above layers with a relatively low
refractive index. This inversion of the normal exponentially
decaying refractive index profile with height is rare and leads
to anomalous propagation conditions. Anomalous propagation, and the enhanced signal levels it causes at unlikely distances from the transmitter, is key to the estimation of potential interference between terrestrial systems caused by ducts.
A signal will become trapped in a duct if the initial direction
of the transmission (the angle of incidence) is very close to
that of the horizontal refractive index boundaries of the duct.
Typically, the angle of incidence of the signal, as it meets the
boundary between the inversion layer and the atmosphere
above, has to be smaller for thinner ducts. Typical values for
a refractivity gradient of 300 N/km are (2):
Critical Angle of Incidence
(deg)
Duct Thickness
(m)
0.1
0.14
0.22
10
20
50
Frequency also plays a part in the ducting. The higher the
frequency, the smaller the duct thickness consistent with
trapping. For the same 300 N/km profile assumed above,
the minimum trapping frequency for a surface duct increases
approximately from 2 GHz to 2.5 GHz to 4 GHz to 10 GHz to
25 GHz as the duct thickness changes from 40 m to 30 m to
20 m to 10 m, to 5 m, respectively (2). The necessary condition
for ducting to occur at a given elevation angle is given by (5)
< c
From the table above, it is clear that, when dN/dh is less
than 157, the radius of curvature of the earth is greater
than the radius of curvature of a signal that is initially propagating horizontal to the surface of the Earth. The signal will
therefore propagate beyond what would normally be the local
horizon. This is a form of anomalous propagation, and the
mechanism is referred to as ducting.
(11)
where is the elevation angle and c is the critical elevation
angle above which ducting will not occur. The critical angle is
given by
c = (2| M|)1/2
(10)
381
(mrad)
(12)
where M is the modified refractivity (and M N 106 h/a,
with h the height above the ground and a the radius of the
earth), and M is the change in M across the ducting layer.
As an example, for c to equal 1, the required M is 152.3.
Assuming a surface duct 50 m in height, the above M lapse
corresponds to a refractivity gradient of 840 N-units/km. Refractivity gradients of this magnitude are extremely rare, and
so ducting is normally limited to terrestrial radio systems or
satellite-to-ground paths where the elevation angle is below
1. The latter condition is not expected to be used in commercial systems because it leads to poor propagation, but may
have military applications in some situations.
Signal Delay. A signal propagating in the atmosphere will
travel at a speed that is less than the speed of light in a vacuum. If a ranging system is being employed, such as the
Global Positioning System (GPS), which ignores the true
speed of the signal through the atmosphere, an error will be
made in calculating the range. There will be two components
of range error (here, range delay, since the signal will arrive
382
REFRACTION AND ATTENUATION IN THE TROPOSPHERE
later than expected). These are the range delay d due to dry
air and the range delay w due to the moisture in the air.
Typical values (6) are given in the table below for a path that
exits the atmosphere to a satellite where the refractive index
of the atmosphere is assumed to be unity. The range error
values have been increased by 2% from those in Ref. 5 to
allow for the additional atmosphere encountered from the 80
km height used in Ref. 5 for the target. Regression analyses
(7) have shown that these typical range errors can be reduced
in practice by 25% by incorporating site latitude, height above
sea level, and time of day within the calculation procedure
Initial Elevation
Angle (deg)
Elevation Angle
Error (mdeg)
Range Error (m)
0
5
50
595
159
13.4
106
25.5
3.2
Small-Scale Refractive Effects
The air is rarely still: differential heating and cooling effects
within the atmosphere induce wind currents from balmy
breezes to tornado-force hurricanes and tropical cyclones. The
resultant turbulence in the troposphere will cause small-scale
changes in the refractive index. These in turn will lead to
relatively rapid variations in the amplitude, phase, and angle
of arrival of a signal (8). The effect on the signal is usually
described by one of
1. The log amplitude (dB), which equals the ratio of the
instantaneous amplitude of the observed signal to its
mean amplitude
2. The scintillation variance 2 in dB2 or the intensity
in dBrms, which are the variance and root mean square
of the log amplitude , respectively (8)
On terrestrial paths, the turbulence occurs in the boundary layerthe part of the troposphere that is next to the surface of the earth. In general, the signal fluctuations due to
turbulence on terrestrial paths are neither performance- nor
availability-limiting, the major propagation impairments being multipath (due to multiple atmospheric rays and ground
reflections) and (for systems operating at frequencies above
about 10 GHz) rain attenuation. On earthspace paths, the
primary cause of major tropospheric scintillations is cumulus
clouds on warm, humid days. The attendant air entrainment
process caused by the convective activity within the cloud
structure induces turbulence: the more humid and hot the
conditions are, the more severe the scintillations become.
Similarly, the lower the elevation angle, the larger the amplitudes of both signal enhancements and degradations become.
In most systems, tropospheric scintillation will only affect the
performance of the link and not its availability (9). At elevation angles below about 5, however, low-angle fading effects
become evident (10,11), and they can be system-limiting (i.e.
cause a link outage). Figure 2 illustrates the general concept
Tropospheric scintillation and gaseous attenuation
Cloud attenuation and effects of melting layer
;;;;;
Heavy thunderstorm rain
10
Performance
threshold (108)
log (BER)
7
6
5
4
Availability
threshold (106)
Range of degraded
performance
(i.e. < 108
but > 106 BER)
3
2
1
Figure 2. Example of annual statistics of
bit error ratio (BER) versus time percentage for a given link.
0
0.001
0.01
0.1
1
Annual time percentage
10
100
REFRACTION AND ATTENUATION IN THE TROPOSPHERE
of performance and availability thresholds for digital communication systems (12).
Step 9. Calculate the scintillation fade depth for the time percentage p by
A s ( p) = a( p)
Prediction of Tropospheric Scintillation
In most systems, it is the probable signal degradation that is
required for the link budget analysis (signal enhancements
being of relevance only for potential front-end amplifier saturation or for likely interference to other systems). The currently accepted modeling procedure to calculate the effective
signal reduction due to tropospheric scintillations on an
earthspace path (13) uses the wet term of the refractivity at
ground level as the input parameter to predict the scintillation variance. The step-by-step procedure from Ref. 13 is
given below for elevation angles 4:
Parameters required for the method include: t the average surface ambient temperature (C) at the site for a period of one month
or longer; H average surface relative humidity (%) at the site
for a period of one month or longer; f frequency (GHz), where 4
GHz f 20 GHz; path elevation angle, where 4; D
physical diameter (m) of the earth station antenna; and antenna efficiency; if unknown, 0.5 is a conservative estimate.
Step 1. For the value of t, calculate the saturation water vapor
pressure, es (kPa), as specified in Ref. 3
Step 2. Compute the wet term of the radio refractivity, Nwet,
corresponding to es, t, and H as given in Ref. 3
Step 3. Calculate the standard deviation of the signal amplitude, ref , used as a reference:
ref = (3.6 10 3 ) + (10 4 Nwet )
dB
Step 4. Calculate the effective pathlength L according to
L = (2h L )/ {[sin2 + 2.35 10 4 ] 1/2 + sin }
where hL is the height of the turbulent layer; the value of hL
1,000 m
Step 5. Estimate the effective antenna diameter, Deff , from the
geometrical diameter, D, and the antenna efficiency, ,
D eff = () 1/2 D
Step 6. Calculate the antenna averaging factor from
g(x) = {[3.86(x 2 + 1) 11/12 ] sin[(11/6) arctan(1/x)]
7.08x
5/6 1/2
x = 1.22(D eff ) 2 ( f /L)
where f is the carrier frequency in GHz
Step 7. Calculate the standard deviation of the signal for the
considered period and propagation path,
= ref ( f ) 7/12 [ g(x)/(sin ) 1.2 ]
Step 8. Calculate the time percentage factor a(p) from the time
percentage, p, of concern in the range 0.01 p 50,
a( p) = 0.061(log10 p) 3 + 0.072(log10 p) 2
1.71 log10 p + 3.0
dB
The above procedure has been found to predict well at frequencies at least up to 30 GHz, although it tends to underpredict slightly at all frequencies in the hot, humid summer
months. At higher frequencies (30 GHz) it is expected that
the same f 7/12 scaling law will apply as shown in step 7 above,
although this anticipated result has not been tested experimentally. The fading portion of the tropospheric scintillation,
As, is not usually significant for earth-to-space paths at elevation angles above 20 for the frequency bands currently in
operation (4 GHz to 30 GHz). As the elevation angle decreases, or the frequency band increases, tropospheric scintillation fading becomes more significant. For example, at 11
GHz, As is less than 0.5 dB for 0.01% of an average year at
elevation angles on the order of 30. For the same frequency,
As approaches 3 dB at elevation angles of 3.2 (11). These values of As are for nonraining periods and so will affect the performance of the system.
GASEOUS ABSORPTION
When a molecule or atom is nonsymmetrical in its structure,
it will have a preferred orientation if placed in an electric
field. Such molecules or atoms are said to be polar. Polar molecules or atoms cause far more loss to a radio signal than
nonpolar molecules or atoms, since the application of an external field causes the polar molecules or atoms to reorient
themselves. The reorientation (or relaxation) of the dipoles
will remove energy from the applied field and cause heating
of the medium. This is the principle of a microwave oven. The
major constituents of the troposphere, oxygen and nitrogen,
are electrically nonpolar and so no absorption due to electric
dipole resonance will take place. Oxygen however, is a paramagnetic molecule with a permanent magnetic moment
which causes resonant absorption at particular frequencies.
Water and water vapor, which both contain two oxygen
atoms, are therefore polar molecules and so absorption occurs
due to resonance at critical frequencies.
Oxygen and Water Vapor Resonance Lines
The specific gaseous attenuation (dB/km) is given by
= o + w
with
383
(dB/km)
(13)
where o and w are the specific attenuations (dB/km) due to
dry air (essentially oxygen) and water vapor, respectively.
The specific attenuation needs to be multiplied by the length
of the path over which it operates to arrive at total path attenuation. On terrestrial paths, this is simply the distance
between the transmitter and receiver, L (km), and the path
attenuation A (dB) is given by
A = L = (o + w )L
(dB)
(14)
For earthspace paths, the situation is not so straightforward, since the density and make up of the atmosphere
change rapidly with height. The value of L used must be able
384
REFRACTION AND ATTENUATION IN THE TROPOSPHERE
to take account of the density and other variations along the
link through the atmosphere.
Figure 3, abstracted from Fig. 3 of Ref. 14, gives the total
zenith gaseous attenuation at frequencies up to 1 THz (1012
Hz) for two conditions: a standard atmosphere and a dry atmosphere. A standard atmosphere is defined as having the
following surface characteristics: pressure 1013 hPa, temperature 15C, and water vapor density 7.5 g/m3. In general, the
dry atmosphere curve in Fig. 3 shows the resonance absorption lines of oxygen, while the standard atmosphere curve is
a summation of the resonance absorption lines of water vapor
and oxygen. A software code (in Matlab) that calculates the
total gaseous attenuation in a line-by-line fashion up to a frequency of 1 THz is available from the Radiocommunication
Bureau of the ITU in Geneva.
For nonzenith paths, it is necessary to know not only the
specific attenuation at each point in the link but also the
length of the path that has this specific attenuation. To derive
the incremental path lengths, ray bending through the atmosphere must be considered [see Fig. 1(a)]. A full procedure to
accomplish this is given in Ref. 14. In this procedure, formulae are given for the precise calculation of o and w, and the
total zenith attenuation is calculated from
Az = o ho + w hw
(dB)
(15)
where ho is the equivalent height of the dry gaseous absorption and hw is the equivalent height of the water vapor absorption. Typical values of ho and hw are 6 km and 2.1 km
(during rainy conditions), respectively. For elevation angles
between 10 and 90, a cosecant law gives the total gaseous
attenuation A, namely
A=
Az
sin
(dB)
(16)
where is the elevation angle. For elevation angles below 10,
a more accurate formula that takes account of the real length
of the atmospheric path is required (14). For most earth
space systems, gaseous absorption is not significant compared
with the other impairments. It also changes very little with
time and so is usually relatively easy to factor into a system
design.
RAIN ATTENUATION
Attenuation due to rain is made up of two components: absorption and scattering. Absorption takes place when the incident signal energy is transformed into mechanical energy,
thereby heating the raindrop. Scattering occurs when the incident signal is redirected away from the desired propagation
path without energy loss to the raindrop. The relative importance of scattering and absorption is a function of the complex
index of refraction of the raindrop and the size of the raindrop
relative to the wavelength of the signal (15). The significance
of the scattered component generally increases both with the
signal frequency and with the size of the raindrop. In general,
attenuation due to rain increases with increasing signal frequency for a given rain rate. Typical annual cumulative statistics for frequencies of 12 GHz, 18.7 GHz, 39.6 GHz, and
49.5 GHz on paths at an elevation angle of approximately 38
in Europe are shown in Fig. 4 (some data were extracted from
Ref. 16). These curves show a monotonic increase in attenuation with decreasing time percentage, which is typical of temperate climates. There is evidence, however, that in tropical,
high-rain-rate regions there is a saturation effect in the attenuation and rainfall rate statistics at low time percentages,
leading to a breakpoint in the cumulative statistics (17). This
and other features of high-rain-rate regions make the accurate prediction of attenuation due to rain very difficult for
those locations.
Rain does not usually consist of a homogeneous collection
of raindrops falling at a constant rate. The size, shape, temperature, and fall velocities usually vary in a dynamic manner along the path. To calculate the attenuation of a radio
signal passing through rain it is necessary to invoke a dropsize distribution and to integrate the attenuation contributions of the various raindrops along the path through the
rain. The characteristics of rain vary so much in space and
105
Zenith attenuation (dB)
104
103
102
10
101
Figure 3. Zenith attenuation due to atmospheric gases (from Fig. 3 of Ref. 14).
Curve A: standard atmosphere (7.5
g/m3); curve B: dry atmosphere. ITU.
Reproduced with permission.
102
0
100
200
300
400
500
600
Frequency (GHz)
700
800
900
1000
REFRACTION AND ATTENUATION IN THE TROPOSPHERE
50
Step 1. Calculate the effective rain height, hR, for the latitude
of the station, :
Total path attenuation (dB)
49.5 GHz
5 0.075( 23)
5
5
h R (km) =
5 + 0.1( + 21)
40
39.6 GHz
30
18.7 GHz
0
0
71
20
10
12 GHz
0
0.001
385
0.01
0.1
1
10
Percentage of a year attenuation is exceeded
for
for
for
for
for
> 23 Northern Hemisphere
23 Northern Hemisphere
23 Southern Hemisphere
< 21 Southern Hemisphere
< 71 Southern Hemisphere
Step 2. For 5 compute the slant-path length, Ls, below the
rain height from:
100
L s = (h R h S )/(sin )
Figure 4. Average annual cumulative statistics of total path attenuation at an elevation angle of about 38 in Europe (some data extracted from Ref. 16).
km
For 5, the following formula is used:
L s = [2(h R h S )]/{[sin2 + 2(h R h S )/8500] 1/2 + sin }
Step 3. Calculate the horizontal projection, LG, of the slantpath length from:
time that it is necessary to resort to empirical methods that
employ statistical averaging. The accepted approach is to use
a power law representation based on the rainfall rate R (usually a rainfall rate measured at a point on the ground close to
the communications link). The specific attenuation (dB/km)
in rain is related to the rainfall rate R (mm/h) by the power
law relationship
= kR
(dB/km)
L G = L S cos
r 0.01 = 1/(1 + L G /L 0 )
where L0 35 exp (0.015 R0.01) and, for R0.01 100 mm/h, use
the value 100 mm/h in place of R0.01.
Step 6. Obtain the specific attenuation, R, using the frequency
dependent regression coefficients k and [Ref. 18] by using
where k and are regression coefficients (18) that take account of absorption and scattering.
R = k(R 0.01 )
Prediction of Attenuation
The following parameters are required: R0.01 point rainfall rate
for the location for 0.01% of an average year (mm/h); hs height
above mean sea level of the earth station (km); elevation
angle; f frequency (GHz). The geometry is illustrated in Fig. 5.
The method is applicable up to at least 30 GHz.
km
Step 4. Obtain the rain intensity, R0.01, exceeded for 0.01% of
an average year at the site
Step 5. Calculate the reduction factor, r0.01, for 0.01% of the
time for R0.01 100 mm/h
(17)
Rain-attenuation prediction models fall into two classes: (1)
those that attempt to analyze the physics of the process and
model the constituents of storm cells, and so on; and (2) those
that use empirical approaches with simplified assumptions.
To date, those that use empirical methods have demonstrated
the most consistent accuracy on a global basis (although, with
the strong emphasis currently on the development of accurate, world-wide climatic parameter maps, a physical modeling approach should eventually prevail). For empirical methods, the key to the prediction process is the development of a
procedure that enables the user to move seamlessly from
point rainfall rate on the ground to average rainfall rate along
the path at any elevation angle. This approach is described
below, abstracting directly from the current ITU-R prediction
procedure for calculating rain attenuation on a satellite-toground path, as set out in Ref. 13:
km
dB/km
Step 7. The predicted attenuation exceeded for 0.01% of an average year is obtained from
A 0.01 = R L S r 0.01
dB
Step 8. The estimated attenuation to be exceeded for other percentages of an average year, in the range 0.001% to 1%, is de-
Region of frozen precipitation
Earthspace
path
Rain height
hR
Region of liquid
precipitation
LS
hR-hS
hS
LG
Figure 5. Schematic presentation of earthspace path (from Fig. 1
in Ref. 13). ITU. Reproduced with permission.
386
REFRACTION AND ATTENUATION IN THE TROPOSPHERE
termined from the attenuation exceeded for 0.01% of an average year by using
(A p )/(A 0.01 ) = 0.12 p (0.546+0.043 log
p)
Note that the variability of the weather from year to year will
cause this prediction method to vary similarly. Average accuracy
better than about 35% should not be expected for any given year.
The following parameters are needed: Ap, the rain attenuation
(dB) exceeded for the required percentage of time, p, for the path
in question; tilt angle of the linearly polarized electric field
with respect to the horizontal (for circular polarization use
45); f frequency (GHz); path elevation angle (degrees). The
method is applicable to frequencies between 8 GHz and 35 GHz.
Step 1. Calculate the frequency-dependent term
C f = 30 log( f )
for 8 f 35 GHz
DEPOLARIZATION
Step 2. Calculate the rain attenuation dependent term
The polarization sense of a wave usually refers to the preferred orientation of the electric vector. A perfectly polarized
wave will have no component of its electric field in the orthogonal sense. There are only two states for a perfectly polarized
wave: linear and circular. For linear polarization, it is common to use vertical and horizontal as the two orthogonal (linear) reference axes. For circular polarization, right-hand circular polarization (RHCP) and left-hand circular polarization
(LHCP) are the two orthogonal (circular) reference senses.
Depolarization from one reference polarization sense to the
other occurs when the signal passes through an anisotropic
medium and the polarization sense of the signal is not aligned
with the principal axes of the medium. Rain can be such a
medium.
Raindrops exist in a dynamic medium: the atmosphere.
When the drops are very small, surface tension forces (which
strive to keep the drop perfectly spherical) far exceed the
aerodynamic forces on the raindrop (which seek to shape the
drop to conform with the external pressures). Under these
conditions, the raindrops remain spherical and the rain medium appears symmetrical to the signal. No signal depolarization therefore results. As the drops become larger, however,
through coalescence with other drops on collision in the rain
medium, the surface tension forces can no longer keep the
drops spherical. The drops distort into elliptical shapes and
so exhibit preferred orientation directions along the major
and minor axes of the ellipsoid. Differential attenuation and
phase shift will occur on component signals propagating along
these axes and will result in signal depolarization. The degree
of depolarization measured is called the cross-polarization
discrimination (XPD) and is defined as
XPD = 20 log10
Eco
Ecross
(dB)
(18)
where Eco is the field measured in the wanted (copolarized)
polarization and Ecross is the field measured in the unwanted
(cross-polarized) polarization. If the differential attenuation
(, nepers) and differential phase (, radians) caused by a medium are known, XPD can be calculated from
(+ j )
e
+ 1
(dB)
(19)
XPD = 20 log10 (+ j )
e
1
Generally and are not known, and it is usual to model
XPD against rain attenuation.
Prediction of Cross-Polarization Discrimination
The most often used prediction method for calculating XPD is
that in the ITU-R Recommendation 618 (13). The procedure
is abstracted directly below:
C A = V ( f ) log A p
where
V ( f ) = 12.8 f 0.19
V ( f ) = 22.6
for 8 f 20 GHz
for 20 < f 35 GHz
Step 3. Calculate the polarization improvement factor
C = 10 log[1 0.484(1 + cos 4 )]
The improvement factor C 0 for 45 (which is equivalent
to circular polarization) and reaches a maximum value of 15
dB for linearly polarized signals with 0 or 90.
Step 4. Calculate the elevation angle dependent term
C = 40 log(cos )
for 60
Step 5. Calculate the canting angle dependent term
C = 0.0052 2
is the effective standard deviation of the raindrop canting
angle distribution, expressed in degrees; takes the value 0,
5, 10, and 15 for 1%, 0.1%, and 0.01%, and 0.001% of the
time, respectively.
Step 6. Calculate the rain XPD not exceeded for p% of the time:
XPDrain = C f C A + C + C + C
dB
Step 7. Calculate the ice crystal dependent term
C ice = XPDrain [(0.3 + 0.1 log p)/2]
dB
Step 8. Calculate the XPD not exceeded for p% of the time,
including the effects of ice crystals:
XPDp = XPDrain C ice
dB
For frequencies below 8 GHz the scaling formula below can
be used for 4 f 1, f 2 30 GHz:
XPD2 = XPD1 20 log
f 2 [1 0.484(1 + cos 42 )]1/2
f 1 [1 0.484(1 + cos 41 )]1/2
The component for ice depolarization in the above formulation, Cice, is known to underestimate the degree of ice crystal
depolarization encountered on low-elevation-angle paths.
Depolarization can be system-limiting in the lower frequency bands, where rain attenuation is not significant (e.g.,
in the C band, which is the 6/4 GHz uplinkdownlink pair
for many satellite systems). As the frequency goes up, however, the amount of depolarization observed per decibel of at-
REFRACTION AND ATTENUATION IN THE TROPOSPHERE
tenuation becomes less and less until, by Ka band (the 30/20
GHz uplinkdownlink pair), depolarization is rarely systemlimiting. In analog systems, an XPD margin of 12 dB is usually required to operate successfully. In digital systems, this
margin can be reduced due to the possibility of adding coding
to the signal. When two satellite links are sharing the same
transponder, but are using opposite polarizations (this is referred to as polarization frequency reuse), depolarization
caused by ice crystals can be a problem. This is because the
unwanted (depolarized) component of one link caused by the
ice crystals will enter the channel of the wanted signal in the
satellite transponder without any attenuation having occurred.
ATTENUATION DUE TO CLOUDS AND FOG
Fog or mist is essentially supersaturated air in which some
of the water has precipitated out to form small droplets of
water, usually 0.1 mm in diameter (19). Fog is mainly
formed from two processesradiation and advection (6)and
rarely extends more than 100 m above the ground. For this
reason, fog is usually ignored in satellite-to-ground link design for frequencies below about 100 GHz. Clouds, however,
are much more complicated.
Several models are available for the prediction of cloud attenuation (1923), but there remain two major difficulties in
turning these into link design tools: (1) an accurate cumulative distribution of the vertical and horizontal extent, on a
global basis, of those clouds that contribute to slant-path attenuation; and (2) separating this attenuation contribution
from that due to rain formed in some of the cloud processes.
Many measurements of rain attenuation implicitly include attenuation due to clouds. Understanding, and accurately predicting, the contribution of clouds to path attenuation is a key
element in the modeling of the proposed V-band (50/40 GHz)
satellite services. Of equal importance will be the development of modeling procedures that can predict the combined
effects of all the attenuating phenomena on an earthspace
link (24,25).
ATTENUATION EFFECTS ON INFRARED AND VISIBLE LIGHT
The far infrared (1 THz), infrared (10 THz), and visible
(100 THz) portions of the electromagnetic spectrum have
been utilized for many applications. Figure 3 shows that, for
a dry atmosphere, the total zenith attenuation is only about
1 dB at 1 THz. This value of zenith attenuation is fairly constant for frequencies through visible light (100 THz). A window in the standard atmosphere exists around 890 GHz, the
frequency of HCN masers, but the zenith attenuation at this
frequency is still two orders of magnitude higher than that
in the dry atmosphere (i.e., 100 dB versus 1 dB total zenith
attenuation). For all three frequency ranges, the key attenuating elements in the atmosphere are aerosols, usually cloud
and fog droplets. Thick fog (liquid water density 0.5 g/m3)
will cut visibility to 50 m (19). Most free-space laser links are
therefore short-range (1 km) unless extraordinary powers
are contemplated, such as the 1 MW chemical oxygen iodine
laser (COIL) lasers planned for missile interception. At such
huge power levels, though, the almost instantaneous heating
process caused by the passage of the beam through the atmo-
387
sphere will modify the constituents along the path, thereby
altering the transmission characteristics significantly.
BIBLIOGRAPHY
1. J. Powers, An Introduction to Fiber Optic Systems, 2nd ed., Homewood, IL: Irwin, 1997, Chap. 2, Sec. 2.2.
2. Effects of tropospheric refraction on radiowave propagation, Rec.
ITU-R P.834-2, ITU-R Recommendations, Propagation in Ionized
and Non-ionized Media, 1997.
3. The radio refractive index: Its formula and refractivity data, Rec.
ITU-R P.453-6, ITU-R Recommendations, Propagation in Ionized
and Non-ionized Media, 1997.
4. M. P. M. Hall and C. M. Comer, Statistics of tropospheric radiorefractive index soundings taken over a three year period in the
UK, Proc. IEE, 116: 685690, 1969.
5. H. T. Dougherty and B. A. Hart, Recent progress in duct propagation predictions, IEEE Trans. Antennas Propag., 27: 542548,
1979.
6. W. L. Flock, Propagation effects on satellite systems at frequencies below 10 GHz, NASA Reference Publication 1108, 1983.
7. E. E. Altshuler, Tropospheric range-error corrections for the
Global Positioning System, IEEE Trans. Antennas Propag., 46:
643649, 1998.
8. D. Vanhoenacker-Janvier, Review of tropospheric scintillation
measurements and prediction techniques, CLIMPARA98, Proc.
URSI Commission F Open Symp. on Climatic Parameters in Radiowave Propagation Prediction, Ottawa, Canada, 1998, pp.
9398.
9. E. T. Salonen, J. K. Tervonen, and W. J. Vogel, Scintillation effects on total fade distribution for earthsatellite links, IEEE
Trans. Antennas Propag., 44: 2327, 1996.
10. J. L. Strickland, R. L. Olsen, and H. L. Werstiuk, Measurement
of low-angle fading in the Canadian Arctic, Ann. Telecommun. 32:
530535, 1977.
11. D. L. Bryant, Low elevation angle 11 GHz beacon measurements
at Goonhilly earth station, BT Technol. J., 10 (4): 6875, 1992.
12. J. E. Allnutt and A. W. Dissanayake, The combination of attenuating phenomena in prediction procedures for 1050 GHz satellite-to-ground paths, CLIMPARA98, Proc. URSI Commission F
Open Symp. Climatic Parameters Radiowave Propagation Prediction, Ottawa, Canada, 1998, pp. 183190.
13. Propagation data and prediction methods required for the design
of earthspace telecommunications systems, Rec. ITU-R P.618-5,
ITU-R Recommendations, Propagation in Ionized and Non-ionized
Media, 1997.
14. Attenuation by atmospheric gases, Rec. ITU-R P.676-3, ITU-R
Recommendations, Propagation in Ionized and Non-ionized Media, 1997.
15. G. Brussaard, Radiometry: A useful tool? ESA publication SP1071, 1985.
16. R. Polonio and C. Riva, ITALSAT propagation experiment at
18.7, 39.6, and 49.5 GHz at Spino DAdda: Three years of CPA
statistics, IEEE Trans. Antennas Propag., 46 (5): 631642, 1998.
17. G. H. Bryant, The structure of tropical rain from attenuation and
rain exceedences, Proc. ISAP92, Sapporo, Japan, 1992, pp.
877880.
18. Specific attenuation model for rain for use in prediction methods,
Rec. ITU-R P.838, ITU-R Recommendations, Propagation in Ionized and Non-ionized Media, 1992.
19. Attenuation due to clouds and fog, Rec. ITU-R P.840-2, ITU-R
Recommendations, Propagation in Ionized and Non-ionized Media, 1997.
388
RELATIONAL DATABASES
20. D. S. Slobin, Microwave noise temperature and attenuation of
clouds: Statistics of these effects at various sites in the United
States, Alaska, and Hawaii, Radio Sci., 17: 14431454, 1987.
21. E. E. Altshuler and R. A. Marr, Cloud attenuation at millimeter
wavelengths, IEEE Trans. Antennas Propag., 37: 14731479,
1989.
22. F. Dintelmann and G. Ortgies, A semi-empirical model for cloud
attenuation prediction, Electron. Lett., 25: 14871488, 1989.
23. E. Salonen, Prediction models of atmospheric gases and clouds
for slant path attenuation, Olympus Utilization Conf., Seville,
Spain, 1993, pp. 615622.
24. G. Feldhake, Estimating the attenuation due to combined atmospheric effects on modern earthspace paths, IEEE Antennas Propag. Mag., 39: 2634, 1997.
25. A. W. Dissanayake, J. E. Allnutt, and F. Haidara, A prediction
model that combines rain attenuation and other propagation impairments along earthspace paths, IEEE Trans. Antennas Propag., 45: 15461558, 1997.
JEREMY E. ALLNUTT
Virginia Tech/Northern Virginia
Center
REFRACTION CAUSED BY ILLUMINATION. See
PHOTOREFRACTIVE EFFECT.
REGULATORS, SELF-TUNING. See SELF-TUNING REGULATORS.
REJECTION. See TESTING FOR ACCEPTANCE-REJECTION.
128
RADIO NOISE
RADIO NOISE
Radio noise can be defined as unwanted and unavoidable electromagnetic fluctuations that tend to obscure the information
content of a desired radio signal. In a radio receiver, the noise
disturbances are fluctuating voltages and currents that alter
the desired signals original frequency, amplitude, or phase in
some unpredictable, unwanted manner. In early radios, much
of the noise stemmed from the quality of the equipment; for
example, poor contacts and unreliable connections, sensitivity
to mechanical vibrations, and unstable detectors and amplifiers all contributed to the degradation of the radio signal.
Over the years, there was steady progress in improving transmitters and receivers and greatly reducing or completely
eliminating these and other noise sources. It was soon realized, however, that all sources of noise could not be totally
eliminated and that a fundamental residual noise remained
as a lower limit. This article focuses on such noise emanating
from both natural sources and human activities that affect
the performance of narrowband radio frequency (RF) communications systems.
Typically radio noise is divided into two components
internal or receiver noise (that is, intrinsic noise generated by
the receiving system components) and external or environmental radio noise (noise collected by the receiving system
antenna). Receiver noise is largely due to natural processes
such as thermal noise, arising from the motion of thermally
agitated free electrons in resistors, and shot noise, resulting
from current fluctuations superimposed on the steady current
flow in both vacuum tubes and solid-state devices. Environmental noise is an intrinsic part of a wireless communications
channel and consists of natural radio noise (i.e., noise originating from natural sources such as atmospheric noise and
sky noise), and unintentional RF emissions due to human activity, which are commonly referred to as man-made noise.
Atmospheric noise is largely due to lightning and related phenomena, such as precipitation static. Sky noise is a generic
term used to describe noise from a variety of terrestrial and
extraterrestrial sources, such as cosmic noise and absorption
of radio waves in the earths atmosphere due to rain, water
vapor, and oxygen. Intentional RF emissions that result in
adjacent- and co-channel radio interference, including intermodulation and jamming, are not covered in this article.
While such noise cannot be completely eliminated, understanding the character of radio noise allows engineers and
system designers to evaluate radio system performance and
devise means to lessen the adverse effects. The most important design parameter used to characterize noise performance
is the predetection signal-to-noise ratio (SNR), and hence,
noise power is generally the most significant parameter in relating the interference potential of noise to system performance. Noise power calculations must include noise generated in all components of the receiving system as well as the
radio channel. This is usually accomplished by determining a
single parameter, the operating system noise factor, which is
essentially the ratio of the receiving system predetection
noise power referenced to the terminals of an equivalent lossless receiving antenna and the noise power available from a
resistor at standard temperature (288 K).
In this article, important sources of receiver noise are described and the concept of receiver noise factor is developed.
The noise factor concept is then extended to cover the entire
radio receiving system, including the receiving antenna and
environmental noise. This is followed by a description of important environmental noise sources.
RECEIVER NOISE AND NOISE FACTOR
The two most important receiver noise sources, used as noise
standards and for noise characterization, are thermal and
shot noise. Thermal noise is characterized as the spontaneous
fluctuations in voltage across a resistor due to the random
motion of thermally agitated electrons. Thermal noise is always present in any radio system and often establishes the
lower limit of signal detection. Shot noise, on the other hand,
originates in a flow of discrete charges not in equilibrium
where the mean-square noise current is proportional to the
product of elementary charge, the average current, and the
bandwidth or 2eIb (1). The cathode-to-anode stream of a temperature-saturated thermionic diode is an example of a pure
shot noise generator. Shot noise currents in semiconductors
such as diodes and transistors are due to the diffusion of
charge carriers in pn junctions.
Since thermal and shot noise may be described as being
the sum of a very large number of random, short-lived disturbances, it follows from the central limit theorem that the processes are Gaussian. More precisely, observed voltages are
white Gaussian noise. It was shown experimentally by Johnson (2) and theoretically by Nyquist (3) that the mean thermal noise power density available from a resistor is simply
J. Webster (ed.), Wiley Encyclopedia of Electrical and Electronics Engineering. Copyright # 1999 John Wiley & Sons, Inc.
RADIO NOISE
kT, where k 1.38 1023 W/Hz/K is Boltzmanns constant
and T is the absolute temperature of the resistor. The characterization as white noise is an approximation valid for typical
narrowband RF communication systems operating at or near
the standard ambient temperature T0 288 K. When quantum mechanical considerations are included, the mean power
density kT is replaced by the more accurate expression obtained by Max Planck (4):
h
(W/Hz)
e h /kT 1
(1)
where h 6.626176 1034 J/Hz is Plancks constant and
represents frequency. Clearly, the flat power spectral density
approximation is inaccurate when high frequencies are combined with cryogenic temperatures. Such systems are beyond
the scope of this article.
Since both thermal and shot noise constitute white
Gaussian noise sources, the properties of the noise voltage
they deliver to an output circuit are indistinguishable. The
significant difference is that thermal noise is proportional to
temperature and shot noise is proportional to the average current. Hence, an equivalent noise temperature Te can be used
to characterize the noise of a device with thermal sources at
temperature T and shot noise sources with mean power dws
in an incremental bandwidth d as follows:
kTe = kT +
dws
d
(2)
Solid-state shot noise sources such as noise diodes are commonly used in noise measurements. To achieve the required
precision in Te, these devices are calibrated against a primary
standard. Noise diodes are commonly characterized in terms
of excess noise ratio defined as
NR = 10 log10
T
T0
(dB)
(3)
Thermal noise generated within or passing through a bandlimited receiver circuit is colored and perhaps amplified.
Since thermal noise power is independent of frequency over
the bandwidth of most narrowband RF communication systems, it is useful to define a noise-equivalent bandwidth b0.
When a resistor is attached to the input terminals of a linear
receiver with gain g(), the total noise power available to the
receiver wa in the system band can be expressed in terms of
the noise equivalent bandwidth as follows:
wa = kT
0
g()
d = kTb0
g0
(4)
where g0 is the nominal gain in the system bandwidth.
Statistical Characterization of Narrowband
White Gaussian Noise
If the receiver input termination is a resistor, the receiver
output noise voltage is a random function of time whose behavior can only be described statistically. For a typical RF
communication system, the signal is narrowband with respect
to a central radio frequency c and the output noise voltage
129
V(t) can be written as
V (t) = Re{v(t)e j2 c t } = E(t) cos(2ct (t))
(5)
where v(t) is the complex baseband representation, E(t) is the
amplitude, and (t) is the phase. In this representation, v(t)
is a complex random process and the pair (E(t), (t)) are two
associated real random processes. When the observed voltages are white Gaussian noise, their baseband representations are 0-mean complex Gaussian processes with a flat
power spectral density. This means that the co- and quad
phases of v are independent and Gaussian distributed with 0
means and identical variances. It also means that voltages at
different times are independenta statement that must be
emended by the fact that the receiver is narrowband, so that
the actual observed power spectral density is proportional to
the power spectrum of the receiver.
For these complex Gaussian processes, it is also true that
the resulting amplitude and phase are independent of each
other and that the phase is uniformly distributed over the
complete circle. The amplitude is then Rayleigh distributed.
In terms of the instantaneous noise power (which is proportional to the square of the amplitude), the cumulative distribution function (more precisely, the exceedance distribution
function) is
Pr{noise power > w > 0} = ew/w o
(6)
where wo is the average output noise power. This average
equals the total area under the power spectral density curve
and is equal to the sum of the mean power of the input noise
amplified by the receiver and the mean noise power added by
the receiver wr or
wo = g0 kTb0 + wr
(7)
where T is the temperature of the resistor connected to the
input terminals.
To obtain this result, the receiver noise is assumed to be
Gaussian and white, as would be the case when thermal and
shot noise are dominant. More generally, Eq. (7) is valid for
any random processes. If the receiver noise is not Gaussian,
however, the instantaneous power distribution would not be
Rayleigh, as given in Eq. (6). The white Gaussian thermal
noise assumption is reasonable for most narrowband RF receivers. In general, however, wr will depend on frequency.
In noise calculations, it is important to define clearly what
is meant by gain. There are many possible definitions of gain.
Assuming that a signal generator is connected to the receiver
input terminals and a power meter is connected to the receiver output terminals, an obvious choice would be to use the
ratio of the input and output powers, or so-called power gain.
Clearly, power gain depends on the impedance of both the
signal generator as well as the impedance of the power meter,
and hence a noise characterization based merely on power
gain is ambiguous. To circumvent this difficulty, the gain
used in noise calculations is usually taken to be the available
power gain (5), defined as the ratio of the available power
from the receiver to the available power from a signal generator connected to the receiver input terminals so that any mismatch losses are included. With this definition, the gain does
not depend on the receiver load; however, it does depend on
the impedance of the signal generator.
130
RADIO NOISE
In determining noise performance, the signal generator is
often matched to the receiver so that the maximum power is
delivered to the receiver. This is a reasonable choice for most
RF applications, where signal generators, connecting cables,
the loads, and so on are usually 50 devices, or they are
matched waveguide interfaces. It should also be noted that
when there is a complete match at the input and output, the
available gain and the power gain are equal.
In radio engineering it is desirable to have a simple, yet unambiguous method of characterizing the noise properties of a
radio receiver. For this purpose, the concept of noise factor
was developed and defined. There are a variety of ways to
define noise factor. The basic idea is to relate the available
noise power at the input of an active or passive two-port linear network to the noise power available at its output terminals. Assume that a CW signal generator is connected and
matched to the input terminals of a receiver. Let wg be the
available signal generator power, wl be the signal power available at the load, and w0 be the available noise power at the
load; the International Radio Consultative Committee (CCIR)
(6) definition of noise factor is given as follows: The noise factor f r of a radio receiver is defined to be the ratio of the available CW signal-to-reference noise power wg /kT0b0 at the terminals of the signal generator to the corresponding total
signal-to-noise power ratio wl /wo available to the load of the
linear portion of the receiver with the CW signal tuned to the
maximum response of the receiver bandpass characteristic
and with the signal generator output impedance at the reference temperature T0:
wg /kT0 b0
wg wo
=
wl /wo
wl kT0 b0
(8)
The ratio wl /wg is simply the maximum available power gain
and hence,
fr =
which is precisely f r defined previously.
Noise Factor for a Passive Two-Port Linear Network
The Receiver Noise Factor
fr
commonly used to characterize receiver noise performance:
f s ()g() d
w
wo
=
=
f = 0
0
(11)
kT0 g0 b0
g() d
kT0
g() d
wo
go kT0 b0
(9)
If it is assumed that the receiver bandwidth is very narrow
b0 d so that the receiver noise is constant over the bandwidth, Eq. (9) gives the spot noise factor
dw0
d
f s () =
g()kT0
(10)
which is essentially a single-frequency value based on a unit
bandwidth. It should be noted that the noise factor is simply
the available output noise power referred to the input terminals (i.e., divided by the nominal gain) in units of kT0 b (i.e.,
the noise power available from a resistor at standard temperature).
The spot noise factor as given in Eq. (10) is equivalent to
the Institute of Radio Engineers (IRE) definition (7). In practice, for narrowband linear receivers, the weighted average
spot noise factor over the system bandwidth f is the quantity
Communications receivers use transmission lines, attenuators, and other devices that can be generally classified as passive two-port linear networks. It is important to consider the
effects of such devices operating in tandem with the actual
receiver to characterize properly the noise performance of the
receiving system.
The spot noise factor of a linear two-port network is obtained by calculating the power density ratio given in Eq. (10)
when a resistor at reference temperature T0 is connected to
the input terminals of the network. If the network is at the
ambient temperature T0, the available output noise power
density is just kT0. If the network temperature is Tn T0, a
correction arising from the temperature difference T0 Tn
must be included; hence
dw0
= kTn + gk(T0 Tn )
d
(12)
where g is the available gain. From Eq. (10), the spot noise
factor is
fs = 1 +
Tn
Tn
(1/g 1) = 1 +
( 1)
T0
T0
(13)
where 1/g is the available loss factor. Note that the spot
noise factor depends on the input resistance used to determine the available gain (or loss).
Noise Factor for Cascaded Linear Networks
The noise factor for a linear receiving system can be calculated from the noise factors for its individual components. Understanding the relationship between the system noise factor
and the noise factors for various components is important in
system design since it shows which components contribute
most significantly to the noise factor.
Assume that two networks are connected together: network a with spot noise factor f a and gain ga, and network b
with spot noise factor f b and gain gb. Further assume that a
resistor is connected to the input of network a and the entire
system is at temperature T0. The power density available at
the output of network b is equal to the sum of the power density available at the input of network b scaled by gb and the
available noise power density due only to network b or
dwo
= f a ga gb kT0 + ( f b 1) gb kT0
d
(14)
Using Eq. (10), the spot noise factor for two cascaded networks is
f ab = f a +
fb 1
ga
(15)
RADIO NOISE
which is the result originally obtained by Friis (8). Spot noise
factors will usually vary somewhat across the band of an actual network and, for a particular system, the effects of nonuniform noise and gain may require additional consideration.
As described previously, the weighted average spot noise factor is often used to characterize the receiving system.
This analysis is readily extended to several linear networks in cascade. For example, if network c is connected to
the output terminals of network b, the system noise factor is
f abc = f ab + ( f c 1)/gab = f a + ( f b 1)/ga + ( f c 1)/ga gb
(16)
and so on for additional networks. This result shows that if
network a has a high gain, its noise factor will be the most
important in determining the overall noise factor for the system. High-gain low-noise amplifiers are often used to reduce
the noise factor in receiving systems.
Noise Factor for the Linear Portion of a Receiving System
In addition to the actual receiver, a typical radio receiving
system is composed of an antenna, transmission lines, and
other circuits, all of which contribute noise and must be included in the analysis of the system noise performance. Of
particular importance for wireless systems is the receiving
antenna, which collects environmental noise. By using the results of the previous sections, the noise factor can be extended
to characterize any specified portion of an operating system
at any specified set of input terminals. For a radio receiving
system, the most useful reference point for noise factor calculations is the input to the terminals of an equivalent lossfree receiving antenna (6). This reference point provides an
appropriate and unique measure of the performance of the
entire receiving system.
Typically, the linear portion of a receiving system can be
divided into a series of cascaded two-port networks, as shown
in Fig. 1. The receiving antenna is modeled as an equivalent
loss-free antenna connected to an antenna circuit network. A
section of transmission line connects the antenna to the receiver. Using Eq. (16), the noise factor for the receiving system is
f = fa +
where f a, f c, and f t are the spot noise factors; and ga, gc, and
gt are the gains of the antenna, antenna circuit, and transmission line, respectively; and f r is the spot noise factor of the receiver.
The external noise power in the band d that is available
at the terminals of the loss-free receiving antenna can be expressed as kTa d, where Ta is the noise temperature of the
radiation resistance of the receiving antenna at the center frequency of the receiver. The antenna noise factor f a is equal to
the ratio Ta /T0. Since the antenna circuit and transmission
line are passive two-port networks, the spot noise factor can
be expressed as
f = Ta /T0 + (c 1)Tc /T0 + c (t 1)Tt /T0 + c t ( f r 1)
(18)
where the gains have been replaced by the corresponding loss
factors. Strictly speaking, this result is meaningful for narrowband systems where the noise factors do not vary significantly over the operating frequency range. It should be understood that all of the spot noise factors and loss factors are
determined with generator impedances that are the same as
those in the actual receiving system. Also, the formula for
cascaded networks is applicable regardless of the degree of
match or mismatch between the output impedance of one network and the input impedance of the following network. The
magnitudes of these mismatch losses do influence the values
of the spot noise factors and loss factors and will thus affect
the noise factor.
The quantity of interest for finite bandwidth systems is the
weighted average spot noise factor, which is often referred to
as the operating noise factor f op (6). The operating noise factor
can be expressed in terms of the weighted average noise factors of the receiving system components as follows:
f op = f a + f ct 1 + (br /b)( gr /g0 )( f r 1)
(19)
where, denoting the overall system gain at frequency as
g(), the weighted average antenna noise factor is
fa =
fc 1
f 1
fr 1
+ t
+
ga
ga gc
ga gc gt
131
(17)
Ta () g() d
T0
g() d
(20)
SNR, f, and fa defined here
Physically accessible antenna terminals
Loss-free antenna
with available
external noise
kTab
Antenna
circuit
Transmission
line
Receiver
Output
(a)
(b)
(c)
(d)
(e)
la = 1
lc
lt
Ta
Tc
Tt
fa = Ta/T0
fc = 1 + (lc 1)(Tc/T0)
ft = 1 + (lt 1)(Tt/T0)
f = fa + (lc 1)(Tc/T0) + lc(lt 1)(Tt/T0) + lc lt(fr 1)
fr
Figure 1. The receiving system and its
operating noise factor, f.
132
RADIO NOISE
f ct is the weighted average combined noise factor of the antenna circuit and transmission line network,
f ct () g() d
f ct = 0
(21)
g() d
0
b is the equivalent bandwidth of the system, br is the equivalent bandwidth of the receiver, g0 is the nominal gain of the
system, gr is the maximum gain of the receiver, and f r is the
weighted average receiver noise factor, as given in Eq. (11)
and using the receiver gain function.
To apply the foregoing results, noise factors for the receiver, related electronic components, and an equivalent lossless antenna must be determined. Using the definitions given
in the previous sections, measurement methods can be readily
devised and implemented to calculate accurately noise factors
for receivers and other electronic components (see, for example, Ref. 9). The determination of the antenna noise factor is,
in general, much more difficult. For RF communication systems, the antenna noise factor is often a nonstationary random process that varies with time, frequency, geographical
location, and receiving antenna characteristics. As a consequence, researchers have spent many years measuring and
compiling statistics that can be used to estimate the antenna
noise factor for various environmental noise sources. In the
case of man-made noise, not only is there a fast time dependence over fractions of an hour to days, but there is also a
relatively slow time dependence based on advances in technology. Significant changes in the background noise for a particular environment may occur within a few years, and as a
consequence published data can become outdated. Antenna
noise factors based on published noise statistics are described
in the following sections.
So far in this article, lowercase letters have been used to
represent noise factors, gains, and bandwidths. In noise analysis, it is common to express these quantities as decibels and
to use uppercase letters to denote that the quantity is in decibels. Also, when given in decibels, the noise factor is usually
referred to as noise figure; for example, the antenna noise
figure is
Fa = 10 log10 f a (dB)
(22)
years of measurements of natural and man-made noise over
the entire radio spectrum.
Fa for Blackbody Radiation
All objects at temperatures above absolute zero radiate energy in the form of electromagnetic waves. A blackbody is a
perfect absorber and a perfect radiator of electromagnetic energy. It absorbs all incident radiation at all wavelengths, and
the radiation from it is a function of only temperature and
wavelength. Although the concept of a blackbody is an idealization, it can be used to estimate the radio noise emitted
by a variety of objects. For example, the cosmic background,
thought to be the lingering echo of the creation of the universe, is equivalent to radiation from a blackbody at 2.7 K.
The brightness of radiation from a blackbody radiator at
temperature T and frequency is given by Plancks radiation
law (10):
=
2h 3
c2 (eh /kT 1)
(23)
This notation and terminology is used in the following sections.
ENVIRONMENTAL NOISE AND ANTENNA NOISE FIGURE
Environmental noise emanates from both natural and manmade sources and is collected by the receiving system antenna. The determination of noise parameters, such as the
antenna noise figure Fa, requires careful measurement programs that must account for temporal, spatial, and frequency
variations of the particular noise source. In this section, some
of the more important sources of environmental noise are described. The statistical data presented are based on many
(24)
where c 2.99792458 108 (m/s) is the speed of light in free
space. For typical radio frequencies and standard temperatures, where, h kT, this reduces to the RayleighJeans
approximation
2kT
= 2
(25)
where is the free space wavelength. As in the case of thermal noise in a resistor, the radiation power is directly proportional to temperature.
The power received by an antenna from an incremental
blackbody of area dA at temperature T in bandwidth d is
dw =
1 2
kT () d d
() d d =
2 4
4
(26)
where d is the element of solid angle subtended at the receiver by dA and () is the directive gain of the antenna.
Note that the antenna only collects half the power since only
a single polarization is received. The total power collected
when the blackbody radiation at temperature T is received
from all directions is
and in decibels, the notations for gain and bandwidth are
G = 10 log10 g (dB); B = 10 log10 b (dB/Hz)
(W/m2/Hz/sr)
w = kT d
1
4
() d = kT d
(27)
which is the same as the total power available from a resistor
at temperature T. Hence, the antenna temperature is simply
T and f a T/T0 independent of the antenna gain.
Fa for Common Natural and Man-Made Radio Noise Sources
Both natural and man-made radio noise have been measured
and carefully studied by many scientists and engineers in the
latter half of the twentieth century. The results of these efforts have been published in various journals, conference
proceedings, and reports and recommendations of radio engineering organizations, such as the International Telecommunication Union (ITU) (1114). In this article, statistical data
from these studies are presented for what are considered to
RADIO NOISE
260
240
A:
B:
C:
D:
E:
F:
A
B
220
G:
200
H:
I:
Fa (dB above kT0b)
180
J:
K:
LQ:
M:
N:
160
140
Maximum expected value of atmospheric noise.
Minimum expected value of atmospheric noise.
Atmospheric noise: value exceeded 0.5% of time.
Atmospheric noise: value exceeded 99.5% of time.
D-region daytime noise temperature.
Noise from galactic center. Part below 10 MHz
represents nighttime conditions.
Noise from galatic pole. Part below 10 MHz
represents daytime conditions.
Emission by moist (17 g/m3) atmosphere for 0
elevation angle.
Emission by dry (1 g/m3) atmosphere for 90
elevation angle.
Cosmic background (2.7 K).
Quantum limit.
Quiet sun, LD Disturbed sun.
Heavy rain (50 mm/h over 5 km).
Light rain (1.235 mm/h over 5 km0.
120
100
2.9 1026
2.9 1023
2.9 1020
2.9 1017
2.9 1014
2.9 1011
80
LQ
2.9 108
60
40
LQ
2.9 105
0
E
20
10
100 1 kHz 10
100 1 MHz 10
Frequency
H
M
290
N
J
100 1 GHz 10
be some of the more important environmental radio noise
sources. For a more detailed treatment of these and other
sources of radio noise, the reader is referred to the references.
The antenna noise figure Fa for background natural radio
noise from 1 Hz to 1 THz is illustrated in Fig. 2 (15). These
data show that natural radio noise depends strongly on frequency over the radio spectrum (nominally 3 kHz to 300
GHz). In addition, several noise sources are nonstationary in
time and space (e.g., atmospheric, sun, rain). Of particular
interest for communications systems operating at or below
about 30 MHz is atmospheric noise, where Fa is random and
is characterized by its statistics. Atmospheric noise is also
non-Gaussian. The other noise sources shown in this figure
are essentially Gaussian.
For RF systems operating at frequencies of several hundred megahertz and below, man-made noise is an important
source of radio noise. Like atmospheric noise, man-made
noise is both nonstationary and non-Gaussian. Figure 3 (16)
shows the median antenna noise figure Fam for man-made
noise in four environments and galactic noise as compared
with the expected day-time and night-time levels for atmospheric noise. Man-made noise is strongly dependent on frequency and, in general, the Fam curves have a slope of 27.7
dB/decade of frequency.
Figure 4 (15) shows the details of natural radio noise over
the frequency range of 100 MHz to 100 GHz. The estimated
median business-area man-made noise has also been included. The E(90) curve shows sky noise measured with a
narrow beam antenna at zenith. The water and oxygen absorption bands are clearly visible. The E(0) curve is sky noise
0.29
100 1 THz
Figure 2. Natural radio noise (1 Hz to 1
THz) (15).
with a narrowbeam antenna directed along the earths
surface.
It was shown that when an antenna receives blackbody
radiation at a uniform temperature from all directions, Fa
does not depend on the receiving antenna gain. For most environmental noise sources, however, Fa does depend on the antenna gain and on several other factors. Appropriate corrections must be applied when the radio system-receiving
antenna differs significantly from that used to measure the
noise.
120
Atmospheric
100
Fam (dB above kT0b)
20
1 Hz
2.9 1029
Brightness temperature (K)
280
133
Night
80
60
Day
Business
Residential
Rural
40
20
0
0.1
Quiet rural
Galactic
1
10
Frequency (MHz)
100
Figure 3. Median values of Fa (16).
1000
RADIO NOISE
40
2.9 106
30
2.9 105
0
2.9 104
20
A
2.9 103
Fa (dB)
10
Figure 4. Fa versus frequency (100 MHz
to 100 GHz), where A estimated median business area man-made noise, B
galactic noise, C galactic noise (toward
galactic center with infinitely narrow
beamwidth), D quiet sun ( degree
beamwidth directed at sun), E sky
noise due to oxygen and water vapor (very
narrow beam antenna); upper curve 0 elevation angle, lower curve 90 elevation
angle, F cosmic background, 2.7 K (15).
C
B
10
F
20
Sun
T ()p() d
= 0 Ts
2.9
2.9 101
30
40
108
109
(1 GHz)
1010
2.9 102
1011
Frequency (Hz)
where 0 is the gain and p() is the pattern of the receiving
antenna; that is, () 0p(). For example, the sun has a
beamwidth of about . If a receiving antenna with gain 0 is
aimed at the sun and the pattern is essentially constant over
the intersection with the suns beam, the antenna temperature is
0
4
2.9 10
E (90)
Curves LD, LQ, F, H, and M in Fig. 2 all refer to very narrowbeam antennas pointing directly at the source. Noise from
such sources (sun, atmospheric gasses, the earths surface)
are also expressed in terms of brightness temperature. These
curves can be used to calculate the antenna temperature of a
particular receiving antenna by integrating Eq. (25) in terms
of temperature over the region occupied by the noise source:
Ta = 0
T ()p() d
(28)
4 sources
Ta =
2.9 102
E (0)
ta (K)
134
2
1440
(29)
where Ts is the brightness temperature of the sun at the desired frequency.
In Fig. 4, there are two curves associated with galactic
noise. Curve B is for an omnidirectional antenna, while curve
C is for an infinitely narrow beam aimed toward the galactic
center. Because of the relative motion of the earth and galaxy,
galactic noise is not constant in time. A more accurate determination of galactic noise for other types of antennas can be
obtained by using published radio sky data, which gives the
brightness temperature as a function of position in the sky.
Such data are available in CCIR Report 720-2 (14), which contains maps of the brightness temperature of the radio sky at
408 MHz and an approximate expression for the frequency
dependence of the temperature.
Atmospheric and Man-Made Noise
The most significant sources of environmental radio noise at
frequencies below 1 GHz are man made and atmospheric. For
these sources, the noise data were measured with a grounded
electrically short monopole antenna. Since this type of noise
most probably arrives at the receiver at relatively low elevation angles and from random directions, such an azimuthally
omnidirectional antenna is well suited for noise measurements. Predicting the antenna noise figure for other types of
receiving antennas requires an assessment of the differences
between the ideal short monopole antenna and the desired
receiving antenna. Factors that should be considered are antenna efficiency, directivity, polarization, and height above
the ground.
The direction of arrival for both atmospheric and manmade noise has been shown to be nonuniform, varying by as
much as 10 dB with direction (17). Since the noise is nonstationary, predicting Fa for high-gain antennas would likely be
arduous if worst-case estimates based on the measured data
do not provide sufficient accuracy. For azimuthally symmetric
antennas such as a half-wave dipole, a correction factor based
on the ratio of the desired antenna gain to the reference antenna gain can be applied to obtain the appropriate value
for Fa.
Since these noise processes are nonstationary, the usual
design parameter, SNR, is random and the underlying statistics of the noise process as a function of time and geographical
location must be understood to assess radio performance
properly. These characteristics are discussed in more detail
in the following sections.
Another important consideration is that both atmospheric
and man-made noise are non-Gaussian. Typically, communication system performance is calculated based on Gaussian
noise. A more detailed analysis incorporating the statistics of
the actual non-Gaussian noise process may be required in radio design and performance evaluations. Several publications
listed in the references provide information regarding the impulsive nature of these noise sources and its effect on radio receivers.
Statistics of Fa for Atmospheric Noise. Atmospheric noise is
an important consideration for wireless communication systems operating below 30 MHz. The main source of atmospheric noise is lightning. The electromagnetic energy emit-
RADIO NOISE
ted by electrical storms couples into the earth-ionosphere
waveguide, and hence, local noise levels can be significantly
influenced by distant thunderstorms. Because of ionospheric
interactions, overall atmospheric noise levels are greater at
night, as shown in Fig. 3.
In Fig. 2, curves A, B, C, and D represent the expected
range of Fa at the surface of the earth. These data are of the
average background, taking into account all times of the day,
seasons, and the entire surface of the earth. Curves A and B
give the maximum and minimum values of Fa from 1 Hz to
10 kHz. In this frequency range, there is very little seasonal,
diurnal, or geographic variation. Note that the variation of
Fa begins to increase significantly at about 100 Hz. This is
due to the variability of the Earth-ionosphere waveguide cutoff. Curves C and D give the atmospheric noise from 10 kHz
to about 30 MHz, above which the noise levels are quite low.
Curve C is the value of Fa exceeded 0.5% of the time, and
curve D is the value of Fa exceeded 99.5% of the time. These
results are derived from background atmospheric noise and
do not include effects of nearby electrical storms. A compilation of measurements showing the peak field strength for 1
mile distant lightning as a function of frequency is given in
Fig. 5 (18).
The variability of Fa, particularly in the medium frequency
(MF) and high frequency (HF) communication bands (300
kHz to 30 MHz), is so large that the bounds given in Fig. 2
alone cannot be used to obtain a useful characterization of
radio system performance. It is important, therefore, to know
how Fa and other noise statistics vary with time and location.
Starting in 1957, the average power levels and other relevant
statistics were measured on a worldwide basis using a network of 15 stations. These measurements spanned 13 kHz to
20 MHz and considered both the time of day and the season.
The results of several years of measurements were published
Peak field intensity (dB V/m, 1 kHz BW, 1 S Mile)
140
130
120
110
100
90
80
70
60
in the National Bureau of Standards (NBS) Technical Note
Series 18 (19) and later published in CCIR Report 332 (12). A
numerical representation of the data contained in Report 332
is also available (20).
The published data give, for each frequency, location, season, and time of day (measured in 4-h increments), the
month-hour median value of Fa along with values exceed 10%
(upper decile, Du) and 90% (lower decile, Dl) of the time. As
an example of these data, Fig. 6 shows worldwide values for
the median antenna noise figure Fam in the winter between
0000 and 0400 local time. The median noise figure at other
frequencies, Du, Dl, and related statistics are obtained using
the curves shown in Fig. 7.
The statistical distribution of Fa and hence the radio system SNR is readily obtained from the published data. For a
given season and measurement time block (4 h) it has been
shown that Fa is adequately represented by two log-normal
distributions (21), one above the median value and one below.
As an example, the distribution of Fa for 3 MHz at Boulder,
Colorado in the winter at 0000 to 0400 can be determined
using the data from Figs. 6 and 7. First, the 1 MHz value of
Fam at the geographic location of interest is obtained from Fig.
6 and corrected to 3 MHz using Fig. 7. Then Du and Dl as well
as their standard deviations are obtained from Fig. 7. Using
normal probability paper, these three points define the two
intersecting lines that give the two desired log-normal distributions. The resulting distribution is shown in Fig. 8. Hence,
if a radio system is operating at 3 MHz, the system performance can be conveniently specified in terms of the percent
of time that the required SNR will be available at a particular
geographic location, season, and time.
Statistics of Fa for Man-Made Noise. In 1974, Spaulding and
Disney (22) presented results from many years of measure-
Horner and Bradley (1964)
Watt and Maxwell (1957)
Horner
(expected average combined
(1962)
main stroke and precursors)
Horner and Bradley (1964)
Estimated average level of spectral amplitude
Watt and Maxwell (1957)
Horner and Bradley (1964)
(using Florman's data)
Horner (1962)
Japan URSI
Horner and Clarke (1964)
(1963)
Hallgren and MacDonald
1
Slope /f
(1963)
50
Aiya (1955)
Aiya and Sonde (1963)
Hallgren and MacDonald (1962)
Schafer and Goodall (1939)
Japan URSI (1963)
40
30
20
10
Ep
0
10
Ep
20
30
1
Distance
Horner
and Bradley
(1941)
Brook and Kitagwa (1964)
Hewitt (1957)
BW BW < 10 kHz
BW BW > 10 kHz
Slope 1/f 5/2
40
2
1 kHz
4 68
2
10 kHz
4 68
2
100 kHz
135
4 68
2 4 68
2
1 MHz
10 MHz
Frequency
4 68
2
100 MHz
4 68
2
1 GHz
4 68
10 GHz
Figure 5. Lightning emission peak field strength, 1 mile distant. (Reprinted from p. 369 of Ref.
18, by permission, 1982 IEEE.)
136
90
RADIO NOISE
60
75
90
105 120 135 150 165 180 165 150 135 120 105
90
75
60
45
30
15
15
45
60
90
40
35
80
30
50
55
80
70
70
60
60
60
60
50
50
45
55
40
40
70
30
30
60
20
20
10
90
95
85
10
75
85
90
0
10
8580
55
90
75
7085
10
80
20
20
30
30
75
75
40
60
55
50
70
65
40
70
50
60
60
50
45
70
70
80
80
90
40
60
75
90
105 120 135 150 165 180 165 150 135 120 105
90
75
60
45
30
15
15
30
45
60
90
Figure 6. Expected values of atmospheric radio noise, Fam (dB above kT0b at 1 MHz) (Winter
00000400 LT) (12).
ments of man-made radio noise. They devised methods for estimating the noise power and noise amplitude statistics that
are important in the design of radio systems. These methods
are described in the CCIR Reports (13) and have been widely
used by industry. Figure 3 summarizes these results in terms
of the median antenna noise figure Fam. As with atmospheric
noise, man-made noise is both nonstationary and nonGaussian and is a significant source of radio noise for frequencies below a few hundred megahertz. The antenna noise
figure Fa varies both in time and location. The noise level depends on the type and extent of human activities, which are
conveniently classified into four man-made noise environments (13) described in Table 1.
The within-the-hour time variability of Fa is commonly described by two log-normal distributions (21), as described previously for atmospheric noise. Values of Du and Dl are given
in CCIR Report 258 as a function of frequency and environment. More recently, Spaulding and Stewart (21) have analyzed the data used to obtain these decile values and have
found that it is appropriate to use the values Du 9.7 dB
and Dl 7 dB, independent of environmental category and
frequency. Other proposed noise models described in Report
258 include a simple Gaussian model that does not describe
the skewness observed in measured noise data and a more
complex -square model.
As an example, the distributions of Fa using Du 9.7 dB
and Dl 7 dB at 137 MHz for business, residential, rural,
and quiet rural noise environments are shown in Fig. 9. These
data include the contribution of Galactic noise, which is only
significant in the quiet rural noise environment.
Location variability is also an important consideration
when characterizing Fa. The usual assumption (22) is that
Fam is the noise figure exceeded 50% of the time at 50% of the
locations. Hence, the time distribution of Fa as shown in Fig.
9 is the noise power exceeded at 50% of the locations for a
particular environment. If it is assumed that the location
variability is Gaussian, then the value Fa that is exceeded at
other than 50% of locations is obtained from
Fa = Fa +
1 % locations
2L erfc
50
(30)
137
Fam (dB above kT0b)
20
0.01
20
40
60
80
100
120
140
160
180
0.02 0.03 0.05 0.07 0.1
0.2 0.3
50
7 10
20 30
50 70 100
0
0.01
10
12
14
16
18
20
Du
Fam
0.02 0.03 0.05 0.07 0.1
Vdm
0.5 0.7 1
De
Dt
Du
Frequency (MHz)
0.2 0.3
Vd
Figure 7. (a) Variation of radio noise with frequency (winter; 0000400 LT) (12).
Expected values of atmospheric noise. Expected values of galactic noise. (b) Data
on noise variability and character (winter; 0000400 LT) (12). Fam: Standard deviation of values
of Fam. Du: Ratio of upper decile to median value, Fam. Du: Standard deviation of values of Du.
Dl: Ratio of median value, Fam, to lower decile. Dl: Standard deviation of values of Dl. Vdm: Expected value of median deviation of average voltage. The values shown are for a bandwidth of
200 Hz. Vd: Standard deviation of Vd (12).
Frequency (MHz)
0.5 0.7 1
20
30
40
60
70
80
90
100
(dB)
7 10
20 30 40
138
RADIO NOISE
75
Fam + D + D
70
Fam + D
65
Fa (dB)
60
Fam
Fam + D D
55
Fam
Fam Dl + Dl
Fam Dl
50
45
40
Figure 8. The distribution of Fa values
for atmospheric radio noise at Boulder,
Colorado. 3 MHz, for the winter season,
00000400 hours (21).
35
.1 .2
.5
Table 1. CCIR Report 258 Definitions of Man-Made Noise
Environments
Environment
Business
Residential
Rural
Quiet Rural
Characteristics
Areas where predominant usage is for any type
of business
Areas used predominantly for single or multiple
family dwellings (at least five single-family
units per hectare), no large or busy highways
Areas where dwelling density is no more than
one every two hectares
No definition given
2.9 107
Fa (dB)
2.9 106
2.9 105
2.9 104
20
2.9
2.9 102
0
0.1
103
5 10 20 30 50 70 80 90 95 99
Percent of time Fa is exceeded
99.9
Figure 9. Distribution of Fa for man-made and galactic noise.
Temperature (K)
Business
Residential
Rural
Quiet rural
Galactic
40
10
20
30 40 50 60 70 80
Percent exceeded
90
95
98 99 99.5 99.9
99.8
where erfc1 is the inverse complimentary error function and
L is the standard deviation of the location distribution.
The location variability in terms of the standard deviation
L of the median value as a function of frequency and environment is given in Table 2 (13). As may be expected, L for the
business environment is much larger than either the residential or rural environment.
APPLICABILITY OF PUBLISHED MAN-MADE NOISE
STATISTICS TO CONTEMPORARY ENVIRONMENTS
The man-made noise statistics presented are largely based on
measurements that were made more than 20 years ago in
North America by Spaulding and Disney (22). More recently,
Spaulding has warned that the CCIR data may now be inaccurate due to technological advances (23). This is largely
based on the fact that emissions from newer automobile ignition systems, a major contributor to man-made noise in urban
Table 2. Location Variability in Terms of the Standard
Deviation for Various Environments
Frequency
(MHz)
L
Business
L
Residential
L
Rural
0.25
0.50
1.00
2.50
5.00
10.00
20.00
48.00
102.00
250.00
6.1
8.2
2.3
9.1
6.1
4.2
4.9
7.1
8.8
3.8
3.5
4.3
2.5
8.1
5.5
2.9
4.7
4.0
2.7
2.9
3.9
4.4
7.1
8.0
7.7
4.0
4.5
3.2
3.8
2.3
RADIO NOISE
40
Noise power (dB above kT0b)
areas, have decreased dramatically over the years. After reviewing more recent measurements and trend analyses,
Spaulding concluded (23) that in the business environment
at 100 MHz in the 1970s time-frame, Fam was on the order
of 20 dB but now is probably approximately 20 dB less. This
conclusion, however, is not based on a comprehensive set of
noise measurements as would be necessary to update the previous survey described in Ref. 23.
While the improvements in automobile ignition systems
have likely affected the noise levels in business and residential environments, emissions from gap discharge and corona
in power transmission and distribution lines have probably
not decreased with time. Figure 10 (22) shows Fam under, and
one-quarter mile from, a 115 kV line in rural Wyoming. It is
interesting to note that the noise measured one-quarter mile
from the power line is about the same as that predicted for a
rural environment. A possible conclusion is that if power and
distribution lines are the primary noise source in rural environments, rural man-made noise is not likely to have decreased. However, one would not expect noise in an urban
environment to be less (than rural), as would be the case with
the estimated 20 dB reduction in Fam.
Another factor that could significantly affect the level and
character of man-made radio noise is the proliferation of electronic devices (e.g., computers, electronic switching devices,
microwave ovens, etc.) that are unintentional RF emitters.
Such devices have become ubiquitous in business, residential,
and rural environments and could affect both the magnitude
of the noise power as well as its frequency dependence.
The man-made noise data presented in the previous sections are applicable to North America; the validity of extension to other parts of the world cannot be determined precisely. CCIR Report 258 describes very high frequency (VHF)
measurements made in business and residential areas of the
United Kingdom where the noise power was found to be some
10 dB below that shown in Fig. 3 (16). This is attributed to
differences in patterns of utilization of electric and mechanical appliances and regulation of interference. The report also
states that due to such differences, the noise statistics should
be used with caution. It should be noted, however, that if an
overall 10 dB reduction in urban noise can be justified, the
30
20
10
12
15
Hour (MST)
18
21
24
man-made noise environments near 100 MHz would be
bounded by what are now classified as rural (worst) and quiet
rural (best) environments, as shown in Fig. 3.
Relatively recent noise measurements at 137 MHz (24)
show that the statistics of man-made noise are significantly
different from what is predicted by CCIR Report 258. For example, Fig. 11 shows the median, mean, and peak (exceeded
0.01% of the time) values of Fa measured over a 24-h period
in a business environment. Diurnal variations corresponding
to human activity are clearly evident. The relatively steady
within-the-hour values of the mean power (Fa) are not consistent with the predicted within-the-hour distribution of Fa for
a business environment (see Fig. 9). Figure 12 shows the distribution of Fa measured at six urban sites plotted on normal
probability paper. The distribution at a particular site was
obtained by collecting statistics measured within two-minute
intervals spaced about an hour apart from hours of continuous measurements made at that particular location. Hence,
the results should correspond to the hour-to-hour time variability, which, for the most part, is relatively low at most of
40
Office park, edge
Office park, middle
Office park, edge
Downtown Denver
Downtown Boulder
Denver at I-25
115 kV lineWyoming
wooden towers
100
80
dB above kT0b
30
Fam (dB)
Figure 11. Median, mean, and peak noise power near an office
park (24).
120
Under power line
60
40
139
1/4 mile from
power line
20
10
Rural areas
20
0
0.1 0.2
0.5
5 10 20
50 100 200
Frequency (MHz)
500 1000
Figure 10. Power line noise measurements near a 115 kV line in
rural Wyoming (22).
0
0.1
5 10 20 30 50 70 80 90 95
Percent exceeding ordinate
99
99.9
Figure 12. Power averages from measurements at six urban sites
(24).
140
RADIONUCLIDE IMAGING
the locations. Location variations however are quite large, exceeding 12 dB in some cases. More importantly, these measurements show that there are business environments (downtown urban areas) where Fam is still nearly 20 dB.
In summary, the 137 MHz measurements demonstrate
that important changes have occurred in both the level and
character of man-made noise since the comprehensive noise
survey described by Spaulding and Disney (22). While these
measurements can only be considered as a spot check, they
do show that standard methods used to predict man-made radio noise are probably outdated. It is concluded that additional comprehensive man-made noise measurements at RF
frequencies into the ultrahigh frequency (UHF) band will be
necessary to provide radio system designers and engineers
with the required tools to effectively design modern radio
systems.
19. National Bureau of Standards, Quarterly radio noise data, Technical Note 18 (132), US Dept. of Commerce, Washington, DC,
19591966.
20. D. L. Lucas and J. D. Harper, A numerical representation of
CCIR Report 332: High frequency (3-30 Mc/s) atmosphere-radio
data, NBS Tech. Note 318, US Dept. of Commerce, Washington,
DC, 1965.
21. A. D. Spaulding and F. G. Stewart, An Updated Noise Model for
Use in IONCAP, US Dept. of Commerce, Boulder, CO: NTIA Rep
ort 87-212, May 1997.
22. A. D. Spaulding and R. T. Disney, Man-Made Radio Noise Part
1: Estimates for Business, Residential, and Rural Areas, US Dept.
of Commerce, Boulder, CO: OT Rep., 74-38, June 1974.
23. A. D. Spaulding, The roadway natural and man-made noise environment, IVHS J., 2: 175211, 1995.
24. R. Dalke et al., Measurement and analysis of man-made noise in
VHF and UHF bands, 1977 Wireless Commun. Conf., Boulder,
CO, Aug. 1113, 1997, pp. 229233.
BIBLIOGRAPHY
ROGER DALKE
1. W. R. Bennett, Electrical Noise, New York: McGraw-Hill, 1960,
Chap. 5.
2. J. B. Johnson, Thermal agitation of electricity in conductors,
Phys. Rev., 32: 97109, 1928.
3. H. Nyquist, Thermal agitation of electric charge in conductors,
Phys. Rev., 32: 110113, 1928.
4. C. Kittel and H. Kroemer, Thermal Physics, San Francisco, CA:
Freeman, 1980, p. 91.
5. R. Pettai, Noise in Receiving Systems, New York: Wiley, 1984,
Chap. 7.
6. CCIR, Operating Noise-Threshold of a Radio Receiving System,
CCIR Rep. 413, Geneva, Switzerland: International Telecommunications Union, 1966.
7. IRE standards on electron tubes: Definition of terms, 1957, Proc.
IRE, 45: 9831010, 1975.
8. H. T. Friis, Noise figures of radio receivers, Proc. IRE, 32: 419
422, 1944.
9. R. Pettai, Noise in Receiving Systems, New York: Wiley, 1984,
Chap. 11.
10. J. D. Kraus, Radio Astronomy, Powell, OH: Cygnus-Quasar, 1986,
Chap. 3.
11. CCIR, Worldwide Minimum External Noise Levels, 0.1 Hz to 100
GHz, Rep. 670, Geneva, Switzerland: International Telecommunications Union, 1990.
12. CCIR, Characteristics and Applications of Atmospheric Radio
Noise Data, Rep. 332-3, Geneva, Switzerland: International Telecommunications Union, 1986.
13. CCIR, Man-Made Radio Noise, Rep. 258-5, Geneva, Switzerland:
International Telecommunications Union, 1990.
14. CCIR, Radio Emission from Natural Sources in the Frequency
Range above About 50 MHz, Rep. 720-2, Geneva, Switzerland:
International Telecommunications Union, 1986.
15. A. D. Spaulding, The Natural and Man-Made Noise Environment
in Personnel Communications Services Bands, US Dept. of Commerce, Boulder, CO, NTIA Rep. 96-330 (revised), May 1997.
16. R. Achatz et al., Man-Made Noise in the 136138-MHz VHF Meteorological Satellite Band, US Dept. of Commerce, Boulder, CO:
NTIA Rep. 98-355, Sept. 1998.
17. H. Volland, CRC Handbook of Atmospherics, Boca Raton, FL:
CRC Press, 1982, p. 294.
18. T. H. Shumpert, M. A. Honnel, and G. K. Lott, Jr., Measured
spectral amplitude of lightning sferics in the HF, VHF, and UHF
bands, IEEE Trans. Electromagn. Compat., EMC-24: 368369,
1982.
US Department of Commerce
Institute for Telecommunication
Sciences
J. Webster (ed.), Wiley Encyclopedia of Electrical and Electronics Engineering
c 1999 John Wiley & Sons, Inc.
Copyright
IONOSPHERE CHARACTERISTICS
The ionosphere poses an interesting challenge for many radio systems that make use of signal transmission
through all or some portion of that medium. Being a magnetoionic medium imbedded in a background neutral
atmosphere, it exhibits very interesting refractive properties, including anisotropy, dispersion, and dissipation.
The laminar ionosphere introduces an array of effects, which are related to the ionospheric component of radio
refractivity. These include ray path bending, phase path increase, group path delay, absorption, Doppler shift,
pulse dispersion, Faraday rotation, and magnetoionic path splitting. Inhomogeneities in the ionosphere give
rise to temporal and spatial variations in the effects just cited. An understanding of the ionospheric personality
provides information about a wide range of solar-terrestrial interactions, and it has significant space-weather
implications. Space weather is a new discipline that includes a wide range of exoatmospheric phenomena of
major importance to space systems and their operational effectiveness.
The main features of the ionosphere are well known, although details are subjects of continuing research.
There are many excellent sources of information about the ionosphere, from both a theoretical and an experimental perspective. The books by Davies (1,2,3), Ratcliffe (4), and Giraud and Petit (5) should be consulted.
Theoretical and plasma-physics aspects of the ionosphere have been discussed in a book by Kelley (6). A readable account of the basic physics of the ionosphere has been given by Rishbeth (7). Other useful references,
which place the ionosphere within a larger context of the geospace weather system, include the Air Force
Handbook of Geophysics and the Space Environment (8), and an Introduction to the Space Environment by
Tascione (9). Various techniques for probing the ionosphere have been described in a monograph by Hunsucker
(10). From a practical perspective, Goodman (11), Johnson et al. (12), and McNamara (13) have published expositions on the ionosphere in connection with radio system applications. There are also proceedings of topical
conferences and workshops. The Ionospheric Effects Symposia (14) have chronicled ionospheric research activities and applications since 1975; and the Commission of the European Communities has published reports
dealing with ionospheric prediction and modeling (15,16).
The purpose of this article is to provide a general understanding of the ionosphere. The emphasis is on
those ionospheric processes and phenomena that are encountered by users of radio propagation systems. More
complete descriptions of the underlying physical processes may be found in various references cited in the text.
A final section on the current status of ionospheric research is provided as an aid to specialists and graduate
students.
General Properties of the Ionosphere
Basic Structure. The ionosphere is an ionized region in the upper atmosphere that, by generally accepted convention, lies between an altitude range of 60 km to 1000 km. Nevertheless, the region above 1000 km
but below 2000 km, called the protonosphere, is also ionized and may be important when considering the totality of ionization effects on radio systems. As a matter of convenience, some specialists have combined the
ionosphere and protonosphere into a single region of ionization. For example, the integrated electron density
from a ground station to a geosynchronous satellite (referenced to the vertical) is referred to as the total electron
1
IONOSPHERE CHARACTERISTICS
Fig. 1. Depiction of the ionospheric layers and the diurnal variation. [By permission of J. M. Goodman and Kluwer
Academic Publishers, Norwell, MA (11).]
content of the ionosphere (TEC), even though both ionospheric and protonospheric electrons contribute to the
integral. For the purpose of this article, we shall use the more restricted definition for the ionosphere, generally
placing the upper limit at approximately 1000 km. While there are equal numbers of free electrons and positive ions within the ionosphere, it is the electron number density that characterizes the array of interesting
phenomena associated with the region. The ionosphere is imbedded in the earths magnetic field, and this
situation influences the distribution of the ionized constituents. A clear indication of this may be seen in the
worldwide distribution of electron density in the upper ionosphere, which tends to be described by geomagnetic
rather than geographic coordinates. Moreover, being a magnetoionic medium, the ionosphere has a profound
effect upon radiowaves that interact with the medium.
The ionospheric electron density distribution is logically evaluated first in terms of its height profile,
followed by its geographical and temporal variabilities. Though there is abundant evidence suggesting a
rather complex electron density profile comprised of several peaks and valleys, the basis for understanding
fundamental properties of the ionosphere comes from a simple picture of an ionized medium dominated by a
single region, or layer, having a distinct maximum in electron density. This is not without justification, since the
highest and thickest component region, the so-called F layer, typically exhibits the greatest electron density.
Moreover, in many radiowave applications, it is the F layer that exhibits the dominant interaction. Figure 1
IONOSPHERE CHARACTERISTICS
Fig. 2. Various atmospheric and ionospheric layers, the depth of penetration of solar radiation, and the thermospheric
temperature profile. [From a National Research Council report (62).]
depicts the various regions or layers of the ionosphere in terms of the electron number density. It has been
observed that the height profile varies diurnally, seasonally, and as a function of solar activity.
Formation of the Ionosphere. The sun exerts a number of influences on the upper atmosphere, but
the interactions of most importance for our discussion are photodissociation and photoionization. Figure 2
depicts the neutral atmosphere, its various regions, and the depth of penetration of the various components of
solar flux.
In the lower atmosphere, species such as N2 and O2 dominate the constituent population, though other
species such as water vapor, carbon dioxide, nitric oxide, and trace element gases are influential in specific
contexts. In the upper atmosphere, however, molecular forms are dissociated by incoming solar flux into
separate atomic components. Formally the lowest portion of the ionosphere is the so-called D layer at an
altitude of 60 km 20 km, but the free-electron and ion population rises dramatically at an altitude of
100 km, which is the median altitude of the E layer. Two things occur at this altitude. First, oxygen becomes
dissociated as a result of solar UV radiation. Secondly, the mixing of the atmosphere, so efficient below 100 km,
ceases rather dramatically, and the region where this occurs is called the turbopause.
The process of dissociation is so efficient that we treat the distribution of neutral species in a vast segment
of the upper atmosphere (i.e., above 200 km) as that of a monatomic gas. In the lower atmosphere (i.e., below
IONOSPHERE CHARACTERISTICS
Fig. 3. (a) Profiles of ion concentrations, as a function of height, for daytime conditions. (b) Electron density distributions
for day/night and solar maximum/minimum conditions. [From Jursa (8).]
roughly 200 km), the gas is largely polyatomic, although the transition between the two regimes is rather
gradual between 100 km and 200 km. This has implications for the lifetime of ionelectron pairs created
through photoionization. Also, in the altitude regime above about 200 km and well above the turbopause,
collisions become a rarity, so that mixing of the various species becomes unimportant in comparison with
diffusive forces. As a consequence, diffusive separation occurs, with constituents of the neutral gas seeking
their own height distributions dictated by their atomic masses, the gas temperature, and the acceleration of
gravity. Figure 3(a) shows height profiles of ionic species in the upper atmosphere, and Fig. 3(b) shows typical
IONOSPHERE CHARACTERISTICS
distributions of midlatitude electron density for daytime and nighttime under solar maximum and minimum
conditions.
It may be seen that ionized monatomic oxygen is the majority ion between roughly 180 km and 800 km,
and is wholly dominant between about 200 km and 500 km. Atomic hydrogen ions become important above
500 km, and the region from about 800 km to 2000 km is called the protonosphere. It should also be noted that
above 500 km (i.e., the base of the exosphere), the neutral atmosphere is virtually collisionless and particles
tend to move about freely. On the other hand, electrons and ions in the exosphere are still influenced by the
earths magnetic field and electrodynamic forces.
The electron density distributions in the ionosphere and protonosphere are variable. Because of this,
the boundary between the ionosphere and the protonosphere is not sharply defined, being dependent upon
a number of factors including time of day, season, and solar activity. The protonosphere is often referred to
as the plasmasphere, especially by magnetospheric scientists and those engaged in transionospheric TEC
measurements.
Ionospheric Layering. Table 1 provides information about the various ionospheric layers, the altitude
ranges of each, the principal ionic constituents, and the means of formation. A comment is appropriate here
on the nature of ionospheric layering, with some emphasis on the historical distinctions made between the
words layer and region as they pertain to the ionosphere. Often the terms are used interchangeably, and while
neither is generally preferred, region is the more accurate description. This is because it does not convey
the incorrect impression that sharp discontinuities in electron density exist at well-defined upper and lower
boundaries. This is especially the case for the F region, and to a lesser extent for the D and E regions. From a
historical perspective, the concept of layering derives from the appearance of the ionospheric regions on verticalincidence ionospheric soundings, called ionograms (see the subsection Sounder Measurement Method below).
Furthermore, the alphabetic designation of the ionospheric regions was also based upon the early sounding
studies. On the other hand, there are certain situations for which the restrictive term layer is acceptable. For
example, the normal E region may occasionally be characterized by an electron density profile displaying a
degree of boundary sharpness. Aside from this, the most significant localized concentration of free electrons
in the ionosphere, called sporadic E (or Es), exists as an isolated layer within the boundaries of the normal E
region (see the section Sporadic E below). It is termed sporadic because it exhibits a generally unpredictable
temporal and geographical distribution, and because of its limited geographical extent, it is sometimes referred
to as a sporadic E patch.
As indicated above, the ionosphere is often described in terms of its component regions or layers. These
were the so-called D, E, and F regions. These designations are largely based upon data obtained from crude
sounder (i.e., ionogram) measurements undertaken in the 1920s and 1930s. These early measurements often
exhibited evidence for an additional layer between regions E and F in the daytime ionosphere. This led to the
notion that the F region is actually composed of two distinct regions (F1 and F2 ) having different properties.
The lowest region of the ionosphere, the D region, is important in the characterization of absorption losses
for short-wave systems, and also as a reflecting layer for long-wave communication and navigation systems.
There is also evidence for a bifurcation in the D region, with the upper portion (above 60 km) being produced
by solar flux, and with the lower portion (below 60 km) being produced by galactic cosmic rays.
Ground-based vertical-incidence sounder measurements have provided the bulk of our current information about ionospheric structure (see the subsection Sounder Measurement Method below). Through application of ionogram inversion technology to allow for the radio-wave interaction effects, individual sounder stations
provide information about the vertical distribution of ionization to the altitude of the F2 maximum (i.e., 300 km
to 400 km). In addition, the worldwide distribution of these systems has allowed a good geographical picture
to be developed using sophisticated mapping algorithms. These measurements are somewhat limited in the
characterization of certain features such as the so-called EF valley, and they cannot evaluate ionization above
the F2 maximum. There is also a paucity of data over oceanic regions. Satellite measurements (viz., topside
sounders and in situ probes) have been invaluable in the characterization of the F-region ionization density
IONOSPHERE CHARACTERISTICS
over oceanic regions. Rocket probes and incoherent backscatter radar measurements, which provide a clearer
representation of the true electron density profile, typically reveal a relatively featureless profile exhibiting a
single F-region maximum with several underlying ledges or profile derivative discontinuities. Nevertheless,
a valley of ionization may often be observed between the E and F regions. Ionization above the F2 maximum
may be deduced from satellite probes and Thomson-scattering radars, but a large amount of information has
been derived from total electron content measurements using Faraday rotation or group path measurements of
signals from geostationary satellites or Global Positioning System (GPS) satellites. Hunsucker (10) describes
various ionospheric measurement techniques.
Simple layering occurs as the result of two factors. First, the atmospheric neutral density decreases exponentially with altitude, while the solar ionizing flux density increases with height above sea level. This leads
to the formation of single region for which the ionization rate is maximized, and ultimately results in a layer
having the so-called Chapman shape. This shape is based upon a simple theory advanced by Sidney Chapman
in 1931 (Ref. 17; see Fig. 4). We observe nonetheless a degree of structure in the ionosphere, which suggests
more than one layer. One cause for multilayer formation is the existence of a multicomponent atmosphere,
each component of which possesses a separate height distribution at ionospheric altitudes. But there are other
factors. Solar radiation is not monochromatic as suggested in the simple Chapman theory, and it has an energy density that is not evenly distributed in the wavelength domain. Furthermore, its penetration depth and
ionization capability depend upon wavelength and atmospheric constitution. All of this results is a photoionization rate, and an associated electron density profile, that are structured functions of altitude. It has been
shown that the Chapman model is valid for the D, E, and F1 regions but is not generally valid for the F2 region.
Chapman Layer Theory. One of the basic tenets of Chapman theory is that solar radiation will
penetrate to an altitude for which the total number of atoms or molecules, P (populating a column of unit
cross sectional area directed toward the sun) is equal to the reciprocal of the absorption (or interaction) cross
IONOSPHERE CHARACTERISTICS
Fig. 4. An idealized representation of ionization production in the atmosphere as the solar radiation encounters a neutral
gas with exponentially increasing density. [By permission of J. M. Goodman and Kluwer Academic Publishers, Norwell,
MA (11).]
section a that is P = 1/a. The peak in ionization will be produced in the neighborhood of that altitude, and the
concept is valid for oblique solar illumination as well as for the case in which the sun is directly overhead. It
is convenient to look at the production rate in terms of its deviation from the peak (overhead) value at height
h0 . For this it is useful to define a reduced height z, corresponding to the normalized departure of an arbitrary
value of ionospheric height h from h0 .
where h0 is the peak height for vertically incident radiation from the sun, and H is the neutral scale height
given by the following expression:
where k is Boltzmanns constant, T is the absolute gas temperature, m is the atomic or molecular mass, and g
is the acceleration of gravity. Within the thermosphere (with h 100 km), the gas temperature is monotonically
increasing, reaching an asymptotic level near the base of the exosphere. The temperature rises from 180
K at the mesopause (and incidentally near the turbopause) to levels approaching a diurnal range of 600 K to
1100 K at solar minimum and 800 K to 1400 K at solar maximum. The heat sources include solar radiation,
the dissipation of atmospheric gravity waves, and particle precipitation. The asymptotic levels of T are due to
limits on the thermal conductivity of the gas. The scale height H is a convenient parameter, since it may be
used as a measure of layer thickness for an equivalent fixed-density slab. More importantly, it has a physical
meaning. If the atmosphere is in diffusive equilibrium governed by the force of gravity and the gas pressure
IONOSPHERE CHARACTERISTICS
gradient, and N is the atomic or molecular gas density (as appropriate), we have
where N 0 is the atomic or molecular density at some reference height.
In a diffusively separated atmospheric environment, each constituent has its own unique scale height
governed by its own molecular (atomic) mass. In an ionized gas in which the electrons and ions are coupled by
electrostatic forces, the effective value of the mean molecular mass is 12 the mass of the positive ion. This is
because the mass of the electron is essentially zero in comparison with the ion mass.
Figure 5 depicts the production-rate curves associated with an ideal Chapman-like production profile and
a range of solar zenith angles . It is seen that there are a number of curves, parametrized in terms of ,
for which production rate maxima qmax , may be observed. The largest qmax occurs for = 0 (overhead case
corresponding to q = q0 ), and we see that other values for qmax , corresponding to oblique geometries wherein
= 0, will decrease in magnitude and occur at increasing heights as becomes larger (i.e., the sun moves toward
the horizon). Chapman theory yields the following rate-of-production formula:
At altitudes well above the peak in q, the rate of electron production drops off in an exponential fashion
imitating the exponential decrease in gas pressure with height. In order to relate Chapman production curves
to actual electron density distributions, we must examine loss processes and certain dynamic factors.
The Continuity Equation and Equilibrium Processes
The equation that expresses the time rate of change of electron concentration, N e , is the continuity equation:
where N e is the electron density, L(N e ) is the loss rate, which is dependent upon the electron density, div stands
for the vector divergence operator, and V is the electron drift velocity.
The divergence of the vector in Eq. 6 is the transport term, sometimes conveniently called the movement
term. The continuity equation says that the time derivative of the electron density within a unit volume is equal
to the number of electrons that are generated within the volume (through photoionization processes) minus the
number that are lost (through chemical recombination or attachment processes), and finally adjusted for those
electrons that exit or enter the volume (as expressed by the transport term). To first order, the only derivatives
of importance in the divergence term are in the vertical direction, since horizontal N e gradients are generally
smaller than vertical ones. In addition, there is a tendency for horizontal velocities to be small in comparison
with vertical drift velocities. Consequently, we may replace div (N e V) by (d/dh) (N e V h ), where V h is the scalar
IONOSPHERE CHARACTERISTICS
Fig. 5. Curve illustrating the rate of electron production as a function of reduced height (h h0 ) and for selected values
of the solar zenith angle. [From Davies (1).]
velocity in the vertical direction. We rewrite Eq. 6 as follows:
10
IONOSPHERE CHARACTERISTICS
Now let us look at some special cases. If V h = 0 (no movement), then the time variation in electron
concentration is controlled by a competition between production q and loss L. At nighttime, we may take q =
0, and this results in
In principle, there are two mechanisms to explain electron loss: attachment of electrons to neutral atoms
(in the upper ionosphere), and recombination of electrons with positive ions (in the lower ionosphere). The
attachment process is proportional to N e alone, while recombination depends upon N e with N i , where N i is the
number of ions. Attachment involves radiative processes and has an extremely low cross section (probability
of occurrence). We may ignore it in many practical situations and take recombination as the major source
for electron loss. Since N e = N i , the recombination process obeys the equation L = N e 2 , where is the
recombination coefficient. Recombination is very rapid in the D and E regions, the process being accomplished
in a time on the order of seconds to minutes. Attachment, the electron loss process for the upper ionosphere, has
a time constant on the order of hours. This is the primary reason that the ionosphere does not entirely disappear
overnight. Another reason is that there exists a second source of electrons associated with the plasmasphere.
This reservoir of ionization is built up during the daytime through vertical drift, but bleeds into the ionosphere
during nocturnal hours.
In the vicinity of local noon, dN e /dt = 0 and we may analyze the quasiequilibrium conditions suggested
by Eq. (7) when the left-hand side of the equation equals 0. The two main types of equilibrium processes are
given in Table 2.
The equilibrium processes identified in Table 2 are the dominant possibilities during daytime when
photoionization is significant. During nocturnal hours, equilibrium is seldom achieved at F-region heights,
although it is approached in the period before sunrise.
While the continuity equation appears quite simple, the generic terms (i.e., production, loss, and transport)
represent a host of complex photochemical and electrodynamic processes, which exhibit global variations and
are influenced by nonstationary boundary conditions within the atmosphere and the overlying magnetosphere.
Notwithstanding these complications, the equation provides a remarkably clear view of the basic processes
that account for ionospheric behavior. In fact, the relative contributions of terms in the continuity equation
will account for the majority of the anomalous ionospheric properties; that is, those ionospheric variations that
depart from a Chapman-like characteristic. This is especially true for the F2 layer, within which the movement
term attains paramount status. In the E and F1 regions, where the movement term is small compared with
production and loss (through recombination), photochemical equilibrium exists in the neighborhood of midday.
All of this has had a significant bearing on the development of ionospheric models and prediction methods.
IONOSPHERE CHARACTERISTICS
11
Indeed, as it relates to the F region of the ionosphere, it may be said that the existence of a nonvanishing
divergence term in the continuity equation has been the primary impetus for the development of statistical
modeling approaches. Nevertheless, efforts to account for all terms in the continuity equation through physical
modeling are ongoing.
The underlying assumptions used by Chapman in his theory of layer production are in substantial
disagreement with observation. The Chapman layer was based upon an isothermal atmosphere, and it is well
known that the atmosphere has a scale height, kT/mg, which varies with height. Moreover, the basic theory
assumes a monochromatic source for photoionization and a single constituent gas. Corrections and extensions
to the early Chapman theory have led to better agreement with observation, and to this day the Chapman
layer provides a fundamental baseline for ionospheric profile modeling.
Description of the Ionospheric Layers
Sounder Measurement Method. In any discussion of the ionospheric electron density distribution, it
is important to recognize that many experimental methods have been used to arrive at our current understanding. The major ones include ground-based vertical-incidence sounding (VIS), topside sounding using satellite
platforms, incoherent backscatter radar, the Faraday rotation and signal delay of satellite signals, and in situ
measurements using rocket probes and satellites. The VIS method, which employs the high-frequency (HF)
band, was the earliest method and has provided the most comprehensive picture of the lower ionosphere and its
worldwide distribution. The ionospheric D layer is an exception, and special methods are needed to determine
the electron densities in that region. Much of the current nomenclature involving ionospheric structure and
phenomena is a carryover from early VIS investigations. As a consequence we shall briefly examine the VIS
method. For a discussion of other methods, the reader is referred to Hunsucker (10).
The plasma frequency associated with an electron gas, f p (a natural resonant frequency), is proportional
to the square root of the electron density of the gas:
where f p is in hertz and N e is in electrons per cubic meter.
It may be shown that a radio wave, propagating vertically upward into the ionosphere, will penetrate the
region until it reaches a point at which the sounding frequency matches the plasma frequency. All frequencies
less than this value will be reflected back to ground. An ionospheric sounder is essentially a radar, which maps
out the height-dependent ionospheric electron concentration versus transmission frequency, where the probing
frequency is typically a stepwise increasing function of time. A plot of signal echo time delay versus transmission
frequency is called an ionogram. A typical ionogram and the corresponding ionospheric profile are given in
Figure 6. If N max is the maximum electron density of a layer, then we define a so-called critical frequency of
reflection, f c , which is the maximum plasma frequency within the layer. If the sounder transmission frequency
exceeds f c , then the signal is not reflected and penetrates the layer. There are as many critical frequencies in
the ionosphere as there are layers or regions. A more complete treatment of the theory of radio propagation
in the ionosphere shows that a magnetoionic medium supports two modes of propagation, the ordinary (O
mode) and the extraordinary (X mode). These modes encounter slightly different indices of refraction and
thus travel with slightly different velocities and directions. As a consequence, each ionogram consists of two
traces, corresponding to O- and X-mode echoes. These traces may be closely aligned over a large portion of
their respective propagation bands but can depart significantly at their respective critical frequencies, with
the X mode supporting somewhat higher-frequency signal reflections. By convention, the O-mode trace is used
for conversion of ionogram critical frequencies into maximum electron densities. The following convenient
12
IONOSPHERE CHARACTERISTICS
Fig. 6. Typical vertical-incidence ionosonde recording (i.e., ionogram) and the corresponding plasma frequency profile
fp(h). The electron density profile is related to the plasma frequency profile by the Eq. 9 in the text. [This ionogram was
derived from U.S. Government web site (http://www.ngdc.noaa.gov/stp/), and the instrument was developed by University
of Massachusetts-Lowell for the U.S. Air Force.]
expression is used:
where f O is the ordinary-ray critical frequency (MHz) and N max is the maximum electron density of the given
layer (e/m3 ). Equation (10) is equivalent to Eq. (9).
From a historical perspective, it is interesting to note that the concept of radar detection of aircraft
derived from the early work of ionospheric specialists who were already using ionospheric sounders as a means
to detect ionospheric layers.
The D Region. The D region is responsible for most of the absorption encountered by HF signals, which
exploit the sky-wave mode. In most instances, D-region absorption is a primary factor in the determination
of the lowest frequency, which is useful for communication over a fixed sky-wave circuit. In addition, the D
region supports long-wave propagation at very low frequency (VLF) and low frequency (LF), and the medium
is exploited in certain legacy navigation systems and strategic low-rate communication systems. The sounder
method as described in the previous section is not useful for measurement of the D region, since the electron densities are relatively low. Details of D-region electron concentration are sketchy in comparison with
IONOSPHERE CHARACTERISTICS
13
information available about the E and F regions, principally because of the difficulty in making diagnostic
measurements. Moreover, analysis is hampered because many photochemical processes with poorly defined
reaction rates take place in the D region. Over 100 reactions have been compiled.
Table 1 shows that the D region lies between 70 km and 90 km. In fact, the upper and lower levels are
not precisely defined. It is evident that more than one source of ionization gives rise to the D-region electron
density distribution. Sources include solar radiation at the upper levels and galactic cosmic rays at lower
levels. In addition, relatively rare polar-cap absorption (PCA) events are characterized by highly energetic
solar protons that provide an additional source for ionization of the lower D region within the polar cap. Some
investigators place the lower boundary of the D region at 50 km to allow for the contribution of galactic cosmic
rays in the neighborhood of 50 km to 70 km. This altitude regime, termed the C region, is not produced by
solar radiation. It exhibits different characteristics from the region between 70 km and 90 km. Specifically, a
minimum in electron concentration is observed during solar-maximum conditions for the lower portion (viz.,
region C), while the reverse is true in the upper portion (viz., region D). This can be explained if we assume
that the galactic cosmic rays are partly diverted from the earth by an increase in the interplanetary magnetic
field (IMF), which occurs during solar maximum conditions.
The E Region. In an () Chapman layer for which photochemical equilibrium has been established,
the following equation represents the electron density distribution as a function of reduced height z:
where is the recombination coefficient, is the solar zenith angle, and q0 is the maximum production rate
in the layer. Recall that is the recombination coefficient (see the section The Continuity Equation and
Equilibrium Processes above). The quantity q0 / is dependent upon the sunspot number and is specific to the
region involved, in this case the E region. The maximum rate of electron production q0 occurs only for the
overhead sun. However, it may be shown that actual maxima for other zenith angles are simply related by this
expression:
It may be shown that the ordinary-ray critical frequency for the E region, which is directly related to the
E-region maximum electron density through Eq. (10), may be found from Eq. (11), and is given by
where is a constant of proportionality, which is dependent upon the sunspot number. The exponent n tends to
a value 0.25 for long-term seasonal behavior, and in compliance with Chapman theory, but some workers have
found that a value for n 0.3 better represents the diurnal dependence. The constant of proportionality ranges
between about 3 MHz and 4 MHz, bearing in mind that Eq. (13) represents a climatological median value.
The solar-activity dependence of the ratio of peak production to the effective loss (recombination) coefficient has been studied by a number of workers, and the results enable values of foE to be deduced. There have
also been direct measurements of foE using vertical incidence sounders. While there is some variability to be
considered, it is possible to develop a relationship connecting the median value of foE, the solar zenith angle,
and the 12-month running-mean sunspot number. A generally accepted candidate for the daytime E-region
critical frequency is
where R12 is the running 12-month sunspot number, which may range between roughly 10 and 150.
Equation (14) provides excellent agreement with observation during the daytime, but alternative expressions are found to be more appropriate during the nighttime hours (3). Moreover, it has been found that
14
IONOSPHERE CHARACTERISTICS
Fig. 7. (a) Depiction of the local-time (LST) and latitude dependence of foE for solar-maximum conditions in summer
[from Davies (1)](b) Contours of foE at Fort Belvoir, Virginia, in 1958 (solar maximum), showing seasonal variations. The
contours are in megahertz.
Eq. (14) is inaccurate in the very high latitudes, where other means of electron production become important,
invalidating the Chapman hypothesis. Internationally adopted relations for monthly median foE are due to
Muggleton (18); and an alternative relation, specific to the European region, has been published (15).
Figure 7(a) contains an E-region critical-frequency map for summer solstice conditions in 1958, a period
of high solar activity (i.e., R12 large). The contours are representative of median conditions as a function of
IONOSPHERE CHARACTERISTICS
15
geographic latitude and local time. It is seen that the E-region critical frequencies (and consequently the
electron densities) are vanishingly small in regions devoid of solar illumination. This summer solstice behavior
is consistent with Eq. (14), and other seasons have been shown to behave in conformance with (cos )0.25 as
well.
Figure 7(b) shows the monthly variation of foE for one station (Ft. Belvoir, Virginia) for the year 1958.
The solar control is obvious in the median data plotted.
The F1 Region. The F1 region is not unlike the E region in the sense that it obeys many of the
predictions of Chapman theory. We look for a relation for the ordinary-ray critical frequency that is formally
similar to Eq. (13). A relation patterned after Chapman principles may be expressed as
Like the E region, the F1 region exhibits more complicated behavior than that expressed by such a simple
formula. Specifically, it has been found that the geomagnetic latitude tends to exhibit some control over the
F1 -region electron densities. The function f s in Eq. (15) depends upon sunspot number and magnetic latitude.
It is also observed that the F1 region disappears (i.e., merges with the F2 region) at values of the solar zenith
angle exceeding a certain maximum that itself depends upon both the sunspot number and the geomagnetic
latitude. The Radio Sector of the International Telecommunications Union (ITU-R, previously the CCIR) has
developed a method for computing foF1 taking all these factors into account (18). The internationally adopted
monthly median foF1 formulation is based on the work of Ducharme et al. (19). The relation due to Davies (3)
gives a convenient but approximate expression for the F1 -layer critical frequency:
Figure 8 shows the solar-zenith-angle control of foF1 under sunspot maximum and minimum conditions.
The height of the F1 ledge, hF1, is taken to be between 180 km and 210 km. From Chapman theory we
anticipate that hF1 will be lower in summer than in winter and will be higher at midlatitudes than at low
latitudes. Unfortunately, the reverse is true. Explanations for this behavior may be found in a detailed study of
scale-height gradients, a nonvanishing movement term (as expressed in the continuity equation), or gradients
in upper atmospheric chemistry.
The F2 Region. The F2 region is the most prominent layer in the ionosphere, and this significance
arises as a result of its height (it is the highest of all the component layers) and of course its dominant electron
density. It is also characterized by large ensembles of irregularity scales {L} and temporal variations {T}.
The F2 region is a vast zone, which eludes prediction on the microscale ( L < 1 km) and mesoscale (1 km <
L < 1000 km) levels, and even provides challenges to forecasters for global and macroscale (L > 1000 km)
variations. This is largely because of the elusive transport term in the continuity equation. There are also a
host of so-called anomalous variations to consider, and these are the subjects of a succeeding section.
As in the E and F1 regions, we may conveniently specify the behavior of the F2 region in terms of equivalent
plasma frequency rather than the electron density. For the peak of ionization we have
where foF2 is the ordinary-ray critical frequency.
While foF2 exhibits solar-zenith-angle, sunspot-number, and geomagnetic-latitude dependences, simple
algebraic algorithms do not characterize these relationships. As a consequence, mapping methods are used to
describe the F2 region electron density patterns.
16
IONOSPHERE CHARACTERISTICS
Fig. 8. Depiction of the local-time and latitudinal variation of foF1 for two different solar activity conditions: (a) solar
minimum, June 1954; (b) solar maximum, June 1958. The contours are in megahertz. [From Jursa (8).]
The CCIR published its CCIR Atlas of Ionospheric Characteristics, which includes global maps of F2 -layer
properties for sunspot numbers of 0 and 100, for every month, and for every even hour of Universal Time (20).
Figure 9 is an illustration of the global distribution of foF2 for a sunspot number of 100. Such maps are derived
from coefficients based upon data obtained from a number of ionosonde stations for the years 19541958 as
well as for the year 1964. This set of coefficients is sometimes identified by an ITS prefix, but is known more
IONOSPHERE CHARACTERISTICS
17
Fig. 9. Map of foF2 showing the worldwide distribution under the following conditions: 15 November, Sunspot Number =
135, Time = 0000 UTC. The countours of foF2 are developed using the URSI set of ionospheric coefficients. Curves similar
to this are found in the Atlas of Global Ionospheric Coefficients (20). [By permission, Radio Propagation Services, November
2000.]
generally as the CCIR coefficients. Because of the paucity of data over oceanic areas, a method for improving
the basic set of coefficients by adding theoretically derived data points was developed. As a result, a new set of
coefficients has been sanctioned by International Union of Radio Science (URSI), and this is termed the URSI
coefficient set. Many communication prediction codes, which require ionospheric submodels, allow selection of
either set of ionospheric coefficients.
Anomalous Features of the Ionospheric F Region. The F2 layer of the ionosphere is probably the
most important region for many radio-wave systems. Unfortunately, the F2 layer exhibits the greatest degree of
unpredictable variability because of the transport term in the continuity equation. As indicated previously, this
term represents the influences of ionospheric winds, diffusion, and dynamical forces. The Chapman description
for ionospheric behavior depends critically upon the unimportance of the transport function. Consequently,
many of the attractive, and intuitive, features of the Chapman model are not observed in the F2 region.
The differences between actual observations and predictions derived on the basis of a hypothetical Chapman
description have been termed anomalies. In many instances, this non-Chapman-like behavior is not anomalous
at all, but rather typical.
The following list represents the major forms of anomalous behavior in the F2 layer: diurnal, Appleton,
December, winter, and the F-region trough. A few comments are provided for each major form.
The Diurnal Anomaly. The diurnal anomaly refers to the situation in which the maximum value of
ionization in the F2 layer occurs at a time other than at local noon as predicted by Chapman theory. On a
statistical basis, the actual maximum occurs typically in the temporal neighborhood of 1300 to 1500 LMT.
Furthermore, there is a semidiurnal component that produces secondary maxima at approximately 1000 to
18
IONOSPHERE CHARACTERISTICS
1100 LMT and 2200 to 2300 LMT. Two daytime maxima are sometimes observed (one near 1000 and the other
near 1400), and these may cause the appearance of a minimum at local noon. This feature, when observed, is
called the midday biteout.
Appleton Anomaly. This feature is symmetric about the geomagnetic equator and goes by a number of
other names, including the geographic anomaly, the geomagnetic anomaly, and the equatorial anomaly. The
Appleton anomaly is associated with the significant departure in the latitudinal distribution of the maximum
electron concentration within 20 to 30 on either side of the geomagnetic equator. Early in the morning a
single ionization peak is observed over the magnetic equator. However, after a few hours the equatorial F
region is characterized by two distinct crests of ionization that increase in electron density as they migrate
poleward. This phenomenon is described as an equatorial fountain initiated by an E B plasma drift (termed
a Hall drift), where E is the equatorial electrojet electric field and B is the geomagnetic field vector. This drift is
upwards during the day, since the equatorial electric field E is eastward at that time. As the electrojet decays,
the displaced plasma is now subject to downward diffusion when the atmosphere begins to cool. This diffusion
is constrained along paths parallel to B, which map to either side of the geomagnetic equator. The poleward
extent of the anomaly crests is increased if the initial Hall-drift amplitude is large. This anomalous behavior
accounts for the valley in the parameter foF2 (with peaks on either side) seen at the geomagnetic equator in
Figure 9. There are significant day-to-day, seasonal, and solar-controlled variations in the onset, magnitude,
and position of the anomaly. There are also asymmetries in the anomaly crest position and electron density.
Asymmetries in the electron density in the anomaly crests appear to be the result of thermospheric winds that
blow across the equator from the subsolar point. The effect of magnetic activity on the anomaly is to constrain
the electron density and latitudinal separation of the crests. Magnetic activity is monitored worldwide, and the
quasilogarithmic index Kp is used to represent the level of worldwide activity (21). When Kp 5 (on a scale
from 0 to 9), the anomaly disappears.
The December Anomaly. This term refers to the fact that the electron density at the F2 peak over
the entire earth is 20% higher in December than in June, even though the solar-flux change due to earth
eccentricity is only 5% (with the maximum in January).
The Winter (Seasonal) Anomaly. This is the effect in which the noontime peak electron densities are
higher in the winter than in the summer despite the fact that solar zenith angle is smaller in the summer
than it is in the winter. This effect is modulated by the 11-year solar cycle and virtually disappears at solar
minimum.
The F-Region (High-Latitude) Trough. This is representative of a number of anomalous features that are
associated with various circumpolar phenomena, including particle precipitation, the auroral arc formations,
etc. The high-latitude trough is a depression in ionization, occurring mainly in the nighttime sector, and it is
most evident in the upper F region (22). It extends from 2 to 10 equatorward of the auroral oval, an annular
region of enhanced ionization associated with optical aurora (see the section The High-Latitude Ionosphere).
The trough region is associated with a mapping of the plasmapause onto the ionosphere along geomagnetic
field lines (see Fig. 17). The low electron density within the trough results from a lack of replenishment through
candidate processes such as antisunward drift, particle precipitation, or the storage effect of closed field lines.
The latitudinal boundaries of the trough may be sharp, especially the poleward boundary with the auroral
oval. A model of the trough is due to Halcrow and Nisbet (23).
Irregularities in the Ionosphere. In addition to the various anomalous features, irregularities in the
electron density distribution may be observed throughout the ionosphere. The size, intensity, and location of
these irregular formations are dependent upon a number of factors, including geographical area, season, time
of day, and levels of solar and magnetic activity. The traveling ionospheric disturbance (TID; see the subsection
Short-Term Variations and the section Ionospheric Predictions below) belongs to a special class of irregular
formations that are generally associated with significant changes in the electron density (more than a few
percent) over large distances (> 10 km). The remaining irregularities, loosely termed ionospheric inhomogeneities, typically develop as the result of ionospheric instability processes and are not directly associated
IONOSPHERE CHARACTERISTICS
19
with TIDs. On the other hand, TIDs have been shown to be a possible catalyst in the formation of ionospheric
inhomogeneities, especially in the vicinity of the Appleton anomaly. Relatively small-scale ionospheric inhomogeneities are important, since they are responsible for the rapid fading (scintillation) of radio signals from
satellite communication and navigation systems. Such effects may introduce performance degradations or
outages on systems operating at frequencies between 100 MHz and several gigahertz. Models of radiowave
scintillation have been developed, and these are based upon a basic understanding of the global morphology of
ionospheric inhomogeneities.
There are inhomogeneities in all regions of the ionosphere, but the equatorial and high-latitude regions are
the most significant sources. Hunsucker and Greenwald (24) have reviewed irregularities in the high-latitude
ionosphere, and Aarons (25) has examined the equatorial environment.
Equatorial inhomogeneities tend to develop following sunset and may persist throughout the evening, but
with decreased intensity after local midnight. The irregularities are thought to be the result of an instability
brought about by a dramatic change in F-region height at the magnetic equator following sunset. The scale
lengths of the irregularities may range between roughly a meter and several kilometers, and the spectrum of the
irregularities has been observed to exhibit a power-law distribution. There is a tendency for the irregularities to
be field-aligned with an axial ratio of roughly 20 to 1. In addition, the irregularities are organized in distended
patches. Though the situation is variable, the patch sizes range between 100 km and several thousand
kilometers in the upper F region, and average 100 km in the lower F region. The equatorial irregularities
tend to be more intense and widespread at the equinoxes and at solar maximum, but magnetic activity tends
to suppress the growth of the irregularities.
High-latitude irregularities exist within the polar cap and the auroral zone, with the latter being primarily
associated with the bottomside F region. The high-latitude F region is quite variable, and unlike midlatitudes,
it may have an electron density less than the E-region during nocturnal hours. In the wintertime, structured
auroral arcs may migrate within the polar cap, and the electron density enhancements within these formations
may be several orders of magnitude greater than the normal background, especially during elevated solar
activity. During disturbed geomagnetic conditions, structured electron density patches have been observed to
travel across the polar cap in the antisunward direction. These irregularities may have a significant effect on
communication systems. For both the auroral zone and the polar cap, increased geomagnetic activity has a
dramatic influence on the growth of irregular ionospheric formations. Moreover, for large and sustained values
of Kp, it has been observed that the high-latitude irregularity patterns tend to migrate equatorward, replacing
the background midlatitude properties (see the section The High-Latitude Ionosphere).
Diurnal Behavior of the Ionospheric Layers
Mean Variations. As indicated by the Chapman representation for N max , the respective critical frequencies for the layers D, E, F1 , and F2 will generally peak during the daytime. Moreover, all layers, with the
occasional exception of the F2 region, closely follow the tendency for the existence of a peak in the mean value
electron density in the neighborhood of local noon. Figure 10 shows the mean diurnal variation of the E, F1 and
F2 critical frequencies at solar maximum for a midlatitude site. Day-to-day F-region variability is exhibited in
Fig. 11 for a period of maximum solar activity and midlatitudes.
Short-Term Variations. Variations in layer critical frequencies will occur from hour to hour and from
day to day, especially for the F2 region. It appears that much of this variability owes its existence to the impact
of geomagnetic storms, TIDs, and miscellaneous F-region dynamic effects. TIDs are one of the more fascinating
features of the ionosphere. They are the ionospheric tracers of neutral atmospheric gravity waves, which derive
from a number of sources in the upper atmosphere. These sources include localized heating effects, atmospheric
explosions, enhanced auroral activity, and other atmospheric phenomena that are associated with rapid and
20
IONOSPHERE CHARACTERISTICS
Fig. 10. Mean diurnal variation of foE, foF1, and foF2 for summer and winter under northern-hemisphere- and solarmaximum conditions. [From Jursa (8).]
nonuniform changes in atmospheric pressure. Figure 12 shows the variation of foF2 as a function of time,
showing the impact of TIDs. Figure 13 shows the effect of a large geomagnetic storm.
Long-Term Solar-Activity Dependence of the Ionospheric Layers
There is a clear tendency for the ionospheric critical frequencies to increase with sunspot number. Figure 14
shows the long-term variation of R12 , foF2, and foE, and the D-layer absorption level (at 4 MHz), for noontime
conditions. The D region is best characterized by the amount of absorption it introduces (see the subsection
Ionospheric Layering above). A device for monitoring the D-region absorption is the riometer, which evaluates
it as the product of D-region electron concentration and the electron collision frequency. From Figure 14, a slow
11-year modulation in the ionospheric parameters is evident. After smoothing, the results correlate well with
sunspot number. Superimposed on this solar epochal variation is an annual variation, with D-region absorption
and foE exhibiting summertime maxima, while foF2 exhibits a wintertime maximum (i.e., seasonal anomaly).
The slow but definite dependence upon mean sunspot number is illustrated in Fig. 15. This plot is unusual
in that it presents running 12-month averages of the specified ionospheric parameters as well as of the sunspot
number. This obscures the seasonal effects observed in Fig. 14.
IONOSPHERE CHARACTERISTICS
21
Fig. 11. Variations in the hourly values of foF2 as a function of the time of day, for January solar maximum conditions at
a northern-hemisphere midlatitude site. The range of day-to-day variability in foF2 is 10%, suggesting a variation in
NmaxF2 of 5%. [From Davies (1).]
Sporadic E
General Characteristics. Even though the normal E region is Chapman-like in nature, isolated forms
of ionization are often observed in the E-region, having a variety of shapes and sizes. These ionization forms
have been termed sporadic E, because they appear quasirandomly from day to day, and they generally defy
deterministic prediction methods. Sporadic E (Es) ionization has been observed during rocket flights and with
incoherent backscatter radar, and a layer thickness of the order of 2 km has been observed. It generally takes
the form of large-scale structures, having horizontal dimensions of hundreds of kilometers at middle latitudes.
22
IONOSPHERE CHARACTERISTICS
Fig. 12. Variations in the ionosphere thought to be associated with traveling ionospheric disturbances. The foF2 variations
shown here are of the order of 2% and have periods of 20 min. The NmaxF2 variations are 1%. [From Paul (63).]
Fig. 13. Effect of a large geomagnetic storm on N max . [By permission of J. M. Goodman and Kluwer Academic Publishers,
Norwell, MA (11).]
IONOSPHERE CHARACTERISTICS
23
Fig. 14. Variation in R12 , foF2, foE, and 4 MHz absorption at noontime. The seasonal effects are clearly evident, the
foE and D-layer variations being out of phase with the foF2 variations (i.e., seasonal anomaly). [By permission of J. M.
Goodman and Kluwer Academic Publishers, Norwell, MA (11).]
Polar and equatorial forms have different structures and causal mechanisms. Although sporadic E consists
of an excess of ionization (against the normal E-region background), it does not appear to be strongly tied
to solar photoionization processes. Still, midlatitude Es occurs predominantly during summer days. Sporadic
E does exhibit seasonal and diurnal tendencies, which have been examined statistically, and at least three
different types of sporadic-E ionization have been discovered with distinct geographical regimes: low-latitude
24
IONOSPHERE CHARACTERISTICS
Fig. 15. Long-term variation in R12 , foF2, and foE at noontime. Since running 12-month averages were taken, the seasonal
effects observed in Fig. 14 are smoothed out. [By permission of J. M. Goodman and Kluwer Academic Publishers, Norwell,
MA (11).]
(or equatorial), midlatitude (or temperate), and high-latitude ionization. Figure 16 depicts the probability of
Es occurrence.
Formation of Midlatitude Sporadic E. It has suggested that wind-shears in the upper atmosphere
are responsible for the formation of sporadic E at midlatitudes. We shall review this process briefly.
It should be recalled from the examination of photochemistry in the ionosphere that molecular ions such
as those that exist in the E region introduce rapid electron loss by recombination. At the same time it is
recognized that an enormous number of meteors burn up in the E region. This meteoric debris is largely
comprised of metallic ions, which are monatomic. Their presence has been confirmed by mass spectroscopy
measurements using rockets, and they include iron, sodium, magnesium, etc. Since monatomic ions exhibit a
small cross section for electron capture, the process by which atomic ions become concentrated in well-defined
layers will lead to reduced loss rates for ambient free electrons in the interaction region.
The influx of this foreign mass of metallic ions, when distributed over the whole of the E region, would
be insufficient to overwhelm the omnipresent molecular species (such as NO+ ), which are in a state of photochemical equilibrium, were it not for a mechanism that preferentially concentrates the meteoric debris ions.
Apparently wind shear is this mechanism. The basic wind shear theory was proposed by Whitehead (26), but
it remained for Gossard and Hooke (27) to outline a process for meteoric ion concentration based upon the
interaction of the meteoric debris with atmospheric gravity waves, the latter wave structures being responsible
for the development of TIDs as well. The ultimate process involves a corkscrew propagation of atmospheric
gravity waves and atmospheric tides, which results in a rotation of wind velocity as a function of altitude. This
IONOSPHERE CHARACTERISTICS
25
Fig. 16. Probability of Es occurrence as observed in the period 19511952. It is representative of the global, seasonal, and
diurnal variation of sporadic-E ionization. [From Davies (1).]
effect can cause the wind to change direction over an altitude of only a kilometer or so, so as to trap meteoric
ions at an intermediate point having zero velocity. This buildup in a narrow region is sufficient to generate an
intense sporadic-E patch.
Sporadic E at Non temperate Latitudes. The high-latitude sources are evidently of two types,
depending upon whether the observation is made in the neighborhood of the auroral oval or poleward of it
26
IONOSPHERE CHARACTERISTICS
Fig. 17. Idealized picture of ionospheric plasma frequencies in a northsouth plane through Fairbanks and Anchorage,
Alaska. E, region equatorward of trough; B, equatorward edge of trough: C, plasma frequencies (MHz); D, trough minimum;
E, plasmapause field line; F, poleward edge of trough; G, F-region blobs; H, enhanced D-region absorption; I, E-region
irregularities. [By permission of J. M. Goodman and Kluwer Academic Publishers, Norwell, MA (11), after Hunsucker (28).]
(i.e., in the polar cap region). It has been found that auroral Es is basically a nocturnal phenomenon; it is
associated with the optical aurora and is due to auroral electron precipitation. Because of its proximity to the
seat of auroral substorm activity, it is not surprising to find some correlation between auroral Es and some
appropriate magnetic index. Indeed, it has been found that auroral Es is positively correlated with magnetic
activity. On the other hand, polar-cap Es may be relatively weak, and is negatively correlated with substorm
activity.
Turning equatorward, it has been found that equatorial Es is most pronounced during daylight hours, and
evidence points to the formation of ionization irregularities within the equatorial electrojet as the responsible
agent at low latitudes.
The High-Latitude Ionosphere
From a morphological point of view, the high-latitude region is the most interesting part of the ionosphere.
It has been said that the auroral zone and associated circumpolar features, are our windows to the distant
magnetosphere, and the presence of visible aurorae has fascinated observers for centuries. The interplanetary
magnetic field, which may be traced to its solar origins, has a significant impact on the geomorphology of the
high-latitude ionosphere and its dynamics, including magnetic substorm development.
The high-latitude region of the ionosphere is characterized by a hierarchy of phenomena that are largely
orchestrated by magnetospheric and interplanetary events (of a corpuscular nature) rather than solar (electromagnetic) flux variations. Hunsucker (28) has examined the salient features and they are depicted in
Fig. 17, with particular emphasis on the high latitude trough. In Fig. 18, from Bishop et al. (29), many of the
same features are depicted and compared with worldwide features.
IONOSPHERE CHARACTERISTICS
27
Fig. 18. Depiction of various ionospheric features at a given time such that the daynight terminator is passing through
the middle of the United States (i.e., 2300 GMT). [By permission of J. M. Goodman and Kluwer Academic Publishers,
Norwell, MA (11), from Bishop et al. (29).]
The magnetic activity index K p is generally available and is typically used as the parameter of choice
to determine the statistical position of the auroral zone. The concept of the auroral oval was developed by
Feldstein and Starkov (30) on the basis of a set of all-sky camera photographs that were obtained during the
International Geophysical Year. Other models exist, but the Feldstein picture is found in most models that
attempt to include auroral effects in some way. The position of the oval is important, not only as an ionospheric
feature in itself, but because it also represents a boundary between the decidedly different geophysical regimes
that are poleward of it (the polar cap) and equatorward of it (the midlatitudes). Because the position of the
auroral zone varies diurnally as well as with the index K p , there are some sites that may be characterized by
all four regimes at any given time: polar, auroral, trough, and midlatitude. Iceland is such a location.
One of the most fascinating properties of the various circumpolar features is their latitudinal motion as
a function of magnetic activity. The ionospheric plasma is best organized in terms of some form of geomagnetic
coordinates, but the high-latitude plasma patterns are not fixed in that frame of reference either. The equatorward boundary of the region of precipitating electrons has been deduced from DMSP satellite instruments,
and it takes a form due to Gussenhoven et al. (31):
where corrected geomagnetic coordinates are used, L(t) and L0 (t) are specified in degrees, and L0 (t) is the
equatorward boundary of the oval when K p = 0. It is emphasized that L0 (t) and a(t) are functions of time
in the magnetic local time (MLT) system.. Both functions are smoothly varying over the diurnal cycle: L0
ranges between 65 at 0100 MLT and 72 at 1700 MLT, and a(t) varies between 2 at 2400 MLT and
28
IONOSPHERE CHARACTERISTICS
Fig. 19. (a) Descent of the auroral oval as a function of magnetic activity; (b) position of the auroral oval and its thickness
versus the magnetic index K p ; (c) position of the auroral arc formations versus the magnetic activity index Dst . [By
permission of J. M. Goodman and Kluwer Academic Publishers, Norwell, MA (11).]
0.8 at 1500 MLT. Therefore the statistical representation of the oval has its greatest equatorward descent
during nocturnal hours. Moreover this equatorward boundary is greatly influenced by magnetic activity. Chubb
and Hicks (32) have found that the daytime aurora descends approximately 1.7 degrees per unit K p , and the
nighttime aurora descends at a rate of 1.3 degrees per unit K p . The auroral oval and thickness are depicted in
Fig. 19. Ultimately the auroral arcs, which reside within the auroral oval, are tied to interplanetary phenomena.
IONOSPHERE CHARACTERISTICS
29
Fig. 20. Hierarchy of solarterrestrial effects. [By permission of J. M. Goodman and Kluwer Academic Publishers, Norwell,
MA (11).]
Workers have shown that the magnitude of the southward component of the interplanetary magnetic field is a
key factor is the development of so-called geomagnetic substorms, wherein K p exhibits large enhancements.
The US Air Force prepares daily summaries of an index Q in order to provide a basis for various analyses of
the high-latitude ionosphere. The index Q ranges between 0 and 8, with larger values associated with a widening
of the oval region and a general increase in intensity of activity within the oval. Moreover, the equatorward
boundary of the auroral oval moves to lower latitudes as Q increases. Since Q, viewed as a parameter, defines
the shape and location of the auroral zone, it is a convenient index for transmission to communication facilities
and forecasting facilities. Its utility is dependent upon timeliness and accuracy. As originally designed by
Feldstein, Q defines only a statistical relationship between the oval position and magnetic activity, the latter
being parameterized by the planetary index K p . Nevertheless, the Feldstein oval concept has been shown to
have some utility under real-time circumstances. Satellite imagery is used to deduce an effective Q.
Auroral physics is an exceedingly rich and complex subject. Not all phenomena in the high-latitude region
are understood, and insufficient data are available to fully characterize even those factors for which a general
understanding exists.
Ionospheric Response to Solar Flares
Now we shall take note of a special class of effects called sudden ionospheric disturbances (SIDs). These
constitute those events that arise as a result of the atmospheric interaction with electromagnetic flux from
solar flares. A book by Mitra (33) is an excellent treatise on the ionospheric effects of solar flares.
We recognize that the sun is the ultimate source for a large variety of ionospheric and magnetospheric
effects. Fig. 20 exhibits the hierarchy of solar-induced ionospheric effects. There are many types of SID observed;
one of the most important is the short-wave fade (SWF), which affects HF communication circuits on the sunlit
side of the earth. The source of the enhanced D-region ionization responsible for the SWF is typically an impulse
burst of X-ray energy from within an active region on the sun (generally a sunspot). An X-ray flare generates
30
IONOSPHERE CHARACTERISTICS
a significant increase in D-layer ionization with a temporal pattern that mimics the flare itself. This results in
an increase in the product of the electron density and the collision frequency. It is the growth of this product
that accounts for the absorption of HF signals passing through the D region. Flares tend to be more prevalent
during the peak in sunspot activity, and the individual-flare duration distribution ranges from a few seconds
to roughly an hour.
The Ionospheric Storm
The magnetic storm is a fascinating geophysical phenomenon, which goes far beyond the visible evidence
corresponding to auroral displays at high latitudes. It is central to the issues surrounding what is now referred
to as space weather. A discourse on this subject is beyond the scope of this article, but the reader is referred to
an excellent geophysical monograph edited by Tsurutani et al. (34).
The ionospheric storm is the ionospheres response to a geomagnetic storm. While the ionospheric response
to magnetic storms is varied, it has been shown that they may be conveniently classified as either positive or
negative in nature. The main attribute of so-called negative storms is that they are generally associated
with decreases in foF2. Positive storms exhibit the opposite behavior. At midlatitudes the ionospheric storm
signature is typically commensurate with the main features of a negative storm, although variations may occur.
Often the temporal (or storm-time) pattern is complex. For example, the midlatitude ionospheric response to a
large magnetic storm is generally characterized by a short-lived increase in the F-region electron concentration
in the dusk sector following storm commencement (SC), after which it decreases dramatically (see Fig. 13). The
initial short-lived enhancement is observed in foF2 records, and it is correlated with the initial positive phase
of the geomagnetic storm. The main phase of the geomagnetic storm is correlated with a concomitant foF2
diminution, and this reduction in foF2 may last for a day or longer. It is thought that the initial enhancement
in foF2 is a result of electrodynamic forces, while the long-term reduction in foF2 is associated with changes in
upper-atmospheric chemistry and modification of thermospheric wind patterns. A key factor in this process is
ionospheric heating through dissipation of storm-induced atmospheric gravity waves. This heating effect will
cause the thermosphere to expand, and ionospheric loss rates will increase.
Ionospheric Current Systems
Current systems are important in an understanding of ionospheric perturbations associated with the onset
of geomagnetic storms and the progression of these events. There are four principal current systems in the
ionosphere that give rise to relatively rapid fluctuations in the geomagnetic field: the ring current, the magnetopause current system, the atmospheric dynamo, and various high-latitude current systems. The first two are
associated with magnetic storms and occur at magnetospheric distances. The atmospheric dynamo is important
in an understanding of tidal-driven forces, which interact with the ionospheric plasma, causing a vertical drift
of the F-region ionization. Excellent descriptions of ionospheric current systems and dynamo theory may be
found in a monograph by Rishbeth and Garriott (35) and in a book by Ratcliffe (4). High-latitude (i.e., polar and
auroral) currents and atmospheric dynamo currents are observed at lower ionospheric heights in the vicinity
of the E layer. Brekke (36) provides a good treatment of relevant high-latitude current systems. There is also
a current system within the neighborhood of the magnetic equator: the equatorial electrojet, a current that
flows along the geomagnetic equator, eastward by day and westward by night. It is associated with a class of
discrete ionospheric formations that are termed equatorial sporadic E.
IONOSPHERE CHARACTERISTICS
31
Ionospheric Models
As in many areas of geophysical study, ionospheric modeling may assume a number of forms, ranging from the
purely theoretical to the totally empirical. Approaches may also include a combination of these forms, although
empirical models dominate the field. Recent developments include allowance for adaptivity within the models
to accommodate exploitation in the near-real-time environment for special applications. While physical or
theoretical principles are the inspiration for a number of models, in fact most models in use today are largely
specified on the basis of semiempirical relationships derived from observational data.
Ionospheric models fulfill a variety of needs beyond basic research, with the most prominent application
being the assessment and prediction of radio-system performance. For example, ionospheric models are the
engines that drive HF system performance models such as IONCAP (37). Related models are supported by the
U.S. Department of Commerce, including VOACAP, ICEPAC, and REC533 (38). Other applications include evaluation of transionospheric signal parameters and errors in ranging or geolocation introduced by the electron
content of the ionosphere. A general discussion of the status of ionospheric modeling in the context of HF communication systems has been covered by Goodman (11), and recent information regarding telecommunication
system planning has been published by the Commission of European Communities (16).
Ionospheric profile models are based upon the superposition of various submodels of the ionospheric layers or regions (i.e., D, E, Es, F 1 , and F 2 ).The basic purpose of modeling is to represent the electron density
profile under a variety of conditions [see Fig. 3(b)]. These submodels may represent the respective layers as
thin horizontal sheets (e.g. sporadic E) or quasiparabolic regions in the vicinity of maximum ionization. The
submodels are specified by the maximum electron density of the layer, the layer height, the layer thickness,
and a functional representation of the layer shape. There are a number of models for the height profile, the
main differences being the manner in which the component layers are combined. Figure 21 depicts the International Reference Ionosphere (39) and the ionospheric model in the computer program IONCAP. There
are also geographical, seasonal, and solar epochal variations in the specified ionospheric profiles and the
parameters upon which they are built. An example of the geographical variations in foF2 was shown in
Fig. 9, and the Global Atlas of Ionospheric Coefficients was discussed in the section Description of the Ionospheric Layers. Ionospheric coefficients used to produce maps similar to Fig. 9 are common to virtually all
global models of the ionosphere. Currently there are two sets of ionospheric coefficients that may be specified:
the original CCIR (or ITU-R) set, which is sanctioned by the ITU-R, and the newer URSI set (40).
A simplistic model of the ionosphere consists of a parabolic E layer, a linear increase in electron density in
the F 1 layer a parabolic F 2 layer (41). At nighttime, the E and F 1 layers effectively disappear. A newer ITU-R
recipe, consisting of multiple quasiparabolic layers to provide continuity of the overall profile and its height
derivatives (42), has replaced this so-called BradleyDudeney profile model.
Significant improvements in empirical ionospheric modeling have been promoted by military agencies
around the world, including the US Department of Defense, the UK Ministry of Defence, and others. This is not
surprising in view of the large number of applications of ionospheric specification in radio-wave systems used
by the military. The original ICED model was intended to be a northern-hemisphere ionospheric specification
model to serve the requirements of the US Air Force. It was only a regional model, descriptive of midlatitude
behavior but extending into the auroral zone. It was designed to allow for recovery of some of the dynamic
features embodied in auroral climatology that are smeared out in most mapping procedures. The model, as
described by Tascione et al. (43,44), is driven by an effective sunspot number and an index derived from auroral
oval imagery. The effective sunspot number is not based on solar data at all, but is derived from ionospheric
data extracted from the US Air Force real-time ionosonde network. This sunspot number is similar to an
ionospheric T index developed by Australian workers, and to the pseudoflux concept used by the US Navy
for HF predictions (11). The ICED model has been generalized to incorporate global considerations, while
emphasizing near-real-time applications. Anderson et al. have developed a low-latitude ionospheric profile
model, SLIM (45) and a fully analytic ionospheric model, FAIM (46) in order to eliminate the use of limiting
32
IONOSPHERE CHARACTERISTICS
Fig. 21. (a) International Reference Ionosphere (IRI); (b) IONCAP ionospheric submodel. [By permission of J. M. Goodman
and Kluwer Academic Publishers, Norwell, MA (11).]
simplifications in the driving parameters associated with prediction models. A discussion of SLIM and FAIM
may be found in a paper by Bilitza (47). Other developments supporting Air Force requirements include PIM
and PRISM. PIM (Parameterized Ionospheric Model) is a global model of theoretical and empirical climatology,
which specifies the ionospheric electron and ion densities from 90 km to 25,000 km. PRISM (Parameterized
Real-Time Ionospheric Specification Model) uses ground-based and space-based data available in real time
IONOSPHERE CHARACTERISTICS
33
to modify PIM, thereby providing a near-real-time ionospheric specification. Another model, RIBG, developed
by Reilly (48), combines ICED and several other models with a general ray-tracing utility. Current versions
of these models and validation of PRISM are discussed by Doherty et al. (49). A survey of computer-based
empirical models of the ionosphere has been published by Secan (50).
The IRI mentioned previously (see Fig. 21), is a global empirical model that specifies monthly averages
of electron, ion, and neutral temperatures, in addition to electron and ion densities, from about 50 km to
about 2000 km (39). The IRI development is a joint project of URSI and the international Committee of
Space Research (COSPAR), and has proven to be a useful model for scientific research. Another model of
primary interest to workers studying transionospheric propagation effects is the so-called Bent model, a profile
model based upon topside and bottomside sounder data (51). Simplicity is not always important in the age of
sophisticated computers, but the ChingChiu model (52) has found a number of scientific applications in cases
in which detailed ionospheric specification is not paramount.
Aside from global modeling of the ionosphere, there have been attempts to model selected regions of the
world more accurately. During the decade of the 1990s, European scientists affiliated with the COST program
have taken a lead in regional modeling and mapping of the ionosphere (15,16)
Ionospheric Predictions
Ionospheric predictions influence several disciplines, including the prediction of radio system performance, a
matter of some interest in planning as well as ultimate operations. Long-term predictions are generally based
upon predictions of driving parameters such as the sunspot number, the 10.7 cm solar flux, and magnetic
activity indices. Unfortunately, these parameters are not easy to predict. Moreover, the functions relating
these parameters to the ionosphere are imprecise. Therefore, long-term predictions needed for system design
are subject to a considerable amount of uncertainty. To first order the uncertainty in the median value of foF2
for a particular time and location is proportional to the uncertainty in the sunspot number.
In addition to the uncertainty in the mean parameters, we must allow for the fact that ionospheric
parameters have real distributions, and with few exceptions the spread of these distributions is such that errors
about the mean may be a dominant contribution. Short-term ionospheric predictions (or forecasts) generally
refer to departures from the median behavior, the latter being well characterized by running averages of solar
flux and related parameters (e.g., sunspot number). The short-term fluctuations may be specified in terms of
hour-to-hour, day-to-day, and week-to-week variabilities. There are also second-to-second and minute-to-minute
variations, they are generally unpredictable. Compensation for such fluctuations is quite difficult, but may be
accommodated through use of system protocols that enable real-time channel evaluation (RTCE) measures to
be initiated, such as channel sounding or probing. These very short-term forecasts are generally referred to as
nowcasts.
There are four ITU-R documents that are pertinent to the investigation of the ionospheric forecasting
problem. The first deals with the exchange of data forecasts (53), the second outlines various measures for
forecasting of ionospheric parameters (54), the third deals specifically with solar-induced ionospheric effects
(55), and the fourth outlines various real-time channel evaluation schemes (56). These reports should be
consulted.
The distributions of parameters such as foF2, foEs, and hF2 are important, since these parameters depart
significantly from fundamental intuition and from rules set forth by Sidney Chapman in his classic theory.
Distributions of foF2 and foEs are available (57,58), but F 2 -layer height distributions are not directly available.
Ionospheric predictions in the short and intermediate terms provide the most exciting challenge for ionospheric
researchers.
Observational data have shown that TIDs are the ionospheric tracers to a class of atmospheric gravity
waves, and these disturbances are a major contribution to ionospheric variability, especially at F-region heights.
34
IONOSPHERE CHARACTERISTICS
TIDs have a major impact on layer height as well as peak electron density, and possess a variety of scales,
from kilometers to thousands of kilometers. The small- to intermediate-scale TIDs, having wavelengths of less
than a few hundred kilometers and periods of the order of 10 min to 20 min, arise from local sources and have
relatively small amplitudes away from the source region. The large-scale TIDs have sources that are located at
great distances, and there is a strong correlation between this class of disturbances and geomagnetic storms.
Evidence suggests that large-scale TIDs have an impact over global distances and originate within the auroral
zone as a result of atmospheric modifications associated with precipitation and auroral arc formation. A survey
of the effects of TIDs on radiowave systems may be found in a review paper by Hunsucker (59).
The field of ionospheric predictions is undergoing continuous evolution with the introduction of new
scientific methods and instruments, which are providing fresh insight. The requirements for quasi-real-time
products based upon current ionospheric specification has led to increased importance of so-called real-time
ionospheric models. This class of models, in turn, is driven by a hierarchy of solarterrestrial observations,
which enables the analyst to examine the space-weather environment as an integrated complex of phenomena.
This general approach is leading to an improvement in our understanding of ionospheric structure and it
variations, if not better short-term forecasts. In the immediate future, it is anticipated that the primary
ionospheric specification tools will consist of terrestrial sounding systems, including real-time networks of
ionospheric sounders (60). Real-time data services based on these approaches are becoming available (61).
Fundamental Science Issues and Challenges in Ionospheric Research
There are a number of challenges facing ionospheric specialists and aeronomists. While theories explaining
most facets of ionospheric behavior exist and are generally accepted, the theories do not always provide a good
basis for prediction. This is because the driving forces and boundary conditions needed in a physical model are
not always known, and estimates must be used. This has led to the development of semiempirical models for
the purpose of system design, and these are used for operations as well. By and large these models exploit large
ionospheric databases and yield only median representations of ionospheric parameters. To fix this problem,
various update schemes have been developed to make the specification of the ionospheric state as current as
possible. The physics is then used to let the system evolve. All of this can be very unsatisfactory unless an
understanding of the nature of ionospheric variability (in both space and time) is established. There are many
sources within the earthsun system that contribute to the growth of ionospheric structure. While these have
been characterized to some extent, the characterizations are not sufficient to provide predictions acceptable
for many users of the ionospheric channel. Currently this is a major challenge facing the ionospheric research
community.
The following topics require more attention from ionospheric specialists: (a) the driving forces of upper
atmospheric winds and their impact on ionospheric structure and dynamics; (b) the hierarchy of energy sources
within the earthsun system that influence ionospheric behavior; (c) the development of geomagnetic storms
and the impact that storms have on ionospheric behavior; (d) the development and evolution of ionospheric
inhomogeneities; and (e) various methods for ionospheric prediction.
Finally, in the new millennium, the researcher is confronted with an enormous amount of data, both
near-real-time and archived, that may be accessed via the Internet. Harnessing this information stream,
and using the state-of-the-art computational assets, it should be possible to leverage ongoing science efforts,
organize more efficient experimental campaigns, and enhance collaborative efforts, all resulting in a fuller
understanding of ionospheric physics. Some examples include programs such as the SolarTerrestrial Energy
Program (STEP) and the STEP Results, Applications, and Modeling Phase (S-RAMP). These programs are
international efforts run under the aegis of the Scientific Committee on SolarTerrestrial Physics (SCOSTEP),
in cooperation with the International Council of Scientific Unions (ICSU).
IONOSPHERE CHARACTERISTICS
35
BIBLIOGRAPHY
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
22.
23.
24.
25.
26.
27.
28.
29.
30.
31.
32.
33.
K., Davies Ionospheric Radio Propagation, NBS Monograph 80, Washington: USGPO, 1965.
K., Davies Ionospheric Radio Waves, Waltham, MA: Blaisdel, 1969.
K. Davies Ionospheric Radio, IEE Electromagnetic Wave Series 31, London: Peregrinus, Ltd., IEE, 1990.
J. A. Ratcliffe An Introduction to the Ionosphere and the Magnetosphere, London: Cambridge Univ. Press, 1972.
A. Giraud M. Petit Ionospheric Techniques and Phenomena, Dordrecht: Reidel, 1978.
M. C., Kelley The Earths Ionosphere, Plasma Physics and Electrodynamics, San Diego, CA: Academic Press, 1989.
H., Rishbeth Basic physics of the ionosphere, in M. P. M. Hall and L. Barclay, (eds.), Radiowave Propagation IEE
Electromagnetic Series 30, Stevenage, Peregrinus UK, 1991.
A. S. Jursa (scientific ed.), Handbook of Geophysics and the Space Environment, Springfield, VA: Air Force Geophysics
Laboratory, Air Force Systems Command, US Air Force, National Technical Information Service (NTIS), 1985.
T. Tascione Introduction to the Space Environment, Malabar, FL: Orbit, 1988.
R. D. Hunsucker Radio Techniques for Probing the Terrestrial Ionosphere, New York: Springer-Verlag, 1991.
J. M. Goodman HF Communications: Science & Technology, New York: Van Nostrand Reinhold, 1991.
E. E. Johnson Advanced High Frequency Radio Communications, Boston: Artech House, 1997.
L. McNamara The Ionosphere: Communications, Surveillance, and Direction Finding, Malabar, FL: Orbit, Kreiger,
1991.
J. M. Goodman (ed.) Ionospheric Effects Symposia, Springfield, VA: National Technical Information Service. 1975, 1978,
1981, 1984, 1987, 1990, 1993, 1996, 1999,
P. A. Bradley (chair, COST 238 Mgmt.Committee), PRIME: Prediction and retrospective ionospheric modeling over
Europe, Final Report, Commission of European Communities, Chilton, Didcot, Oxfordshire, UK: Rutherford-Appleton
Laboratory, 1999.
R. Hanbaba Improved quality of service in ionospheric telecommunication systems planning and operations, COST 251
Final Report, Warsaw: Space Research Centre, 1999.
S. Chapman The absorption and dissociative or ionizing effect of monochromatic radiation in an atmosphere on a
rotating earth, Proc. Phys. Soc, 43: 26, 1931.
L. M. Muggleton A method for predicting foE at any place and time, Telecomm. J., 42 (7), 413418, 1975.
E. D. Ducharme L. E. Petrie R. Eyfrig A method for predicting the F1 layer critical frequency based upon zurich
smoothed sunspot number, Radio Sci., 8: 837839, 1973.
CCIR, CCIR Atlas of Ionospheric Characteristics, Report 340, General Assembly held in Oslo, Geneva: ITU, 1966.
P. N. Mayaud Derivation, Meaning, and the Use of Geomagnetic Indices, Washington: American Geophysical Union,
1980.
D. F. Muldrew F-layer ionization trough deduced from alouette data, J.Geophys. Res. 70: 2635-2650, 1965.
B. W. Halcrow J. S. Nisbet A Model of F2 peak electron densities in the main trough region of the ionosphere, Radio
Sci., 12: 815-820, 1977.
R. D. Hunsucker R. A. Greenwald, (eds.) Special issue, radio probing of the high latitude ionosphere and atmosphere
new techniques and new results, Radio Sci., 18: 1983.
J. Aarons Equatorial scintillations: A review, IEEE Trans. Antennas Propag., 25: 729736, 1977.
J. D. Whitehead Production and prediction of sporadic E, Rev. Geophys. Space Phys., 8: 65144, 1970.
E. E. Gossard W. H. Hooke Waves in the Atmosphere, Amsterdam: Elsevier Science, 1975.
R. D. Hunsucker Anomalous propagation behavior of radio signals at high latitudes, in H. Soicher (ed.), Propagation
Aspects of Frequency Sharing, Interference and System Diversity, AGARD-CF-332, Springfield, VA: NTIS, 1983.
G. J. Bishop et al. A modern trans-ionospheric propagation sensing system, in Operational Decision Aids for Exploiting
or Mitigating Electromagnetic Propagation Effects, NATO-AGARD-CP-453, UK: Specialised Printing Services, 1989.
Y. I. Feldstein G. N. Starkov Dynamics of auroral belts and polar geomagnetic disturbances, Planet. Space Sci., No. 15,
p. 209, 1967.
M. S. Gussenhoven D. A. Hardy N. Heinemann Systematics of the equatorward diffuse auroral boundary, J. Geophys.
Res., 88: 5692, 1983.
T. A. Chubb G. T. Hicks Observations of the aurora in the far ultraviolet from OGO 4, J. Geophys. Res., 75: 12901311,
1970.
A. P. Mitra Ionospheric Effects of Solar Flares, Dordrecht: Reidel, 1974.
36
IONOSPHERE CHARACTERISTICS
34. B. T. Tsurutani et al. (ed.) Magnetic Storms, Geophysical Monograph 98, Washington: American Geophysical Union,
1997.
35. H. Rishbeth O. K. Garriott Introduction to Ionospheric Physics, New York: Academic Press, 1969.
36. A. Brekke Currents in the auroral zone ionosphere, in The Physical Basis of the Ionosphere in the SolarTerrestrial
System, AGARD-CP-295, London: Tech. Edit. & Reprod., pp. 13.113.9, 1980.
37. L. R. Teters, et al. Estimating the performance of telecommunication systems using the ionospheric transmission
channel, IONCAP users manual, NTIA Report 83-127, PB84-111210, Springfield, VA: U.S. Dept. of Commerce, 1983.
38. ITS, HF Propagation Models: [Online], 2000. Available www: http://elbert.its.bldrdoc.gov/hf.html
39. D. Bilitza International reference ionosphere, NASA, NSSDC 90-22, Greenbelt, MD: World Data Center A (Rockets
and Satellites), 1990.
40. C. M. Rush et al. Ionospheric mappingan update of fof2 coefficients, Telecomm. J., 56: 179182, 1989.
41. P. A. Bradley J. R. Dudeney A simple model of the vertical distribution of electron concentration in the ionosphere, J.
Atmos. Terrest. Phys., 35: 21312146, 1973.
42. M. I. Dick P. L. Bradley The RAL quasi-parabolic model ionosphere Proc. COSTPRIME Workshop on Data Validation
of Ionospheric Models and Maps, COST 238TD (93) 001, 1992, pp. 6783.
43. T. F. Tascione, et al. A technical description of the ionospheric conductivity and electron density profile model (ICED,
Version 196-11), Scott AFB, IL: Hqrtrs Air Weather Service, US Air Force, 1987.
44. T. F. Tascione et al. New ionospheric and magnetospheric specification models, Radio Sci., 23: 211222, 1988; also in
Proc. IES87.
45. D. N. Anderson M. Mendillo B. Herniter A semi-empirical low-latitude ionospheric model Radio Sci., 22: 292, 1987.
46. D. N. Anderson J. M. Forbes M. Codrescu A fully analytic, low latitude ionospheric model, J. Geophys. Res., 94: 1520,
1989.
47. D. Bilitza Solar terrestrial models and application software, Planet. Space. Sci., 40(4): 541579, 1992.
48. M. H. Reilly M. Singh A transionospheric radio propagation model, in J. M. Goodman, (ed.), IES93 Proc. Springfield,
VA: NTIS, 1993.
49. P. H Doherty et al. Validation of PRISM: The climatology, in J. M. Goodman, (ed.), Proc. IES99 Springfield, VA: NTIS,
1999.
50. J. A. Secan A survey of computer-based empirical models of ionospheric electron density, Report NWRA CR-89-11038,
Bellevue, WA: Northwest Research Associates, 1989.
51. R. B. Bent et al. The development of a highly-successful worldwide empirical ionospheric model and its use in
certain aspects of space communications and worldwide total electron content investigations, in J. M. Goodman
(ed.), Effect of the Ionosphere on Space Systems and Communications, IES75, USGPO, Springfield VA: NTIS, 1976,
pp. 1328.
52. B. K. Ching Y. T. Chiu A phenomenological model of global ionospheric electron density in the E, F1 , and F2 -regions,
Atmos. Terrest. Phys., 35: 16151630, 1973.
53. ITU-R, Exchange of information for short-term forecasts and transmission of ionospheric disturbance warnings, ITU-R
Recommendation P.313, Geneva: International Telecommunications Union, 1995.
54. ITU-R, Short-term forecasting of critical frequencies, operational maximum useable frequencies and total electron
content, ITU-R P.888, Geneva: International Telecommunications Union, 1994.
55. ITU-R, Short-term prediction of solar induced variations of operational parameters for ionospheric propagation, ITU-R
P. 727, Geneva: International Telecommunications Union, 1994.
56. CCIR, Real-time channel evaluation of Hf ionospheric radio circuits, Report 889-2, Geneva: International Telecommunications Union, 1990.
57. D. Lucas G. W. Haydon Predicting statistical performance indices for high frequency telecommunication systems,
ITSA-1, Boulder, CO: US Dept. of Commerce, 1966.
58. M. Leftin S. M. Ostrow C. Preston Numerical maps of foEs for solar cycle minimum and maximum, ERL 73-ITS 63,
Boulder CO: US Dept. of Commerce, 1968.
59. R. Hunsucker Atmospheric gravity waves and traveling ionospheric disturbances: thirty years of research, in J. M.
Goodman (ed.), Effect of the Ionosphere on Radiowave Signals and System Performance, IES90, USGPO, Springfield,
VA: NTIS, 1990.
60. I. A. Galkin B. W. Reinisch D. Kitrosser Advances in digisonde networking, in Proc. IES99, J. M. Goodman (ed.),
Springfield, VA: NTIS, 1999.
IONOSPHERE CHARACTERISTICS
37
61. J. M. Goodman J. W. Ballard Dynacast-assisted frequency management for HF communication and broadcasting
systems, in J. M. Goodman (ed.), Proc. IES99, Springfield, VA: NTIS, 1999.
62. NRC, SolarTerrestrial Research for the 1980s, Washington: National Research Council, National Academy Press, 1981.
63. A. Paul Ionospheric Variability, Technical Report 1277, San Diego, CA: Naval Ocean Systems Center, 1989.
64. F. Lied (ed.) High Frequency Radio Communications with Emphasis on Polar Problems, AGARDograph 104, NATO,
Maidenhead, UK: Technivision, 1967.
JOHN M. GOODMAN
Radio Propagation Services, Inc.
494
ELECTROMAGNETIC WAVES IN THE IONOSPHERE
ELECTROMAGNETIC WAVES IN THE IONOSPHERE
The terrestrial ionosphere is a roughly spherical shell of
weakly ionized plasma that surrounds the earth. A plasma is
a gas that has been ionized by radiation or by charged particles, so that it consists of free electrons, ions, and neutrals; it
is sometimes referred to as the fourth state of matter (1).
The main ionizing agents are solar radiation in the extreme
ultraviolet (EUV) region, soft X rays, Lyman- radiation, and
hard X rays, as well as cosmic rays. This spherical shell is
stratified into distinct layers, the lowest region being the D
layer starting at about 50 km height, the E region starting at
about 100 km, the F1 layer (during the day) near 250 km,
and the F2 layer at about 350 km. At high geomagnetic latitudes, solar energetic charged particles are also important
ionizing agents. Figure 1 is a plot of log density versus log
kinetic temperature showing the relative state of ionization
of various plasmas.
THE RADIO SPECTRUM
A considerable portion of the radio spectrum (ELF through
HF) is affected by our ionosphere, as shown in Table 1.
IONOSPHERIC INTERACTION
The basic interaction mechanism between radio (EM) waves
and the ionosphere involves the oscillation of the electric component of the wave acting on free electrons, which are 1800
times less massive than the ionospheric neutrals or ions. The
E wave induces motion of the electrons, and at the same time
the ionosphere abstracts energy from the electronsresulting
in a bending of the radio wave and some energy loss from the
wave. Figure 2 illustrates the attenuation of radio waves in
J. Webster (ed.), Wiley Encyclopedia of Electrical and Electronics Engineering. Copyright # 1999 John Wiley & Sons, Inc.
ELECTROMAGNETIC WAVES IN THE IONOSPHERE
25
Wavelength (m)
108
50% ionization
hydrogen plasma
20
Highpressure
arcs
Shock
tubes
Attenuation (dB/km)
Laser
plasma
107
101
ii, ee > 1 cm
(4 /3)n Ds < 1
log10 [n (cm3)]
495
Focus
Z pinches
Fusion
reactor
15
106
105
ELF
104
VLF
103
LF
102
MF
101
HF
Day
102
Night
10
104
Alkali
metal
plasma
10
Glow
discharge
Flames
Fusion
experiments
Low
pressure
103
D, > 1 cm
Earth
5 ionosphere
0
2
Earth
plasma sheet
1
2
log10 [T (eV)]
101
100
101
102
Frequency (kHz)
103
Figure 1. Logarithmic plot of approximate magnitudes of some typical laboratory and natural plasmas.
the ionosphere as a function of frequency from 1 Hz to 30
MHz.
As may be seen in Fig. 2, there is a variation in radio wave
attenuation from day to night. The virtual height (the height
at which radio waves at vertical incidence are reflected) varies with time of day, with season of the year, and with geomagnetic activity. The most regular variation is the local time
variation, as shown in Fig. 3.
There are many techniques used to investigate the characteristics of the ionosphere (24). The propagation of radio
waves in the ionosphere is described in considerable detail in
Ref. 5, and solarterrestrial relations and their effects on radio propagation are covered in Ref. 6. Radiowave propagation
at all frequencies depends to different degrees on the geo-
Table 1. The Radio Spectrum as Defined by the International Telecommunications Union, (ITU);
Primary Modes of Propagation, and Effects of the Terrestrial Ionosphere
ITU Designation
Extra low frequency
(ELF)
Very low frequency
(VLF)
Low frequency
(LF)
Medium frequency
(MF)
High frequency
(HF)
Very high frequency
(VHF)
Ultra high frequency
(UHF)
Super high frequency
(SHF)
a
Frequency
Range
30300 Hz
Principal
Propagation Modes
330 kHz
30300 kHz
Same as above
Navigation, standard-frequency
and -time dissemination
Navigation LORAN-Ca
3003000 kHz
Primarily ground wave, but sky
waveb at night
Primarily sky wave, some ground
wave
Primarily LOS,c some sky wave at
lower VHF
Primarily LOS, some refraction and
scattering by the ionosphere
Same as above
AM broadcasting, maritime,
aeronautical communication
Shortwave broadcasting, amateur, fixed services
FM broadcasting, television,
aeronautical communication
Television, radar, navigation,d
aeronautical communication
Radar, space communication
330 MHz
30300 MHz
3003000 MHz
The LORAN-C system will probably be superseded by the GPS system.
Sky wave denotes the earthionosphereearth reflection mode.
c
Line of sight.
d
Global Positioning System satellite constellation.
b
Principal Uses
Ground wave and earthionosphere
waveguide mode
Same as above
330 GHz
104
Figure 2. Nightday variation of attenuation on radio paths as a
function of frequency from 1 Hz to 30 MHz. (Courtesy of D. Llanwyn Jones.)
Solar
corona
Solar wind
(1 AU)
102
Submarine communication
496
ELECTROMAGNETIC WAVES IN THE IONOSPHERE
Virtual height (km)
Winter
400
Summer
F2
400
F2
200
F
F1
200
F1
E
E
12
18
Local time
24
12
18
Local time
24
Figure 3. Average variation of ionospheric layer height as a function
of season and local time. Note the large change in height of the F2
layer in summer.
graphic and geomagnetic latitudinal region of the ionosphere.
The most benign latitudinal region is the midlatitudes; the
most disturbed regions are the auroral, equatorial, and polar
regions, as described in detail in Ref. 7. In the next section
we describe qualitatively the salient propagation modes and
ionospheric effects on radio waves as a function of frequency.
The section after provides a mathematical description and the
physical principles of the interaction of radio waves with the
ionosphere. The final section will introduce the reader to the
frontiers of ionospheric research at the end of the twentieth
century.
modes in an ideal earthionosphere mode is presented in
Fig. 5.
In reality, the ELFVLF waveguide mode is considerably
more complicated because of its spherical nature and the electrical characteristics of the upper and lower boundaries. At
ELF frequencies, the wavelength is of the same order of magnitude as the transverse dimensions of the waveguide, and
the signal propagates deeply into both land and sea because
of the skin depth effect.
At global distances, the signal is very stable, but extremely
long antennas and high transmitter powers are required and
the signaling rate is extremely slow. One unique advantage
of ELF is the ability of the signal to penetrate relatively
deeply into sea water (at 100 Hz, the attenuation in sea water
is 0.3 dB/m, which is of the attenuation in the waveguide).
The attenuation of ELF signals penetrating normal earth is
approximately of that in sea water, so ELF signals can be
used to probe or communicate into the solid earth. There is
also some evidence that high latitude effects such as D-region
irregularities can cause some phase anomalies in ELF and
VLF transmissions (9).
VLF (3 kHz to 30 kHz) transmissions also require large
antennas and high-power transmitters, but are used more
than ELF for time- and frequency-standard dissemination
and navigational systems. VLF signals are also influenced by
anomalies on the earths surface (10) and in the high-latitude
ionosphere (4,5).
Low Frequencies
EFFECTS UPON SPECIFIC RADIO SERVICES
Extremely Low and Very Low Frequencies
As indicated in Table 1, at the lowest frequencies (ELFVLF)
the basic propagation mode is a spherical waveguide mode,
with the D and E regions of the ionosphere forming the upper
boundary, and the surface of the earth the lower boundary. A
simplified earthionosphere waveguide geometry is shown in
Fig. 4 (8). A schematic diagram of the first two waveguide
Moving up in frequency to the LF band (30 kHz to 300 kHz),
the basic propagation mode below 100 kHz is by the ground
(surface) wave, which follows the earths curvature, and above
100 kHz is the sky wave and the waveguide mode. The sky
wave is, of course, influenced by the ionospheric diurnal, seasonal, and latitudinal variations.
Perfect reflector
Guide wavelength
Height
EV
2h
image
Source
2 h cos n
Ionosphere
R = 1
EH
E
First-order (TM01) mode
Phase front
EV
Ionosphere
R = 1
Ionosphere z = h
Ground
Perfect reflector
R = +1
Ray direction
z=0
Height
4h
image
R = +1
EH
2h
image
E
Figure 4. Simplified ray geometry for the first-order and secondorder VLFELF modes. The two conducting planes are representative
at the earth surface and the ionosphere.
R = +1
Second-order (TM02) mode
Figure 5. The E field for ideal earthionosphere waveguide modes.
ELECTROMAGNETIC WAVES IN THE IONOSPHERE
Medium Frequencies
Propagation during the daytime in the MF (300 kHz to 3000
kHz) band is by ground wave, and for frequencies at night
above 500 kHz by sky wave. At geomagnetic latitudes
greater than 55, the auroral ionosphere introduces some
anomalous sky-wave propagation modes (11).
High Frequencies
The ionosphere has the most profound effect on signals in the
HF (3 MHz to 30 MHz) band, making the sky-wave mode the
dominant means of propagation. The ground wave at HF is
sometimes used in the frequency range of 3 MHz to 6 MHz,
especially over sea water, whose conductivity is much greater
than that of ordinary land. At HF wavelengths relatively high
efficiency, gain, and directivity can be achieved in the antenna systems, so directive communications and broadcasting
are realizable. Above 6 MHz, the sky wave is dominant, so
one must really understand ionospheric behavior and phenomenology in order to predict propagation. Since the ionosphere varies with time of day, season, solar activity, and
sunspot cycle, predicting HF propagation over a specific path
can be somewhat complicated. Propagation paths up to
10,000 km are quite common for shortwave (SW) broadcasters, who use antennas with gains of up to 20 dBi (dBi is referenced to an isotropic source) and transmitter powers of 250
kW and higher. With much less reliability, amateur radio operators (hams) sometimes also achieve two-way communications over similar path lengths using antenna gains of 3 dB
to 12 dB and transmitter powers of 5 to 1000 W.
The ionosphere also behaves differently in the equatorial,
midlatitude, auroral, and polar latitudinal regions. Fortunately, several fairly reliable and easy-to-use HF propagation
prediction programs are now available for PCs (IONCAP,
ASAPS, VOACAP, AMBCOM, etc.). The sources of these programs may be found in recent books and articles (5,1214)
and in the amateur radio magazines (QST, CQ, World Radio).
Unfortunately, none of the existing prediction programs gives
very reliable results in the auroral regions. A following section describes the essentials of ionospheric propagation in
considerable detail.
Very High Frequencies
Propagation in the VHF band (30 MHz to 300 MHz) is primarily by line of sight (LOS) to the optical horizon, so if the
antenna patterns direct most of the RF power in the horizontal plane, there are essentially no ionospheric effects. For
earthspace propagation paths, however, the ionosphere can
affect the signal adversely by refraction, diffraction, scattering, or reflection. These effects can be especially important
when the path traverses the equatorial, auroral, and polar
ionosphere. The amplitude, phase, and polarization of the signal may change measurably. These effects will be quantified
in the following section.
Extrahigh Frequencies
At EHF and above, propagation is primarily LOS, and because of the higher frequencies ( f 300 MHz), these signals
are less affected by the ionosphere than lower frequencies. On
earthspace paths that traverse the equatorial and/or high-
497
latitude ionosphere, however, the signal quality can be significantly degraded. These effects will be described below.
To summarize, the radio services most affected by the ionosphere lie in the frequency range of 1 MHz to 150 MHz
[fixed communication services, AM (amplitude modulation)
and SW (shortwave) broadcasting, amateur radio]. To a lesser
degree, services in the 20 kHz to 300 MHz region (mainly
some of the navigation services] suffer some ionospheric perturbation effects.
There is a plethora of radio instrumentation currently deployed globally that operates routinely or on a campaign basis
to measure characteristics of the terrestrial ionosphere. It is
beyond the scope of this article to describe these techniques,
but they have been described in considerable detail in the literature (24).
PHYSICAL PRINCIPLES AND MATHEMATICAL
DESCRIPTION OF ELECTROMAGNETIC
INTERACTION WITH THE IONOSPHERE
Because of the complexity of the terrestrial ionosphere (a
weakly ionized plasma with a superimposed magnetic field in
which electric currents flow), we must utilize the magnetoionic theory to quantify the ionosphere physical parameters.
The most successful formulation of the appropriate magnetoionic theory was derived by Appleton and others in the mid1920s (1517). We can obtain some first-order properties of
the ionosphere by ignoring the magnetic field (18). A simple
dispersion equation for electromagnetic (EM) waves in the
ionosphere is
Ne2
m f 2
(1)
where
refractive index of the ionosphere (real part of the complex refractive index n)
N electron number density of the ionosphere (e/cm3 or
e/m3)
e electronic charge 1.6 1019 C
m mass of the electron 9.1 1031 kg
f frequency of the radio wave in the ionosphere (Hz)
For reflection at vertical incidence, 1 and
N = m f 2 /e2
= 1.24 104 f 2 e/cm3
( f in MHz)
= 1.24 10 f e/m
( f in MHz)
10 2
(2)
Another useful quantity is the plasma frequency,
fn =
r Ne
(N in cm3 )
= 9 N kHz
= 9 103 N MHz
(N in e/cm3 )
(3)
The Virtual Height Concept
If we consider an RF pulse traveling vertically upward into
the ionosphere at the speed of light, v c, it will be reflected
498
ELECTROMAGNETIC WAVES IN THE IONOSPHERE
at the virtual height, h. The time required for the pulse to be
reflected from an ionospheric layer and return to the earth is
2
c
t=
h
0
dz
space velocity c. Referring to the geometry shown in Fig. 6,
we can write the expression
dx
1
c T ER sin 0
D
sin 0
=
c
TE + ER
=
c
t=
(4)
then the virtual height can be found from h( f) ct, or
h ( f ) =
dz
(5)
1 f n2 / f 2
Martyns theorem may be written concisely as
Since the pulse always travels more slowly in the layer than
in free space, the virtual height of a layer is always greater
than the true height. The true height and virtual height are
related by the integral equation
h ( f ) =
Z max
0
dz
( f, z)
(6)
where z is the true height, Zmax is the maximum height
reached by the frequency f, and n is the refractive index at
Zmax for the frequency f. A good discussion of the relation between true height and virtual height is given in Ref. 19.
Vertical and Oblique Propagation
Before considering the behavior of a radio signal in a magnetoionic medium, we will state three theorems that relate
oblique and vertical incidence propagation as depicted in Fig.
6. The first is the secant law, which relates the vertical-incidence frequency f v reflected at B to the oblique-incidence frequency f ob reflected at the same true height. A typical derivation of this relation is given in Ref. 5, and it is usually written
as
f ob = f v sec 0
E
S O
Radio Propagation in a Magnetized Plasma
Before proceeding with a discussion of the Appleton (magnetoionic) equations, we need to define two quantities contained explicitly in the equations. The first is , the number
of collisions per second (collision frequency) between electrons
and heavier particles (ions and neutrals). Another quantity,
the gyromagnetic frequency or gyrofrequency, is the natural
frequency (Hz) of gyration of an ion or electron in a magnetic
field of strength B0 (Wb/m2) and is given by
fH =
;
;
;;;
;
;;;
D
D
Figure 6. Plane geometry describing vertical and oblique ionospheric propagation.
|e|
B 2.80 1010B0
2m 0
(10)
and the angular gyrofrequency is given by
H =
|e|
B 1.76 1011B0
m 0
(11)
Since electrons are much less massive than ions, the electron
gyrofrequency affects the propagation of HF waves in the
ionosphere more than the ion gyrofrequencies. For example,
since B 0.5 104 Wb/m2, the electron gyrofrequency is
1.40 MHz, which falls at the upper end of the medium
wave band.
The Dispersion Relation. Using the recommended URSI (International Union at Radio Science) notation, the magnetoionic dispersion equation for a radio wave in a homogeneous, partially absorbing ionized gas upon which a constant
magnetic field is impressed is given by
n2 = 1
(9)
Smith (20) devised a set of logarithmic transmission
curves, parametric in range, for the curved earth and ionosphere. They are shown in Fig. 7 and are sufficiently accurate
for the distances shown.
hob = hv
(7)
The secant law, then, relates the two frequencies f v and f ob
reflected from the same true height (the distance BD in
Fig. 6).
In order to determine sec and f ob values from verticalincidence soundings (which measure the virtual height h),
we need two more theorems. Breit and Tuves theorem states
that the time taken to traverse the actual curved path
TABCR in Fig. 6 at the group velocity vg equals the time necessary to travel over the straight-line path TER at the free-
(8)
X
YT2
(1 jZ)
2(1 X jZ)
1/2
YT4
2
+ YL
4(1 X jZ)2
where
n complex refractive index ( j)
angular frequency of the exploring wave (rad/s)
(12)
ELECTROMAGNETIC WAVES IN THE IONOSPHERE
499
Sec (corrected)
800
6.0
5.0
4.0
3.0
2.0
1.0
km
km
00
km
25
00
30
00
00
40
35
Virtual height (km)
600
km
700
500
20
400
300
km
00 km m
00 k
m
18 600 0 k
m
0
1
14 00 k
km
12
0
0
km
10
0
80
m
0k m
60
0k m
0
5
0k
40
km
300
km
200
200
100
m
100 k
12
16
20
24
28
32
36
Angle of departure (deg)
N angular plasma frequency
H angular gyrofrequency B0e/m (rad/s)
L longitudinal angular gyrofrequency (B0e/m) cos
T transverse angular gyrofrequency (B0e/m) sin
X N2 /2
Y H /
YL L /
YT T /
Z /
angle between the wave-normal and the magnetic field
inclination
The Polarization Relation. We begin by defining the polarization ratio R as
R = Hy /Hx = Ex /Ey
(13)
Then we can write the double-valued polarization equation as
j
R=
YL
YT2
1
2 1 X jZ
YT4
1
+ YL2
4 (1 X jZ)2
1/2
(14)
In the upper F region of the ionosphere where the electron
ion collision frequency is very low, we may simplify the dispersion and polarization equations by dropping the Z term
(since 0). Equations (12) and (14) then become (for no
absorption)
n2 = 1
2X (1 X )
2(1 X ) YT2 [YT4 + 4YL2 (1 X )2 ]1/2
(15)
and
R=
Hy
j
=
Hx
YL
1+
X
2
n 1
(16)
If we further simplify Eq. (12) by dropping the Y terms (no
magnetic field), then we obtain n2 1 X, which is equivalent to Eq. (1).
40 44 48 52 56 60 70 80
Figure 7. Logarithmic transmission
curves for curved earth and ionosphere,
parametric in distance between transmitter and receiver.
According to magnetoionic theory, a plane-polarized EM
wave traveling in a medium like the terrestrial ionosphere
will be split into two characteristic waves. The wave that most
closely approximates the behavior of a signal propagating in
this medium, without an imposed magnetic field, is called the
ordinary wave, and the other is called the extraordinary
wave. These terms are taken from the nomenclature for double refraction in optics, although the magnetoionic phenomena are more complicated than the optical ones. The ordinary
wave is represented by the upper sign in the polarization Eq.
(14), except when the wave-normal is exactly along the direction of the magnetic field. Anomalous absorption occurs for
the extraordinary wave when its frequency equals the electron gyrofrequency ( fH B e/me 0.8 to 1.6 MHz). These
frequencies lie in the medium-frequency (MF) band; consequently the absorption of the extraordinary wave [A ( f
f H)2] is large and the polarization of the transmitted wave is
important in the determination of the fraction of the incident
power that goes into the extraordinary wave. This is especially true near the dip equator, where the magnetic field is
nearly horizontal and the field is usually vertical.
In addition to anomalous absorption effects near the electron gyrofrequency, the wave may also experience significant
lateral deviation. This is illustrated for vertical and oblique
propagation in Sections 11.2.2 through 11.2.4 of Ref. 5.
If Eq. (16) is recast as a funtion of and we define f()
(sin2 )/cos and c (B0e/m) f(), then it will be seen to
describe an ellipse. The quantities f() and c play an important part in the description of the polarization behavior of
waves in magnetoionic theory. The magnitude of c is independent of frequency, but varies with the angle between the
wave normal and the magnetic field, , whereas the sign of
c depends on the sign of the charge e and on the direction of
the magnetic field. For longitudinal propagation c 0, and
for transverse propagation c . In the case where X 1,
the quantity c primarily determines the polarization of the
wave. A very complete discussion of R as a function of X and
of the variation of the polarization ellipse is given in Ref. 21.
A more complete understanding of the behavior of EM
waves in the terrestrial ionosphere may be obtained by em-
500
ELECTROMAGNETIC WAVES IN THE IONOSPHERE
ploying two approximations. The quasilongitudinal (QL) approximation applies when the wave is propagating nearly parallel to the geomagnetic field, and the quasitransverse (QT)
approximation applies when the wave propagates in a direction nearly normal to the geomagnetic field. References 21
and 22 contain extended discussions of the QL and QT approximations:
QT: YT4 4(1 X )2YL2
The refractive index n is modified when one introduces collisions between the electrons and heavy particles, and the wave
experiences absorption, which physically is due to the conversion of ordered momentum into random motion of the particles after collision. For each collision, some energy is transferred from the EM wave to the neutral molecules and
appears as thermal energy. We will follow the standard treatment of absorption of radio waves in the ionosphere presented
by Davies (22) and Budden (23).
For the propagation of an EM wave in an unmagnetized
plasma, we can define the absorption index (or coefficient) as
(17)
where is the imaginary part of the refractive index n. For a
magnetized plasma without collisions, we can write
K=
N
e2
20 mc 2 + 2
(18)
On this basis, we can conveniently divide absorption into two
limiting types, commonly called nondeviative and deviative
absorption. Nondeviative absorption occurs in regions where
the product N is large and 1, and is characterized by
the absorption of HF waves in the D region. Deviation absorption, on the other hand, occurs near the top of the ray trajectory or anywhere else on the ray path where significant bending takes place (for small N and 1).
When the refractive index 1, there is essentially no bending of the ray and we can write
K 4.6 102
N
dB/km
(2 + 2 )
(19)
We can further simplify Eq. (19) for the VHF case, since
2 2, as
K = 1.15 103
N
dB/km
f2
(20)
In the MF and HF bands, Eq. (19) may be written as
K = 4.6 102
N
dB/km
(1 2 X 2 )
2c
(22)
In the ionosphere, Eq. (22) reduces to
2c
(23)
where is the group refractive index. For large values of ,
we can write the preceding equation as
Absorption of Radio Waves in the Ionosphere
K=
K=
QL: YT4 4(1 X )2YL2
K=
sorbing layer and there is considerable curvature of the ray
path. The general expression for the absorption index in a
deviating region of a nonmagnetic plasma is
(21)
Unlike nondeviative absorption, deviative absorption occurs when the wave experiences significant group retardation
and consequently spends a relatively long time in the ab-
K=
2c 1 X
(24)
We should remember that the concepts of deviative and nondeviative absorption are limiting cases, and that as a wave
approaches the reflecting level, ray theory breaks down, so we
must employ full wave theory to obtain a complete description
of the behavior of the wave. Extended discussions of application of the QL and QT approximations to ionospheric absorption may be found in Refs. 2, 5, and 23.
Scattering of Radio Waves in the Ionosphere
The principles of scattering of radio waves in general are discussed in the articles ELECTROMAGNETIC WAVE SCATTERING AND
BACKSCATTER. One can qualitatively describe ionospheric scattering as either strong or weak in terms of the received signal
strength of the scattered signal at the receiving radar antenna. An example of the former is VHFUHF backscatter
echoes received from electron density gradients in the auroral
E region, and an example of the latter is incoherent backscatter received by a VHFUHF radar from the undisturbed E or
F layer.
Another way of classifying scattered echoes is in terms of
their backscatter cross section (using a pulsed radar system)
and their temporal stability. A coherent echo exhibits a statistical correlation of the amplitude and phase from one pulse to
another and emanates from quasideterministic gradients in
electron density, which have correlation times usually greater
than 1 ms, corresponding to a spectral width of the radar echo
of less than 1000 Hz (sometimes less than 100 Hz). It also
has a backscatter cross section 104 to 109 times than that from
an incoherent echo. Other important considerations in the
case of coherent backscatter are the relation between the
scattering-irregularity size relative to the backscatter
sounder free-space wavelength, the mean fractional deviation
in electron density of the scatterer, and the aspect angle between the radar line of sight and the major axis of the irregularity. On the other hand, an incoherent echo arises from random thermal fluctuations in the ionosphere, which have
typical correlation times of 20 s, corresponding to a radar
echo spectral width of 50 kHz.
The physical principles governing coherent and incoherent
scattering from the ionosphere are covered in Refs. 2, 5, and
6, while plasma wave theory is covered in detail in Ref. 8, and
extended descriptions of techniques for studying the ionosphere using coherent- and incoherent-scattering sounders
are given in Refs. 2, 3, 5, and 6.
ELECTROMAGNETIC WAVES IN THE IONOSPHERE
Scale
magnetic field (m)
10
100
10
Aggregate
of TIDs
(gravitationally
anisotropic)
Phase
scintillation
One physical principle that makes possible the determination
of ionospheric columnar electron content is Faraday rotation.
This effect (for optics) was discovered by Michael Faraday in
1845, when he subjected a block of glass to a strong magnetic
field. He observed that a plane-polarized monochromatic
beam of light passing through the glass in a direction parallel
to the imposed magnetic field has its plane of polarization
rotated. The amount of rotation is given by the expression
0.1 0.01
= KHl
Atmospheric scale
height
Plasma turbulence
(magnetically anisotropic)
Plasma waves
(magnetically
anisotropic)
106 105 104 103 102 101
Wave number (m1)
(25)
where
Amplitude
scintillation
Radius of earth
Log (spectral density of irregularities)
Wandering of normal
to ionosphere
Mutiple normal
to ionosphere
Faraday Rotation
Electron gyroradius
1000 100
Ion gyroradius
Horizontal
scale (km)
501
angle of rotation
K constant associated with each substance
l length of path of light through the substance (m)
H magnetic field intensity (A/m)
Blur on
ionograms
Strong backscattering
and transequatorial
propagation
at VHF
The Faraday rotation of the electric vector of a radio wave
(see Ref. 45) propagating from a satellite radio beacon in a
direction parallel to the earths magnetic field (as seen by an
observer looking up, in the northern hemisphere) is counterclockwise, as shown in Fig. 9.
Ignoring refraction, the Faraday rotation of the electric
vector is given by
f
=
2c
1
10
X
R
YT4 + 4(1 X )2YL2
(1 X )(1 YL2 ) YT2
ds
(26)
102
Figure 8. Composite spectrum of ionospheric irregularities as a function of wave number over a large spatial scale. (Courtesy of H. G.
Booker.)
Because of charged particle precipitation of solar origin,
ionospheric electric currents and fields, and plasma dynamics,
there exists a wide spectrum of scale sizes of ionospheric irregularities, as shown in Fig. 8. Irregularities are most prevalent at auroral, polar, and equatorial latitudes, although they
also exist at midlatitudes (24). The global morphology of ionospheric irregularities is covered in Refs. 47.
where
Faraday rotation (rad/s)
f wave frequency (Hz)
c 2.998 108 m/s
X kN/f 2
k 80.61
N electron density (e/m3)
and YL, YT are as previously defined.
The integration is between the receiver R, and the satellite S. For VHF frequencies the QL approximation holds and
we can express Eq. (26) as
=
Ionospheric Scintillation
Ionospheric scintillations are fluctuations of amplitude,
phase, and angle of arrival of a VHFUHF signal passing
through irregularities located mainly in the F region. Ionospheric scintillations can have deleterious effects on satellitebased communication and navigation systems. Either extragalactic sources (such as radio stars) or satellite beacon
transmitters may be used as the signal sources for earthobserved studies of ionospheric scintillations, and both geostationary and orbiting satellite beacons have been used.
There is a voluminous body of literature since 1970 describing
the theory, technique, and results of ionospheric scintillation
measurements (25).
K
cf2
f L N ds
(27)
Electric vector
Satellite
antenna
Magnetic
field
Horizontal
Figure 9. Simplified plane geometry of satelliteearth propagation
path in the northern hemisphere to explain Faraday rotation effects.
502
ELECTROMAGNETIC WAVES IN THE IONOSPHERE
where N ds is the ionospheric electron content. Evaluating
the constants yields the relation
8.447 10
Impulse
Whistler
f
S
R
f L N ds rad
(28)
(a)
where f L 2.80 10 BL, the electron gyrofrequency corresponding to the longitudinal component of the geomagnetic
field along the ray path.
Details of the application of Faraday rotation theory and
other techniques to deduce ionospheric columnar electron content may be found in Refs. 2, 5, and 6.
10
f (Hz)
6000
4000
2000
0
Time
(b)
Whistlers
D dispersion (1/2c) s( fN /f 1/2
L ) ds
f N plasma frequency
f L longitudinal component of the plasma frequency
This is the time T for a signal burst to go from one hemisphere to its conjugate point in the opposite hemisphere.
Other natural VLF emissions (called dawn chorus, risers,
hiss, etc.) that are thought to originate in the ionosphere can
also be heard on whistler detection equipment. Since the
1960s, high-power VLF transmitters have been used to generate whistlers to study properties of the magnetosphere
(2628).
FRONTIERS OF IONOSPHERIC RESEARCH
The use of radio waves to explore the terrestrial ionosphere
began with the pioneering efforts of Appleton and of Breit
and Tuve in 1926, when they independently used different
techniques to detect the ionospheric layers. Their work was
founded on Marconis demonstration of transatlantic radio
transmission and on the hypotheses of Kennelly and Heaviside, who independently in 1922 postulated that there must
be radio-reflecting layers in the upper atmosphere to explain
certain experimental results. The foregoing discoveries rested
upon the bedrock of the experimental and theoretical work of
1/f (Hz1/2)
0.04
0.03
0.02
0.01
0
0.5 1.0 1.5 2.0 2.5
Time (s)
(c)
8 hop
6 hop
4 hop
2 hop
Lightning
Time
(d)
Figure 10. Sketch of basic manifestations of a whistler and its initiating disturbance. (a) The frequency spectrum. (b) Frequencytime
curve of a typical whistler. (c) Curve of 1/f with time. Initiating
disturbance and multiple hops when the source and receiver are at
the same end of a magnetic line of force. [After Helliwell (25).]
Heinrich Hertz (1893) and James Clerk Maxwell (1873) respectively.
There seems to have been several peaks in the history of
ionospheric research: first, in the 1920s, following World War
I; second, starting shortly after the end of World War II; and
third, perhaps, starting in the mid-1970s with the advent of
digital techniques, and more recently with the advent of the
National Space Weather Program (see Ref. 29 or 30 or http://
www.nsf.gov/spaceweather/).
While much of the ionospheric research up until about
1960 was in support of HF communications, the advent of satellite communications changed the emphasis to ionospheric
Frequency
where
0.05
Frequency
Whistlers are bursts of EM radiation at VLF that are initiated by lightning discharges and then travel though the ionosphere and magnetosphere in ducts approximately parallel to
geomagnetic lines of force. When translated into sound waves,
whistlers are distinguished by tones of decreasing (or sometimes increasing) frequency, and they may easily be detected
by connecting a suitable antenna to the input of a very sensitive audio amplifier. As a matter of fact, whistlers were first
observed in the last years of the nineteenth century, and were
also heard on the primitive field telephone systems used in
World War I. They have been studied intermittently since
1898, basically as a diagnostic probe of the ionosphere and
magnetosphere (2,25). A graphical representation of whistler
behavior is shown in Fig. 10, and the somewhat rarer nose
whistler behavior is illustrated in Fig. 11. The dispersion relation for whistlers is
f N f L ds
1
D
T=
= 1/2
(29)
1/2
3/2
2c s f ( f L f )
f
tmin
Time
Figure 11. Idealized sketch of the frequency-versus-time characteristics of a nose whistler. [After Davies (5).]
ELECTROMAGNETIC WAVES IN THE IONOSPHERE
research concerning the effects of the ionosphere on transionospheric propagation and research relating the ionosphere to
the magnetosphere. Most current ionospheric research is related to the interrelationship and coupling between regions
over the entire height region of the terrestrial atmosphere
from the troposphere to the magnetosphere and through interplanetary space to the sun.
There are several areas of ionospheric research that currently seem to be producing exciting new results, and these
areas will probably continue to be emphasized well into the
twenty-first century. These areas include (not necessarily in
order of importance) ionospheric modification by using highpower HF transmissions, ionospheric imaging, coherent radars operating from HF through VHF, and incoherent scattering radars (25). Most of these techniques are employed at
high geomagnetic latitudes as part of the Space Weather Program, but some are also deployed in equatorial regions. We
will briefly describe the essentials of each of these areas of
current ionospheric research emphasis.
Ionospheric Modification
In the late 1960s the availability of military surplus equipment such as very high-voltage and -current power supplies
and HF vacuum tubes capable of many kilowatts of RF output, together with advances in antenna array theory and
practice, induced experimenters to design systems to heat or
otherwise modify the ionosphere. As a result of experiments
performed in the early 1970s at the Platteville, Colorado HF
high-power heating facility (31), some 10 new ionospheric
modification facilities were established and have produced
significant results (see Chap. 14 of Ref. 5). The various modification facilities are listed in Table 2. Other ionospheric modification facilities are located in Russia and Ukraine at Kharkov, Moscow, Zimenki, and Monchegorsk. More information
on the HAARP and other heaters may be obtained on internet
at http://www.haarp.alaska.edu/haarp/airf.html.
Ionospheric Imaging by Radio
For over three decades now, ionospheric scientists have investigated using radio methods to image the ionosphere. Rogers
(32) was probably the first to suggest using the wavefrontreconstruction method for this purpose. Many attempts have
been made to produce holographic images of the ionosphere,
but it has not proven to be a very successful technique
probably because of the difficulty in uniformly illuminating a
large enough horizontal slab of the ionosphere, not using a
sufficient number of receivers, and the inability to achieve
precise enough measurements of amplitude and phase of the
reflected wave.
On the other hand, another technique (borrowed from
medical technology), computerized ionospheric tomography
(CIT), has produced quite significant results in imaging the
regular (and some irregular) features of the ionosphere. Basically, this technique utilizes radio beacons on satellites in
near-polar orbits and a latitudinal chain on the earth subsatellite path of carefully calibrated TEC receivers, to make
many measurements of total electron content (TEC). The basic geometry is illustrated in Fig. 12.
Currently, VHFUHF beacons on the TRANSIT, GLONASS, and GPS satellites are the most used as signal sources
to measure TEC to use in CIT ionosphere reconstructions.
One must also use some a priori information (ionospheric
models) and ionosonde data in the algorithms in order to
achieve realistic results. Some recent results are summarized
in Refs. 3338 and on the Internet at http://
www.arlut.utexas.edu/~grk/Mace/mace.html and at http://
sideshow.jpl.nasa.gov:80/gpsiono/.
Another ionospheric imaging technique is the IRIS system
(Imaging Riometer-Ionospheric Studies) (46) which uses an
antenna array of up to 64 elements to provide images of enhanced auroral absorption structure in the D-region.
Coherent Radars
As described in the subsection Scattering of Radio Waves in
the Ionosphere, HFUHF coherent backscatter from ionospheric irregularities can provide very useful information on
the morphology and physics of a wide range of irregularity
scale sizes. At this time there are about 20 of these backscatter sounders deployed, operating on frequencies from 8 MHz
to 200 MHz, distributed mainly in the high-latitude and equatorial regions. These radars are sited so that the main antenna lobe is directed to intercept irregularities at near-normal incidence at E- and F-region heights.
The HF coherent radars are mainly grouped into a large
network, which covers approximately half of the northern polar cap ionospherethe SuperDARN network (40), which is
shown on the map in Fig. 13. Much information has been
gained on the F-region plasma convection patterns in the polar cap, atmospheric gravity waves, and other ionospheric
phenomena related to ionosphere magnetosphere interaction;
see Refs. 4042 or http://sd-www.jhuapl.edu/RADAR. The
VHFUHF coherent radars are documented in Refs. 25, and
some useful Internet sources are to be found at http://dan.sp.
Table 2. Ionospheric Modification Facilities (1970 to 1978)
Facility
First
Used
Latitude
Platteville, CO
Arecibo, PR
SURA, Russia
Tromsoe, Norway
HIPAS, Alaska
HAARP,a Alaska
1970
1980
1980
1980
1977
1997
40.2N
18N
56.1N
69.6N
64.9N
62.4N
503
Longitude
Geomag.
Lat.
Transmit
Power
Freq.
Range
(MHz)
Antenna
Gain (dB)
104.7W
67W
46.1E
19.2E
146.8W
145.2W
49
32
71
67
65
62
1.6 MW
800 kW
750 kW
1.5 MW
800 kW
3.6 MWa
2.725
315
4.59
2.58
2.8, 4.5
2.810
18
25
26
28
17
30a
HAARP is currently under construction. Values given are for the completed facility.
504
ELECTROMAGNETIC WAVES IN THE IONOSPHERE
40
45
50
55
Latitude (deg)
Figure 12. Basic satelliteearth geometry for computer ionospheric tomography, illustrating
the multiple ray paths on which the total electron content is measured.
sp-agency.ca/www/cpus1e.htm, http://thor.ee.cornell.edu/~wes/
CUPRI.
Incoherent Scatter Radars
One of the most powerful earth-based radio methods for
studying the ionosphere is the incoherent scatter radar (ISR)
technique, which has been in use since the early 1960s. ISRs
HF radars
SuperDARN
can reveal the electron density, electron and ion temperature,
plasma velocity, and other ionospheric parameters, even during very disturbed conditions (25). At the present time there
are some seven ISRs in operation, located from the north polar cap to the magnetic equator and spread longitudinally
from Scandinavia to Japan. The newest ISR is located at
Svalbard, Norway (43), and another ISR is being planned for
a polar observatory at Resolute, Canada in the near future
IS radars
Sondrestrom
EISCAT
Magnetometer chains
Alaska
Canopus
MACCS
SuperDARN
Figure 13. Northern hemisphere map showing area coverage of SuperDARN HF radars,
incoherent scattering radars, and magnetometer chains. (Courtesy of R. A. Greenwald.)
EISCAT
Sondrestrom
Greenland
Magic
Image
ELECTROMAGNETIC WAVES IN THE IONOSPHERE
505
(44). Data from ISRs are essential in studying the relation
between the magnetosphere, ionosphere, and middle atmosphere at high and equatorial latitudes (47,48).
26. D. Park and D. Carpenter, Very low frequency radio waves in the
magnetosphere, in L. J. Lanzerotti and C. G. Parrs (eds.), Upper
Atmosphere Research in Antarctica, Res. Ser. vol. 29, Monogr. 72,
Washington: American Geophysical Union, 1978.
BIBLIOGRAPHY
27. D. L. Carpenter, Remote sensing of the magnetospheric plasma
by means of whistler mode signals, Rev. Geophys., 26: 535549,
1988.
1. H. Alfven, Cosmic Plasma, Dordrecht, Holland: D. Reidel, 1981.
2. R. D. Hunsucker, Radio Techniques for Probing the Terrestrial
Ionosphere, Heidelberg: Springer-Verlag, 1991.
3. R. D. Hunsucker, A review of ionospheric radio techniques: Present status and recent innovations, in W. R. Stone (ed.), Review
of Radio Science 19901992, Oxford: Oxford Univ. Press, 1993,
Chap. 22.
4. J. K. Hargreaves and R. D. Hunsucker, The High-Latitude Ionosphere and its Effects on Radio Propagation, Cambridge, UK: Cambridge Univ. Press, 1999, in press.
5. K. Davies, Ionospheric Radio, London: Peregrinus, 1990.
6. J. K. Hargreaves, The Solar-Terrestrial Environment, Cambridge,
UK: Cambridge Univ. Press, 1992.
7. M. C. Kelley, The Earths IonospherePlasma Physics and Electrodynamics, San Diego, CA: Academic Press, 1989.
8. J. R. Wait, Electromagnetic Waves in Stratified Media, Oxford:
Pergamon, 1970.
9. J. R. Wait, EM Scattering from a vertical column of ionization in
the earthionosphere waveguide, IEEE Trans. Antennas Propag.,
39: 10511054, 1991.
10. A. D. Watt, VLF Radio Engineering, Oxford: Pergamon, 1967.
11. R. D. Hunsucker and B. S. Delana, High-latitude field-strength
measurements of standard broadcast band skywave transmissions
monitored at Fairbanks, Alaska, Final Report on FCC Contract
Number FCC-0375, Geophysical Institute/Univ. Alaska, 1988.
12. J. M. Goodman, HF CommunicationsScience and Technology,
New York: Van Nostrand Reinhold, 1992.
13. E. E. Johnson et al., Advanced High-Frequency Radio Communications, Boston, MA: Artech House, 1997.
14. R. D. Hunsucker, Auroral and polar-cap ionospheric effects on
radio propagation, IEEE Trans. Antennas Propag., 40: 818828,
1992.
15. E. V. Appleton, Geophysical influence on the transmission of
wireless waves, Proc. Phys. Soc. London, 37 (2): 16D22D, 1925.
16. H. W. Nichols and J. C. Schelling, The propagation of electric
waves over the earth, Bell Syst. Tech. J., 4: 215, 1925.
17. C. S. Gillmor and W. Altar, Edward Appleton and the magnetoionic theory, Proc. Amer. Philos. Soc., 126 (5): 395440, 1982.
18. W. H. Eccles, Proc. R. Soc. London A, Math. Phys. Sci., 87: 79,
1912.
19. J. M. Kelso, Radio Ray Propagation in the Ionosphere, New York:
McGraw-Hill, 1964.
20. N. Smith, The relation of radio sky-wave transmission to ionospheric measurements, Proc. IRE, 27: 332347, 1939.
21. J. A. Ratcliffe, The Magnetoionic Theory and Its Applications to the
Ionosphere: A Monograph, London: Cambridge Univ. Press, 1959.
22. K. Davies, Ionospheric Radio Waves, Waltham, MA: Blaisdell,
1969.
23. K. G. Budden, The Propagation of Radio Waves. The Theory of
Radio Waves of Low Power in the Ionosphere and Magnetosphere,
Cambridge, UK: Cambridge Univ. Press, 1985.
28. R. A. Helliwell, VLF Whistler Mode Experiments, Cambridge, MA:
Cambridge Univ. Press, in press.
29. J. W. Wright, Jr. (ed.), National Space Weather Program
Strategic Plan, FCM-P30-1995, 1995, OFCMSSR; 8455 Colesville
Rd. Suite 1500; Silver Spring, MD 20910.
30. The National Space Weather ProgramThe Implementation Plan,
FCM-P31-1997, January 1997.
31. W. F. Utlaut (ed.), Special issue: Ionospheric modification by high
power transmitters, Radio Sci., 9: 8811089, 1974.
32. G. L. Rogers, A new method of analyzing ionospheric movement
records, Nature, 177: 613614, 1956.
33. B. D. Wilson, Subdaily northern hemisphere ionospheric maps
using an extensive network of GPS receivers, Radio Sci., 30 (3):
639648, 1995.
34. H. Na, J. Shen, and H. Lee, A Fourier domain technique for ionospheric tomography, Radio Sci., 30 (3): 747754, 1995.
35. M. Hernandez-Pojares, J. M. Juan, and J. Sanz, Neural network
modeling of the ionospheric electron content at global scale using
GPS data, Radio Sci., 32 (3): 10811089, 1997.
36. L. Kersley et al., Imaging of electron density troughs by tomographic techniques, Radio Sci., 32 (4): 16071621, 1997.
37. R. Leitinger, H.-P. Landreiter, and G. Kirchengast, Ionospheric
tomography with data from satellite reception of Global Navigation Satellite system signals and ground reception of Navy Navigation satellite system signals, Radio Sci., 32 (4): 16571667,
1997.
38. C. Coker, R. Hunsucker, and G. Lott, Detection of auroral activity
using GPS satellites, Geophys. Res. Lett., 22 (23): 32593262,
1995.
39. R. A. Greenwald et al., DARN/Superdarn: A global view of the
dynamics of high-latitude convection, Space Sci. Rev., 71: 761
796, 1995.
40. R. A. Greenwald et al., Mesoscale dayside convection vortices and
their relation to substorm phase, J. Geophys. Res., 101 (A10):
21,69721,713, 1996.
41. W. A. Bristow and R. A. Greenwald, On the spectrum of thermospheric gravity waves observed by the SuperDARN network, J.
Geophys. Res., 102 (A6): 11,58511,595, 1997.
42. A. V. Kustov et al., Dayside ionospheric plasma convection, electric fields and field-aligned currents derived from the SuperDARN radar observations and predicted by the IZEMEM model,
J. Geophys. Res., 102 (A11): 24,05724,067, 1997.
43. G. Wannberg et al., The EISCAT Svalbard radar: A case study
in modern incoherent scatter radar system design, Radio Sci., 32
(6): 22832307, 1997.
44. M.-C. Kelley (ed.), A polar cap observatory: The next step in upper
atmosphere science, available from Ms. Sally Bird, The Theory
Center, Room 304, Cornell Univ., Ithaca, NY 14853.
45. W. A. S. Murray and J. K. Hagreaves, Lunar radio echoes and
the Faraday effect in the ionosphere, Nature, 173: 944, 1954.
24. R. D. Hunsucker, Characteristic signatures of the midlatitude
ionosphere observed with a narrow-beam HF backscatter
sounder, Radio Sci., 6: 535548, 1971.
46. J. K. Hargreaves, D. L. Detrick, and T. J. Rosenberg, Spacetime
structure of auroral radio absorption events observed with Imaging-Riometer of South Pole, Radio Sci., 26: 925930, 1991.
25. R. A. Helliwell, Whistlers and Related Ionospheric Phenomena,
Stanford, CA: Stanford Univ. Press, 1956.
47. J. V. Evans , Ionospheric movement measured by ISR: A review.
J. Atmos. Terrestr. Phys., 34: 175, 1972.
506
ELECTROMETERS
48. C. LaHoz (ed.), Special issue: Selected papers from the 6th Int.
EISCAT Workshop, J. Atmos. Terrestr. Phys., 58: 1507, 1996.
ROBERT D. HUNSUCKER
RP Consultants
SKY WAVE PROPAGATION AT MEDIUM AND
HIGH FREQUENCIES
SPECTRUM CONSIDERATIONS
Medium frequencies and high frequencies are usually dened as the frequency bands from 300 kHz to 3000 kHz
and from 3 MHz to 30 MHz, respectively. However, as far
as ionospheric propagation is concerned, there are no sharp
divisions. The medium-frequency band is dominated by
the amplitude modulated (AM) broadcasting band between
about 500 kHz and about 1700 kHz, which is designed primarily for ground-wave usage. On the other hand, highfrequency systems are designed for long-distance sky wave
propagation and for some 50 years (from about 1925 to
about 1975) provided the primary vehicle for global communications. High frequencies are still used extensively
for communications because of the following advantages:
(1) low cost of terminal equipment, (2) low power requirements, and (3) adequate bandwidths. By contrast, medium
frequencies suffer heavy ionospheric absorption during
daytime and, therefore, there is relatively little co-channel
interference; whereas by night, when the absorption is
small, interference between closely spaced channels is common. High-frequency sky waves suffer from several disadvantages brought about by (1) the temporal and geographical variability of the ionosphere, (2) the large number of
possible propagation paths and the consequent time dispersion of the resulting signal, (3) large and rapid amplitude and phase uctuations, (4) high interference because
of spectrum congestion, and (5) frequency distortion of
wideband signals. An important disadvantage is the occurrence of several types of ionospheric disruptions caused by
solar disturbances. These disruptions can be hemispheric,
such as those on the dayside caused by bursts of solar X
rays (sudden ionospheric disturbances) or conned mostly
to high latitudes (e.g., polar cap disturbances and ionospheric storms that originate in the auroral zones and
spread to populated middle latitudes). Ionospheric storms
are major concerns for high-frequency (HF) users. They occur mostly at high sunspot numbers when the higher critical frequencies help to mitigate their adverse effects (see
Propagation of broadcast transmissions).
In this article we shall discuss the following topics:
1. Basic physical properties of sky wave propagation
(namely, refraction, reection, penetration, and absorption)
2. relationships between vertical propagation and
oblique propagation
3. ionospheric models
4. characteristics of medium frequencies
5. characteristics of high frequencies
6. prediction programs for ionospheric sky wave performance
7. real-time channel evaluation
Figure 1. Ionospheric structure on a summer day and night in
middle latitudes, and the main bands of solar and cosmic-ray ionizing radiations.
There is an extensive literature in the eld of ionospheric
radio propagation, and the interested reader is referred to
books by Davies (1), Goodman (2), Hunsucker (3), and McNamara (4) and the bibliographies therein. Groundwaves
are particularly important on medium frequencies (see Radiowave Propagation Concepts).
PROPAGATION CHARACTERISTICS
The Ionosphere
Sky waves result from radio refraction and reection from
the ionosphere. The ionosphere is usually dened as that
part of the upper atmosphere where sufcient ionization
exists to affect the propagation of radio waves. The ionosphere lies between about 50 km and about 2000 km. The
peak electron density usually occurs in the F region (above
140 km) (Fig. 1). The F region often contains two layers:
the lower F1 layer and the upper F2 layer. Below the F
region is the E region (90 km to 140 km), which contains
the normal E layer and sporadic E. The D region (50 km to
90 km) contains the D layer and the C, or cosmic ray, layer.
Above the F peak is the topside, and above about 2000 km is
the protonosphere. The boundaries between these regions
are not well dened. The alphabetic nomenclature was introduced by Appleton (see Ref. 1, Sec. 1.1), who used the
letter E for electric and F for eld. These letters left room
for the discovery of other layers. Ionospheric electron densities vary by orders of magnitude depending on altitude,
time of day, season, sunspot number, solar disturbances,
and geographical location. It is this variability that renders sky waves so difcult to manage.
J. Webster (ed.), Wiley Encyclopedia of Electrical and Electronics Engineering. Copyright 2007 John Wiley & Sons, Inc.
Sky Wave Propagation at Medium and High Frequencies
Refraction, Reection, and Penetration
In an ionized plasma, with electron density N elm2 , in
the absence of collisions and an external magnetic eld,
the radio refractive index of a wave of frequency f Hz, is
given by
where fN is the plasma frequency in Hz. Ionospheric electron densities are such that typical E layer and F layer
plasma frequencies lie in the range from 0.5 MHz to 30
MHz. Thus, for medium and high frequencies, the refractive index of the ionosphere is less than unity and application of Snells law shows that, on entry into the ionosphere, a wave is refracted away from the vertical. When
the electron density is sufcient, the direction of propagation becomes horizontal and reection occurs. The main
sky wave propagation mechanisms are illustrated in Fig. 2:
absorption, reection, scatter, and penetration. When the
maximum electron density is insufcient, the wave penetrates; this is essential for ground-to-satellite communication. Because ionospheric scatter propagation is considered
elsewhere (see Meteor burst communication), it will not
be discussed in detail in this article.
Absorption
The neutral atmosphere affects the propagation of radio
waves because of electron-neutral collisions that convert
the ordered momentum of the wave into heat and electromagnetic noise. To a rst approximation the loss L, in decibels per kilometer, is
where is the number of collisions made by one electron per
second. From Eq. (2) we see that with 1, the nondeviative absorption is inversely proportional to the frequency
squared so that, in general, signals are stronger on high
frequencies than on medium frequencies. The product N
maximizes in the D region. The loss L can be large where
0; this deviative absorption occurs near reection, or
where there is pronounced refraction of the wave.
Effects of the Earths Magnetic Field
The earths magnetic eld affects sky waves by splitting
the incident wave into ordinary and extraordinary waves
that are oppositely polarized and by affecting the global
structure of the ionosphere (see Electromagnetic Waves
in the Ionosphere). The ordinary (o) wave is polarized
such that, with the thumb pointing in the direction of the
earths magnetic eld, the electric eld rotates in a lefthanded sense whereas the extraordinary (x) wave rotates
in a right-handed sense. For example, an incident, vertically polarized wave on exit from the ionosphere will be
elliptically polarized, which depends on the relative amplitudes and polarizations of the emerging o and x waves. The
relative amplitudes of the o and x waves are determined by
the extent to which the incident power is divided between
the two component waves and the relative absorptions of
the two waves thereafter. The relative absorptions are
where the + sign refers to the ordinary wave and the sign
to the extraordinary wave, and A is essentially the same for
both waves. The electron gyrofrequency fH is the natural
frequency of rotation of an electron (right-handed) about
the magnetic eld. In the ionosphere fH varies from about
0.8 MHz at the magnetic equator to about 1.6 MHz near the
magnetic poles. Current models of the geomagnetic eld
are available on the Internet (e.g., Ref. see 9). Equation
(3) shows that the absorption of the ordinary wave is less
than that of the extraordinary wave so that an antenna
should be designed to excite as much of the ordinary wave
as possible (see Ref. 1, Sec. 7.6).
The global effects of the geomagnetic eld are 2-fold.
Near the magnetic equator there is an F2-layer anomaly
in which the peak electron densities maximize in the late
afternoon at magnetic latitudes near 15 . Magnetic latitude is dened in terms of the dip angle I by which the
earths magnetic eld dips below the horizontal.
Another way in which the earths eld affects the ionosphere is via the precipitation of magnetospheric charged
particles into the polar caps and the auroral zones (65 to
70 magnetic latitude), where energy is deposited that produces ionospheric storms. In the auroral zones, D-region
electron densities are enhanced producing auroral absorption. Energetic solar protons enter the atmosphere over the
polar caps (latitudes >70 ) and produce intense polar cap
absorption (PCA) that can black out HF signals for several
days.
VERTICAL AND OBLIQUE PROPAGATION
Equivalence of Vertical and Oblique Reection
Equation (1) shows that, with a radio wave incident at an
angle with the vertical on a plane ionosphere, reection
occurs when
With vertical propagation, = 0 , so f = fN and, therefore,
for a given maximum electron density (i.e., maximum fN ) a
maximum, or critical, frequency is reected. With oblique
propagation, the maximum frequency depends jointly on
the critical frequency and on the incident angle, which is
determined by the ground range and the layer height. The
subject of vertical-to-oblique conversion has been discussed
extensively by numerous authors (e.g., Ref. 1, Chapter 6;
Ref.2, Chapter 4; Ref. 4, Chapter 4). With a at earth and
at ionosphere the relationship between a frequency fo incident obliquely at an angle and a frequency fv reected
vertically from the same true height is
where D is the ground range and h is the virtual height of
reection. The variation of virtual height with frequency
is called an ionogram, and the relationship between sec
Sky Wave Propagation at Medium and High Frequencies
Figure 2. The four main mechanisms in MF and HF sky wave
propagation: absorption, reection, scatter, and penetration.
and h , when presented graphically, is called a transmission curve. The intersection of a transmission curve with
the ionogram trace gives the virtual heights of reection
of the obliquely traveling signal. When the transmission
curve is tangent to the ionogram, we have the maximum
reected frequency. On frequencies below the maximum
frequency there are two intersections of the transmission
curve and the ionogram trace (for a single layer) showing
that, for a given ground range, there is a low-angle ray and
a high-angle ray. Modications to the plane ionosphere geometry are required for ground ranges over about 200 km
and for the extraordinary ray that determines the maximum frequency, but the principles are essentially the same
(the reader is referred to the aforementioned texts and references therein).
Parabolic Layer Theory
Transmission curves were used extensively for the determination of maximum usable frequencies (MUF) for a
number of years (approximately 1940 to 1980). With the
advent of inexpensive and fast computers an alternative
method has come to the fore. Essentially, the method consists of using ionogram data to approximate the electrondensity prole by an analytic function such as a parabola,
or a quasi-parabola, and calculating the maximum frequency. A parabolic prole is dened by its critical frequency fc , the semithickness ym , and the height of the peak
hm . More complicated ionospheres can be approximated by
two (or more) parabolas or by two parabolas joined by linear segments. The International Reference Ionosphere (5)
also uses a segmented prole. The ground range in each
segment is calculated analytically.
Ray Tracing
With realistic proles obtained by ionogram inversion,
which have horizontal and vertical structures and which
cannot be represented by simple analytical formulas, there
is recourse to ray tracing. This involves starting with a ray
from a specied transmitting site and with specied angles
of elevation and azimuth, and plotting the ow of energy
step by step until it returns to earth or escapes into space.
This is a complicated procedure and is feasible only with
adequate computers. A comprehensive three-dimensional
ray tracing program has been constructed by Jones and
Stephenson (6) and is available on the Internet (7). In Fig.
3a we see what happens to the ray paths (on a given frequency) as the angle of elevation slowly increases. For low
angles the ground range is long. As the elevation increases,
the ground range decreases until the skip is reached, after
which the range increases rapidly. Eventually penetration
occurs. Figure 3b shows the equivalent triangular paths
for different angles of elevation. The apexes of the equivalent triangles lie on a smooth curve, called a reectrix,
as shown in Fig. 3c. The ray with its apex at the nose of
the reectrix is the skip-ray and its frequency is the MUF.
Rays reected at lower virtual heights are low-angle rays,
and rays reected on the upper side of the reectrix are
high-angle rays. For a given reectrix, the relationship between virtual height and angle of elevation () is given by
Fig. 3d (8).
Sky Wave Propagation at Medium and High Frequencies
Figure 3. (a) Real rays in a concentric earth and Chapman ionosphere, f = 8 MHz, fc = 4 MHz,
hmax = 300 km, semithickness = 200 km. (b) Equivalent or virtual paths reected at the apexes of
the ctitious triangular paths. (c) The reectrix as the locus of the virtual reection heights. (d)
The reectrix as a function of virtual height, range, and elevation angle. (Adapted from Croft (8)
by permission of the American Geophysical Union.)
Maximum Frequencies
When the ground range, D, is plotted against the angle
we nd that, for f > fc , there is a minimum range, Ds , called
the skip distance, within which no power is received via
the normal reection process. However, signal power in the
skip zone may result from ground waves, ground scatter,
and scatter from ionospheric irregularities. At the edge of
the skip zone the signal frequency corresponds to the maximum usable frequency for the skip distance, MUF(Ds ). The
MUF for a distance D is related to the critical frequency by
where M(D) is called the MUF factor corresponding to the
distance D. The M(3000)F2 is used extensively both as a
reference and as a measure of the height of the F2 layer.
IONOSPHERIC MODELS
Ionospheric models are of two types: (1) empirical models
based on data, and (2) physical models based on production, loss, and movement of plasma by winds and/or electric
elds. For long-term skywave predictions it is customary to
use monthly median models of the main ionospheric characteristics, such as critical frequencies ( fo Ehm E , fEshm F2 ,
fo F1, fo F2) and layer heights (hm E, hm F2) or equivalent M
factors. The diurnal, seasonal, and sunspot number dependencies of the regular E and F1 layers are well behaved and
are represented by analytical expressions (e.g. see Ref. 1,
Chapter 5). However, the most important (F2) layer varies
irregularly, and its characteristics are expressed in the
form of numerical maps of median values. The day-to-day
variability is expressed in terms of upper-decile and lower-
Sky Wave Propagation at Medium and High Frequencies
decile values. D-region absorption, while highly variable in
space and time, is represented by an analytical expression
(see Ref. 1, Eq. 12.12). Electron-density proles can be represented numerically or, for ray tracing, more conveniently
by integrable segments. Several such approximations are
discussed in Davies (Ref. 1, Sec. 5.3.2, 5.3.4, and 5.3.5).
Of particular interest is the International Reference Ionosphere (IRI) which produces electron-density proles, ion
and electrontemperature proles, layer heights, and so on
and is available from World Data Centers on the Internet
(7, 9). Some models, such as PRISM (10), can be updated
using current ionospheric data, such as total electron contents (see Ref. 1, Chapter 8) and/or critical frequencies.
PROPAGATION ON MEDIUM FREQUENCIES
Importance of Medium Frequencies
The medium frequency (MF) band is dominated by the AM
broadcast band ( 500 kHz to 1700 kHz), which during daytime depends on groundwave propagation. This frequency
band has a major economic, social, and political impact on
everyday life and is intensively used. Knowledge of skywave properties in the MF band is restricted because of (1)
the congested spectrum, (2) high ionospheric absorption,
(3) the role of the earths magnetic eld, (4) difculty in
separating deviative ( 0) from nondeviative ( 1) effects, (5) effects of collisions, and (6) the medium not always
slowly varying on the lower frequency end, so that ray theory may be inapplicable. For efcient use of the MF band it
is essential for several users to operate on the same, or adjacent, channels with minimum interference. Thus determination of the signal strength in distant regions is important
in channel sharing (11). Atmospheric radio noise on the MF
band has been modeled by Herman and DeAngelis (12).
It is difcult to collect sky wave data during the daytime
because the signals are weak and/or over short distances
the ground wave may mask the sky wave. For estimating
sky wave eld strengths, the Radio Communications Sector of the International Telecommunications Union (ITU)
has recommended the Wang (11) method for North America
and the CCIR (Consultative Committee on International
Radio) (13) Recommendation 435-8 elsewhere. These two
methods are in reasonable agreement throughout Europe.
Temporal Variations
Diurnal Variations. Because the MF band is dominated
by the broadcast band, relatively little is known about
the sky wave signal structure. Yet this band has the
heaviest usage. During daylight high D-region absorption
suppresses sky wave signals to an extent sufcient for
interference-free broadcasting in most areas. However, in
areas remote from an MF transmitter, weak sky waves
may sufce to produce unwanted interference. Shortly after sunset, when the electron content of the D region essentially disappears, sky wave strengths increase and AM
stations can be received at distances of several thousand
kilometers. During daytime MF waves are reected from
the E layer, but during nighttime sky waves may be received from both E and F layers, especially on the upper
end of the MF band.
Seasonal Variations. Measurements in Europe (see Ref.
14, Report 431-4) show equinoctial maxima of signal
strength and minima in summer and winter, the summer
minimum being the more pronounced. The overall seasonal
variation may be as much as 15 dB at the lower end of
the MF band, decreasing to about 3 dB at the upper end
of the band. Over the western hemisphere there is little
seasonal variation, and over the United States there is a
slight minimum in summer. The sensitivity to length of day
and to magnetic disturbance of broadcast signals received
in Canada is illustrated in Fig. 4. This gure shows the
frequency-integrated power in the AM broadcast band (550
kHz to 1600 kHz) received in Arviat, Northwest Territories
during a six-month interval, September 1994 through April
1995. Local time is on the vertical axis, with local midnight
(0600 UT) near midscale. The gure shows the normal diurnal pattern of MF sky wave reception in that signals are
strong at night (middle of the gure) and weak during the
day (top and bottom of the gure). The span of reception
extends from near sunset ( 00 UT) to near sunrise ( 12
UT), corresponding to the seasonal variations of the terminator. The reception is longest near midwinter. In summer
reception is short lived. There is an asymmetry between
the dawn and dusk terminators: The dawn terminator produces a sharper transition between nighttime conditions
and daytime conditions than is the case near dusk. This
phenomenon results from the prompt production of the D
layer at sunrise versus the relatively slow decay of the D
layer after sunset.
The lower panel in Fig. 4 shows the daily magnetic index from September 1994 to April 1995 and illustrates the
dependence on geomagnetic activity of the high-latitude
reception of distant AM broadcast band transmissions.
Several-day intervals with high magnetic disturbance correlate one-to-one with intervals when the integrated power
in the AM broadcast band at night is nearly the same as
it is during the day. The latter intervals appear as light
vertical bands in the top panel of Fig. 4. On shorter time
scales, prompt absorption of distant AM transmissions is
a sensitive indicator of auroral disturbance (15). Another
indication of the seasonal variation is the average daily
number of hours of MF reception at Fairbanks, Alaska of
signals from ve stations in the United States and Canada:
Hunsucker and DeLana (16) found these hours to be 13.3
h in winter and spring, 4.0 in summer, and 11.0 in autumn.
Daytime eld strengths, though weak, are higher in winter
than in summer. The winter-to-summer ratio is typically 10
dB to 20 dB. For planning purposes annual median eld
strengths are used (Ref. 11, Sec.4.1). Measurements indicate that the annual median value at noon is some 42.5 dB
lower than at 6 hours after sunset, which is the reference
time used by Europeans.
Sunspot Dependence. In Europe Ebert (17) found that
the 11-month smoothed midnight signal, F(11), and the corresponding smoothed sunspot number, R(11), are related
by
Sky Wave Propagation at Medium and High Frequencies
Figure 4. Integrated power in the frequency range 550 kHz to 1600 kHz received in Arviat, North
West Territories, Canada from September 1994 through April 1995. Local meridian (270 E) time
is on the vertical axis. The reception is best near the winter solstice. (Courtesy of J. LaBelle.)
The sensitivity to sunspot number depends on the hour of
observation. In Europe, the reference hour is normally 6
h after sunset, whereas in North America it is customary
to use sunset + 2 h, at which the difference between eld
strengths at sunspot maximum and sunspot minimum during 1944 through 1947 was 14 dB compared with only 8 dB
at sunset + 6 h.
Dependence on Wave Frequency
Some measurements indicate that the signal strength increases with wave frequency, whereas others suggest a
decrease of signal strength with increase of frequency.
Both these conclusions have some basis because, at night,
an appreciable fraction of the ionospheric absorption may
occur near the reection level (i.e., deviative absorption,
which may increase with increasing frequency; see (Ref. 1,
Sec.11.5.3, which gives several formulas for the frequency
dependence of MF eld strengths). For many practical purposes it is sufcient to take the eld strength on 1000 kHz
as representative of the entire band.
Dependence on the Geomagnetic Field
The CCIR (13) basic loss factor is
where the frequency, f, is in kHz and is the magnetic latitude at the center of the path for 60 and = 60 for
magnetic latitudes greater than 60 . Thus reception is poor
near the auroral zones. This is illustrated in Fig. 4, which
shows gaps in the reception of signals on days with high
magnetic indexes. This is the result of increased auroral
absorption on the path due to increased D-region electron
content and equatorial motion of the auroral zone. MF signals are sensitive to auroral absorption and above a certain
threshold of disturbance are blacked out.
For eastwest and westeast propagation in low magnetic latitudes (e.g., in Africa), polarization coupling loss
is important. This arises because the electric eld, from a
vertical antenna, is perpendicular to the (horizontal) magnetic eld and excites only an extraordinary wave, which
suffers high D-region absorption [see Eq. (3), with f fH ].
Predicting Field Strengths
Methods for predicting MF sky wave eld strengths are
given in CCIR [see Ref. 13, pp. 311 to 390; PoKempner
(18); and Davies (1), Sec. 11.7]. The CCIR method takes
into account antenna gain, propagation loss, polarization
coupling, and, where appropriate, sea gain.
Sky Wave Propagation at Medium and High Frequencies
Fading on MF
Nearly all sky waves uctuate, or fade, with time as a result of interference between component echoes, absorption
changes, polarization changes, and so on. To determine a
stations sky wave service area and its interference potential, it is important to know the percentage of time that
a given eld strength is exceeded. For example, elds exceeded for 1% and 10% of the time are about 13 dB and 8 dB,
respectively, higher than the median. In high geomagnetic
latitudes, the corresponding differences are 15 dB and 10
dB (19).
spread sets a limit to the rate of transmission of information because overlapping echoes can result in errors.
Roughly speaking, the maximum rate of transmission (in
binary units per second) is equal to the reciprocal of the
time spread, which is a function of signal frequency, path
length, geographical location, season, sunspot number, and
so on. Ionospheric structures, both vertical and horizontal,
produce echoes with various angles of elevation and azimuth. The interested reader should consult Ref. 4, Chapters 12 through 15, for a detailed discussion of direction
nding.
Fading on HF
PROPAGATION ON HIGH FREQUENCIES
The Available Spectrum
The HF sky wave spectrum is bounded on the upper end by
the MUF, which is essentially determined by the maximum
electron density in the reecting layer, and on the lower end
by the lowest usable frequency (LUF), which is determined
by D-region absorption and/or by E-layer cutoff. Because of
day-to-day uctuations (15% about the monthly median)
of the F2 critical frequencies, operation on the MUF would
provide reception for 50% of the time at a particular hour.
To ensure communications 90% of the time, it is customary
to operate at, or below, the optimum working frequency (or
FOT from the French initials), which is dened empirically
as 0.85 of the monthly median MUF. Since the MUF can
be dened in several ways, the CCIR (Ref. 14, Recommendation 373-5) has adopted the following: (1) Basic MUF is
the highest frequency by which a radio wave can propagate between given terminals, on a specied occasion, by
ionospheric refraction alone; and (2) operational MUF, or
simply MUF, is the highest frequency that would permit acceptable operation of a radio service between given radio
terminals at a given time under specied working conditions (such as antennas, transmitter power, class of emission, information rate, and required signal-to-noise ratio).
The operational MUF refers to propagation in the actual ionosphere and includes effects of scattering, partial
reection, and so on, whereas the basic MUF essentially
depends only on ionospheric refraction. Hence, in general,
the operational MUF is equal to or greater than the basic
MUF.
Path Structure
The layered structure of the ionosphere can produce complicated oblique echoes. A composite echo may consist of
the following: high- and low-angle rays from the E (including Es), F1, and F2 layers (See Characteristics of the Ionosphere for a discussion of ionospheric structure and variability); one-hop, two-hop, and so on echoes; and scatter
from the ground and from ionospheric irregularities. Figure 5 shows a sample oblique ionogram taken, during magnetically quiet conditions, over a geographically eastwest
path. The traces are sharp and the maximum, or junction
(J), frequency of the one-hop F2 trace is well dened. The
high-angle, or Pedersen, trace can be seen for the one-hop
trace. Splitting of the high (H)-angle, two-hop trace into
ordinary and extraordinary traces can be seen. The time
When the ionosphere changes with time and/or space with
moving terminal(s), the relative phases of the component
echoes change and the resultant signal uctuates or fades.
Fading may result from interference between (1) echoes reected from different parts of the ionosphere; (2) ordinary
and extraordinary waves, called polarization fading; (3) reection or penetration, called MUF fading or skip fading;
(4) absorption fading; and (5) focusing. The fading rate depends largely on the type of fading and may range from
100/s (utter fading) to once per day (MUF fading). The
speed of interference fading is related to the width of the
fading power spectrum. The autocorrelation falls to 0.37
after a time, t, called the fading time, given by
where is the signal wavelength and v is a measure of the
velocity of the reecting sources.
For practical purposes, it is customary to make fading
allowances to ensure that the eld strength is exceeded
a certain percentage (e.g., 90%) of the time. Circuit planning requires comparison of the strengths of the wanted
signal relative to (1) natural and synthetic noise, and (2)
cochannel and adjacent channel interference (called electromagnetic compatibility). Hence, in addition to monthly
medians, fading allowances are necessary that depend on
short-term (<1 h) and long-term (day-to-day), fading of the
wanted signal and the background interference. Some proposed fading allowances are discussed in ITU (20), Recommendations 339 and 411.
In high magnetic latitudes, rapid ionospheric motions
(1 kms1 ) result in Doppler frequency shifts and, hence,
frequency spreading on great-circle paths of 0.1 Hz to 0.5
Hz. On nongreat-circle paths, spreading of 5 Hz to 10 Hz
is typical. Some typical time-delay spreads are 100 ms to
200 ms. Sky wave fading models are available (e.g. see (21).
Further, in high latitudes, vertical antennas (which generate vertical electric elds) are preferred to horizontal antennas because they excite ordinary waves, which suffer
less absorption than extraordinary waves.
HF Propagation Programs
Because the ionosphere varies in space and time (local
time, season, sunspot numbers, etc.) so does the usable frequency spectrum and, also, the characteristics of the received signals (e.g., signal strength, fading). Hence, knowledge of the variability of the ionosphere has been invaluable in the design and operation of point-to-point and
Sky Wave Propagation at Medium and High Frequencies
Figure 5. Oblique ionogram for the 2370 km, geographic approximately westeast path from
Boulder, Colorado, to Sterling, Virginia, September 1, 1954, 2112 (90 West Meridian Time). Here the
junction frequency (FJF) or basic MUF is the same as the maximum observed frequency (FMOF).
LOF is the lowest observed frequency.
broadcasting HF systems. For continuous operation of a
given point-to-point circuit a set of frequencies is required
to avoid MUF failure (e.g., at night) and excessive ionospheric absorption by day. For long-term planning the fo F2
is the most important single parameter in controlling the
MUF, and it is highly variable. In spite of the increase in
the number of solar disturbances, HF sky wave propagation is better near sunspot maximum than near sunspot
minimum because of the broader available frequency band.
When the fo F2 and the layer height [or M(3000)F2] are
known, from vertical soundings, the basic MUF can be calculated by graphical methods or by parabolic theory. The
behaviors of the E and F1 layers are regular and are expressed by analytic formulas (see Ref. 1, Sec. 5.2) and are
such that the smaller F1 M factor essentially compensates
for the higher fo F1 so that the products, the MUFs, for both
layers are nearly equal. During daytime the E layer is usually the controlling layer for distances up to about 2000
km and the F1 layer for distances between 2000 km and
3000 km. Over distances of 4000 km and longer, the F2 is
normally the controlling layer. Single-hop propagation is
limited to paths shorter than about 4000 km. Empirically,
it has been found that as the distance increases beyond
the limit, sky wave propagation is maintained by such
mechanisms as ionospheric and ground scatter, high-angle
rays that have longer one-hop limits, and ionospheric tilts
that produce super-modes such as those that occur on
transequatorial circuits. Propagation fails only when the
ionosphere fails to support propagation at one of two control points on the great-circle path at 2000 km from each
end (see Ref. 1, Sec. 12.2). The path MUF is the lower of
the two MUFs for 4000 km with the ionospheric parameters at the control points. In the cases of the E, Es, and F1
layers for paths longer 2000 km, control points 1000 km
from each end are used and the MUFs for a range of 2000
km are calculated (the lower of the two MUFs is taken
as the circuit MUF). Global numerical maps of fo F2 are
available at reference high- and low-sunspot numbers (or
equivalent indexes), and the values for other sunspot numbers are obtained by interpolation or extrapolation. There
is a saturation of fo F2 at sunspot numbers 150 so that
for sunspot numbers greater than 150 the value is set at
150. The monthly median fo F2 correlates better with the
12-month smoothed sunspot number than with the individual monthly number. For prediction purposes, the zero
distance MUF (= fo F2 + 0.5fH ) together with the MUF at a
reference distance (e.g., 3000 km or 4000 km) and, for paths
with other lengths, the MUFs are found by interpolation.
The M(3000) is related to the virtual height, in km, by
The optimum working frequency (FOT) is 0.85 (monthly
median MUF) and it ensures sky wave propagation 90% of
the days of the month at a particular hour (usually local or
universal time). Operating on the monthly median MUF
results in skywave reliability of 50%. The highest probable
frequency, or HPF, is exceeded 10% of the time. The higher
the signal frequency the greater is the signal strength, but
this has to be balanced against the spectrum congestion
caused by many operators using similar circuits. An advantage of using a frequency close to the MUF is that the
time dispersion is a minimum.
In planning a circuit it is necessary to determine (1)
the maximum MUF for the various layers and, hence, the
FOT; (2) the radiation angle for the appropriate layer; (3)
the power delivered to a receiver (using spatial spreading,
ionospheric absorption, fading and polarization losses, antenna gains); and (4) the noise and interference (see Radio
noise; Radio system performance).
Before about 1970, individual circuit evaluations were
necessary using laborious methods, and graphical techniques were customary for the determination of skywave
propagation parameters such as those discussed previously. Today these laborious methods are replaced by convenient user-friendly computer programs, many written
for personal computers, that only require basic circuit information, such as time of day, month, sunspot number
(or equivalent), and path terminals (or area coverage for
broadcasting). The computer programs normally include
numerical maps, or formulas, for calculating the necessary
Sky Wave Propagation at Medium and High Frequencies
ionospheric characteristics and noise and interference; the
more comprehensive models include reference antennas
for calculation of received signal-to-noise ratios. The sophistication of these PC programs ranges from basic frequency outputs to broad coverage. A list of some commercially available programs is given in Goodman (Ref. 2,Table
5.16; see also Ref. 22, Resolution ITU-R25). An example of
a prediction program is the Ionospheric Communications
Enhanced Prole Analysis and Circuit (ICEPAC), which includes, in addition to a global ionospheric model, models of
the subauroral trough, auroral zones, and polar caps. This
software is available (free) on the Internet (23). A sample
output from ICEPAC is shown in Fig. 6, which includes contours of signal strength around a broadcasting transmitter
that has an isotropic antenna.
Besides the basic sky wave characteristics, such as MUF,
FOT, and LUF, the more comprehensive of these computer
programs give signal-to-noise ratio for a given circuit performance (see Radio noise) and the following characteristics that depend on the ionosphere (see Ref. 2, Sec. 5.11,
and references therein). Mode availability, Q (0 < Q < 1), is
the fraction of time that a path is available (see Ref. 1, Sec.
12.6.2). To a rst approximation the standard deviation of
fo F2 about the monthly average is about 15%. On the median MUF, Q = 0.5; on the FOT, Q = 0.9; and on the HPF,
Q = 0.1.
Circuit reliability, , (0 < p < 1) is the fraction of days
that successful communication may be expected at a given
hour of a month on a specied operating frequency
where P is the probability that the mean received signalto-noise ratio exceeds a specied level. The reliability, ,
depends on P increasing with increasing frequency, while
Q decreases with increasing frequency.
Grade of service denes the quality of communication
desired (for example, the percentage of error-free messages
in teletype transmitted or the percentage of satised customers of a given service). Service probability is the fraction of time that a specied grade of service (e.g., signal to
noise) or better is achieved.
Above-the-MUF loss, Lm , accounts for the fact that signals may be received on frequencies above the basic MUF
(e.g., resulting from ionosphere scatter, Es , etc.). The CCIR
(14), Report 252-2, recommends
To account for a wide variety of ionospheric effects, such as
Es, spread F, off-great-circle propagation, focusing, day-today variability, and aurora, ITU (20) Report 252 contains
information on additional system loss resulting from these
extra effects.
10
Sky Wave Propagation at Medium and High Frequencies
Compatibility is dened as the percentage of time during which a specied criterion of service quality is achieved
at a receiver in the presence of interference, relative to the
value that would be obtained if only noise were present.
Digital System Considerations
Digital systems are particularly affected by ionospheric
dispersion, which produces fading (e.g., by multipath, polarization, etc.). The performance of a digital system is
characterized by its bit error rate, which is the probability that a transmitted binary digit is wrongly detected by
the receiver. On HF, intervals with high error densities alternate with intervals of low error densities. Error bursts
occur when the signal-to-noise ratio temporarily falls below
a critical level (e.g., selective fading). Curves are available
in the ITU (20) Report 197, that gives the duration and
probability of fades as a function of the signal level for specic circuits. In the absence of fading the signal-to-noise
ratio (SNR) is the controlling factor for digital transmission speed. The theoretical error-free channel capacity, C,
is
where B is the channel bandwidth (Hz). Because of all the
effects described above these ionospheric channels usually
operate far below the theoretical capacity (Eq. 14).
Real-time Channel Evaluation
While the prediction systems discussed previously are
valuable for circuit planning and frequency allocation, they
have limited value for operational purposes, primarily because of the high hour-to-hour and day-to-day variability
of the F2 layer. To meet operational requirements, several
real-time circuit evaluation (RTCE) systems have been developed (see Ref. 14, Report 889-1). The rst such system
involved oblique-sweep-frequency sounding over the operational circuit and, from the oblique ionogram, selection
of a suitable frequency with low time dispersion, adequate
signal, and minimum noise and interference. The channel
selection is normally limited by the channel allocation. A
critical feature of this technique is the interval between
soundings. When the interval is long, changing ionospheric
conditions can catch the operator unaware. On the other
hand, too frequent soundings generate excessive interference to other services. An alternative, and less bothersome,
approach is channel sounding on the allocated channels
only. This technique has the following advantages: simpler
equipment, lower installation costs, and less interference.
When the optimum channel is identied, a message is dispatched to the sender, who can commence transmission.
This method assumes reciprocity (that is, the sender-toreceiver conditions are the same as the receiver-to-sender
conditions). This is usually a good assumption. Adaptive
systems are essential for RTCE in order to respond to
rapidly changing sky wave conditions. An adaptive system
needs (1) a fast frequency response, (2) antenna agility, and
(3) the ability to adjust rates of information.
Use of RTCE allows adaptive frequency management by
which usage of the HF spectrum can be maximized. Such
an approach is preferable to increasing the transmitter
power and/or transmitting simultaneously on all allocated
frequencies, both of which may be self-defeating. The term
channel estimation is used to describe the process of monitoring channel characteristics with the aim of describing
the states of a set of channels. Real-time adaptive systems
are of particular value when one of the terminals is mobile
(e.g., ship, airplane, vehicle) and the other is xed so that
the xed terminal can avail itself of high power and directive antennas. One advantage of RTCE over prediction is
to maximize above-the-median MUF propagation, in which
interference is usually lower than on the FOT.
COMMUNICATIONS DISRUPTIONS
As mentioned under Spectrum Considerations, medium
frequencies and high frequencies can be adversely affected
by solar events. Such solar events are particularly disruptive on high frequencies. The main solar phenomena that
cause communication outages are as follows: x-ray bursts
that result in sudden ionospheric disturbances SIDs, coronal (charged- particle) emissions that produce ionospheric
storms, and energetic solar protons that cause polar cap absorptions (PCAs). While x-ray bursts last for a few minutes
(less than 1 h), storms and PCAs can endure for several (3
or 4) days. To help minimize these communications outages, forecasting centers are operated in several countries
around the world. These centers observe the suns surface
over a broad range of frequencies from x-rays through radio noise. The data are evaluated by skilled obsevers who
assess the likelihood of a geo-effective result. The leading forecasting service is at the Space Weather Prediction
Center (SWPC) of the National Oceanic and Atmospheric
Administration (NOAA) located in Boulder, Colorado. The
SWPC provides real-time monitoring and forecasting of
solar and geophysical disturbances and it is the United
States and world warning center for a number of disturbances that affect human affairs. The job of the SWPC is
to provide timely warnings of the state of the earths environment and alert users to prevailing conditions, this is
called situation awareness.
Among the numerous services, of importance to HF and
MF communicators, porvided by SEC are:
A.
B.
C.
D.
Global maps of SID absorption
Maps of the geographical extent of PCAs
Maps of total energy deposition in the auroral zones
Forecasts of the state of the geomagnetic eld and
associated atmospheric disturbances
E. Educate users of systems affected by the environment.
Current Space-Weather conditions, including radio
blackouts, can be obtained on the Internet (24).
CONCLUSIONS
The medium-frequency and high-frequency bands have
great commercial, social, and scientic value and are used
extensively for broadcasting and for point-to-point com-
Sky Wave Propagation at Medium and High Frequencies
munications. The pronounced (orders of magnitude) variability of the ionosphere makes sky wave communications
problematic, both as a means of communicating and as a
source of interference. Using HF systems, frequency agility
is essential for continuous operation and, therefore, a set
of suitable assigned frequency channels. Even with such a
frequency allocation, by the appropriate national authority, sky wave propagation is liable to disruption by several solar disturbances (see Propagation of broadcast
transmissions) and synthetic disturbances (e.g., nuclear
explosions and ionospheric modication). For many purposes satellite communications, on gigaHertz frequencies,
have replaced HF for global communications. However, sky
waves will continue to be used well into the foreseeable future.
BIBLIOGRAPHY
1. K. Davies, Ionospheric Radio, London: P. Peregrinus on behalf
of the Inst. Elec. Eng., 1990.
2. J. M. Goodman, HF Communication, Science and Technology,
New York, Van Nostrand Reinhold, 1991.
3. R. D. Hunsucker, Radio Techniques for Probing the Terrestrial
Ionosphere, Berlin: Springer-Verlag, 1991.
4. L. F. McNamara, The Ionosphere: Communications, Surveillance, and Direction Finding, Malibar, FL: Krieger, 1991.
5. D. Bilitza (ed.), International Reference Ionosphere 1990, National Space Science Center, NSSDC 90-20, Greenbelt, MD,
1990.
6. R. M. Jones, J. J. Stephenson, A versatile three-dimensional ray
tracing program for radio waves in the ionosphere, OT Report
75-76, Washington, DC: Government Printing Ofce, 1975.
7. Ionospheric Physics Group, [Online], 1998. Available
http://www.ngdc.noaa.gov/stp/IONO/ionohome.html
8. T. A. Croft, HF radio focussing caused by the electron distribution between ionospheric layers, J. Geophys. Res., 72,
23432355, 1967.
9. National Space Science Data Center, [Online], 1998. Available
http://nssdc.gsfc.nasa.gov/space/model/models/iri.html
10. D. N. Anderson, The development of global, semi-empirical
ionospheric specication models, in J. M. Goodman (ed.), Proc.
Ionospheric Effects Symp., TCI/BR Communications, Alexandria, VA 22314, 1993, pp. 353363.
11. J. C. H. Wang, LF/MF skywave propagation at daytime, IEEE
Trans. Broadcast., 41: 2327, 1995.
12. J. R. Herman, X. A. DeAngelis, Bandwidth expansion effects
on the voltage deviation parameter (Vd) of MF and HF atmospheric radio noise, Radio Sci., 22: 2636, 1987.
13. CCIR, Sky-wave eld-strength prediction method for the
broadcasting service in the frequency range of 150 to 1600
kHz, Recommendation 435-7, 1943, in 1992 CCIR Recommendations, Geneva, Switzerland: Int. Telecommun. Union,
1992.
14. CCIR, Recommendations and Reports of the CCIR 1986,
VI, Propagation in Ionized Media, Geneva, Switzerland: Int.
Telecommun. Union, Report 431-4, 374390, 1986.
15. J. LaBelle et al., The spectrum of LF/MF/HF radio noise
at ground level during substorms, Geophys. Res. Lett., 21:
27492752, 1994.
16. R. D. Hunsucker, B. S. DeLana, High-latitude eld-strength
measurements of standard broadcast skywave transmission
17.
18.
19.
20.
21.
22.
23.
24.
11
monitored at Fairbanks, Alaska, Geophysical Inst., Univ.
Alaska, Fairbanks, AK, 1986.
W. Ebert, Ionospheric propagation on long and medium waves,
Tech. Doc. 3081, European Broadcasting Union, Brussels, Belgium, 1962.
M. PoKempner, Comparison of available methods for predicting medium frequency sky-wave eld strengths, NTIA Report
80-42, Boulder, CO: Nat. Telecommun. Inf. Admin., 1980.
R. D. Hunsucker, B.S. DeLana, J. C. H. Wang, Medium frequency sky-wave propagation at high latitudes: Results of an
FCC sponsored study, Proc. 42nd Broadcast Eng. Meeting,
Nat. Assoc. Broadcasters, 1988.
ITU, Fading allowances in HF broadcasting, Vol. XPart 1,
Broadcasting service (sound), XVII Plenary Assembly, Geneva,
Switzerland: Int. Telecommun. Union, 1990.
C. C. Watterson, J. R. Juroshek, W. D. Bensema, Experimental
conrmation of an HF channel model, IEEE Trans. Com. Tech.,
COM-18 (6): 792803, 1970.
ITU, Computer programs for the prediction of ionospheric
characteristics, sky-wave transmission loss, and noise, III,
Propagation, XVIII Plenary Assembly, Geneva, Switzerland:
Int. Telecommun. Union, 1994.
http://elbert.its.bldrdoc.gov/hf.html
http://www.swpc.noaa.gov
KENNETH DAVIES
National Oceanic and
Atmospheric Administration,
Boulder, CO
J. Webster (ed.), Wiley Encyclopedia of Electrical and Electronics Engineering
c 1999 John Wiley & Sons, Inc.
Copyright
PROPAGATION AT EXTREMELY LOW FREQUENCIES
Extremely low frequency (ELF) electromagnetic waves (frequencies in the range 3 Hz to 3 kHz) are of interest
in the sciences and in communications for a number of reasons. Perhaps most important, lightning is a powerful
natural source of electromagnetic radiation in this frequency range, and the characteristics of lightning can
be studied in considerable detail by making measurements on the waves it generates. In a communications
context, the naturally occurring lightning-generated waves constitute a noise background that can degrade an
ELF communications link. On the other hand, ELF waves travel with little attenuation to large distances in the
space between the earths surface and the ionized upper layers of the earths atmosphere (i.e., the ionosphere),
and thus they can provide communications over large regions of the earth. Furthermore, compared with
the other, higher-frequency forms of electromagnetic waves used for communications, ELF waves are largely
unaffected by disturbances in the ionosphere, and thus they can provide a more reliable communications link
during times of high solar activity. Finally, they penetrate quite well through the conducting materials typically
encountered on the earths surface, and thus they can be used for probing the surface, for prospecting, and for
certain forms of communication through the earth and the sea.
In this article, we provide a brief description of the propagation characteristics of these waves based on
a simplified model of ELF electromagnetic wave propagation in the earthionosphere waveguide. We discuss
some interesting consequences of the long-distance propagation of ELF waves through the waveguide, one of
which is the focusing of the waves at the antipode of the source. Another is the occurrence of resonance in the
waveguide. We also discuss some practical applications of ELF waves and the special considerations that can
affect communication systems.
In order to better understand ELF wave propagation in the earthionosphere waveguide, the electrical
properties of the waveguide walls need to be taken into account, as well as the effect of dynamic changes in
the ionosphere. Another factor affecting the waveguide is the geomagnetic field, which causes the waveguide
to be anisotropic. Finally, because the complexity of the waveguide makes its analytical description difficult,
we discuss the two most common methods for calculating the solution to particular ELF wave propagation
problems.
Simplied Waveguide Model
A simple approximation of the earthionosphere waveguide, which is nonetheless useful in many situations, is
the flat earth model. The waveguide is modeled as an infinite parallel-plate waveguide with the curvature of
the earth and the ionosphere neglected. Figure 1 presents the geometry with a cylindrical coordinate system
(, , z). The ground plane is located at z = 0, and the ionospheric boundary is at z = h.
This model is valid only for distances of up to half an earth radius from the source, because at greater
distances the curvature of the earth affects ELF propagation.
The walls of the waveguide are modeled as perfect conductors, because in the ELF range both the earth
and the ionosphere have very high conductivities.
1
PROPAGATION AT EXTREMELY LOW FREQUENCIES
Fig. 1. Geometry of the flat-earthionosphere waveguide with a cylindrical coordinate system.
The vertical component of the electric field generated by a vertical electric dipole of moment Ids, placed
on the ground at the point ( = 0, z = 0) in a flat waveguide with perfectly conducting walls, is described by
Wait (1) as a modal series:
where
= free-space wavelength
k = 2/
120 is the free-space impedance
=
E0 = i(Ids/) e jk /
= Sn sin n
sine of the eigenangle of the nth mode (which may be complex)
Cn = cos n cosine of the eigenangle of the nth mode
n = eigenangle of the nth mode.
Equation (1) describes the field measured within the waveguide, for 0 < z < h. It is a far-field approximation, valid at big enough distances from the source (h and ), where > 300 km can be taken as a
lower limit. See Bannister (2) for a development of field expressions at shorter distances from the source. The
time dependence is ejt , where = 2c/ is the angular frequency.
The other field components, in addition to Ez [Eq. (1)], can be calculated using Maxwells equations, and
they are described by
The modal phase velocity is c/Re[Sn ], and the modal attenuation factor is k Im [Sn ] per unit distance.
PROPAGATION AT EXTREMELY LOW FREQUENCIES
The eigenangle n can be interpreted as the incidence angle of the wavefront on the ionosphere, as
indicated in Fig. 1. This interpretation of the eigenangle is closely related to the ray-theory description of
electromagnetic wave propagation in the waveguide, which is an alternative to the modal theory discussed
here. However, the ray-theory treatment of the waveguide is of little consequence in the ELF band, so it is not
discussed further.
The modes generated by the vertical dipole are all transverse magnetic (TM) modes, since the magnetic
field is transverse to the direction of propagation of the wave as it propagates away from the source (and thus
H = 0).
Nickolaenko and Hayakawa (3) calculate the spectrum in the ELF band of the vertical electric field
generated by a vertical dipole, from expressions based on the modal series, and demonstrate the Schumann
resonance, a phenomenon that will shortly be described.
A horizontal electric dipole source of moment Ids, placed at (z = z0 , = 0) parallel to the = 0 direction,
generates a magnetic field whose vertical component is
where
Sm = sin m
= sine of the eigenangle of the mth mode.
Cm = cos
m = cosine of the eigenangle of the mth mode
m = eigenangle of the mth mode
Equation (4) is an approximation valid for 0 z < h, h , and , where > 300 km can be taken
as a lower limit for its validity. The time dependence is ejt . The other components of the field are zero, except
for E and Er .
The modes generated by the horizontal dipole in the broadside direction are transverse electric (TE)
modes, since the electric field is transverse to the direction of propagation of the wave (and thus E = 0). In
other directions, the horizontal dipole generates a mixture of TM and TE modes. The TE waves are weak at
low altitudes (near the ground), because the horizontal electric field is extinguished near the conductive earth.
Since most measurements of ELF fields are made by antennas mounted on the ground, the TE component they
receive is weak. This is possibly the reason that the TE modes receive little attention in the scientific literature.
The modal equations (5) and (6) of the simple flat earth model with perfectly conducting walls determine
the eigenangles of the waveguide:
for the TM and TE modes respectively.
The eigenangles for the TM mode are
PROPAGATION AT EXTREMELY LOW FREQUENCIES
and for the TE modes are
The lowest TM mode is the 0th, which propagates over the whole frequency range. This mode is also called
the transverse electromagnetic (TEM) mode, because both the electric and the magnetic fields are transverse to
the direction of propagation. The electric field is vertical (Ez component only), and the magnetic field is in the
direction.
The higher modes, namely the TM modes with n = 1, 2,. . . and all the TE modes, propagate above a certain
frequency, called the cutoff frequency, which characterizes each mode. The cutoff frequency of high-order modes
is higher than that of the lower-order modes. Only the first mode is of interest in the ELF range, because the
cutoff frequencies for all the higher modes are above 3 kHz. The cutoff frequency of these first modes (TM and
TE) varies between 1.5 kHz and 4 kHz, depending on the height of the ionosphere.
When extending the waveguide model from the simple flat, perfectly conducting plate model, the following
features may be considered:
The reflections coefficients of the earth and the ionosphere
The waveguide geometry
The presence of the geomagnetic static field
The effects of these features of the waveguide on the ELF waves propagating within it are discussed in
the section Extensions of the Waveguide Model below.
Applications
ELF waves propagating in the earthionosphere waveguide undergo very low attenuation (sometimes as
low as 1 dB/Mm to 2 dB/Mm). This allows ELF transmissions to propagate great distances, making this
frequency range suitable for communication networks encompassing the whole globe. Bannister (4) discusses
the attenuation in the ELF band and shows measurements of the attenuation rate at different frequencies.
Another important feature of ELF waves is their good penetration into the earth and sea water. The
skin depth, which characterizes penetration of electromagnetic waves into lossy materials, varies inversely as
the square root of the wave frequency ( 1/ ). Thus, ELF waves have a larger skin depth than waves in
higher frequency bands, and they penetrate better into conducting materials. For example, the skin depth of
waves propagating at 70 Hz in sea water is 30.1 m. The large skin depth is the reason for operating submarine
communication systems in the ELF range. An example of such a system is the United States Navys ELF
Communications System for communication with submarines, operating in the frequency range of 70 Hz to 80
Hz. This system consists of two synchronous transmitters connected to 22.5 km quasi orthogonal antennas.
The major difficulties when communicating in the ELF range are the need for large antennas and the
low data rate. In order to achieve efficient transmission, the physical dimensions of the antennas should be
comparable to the wavelength of the communication frequency, so the long wavelengths inherent in the ELF
range (100 km at 3 kHz up to 100,000 km at 3 Hz) make efficient antenna installation very difficult. This
is a problem mainly in the transmitters, since on the receiving side the antenna preamplifier combination is
limited only by atmospheric noise in the frequency range of interest.
PROPAGATION AT EXTREMELY LOW FREQUENCIES
The useful bandwidth in an ELF communication system is limited by the low carrier frequency. Communication systems in the ELF range typically transmit at very low data rates compared to systems utilizing
other frequency bands.
The large antenna dimensions needed for efficient transmission and the inherent low bandwidth are the
reasons that ELF communication systems are few and are usually run by governments for limited military
usage.
Kelly (5) surveys engineering issues related to ELF systems, namely modulation, dispersion, antenna and
system performance, and coverage predictions.
Electrojet Modulation. One scientific application in the ELF band is the modulation of ionospheric
currents, in particular the auroral electrojet. The auroral electrojet is an intense westward current, flowing
in the D and E regions of the ionosphere (the section Ionospheric Reflection below discusses the ionospheric
layers). It flows within the auroral belt, a narrow oval-shaped region encircling the north geomagnetic pole (a
similar belt exists around the south pole).
The intensity of the auroral electrojet is affected by the conductivity of the ionospheric D and E layers.
Changes in the ionospheric conductivity can be induced artificially by heating the ionosphere with radio waves
of appropriate frequency in the HF band, which are efficiently absorbed in the D and E layers. Amplitudemodulated heating of the D and E layers translates into amplitude-modulated conductivity of the medium,
which modulates the intensity of the electrojet. The modulated electrojet reradiates into the earthionosphere
et al. (6) study different modulation methods and polarizawaveguide at the modulation frequency. Villasenor
tions of the heating wave, and Barr and Stubbe (7) discuss the radiation mechanism and evaluate the intensity
of the radiated wave. Papadopoulos et al. (8) discuss the efficiency of the ELF/VLF generation process.
Generation of ELF/VLF signals by modulation of the auroral electrojet has been demonstrated in various
ionospheric heating facilities. Experiments of this kind are discussed by Stubbe et al. (9), Ferraro et al. (10),
Barr et al. (11), Rietveld at el. (12), and Barr et al. (13).
Background Noise
The sources of noise in the ELF range are lightning discharges, man-made noise, and magnetospheric phenomena (such as chorus and hiss); among these, lightning discharges and man-made noise dominate. The low attenuation in this band allows the signals generated by lightning strokes, called atmospherics, or sferics, to travel
great distances around the earth. Thus, a lightning discharge above Africa can be easily detected by a receiver
in North America. It is estimated that on average there are roughly 2000 thunderstorms active around the
earth at any moment, generating an average of 100 to 200 lightning strokes per second. This lightning activity
constitutes the major source of noise in the ELF range. Power-line harmonics are a major source of man-made
noise in the ELF, and their effect is evident throughout the whole frequency band in inhabited areas.
The background power spectrum in the ELF range is generally of the form 1/f [Galejs (14), Chap.
7], where is typically close to 2. Maxwell (15) and Maxwell and Stone (16) show measurements of the
background noise level at various location, and in (15) they discuss the dependence of the background noise
spectrum on latitude. It is shown that the noise spectra correlate well with lightning spectra and propagation
attenuation coefficients. The background noise varies with a period of 24 h because of diurnal variations of
global thunderstorm activity, and the seasonal variation of the noise is related to global thunderstorm activity.
Gustafsson et al. (17) and Smith and Jenkins (18) present background-level measurements recorded in northern
latitudes, and demonstrate that the daytime background level is generally lower than the nighttime level. They
suggest that the reason is enhanced ionospheric absorption during the day.
PROPAGATION AT EXTREMELY LOW FREQUENCIES
Fig. 2. Spectrum of the first eight Schumann resonance modes. The basic resonance mode occurs when the wavelength
approximately equals the circumference of the earth.
The Schumann Resonance
An interesting phenomenon in the lower end of the ELF band is the Schumann resonance, named after W. O.
Schumann, who predicted it in 1952. The basic resonance mode (frequency 7.5 Hz) occurs when the wavelength
of the propagating wave approximately equals the circumference of the earth (40,000 km). Constructive interference at the basic Schumann resonance and its harmonics results in an increased background noise level at
these frequencies (Fig. 2).
Derivations of the Schumann resonance from general field expressions are presented by Wait (19) and
Jones (20), and extended by Large and Wait (21) to include the effects of the daynight inhomogeneity of the
ionosphere. The Schumann resonance was first observed in 1960 by Balser and Wagner (22).
It is thought that the main source of energy for the Schumann resonance is lightning strokes occurring
around the earth, and that, as a result, the resonance level is an indication of global lightning activity [Balser
and Wagner (23), Nickolaenko et al. (24,25), Ogawa et al. (26), and Heckman et al. (27)]. The Schumann
resonance is used to locate thunderstorms [Polk (28) and Galejs (29)] and to study the ionosphere [Sentman
and Fraser (30)]. Sao et al. (31) relate diurnal variations of the Schumann resonance frequency to solar ionizing
radiation, solar zenith angle, and geomagnetic disturbances.
The interest in the Schumann resonance was revived in the 1990s, when Williams (32) showed a correlation between the intensity of the first mode of the Schumann resonance and the fluctuations of the temperature
in the tropical atmosphere. This suggests that the Schumann resonance can be used to monitor global climate
changes. A recent summary on the Schumann resonance with many references is found in Nickolaenko (33).
Extension of The Waveguide Model
The modal theory of the earthionosphere waveguide is extended in this section from the simplified parallelplate model with perfectly conducting walls, which was presented in the section Simplified Waveguide Model.
PROPAGATION AT EXTREMELY LOW FREQUENCIES
The extensions of the waveguide model presented here are:
Consideration of the electrical properties of the earth and ionosphere and calculation of the reflection
coefficients of the waveguide walls
Consideration of the spherical geometry of the earth, which imposes a spherical waveguide instead of a
planar one
Consideration of the geomagnetic field, which causes anisotropic ionospheric reflections.
Reection Coefcients. An accurate model of the earthionosphere waveguide for ELF wave propagation has to take into account the finite conductivities of the two limiting boundaries. The conductivity of
the earth ranges from 10 4 S/m for dry rocky areas to 4 S/m for sea water, and the dielectric constant varies
considerably as well. A list of the conductivity and the dielectric constant values of various types of soil and
water is included in Ref. 34. The earth is usually treated as a uniform sharp boundary, even though this model
does not strictly apply to its changeable surface. The validity of the smooth earth model stems from the small
scale of change of the surface compared to the long wavelengths involved in ELF propagation.
A further extension of the waveguide should take into account the inhomogeneity of the earths surface
conductivity. This extension of the waveguide theory is similar in nature to the theory related to ionospheric
inhomogeneities discussed in the section The Inhomogeneous Ionosphere below.
A realistic model of the ionosphere is more involved, since it depends on the concentration of free electrons
and various positive ions, which in turn depend on complex generation mechanisms. The section Ionospheric
Reflection below discusses the calculation of ionospheric reflection coefficients in more detail.
Another complication of the calculation of the ionospheric reflection coefficients stems from the fact that
the ionospheric boundary is not sharp. In many cases, an effective height is calculated for the waveguide model,
and reflection is treated as though the ionosphere were sharply bounded at this height. Booker and Lefeuvre
(35) calculate the effective ionospheric height for a simple profile of ionospheric ionization density, and show
that it is frequency-dependent. Greifinger and Greifinger (36) present a more complicated reflection model,
which includes two ionospheric reflection layers at different heights.
The reflection coefficients of the earth and the ionosphere affect the eigenangles of the different modes.
These in turn determine the attenuation that each mode undergoes and the relative amplitudes of the modes.
The relation between the eigenangles of the different modes and the attenuation can be appreciated by noting
that if Sn is complex in Equations (1), (2), and (4), then its imaginary part translates into attenuation through
the exponential factor e jkSn . The effect of the reflection coefficients on the eigenangle can be appreciated by
looking at the modal equation, shown here for the TM modes:
where Ri () is the ionospheric reflection coefficient for TM waves, Re () is the earth reflection coefficient for TM
waves, and h() is the frequency-dependent effective reflection height. Both reflection coefficients depend on
the frequency and on the eigenangle n .
Waveguide Geometry. The waveguide geometry affects the various modes [Eqs. (1) and (4)]. It has a
minor effect on the modal equation (9), so the eigenangles are not strongly affected by the flat-earth simplification.
When considering a spherical geometry rather than a parallel-plate geometry, mode coupling is introduced.
This means that TM waves propagating in the waveguide are partially converted to TE waves because of
reflections from the curved walls of the waveguide (and likewise TE waves are converted to TM).
Expressions for the fields in a spherical waveguide are given by Wait (1), using a spherical coordinate
system (r, , ). The earth is modeled as a homogeneous sphere of radius a (= 6370 km), conductivity, g , and
PROPAGATION AT EXTREMELY LOW FREQUENCIES
Fig. 3. Geometry of the spherical earthionosphere waveguide with a spherical coordinate system.
dielectric constant g . The lower edge of the ionosphere is located at r = a + h, its conductivity is i , and its
dielectric constant is i . It is assumed that both the earth and the ionosphere are good but not necessarily
perfect conductors. The source is a vertical electric dipole located at ground level at r = a, = 0, = 0, and the
observation point is located on the ground at r = a, , = 0 as in Fig. 3.
The vertical electric field in this configuration is
where
= free-space wavelength
d = a = arc length between the source and the observer
=
120
= free-space impedance
E0 = the field of the source at a distance d on a flat perfectly conducting earth For d/ 1
where
PROPAGATION AT EXTREMELY LOW FREQUENCIES
Sn = sin n
= sine of the eigenangle of the nth mode (which may be complex)
Cn = cos n cosine of the eigenangle of the nth mode
n = eigenangle of the nth mode.
Equation (10) describes the field measured within the waveguide, for a < r < a + h. This expression is an
approximation that is valid only if is not near 0 or , which means that the source region and the antipode
are excluded. The time dependence is ejt , where = 2c/ is the angular frequency.
As the radius of the earth a tends to infinity, the flat earth approximation [Eq. (1)] is recovered.
A first-order approximation of the effect of the earth curvature on ELF fields is obtained by multiplying
the field expressions for the parallel-plate waveguide [Eqs. (1) and (4)] by the factor 1 + 2 /12. The coordinate
system used here is cylindrical, as in the section Simplified Waveguide Model above:
Using appropriate approximations, it can be shown that the field consists of two components. One of these
components travels along the short great-circle path connecting the source and the receiver, and the other
travels along the complementary long great-circle path, passing though the antipode [Wait (1)].
An interesting feature of the spherical waveguide is the antipodal concentration that causes the vertical
electric field signals measured at the antipodal point to be stronger than signals measured at a similar distance
away from the source in a plane parallel waveguide. This effect is caused by the addition of many coherent
signal components traveling along great-circle paths between the source and the antipodal point. [The actual
antipodal effect is more complex, due to the daynight terminator (see the subsection DayNight Terminator
Effect below) and other propagation effects in the various paths.] The antipodal concentration and other
antipodal effects are discussed further in the section Antipodal Effects below.
Geomagnetic Field. The static geomagnetic field affects the behavior of the ionospheric boundary of
the waveguide. In the presence of the geomagnetic field, the ionosphere becomes anisotropic and the ionospheric
reflection coefficient depends on the angle between the incident wavefront and the static magnetic field (see
Fig. 4 for the definition of ). This adds a new dependence to the ionospheric reflection coefficient, further
complicating the modal equation:
The direction and intensity of the geomagnetic field depend on the geomagnetic location and, as a result,
the description of the earthionosphere waveguide has to take this location into account.
Dinger et al. (37) show that background noise measured in Norway at frequencies near 3 kHz is not
isotropic. The eastwest component of the measured noise (noise propagating along an eastwest or westeast
path) is 1 dB to 1.5 dB lower than the northsouth component during the day, but increases around local sunset
to a level of about 4 dB above the northsouth component. The difference between noise levels measured in
the two components is explained by the anisotropy of the propagation loss, which is caused by the geomagnetic
field.
Nonreciprocity of ELF Propagation. The geomagnetic field introduces a directional dependence into
the attenuation of ELF waves. This is the reason why ELF waves propagating from west to east suffer higher
attenuation than those propagating from east to west in middle latitudes. This difference between the properties
of the propagation between eastwest and westeast paths is called the nonreciprocity of the propagation.
10
PROPAGATION AT EXTREMELY LOW FREQUENCIES
Fig. 4. Geometry of the earthionosphere waveguide with the geomagnetic field present. is the direction of the geomagnetic field relative to the horizontal. is the angle between the incident wave direction and the geomagnetic field.
This effect is analyzed by Barr (38), who shows that the nonreciprocity is most notable for frequencies
above 1 kHz. Barr (39) computes the attenuation factors for east-to-west and west-to-east propagation, and
finds a difference of up to 45 dB/Mm. These results are shown to be in good agreement with measurements.
Antipodal Effects
Some fields measured at the antipodal point of a source are enhanced due to the spherical shape of the
earthionosphere waveguide. The enhancement can be intuitively explained as the coherent addition of many
waves traveling from the source to the antipodal point along great-circle paths. Galejs (14) calculates antipodal
effects on the different field components generated by vertical and horizontal dipole sources. He shows that for a
vertical dipole source the only field component present at the antipode is the vertical component of the electric
field. For a horizontal dipole source the antipodal field contains horizontally directed electric and magnetic
fields, perpendicular to each other, with the electric field parallel to the dipole axis. These results are also
summarized in Ref. 40.
Wait (1) gives a brief description of antipodal effects for a vertical dipole source and shows that waves
traveling through the antipode acquire a /2 phase advance. Wait (41) and Jones (42) analyze the distortion
of ELF pulses after propagation through an antipode. The /2 phase advance of lightning-generated pulses
traveling around the earth is examined.
Measurements of antipodal effects are difficult because of the long distances involved, which dictate
the use of very strong sources. Measurements made on an 82 Hz signal are described in Fraser-Smith and
Bannister (43). The transmitting facility was located in northwestern Russia, and measurements showing
antipodal concentration were made in New Zealand and Antarctica. Similar results, in the VLF band, are
reported by Rogerson (44). In this latter case the source was a U.S. Navy transmitter in Hawaii, and the
measurements were taken by an aircraft in the vicinity of the antipodal point located in southern Africa.
Ionospheric Reection
This section describes the mechanism of reflection of ELF waves from the ionosphere. The ionosphere is
conductive because it contains free electrons and positive ions, which are generated largely by solar ultraviolet
PROPAGATION AT EXTREMELY LOW FREQUENCIES
11
Fig. 5. Typical electron densities at 44.6 N, 2.2 E at noon and midnight on March 21, for solar minimum and maximum
conditions. The electron density determines the behavior of the ionospheric wall of the earthionosphere waveguide.
Adapted from A. D. Richmond, in H. Volland (ed.), Handbook of Atmospheric Electrodynamics, CRC Press, 1995, with
permission (CRC Press home page http://www.crcpress.com).
(UV) and X radiation and by the precipitation of energetic charged particles from the magnetosphere. A
discussion of the ionization generation mechanism and the various positive ions involved can be found in
Richmond (45) and in other literature on the subject [Rees (46), Kato (47), Akasofu and Chapman (48), Ratcliffe
(49,50), Rishbeth and Garriot (51), Whitten and Poppoff (52), and Rawer (53)].
The amount of free charge in the ionosphere is determined by the equilibrium that is reached between
the generation mechanisms and the recombination of the free electrons with the positive ions. The density
of neutral particles in the ionosphere decreases exponentially with increasing altitude, which means that
very little generation occurs at the high altitudes (above 500 km), where the particle density is low. The
energetic particles precipitating from the magnetosphere gradually lose their energy as they penetrate the
ionosphere, generating free charge as they travel downward. The solar radiation penetrating the ionosphere
is likewise gradually absorbed while generating free charge in the ionosphere. The bottom of the ionosphere
(around 50 km) is marked by very little generation of charge, because the ionizing radiation has lost most of its
energy at higher altitudes. The density of free charge in the ionosphere is maximal at middle heights, where
the generating mechanisms are strong and the density of particles is sufficient. The free-electron density is
maximal at an altitude of about 250 km above ground, with a few local maxima of density at lower altitudes.
These local maxima constitute ionospheric layers with enhanced electron density, as seen in Fig. 5.
The layer about the altitude of 100 km is called the E layer, the layer at 170 km, which is sometimes
absent at night, is the F1 layer, and the top layer at 250 km is the F2 or the F layer. The lowest layer, at about
70 km, is the D layer, which is almost always absent at night. A description of the structure and properties of
the ionosphere is presented by Hargreaves (54).
The electron density height profile in the ionosphere differs greatly between day and night conditions,
because of the different ionization generation rates driven by solar radiation. The electron density is also
affected by solar activity, with the ionization being lowest during periods of low solar activity, as seen in Fig. 5.
The ionosphere is a plasma, and its conductivity can be calculated using plasma theory, as described by
Rishbeth and Garriot (51), by Ratcliffe (49), and in other literature on the subject. The ionospheric plasma
parameters (i.e., the types of particles and their concentrations) are altitude-dependent, so the conductivity of
12
PROPAGATION AT EXTREMELY LOW FREQUENCIES
Fig. 6. Typical noontime parallel (0 ), Pedersen (1 ), and Hall (2 ) conductivities at 44.6 N, 2.2 E for solar minimum
conditions on March 21. The different types of conductivity characterize the behavior of the ionosphere as a reflector of ELF
waves. Adapted from A. D. Richmond, in H. Volland (ed.), Handbook of Atmospheric Electrodynamics, CRC Press, 1995,
with permission (CRC Press Home Page http://www.crcpress.com).
the ionosphere also varies with altitude. In addition, the conductivity varies in the lateral plane, with the most
notable variation occurring at the daynight terminator, which marks the dawn and dusk boundaries. Local
disturbances in the ionosphere also cause lateral variations of the conductivity; these variations are discussed
in the section The Inhomogeneous Ionosphere below.
Due to the geomagnetic field, the ionosphere is an anisotropic conductor. The three most commonly used
conductivities apply to specific directions relative to the geomagnetic field, as follows:
The Parallel conductivity 0 relates the current flowing along the static magnetic field to an electric field
component in the same direction. This conductivity is not affected by the presence of the static magnetic
field.
The Pederson conductivity 1 corresponds to current flowing perpendicular to the static magnetic field when
an electric field is applied in the same direction.
The Hall conductivity 2 corresponds to current flowing in a direction perpendicular to both the magnetic
field and the electric field, when the electric field is perpendicular to the magnetic field.
Figure 6 shows typical midlatitude vertical profiles of the daytime conductivity components for low
solar activity. At altitudes above 80 km the parallel conductivity is much larger than the Pederson and Hall
conductivities.
The reflection coefficients of the ionosphere depend on (a) the conductivity profile of the ionosphere, (b)
the local geomagnetic field, (c) the direction of incidence, and (d) the polarization of the incident wave. Due to
the complexity of the reflection coefficients, they are usually calculated numerically or empirically. Simplified
empirical expressions for the reflection coefficients in day and night conditions are given in Volland (55) for
a simplified model characterized by an exponential electron density and an exponential collision rate in the
lower layers of the ionosphere. The effects of the geomagnetic field are neglected. The reflection coefficient for
TM waves under these simplified conditions is
where bi and ci vary exponentially with frequency and is the incidence angle.
PROPAGATION AT EXTREMELY LOW FREQUENCIES
13
Harth (34) calculates the ionospheric reflection coefficient at 3 kHz using a thin-layer method (see the
final section for a discussion of the thin-layer method) with similar assumptions to Vollands, and the results
are presented graphically. Wait and Spies (56) present graphs describing the reflection coefficient for an
exponential electron density profile under various conditions. They show the effects of the magnetic field, angle
of incidence, frequency, reflecting height, and ground conductivity on the reflection coefficient. An experiment
relating electron density profiles to VLF reflection coefficients is described by Mambo et al. (57).
A different approach to the calculation of the reflection coefficients is taken by Budden (58), who presents
a plasma physics approach to understanding the ionosphere. The calculation of the refractive index and
the characteristic polarization of waves traveling through the ionosphere is the basis of this approach. The
admittance matrix of the ionosphere is calculated, which relates the horizontal components of the electric and
magnetic field, and the reflection coefficients of the ionosphere are then derived from the matrix expressions.
Two types of ionospheric boundaries are considered: a sharp boundary and a slowly varying one [Wentzel
KramersBrillouin (WKB) model]. A similar approach is taken by Galejs (14), where the modal equations are
presented in terms of the wave impedance rather than the reflection coefficient.
The ionosphere can usually be modeled as a sharp boundary for frequencies in the ELF range, since the
wavelengths are large compared to the scales of length over which the ionospheric properties change. As a
result, a simplified model of the ionosphere with a sharp bottom boundary is often sufficient.
The Inhomogeneous Ionosphere
A realistic picture of ELF propagation in the earthionosphere waveguide has to take into account the effects of
inhomogeneities in the ionosphere. The vertical structure of the ionosphere, as represented by the ionospheric
layers, was described in the subsection Reflection Coefficients. The following section describes the effects of
the lateral structures and variations in the ionosphere on ELF propagation.
The lateral structure of the ionosphere and various inhomogeneities affect the propagation loss in the
ELF range, and some affect the propagation path. However, compared to other frequency bands, propagation
in the ELF band appears more stable under ionospheric disturbances.
DayNight Terminator Effect. The line marking the transition between day and night is called the
daynight terminator, or simply the terminator, and the transition has a profound effect on propagating ELF
radiation. Measurements show that the phase velocity of ELF waves traveling across the daynight terminator
can be approximated as a weighted average of day and night phase velocities, with each velocity given a weight
according to the length of the path in the two regions [Hughes and Gallenberger (59)]. The loss in mixed
daynight paths varies in a more complex manner with the percentage of the sunlit section in the path.
Measurements at 400 Hz show deep fades when the daynight terminator is located above the receiver [Davis
(60)], which are not explained by a smooth transition between day and night regions.
ELF waves are also reflected from the daynight terminator. Bezrodnyi et al. (61) measure a reflection
coefficient for VLF signals using the Doppler shift produced by the terminator motion. Their measurements
show that the reflection coefficient is on the order of 10 2 to 10 3 . Field and Joiner (62) calculate the lateral
refection and transmission coefficients of ELF TEM waves from the terminator.
Theoretical analysis of the effects of the terminator on ELF propagation is given by Nikolaenko (63) and by
Field and Joiner (62). Nikolaenko (63) calculates the effect of the terminator in three geometrical arrangements,
describing different directions of wave propagation relative to the terminator (parallel, perpendicular, and
inclined). Figure 7 shows the perturbation in the vertical electric field from the homogeneous ionosphere model
for an inclined propagation path relative to the terminator.
14
PROPAGATION AT EXTREMELY LOW FREQUENCIES
Fig. 7. Amplitude of the vertical electric field in a propagation path inclined 45 from the terminator, relative to the
unperturbed field in the models of the smooth and sharp terminators. The day night terminator is the main source
of lateral variation in the earthionosphere waveguide. Adapted from A. P. Nikolaenko, in Radiophysics and Quantum
Electronics, Vol. 29, No. 1, Plenum Publishing Corporation, 1986, with permission.
Ionospheric Disturbances.
Varieties of Disturbances. Three major varieties of ionospheric disturbances are: (1) sudden ionospheric
disturbances (SIDs), (2) magnetic storms, and (3) solar proton events (SPEs), also called proton storms (51).
They can be defined as follows:
Sudden ionospheric disturbances are related to solar flares, electromagnetic phenomena that enhance the
electron density of the D region by up to a factor of ten.
Magnetic storms are caused by enhancements of the solar wind, a stream of particles traveling away from
the sun. An increase in the particle concentration of the solar wind, accompanied by an increase in their
velocity, causes ionospheric modifications where it reaches the magnetosphere, which result in magnetic
storms. The solar wind enhancements are usually associated with the solar events known as coronal mass
ejections or CMEs.
Solar proton events are caused by energetic streams of solar protons (accompanied by electrons). On arrival
at the earth the protons are deviated by the geomagnetic field, and they impinge on the polar caps, so solar
proton events are related to polar cap absorption (PCA) events.
Other ionospheric disturbances include the effect of sporadic E layers and small-scale ionospheric irregularities. A summary of ionospheric disturbances and their effect on radio propagation is given by Hargreaves
(54).
General Considerations. The theoretical analysis of ionospheric inhomogeneities often involves calculations of the electromagnetic field scattered from the inhomogeneities. Pappert (64) calculates the field scattered
from a cylindrically symmetric ionospheric disturbance and compares the expected effect of disturbed polar-cap
PROPAGATION AT EXTREMELY LOW FREQUENCIES
15
boundaries with measurements. The cylindrically symmetric ionospheric disturbance model is also useful in
describing small-scale ionospheric irregularities.
The effect of local ionospheric disturbances on ELF propagation is analyzed by Nickolaenko (65,66) using
a perturbation theory. The disturbances analyzed are (1) local variations of the electrical properties of the
ionosphere and (2) changes in the effective ionospheric reflection height. The analysis shows that the effects
of these two types of disturbances on ELF propagation are similar and that the effect of local disturbances is
greatest when the irregularity is situated above the source or the observer.
Random distortions in ionospheric height and fluctuations of electrical properties cause fluctuations of
ELF waves propagating in the earthionosphere waveguide; these are analyzed by Kostigov (67) using a
perturbation method.
Specic Effects. Ionospheric inhomogeneous phenomena affect ELF waves traveling between the ionosphere and the earth. Bannister (4) reviews measurements of various ELF propagation phenomena related to
ionospheric irregularities, some of which are mentioned below.
The effects of sudden ionospheric disturbances caused by solar flares on ELF propagation are analyzed
by Field (68), and the resulting modal attenuation coefficients are compared with measurements. The X-ray
radiation associated with solar flares causes a lowering of the D layer, which reduces the propagation loss in
the ELF range relative to nondisturbed conditions.
Precipitation of energetic electrons in the several days following magnetic storms may cause enhanced
loss in the ELF range at night. Measurements showing this effect are described by Davis (69) and Bannister (4),
but the theoretical explanation is incomplete. Three different groups analyzed a magnetic storm that occurred
in December 1971 using satellite measurements, and all of them measured enhanced electron precipitation
and enhanced levels of ELF radiation [Parady and Cahill (70), Vakulov et al. (71), and Kovner et al. (72)]. These
results suggest that the relationship between the electron precipitation and the ELF fields is connected with
the location of the magnetopause, that is, the outer termination of the geomagnetic field.
Occasional nighttime decreases in ELF signal levels may be explained by the presence of sporadic E
layers [Bannister (4,73)]. The measurements described by Bannister include ELF anomalies at two different
locations, appearing 2 h to 4 h apart, which may be the result of a traveling sporadic E layer. Further analysis
of the measurements from these two locations yields an estimate of the velocity and path of the sporadic E
layer [Pappert (74)].
Theoretical analyses of the effect of sporadic E layers on ELF propagation are given by Pappert (75) and
by Pappert and Moler (76). Their results indicate that a sporadic E layer 1 Mm by 1 Mm may cause 6 dB to 8
dB attenuation in the ELF range. Patches 1 Mm by 0.5 Mm can account for fades in the 3 dB to 4 dB range. The
deepest fades require the center of the disturbance to be close either to the source or to the receiver, but large
patches cause significant attenuation at distances of up to 10 Mm from the direct propagation path between
the source and receiver.
The main effect of a SPE is an increase in the attenuation in the ELF range for paths passing near the
boundary of the polar cap [Katan and Bannister (77), Field et al. (78) and Fraser-Smith and Helliwell (79)].
The additional attenuation may be caused by lateral refraction, which bends the signal path away from the
polar-cap boundary and into the central cap zone. Field et al. (78) predict enhanced field levels inside the
polar-cap boundary, but this was not verified by measurement.
The expected effects of nuclear events on ELF propagation via ionospheric changes is surveyed by Bannister (4).
ELF Calculation Methods
This section describes the approaches used to calculate ELF fields in the earthionosphere waveguide. As
already indicated, an accurate representation of the earthionosphere waveguide is extremely complex, and
16
PROPAGATION AT EXTREMELY LOW FREQUENCIES
calculations are often done numerically, incorporating information about various ionospheric and ground parameters.
The two main computational approaches presented here, namely (a) the full-wave approach and (b) the
thin-layer approach, are suitable for a horizontally stratified ionosphere with no lateral variation. Calculation
methods for laterally varying ionospheres are also surveyed.
A noniterative approximate method for calculating eigenvalues of the ELF modes, based on a two-layer
ionospheric model, is presented by Greifinger and Greifinger (36) and elaborated by them in Ref. 80. The
physical significance of a reflecting layer for ELF waves within the ionosphere is discussed in those papers.
The Full-Wave Approach. The full-wave approach involves the numerical integration of four coupled
first-order complex differential equations containing an eigenvalue parameter [Greifinger and Greifinger (80)].
The calculated quantities may be (a) the components of the electromagnetic field, (b) ratios of field components,
which may represent impedances, and (c) reflection coefficients. The full-wave integration starts from the top
of the ionosphere, where propagation is assumed to be upward only. An initial value of propagation parameters
is assumed at this height. Integration is performed downward along a vertical path, and boundary conditions
are checked at the surface of the earth. Inconsistencies of the boundary conditions at the surface of the earth
lead to corrections of the initial value (of propagation parameters at the top of the ionosphere), and the whole
process repeats with new initial values, until the boundary conditions are matched to some chosen accuracy.
A full-wave calculation of the ionospheric reflection coefficients is described by Pappert and Moler (81).
The ionosphere is assumed to vary slowly along the lateral propagation path, and the WKB approximation is
used.
Field (82) uses a full-wave method to calculate the ionospheric wave admittance for normal and disturbed
ionospheric conditions, and calculates the fields within the waveguide.
Aksenov and Nazarova (83) use a full-wave method to calculate the dependence of ionospheric reflection
and transmission coefficients on the angle of incidence, frequency, and geomagnetic field for ionospheric models
of the day and night.
The Thin-Layer Approach. The thin-layer approach is similar in principle to the full-wave approach,
but the ionosphere is taken to be a collection of thin homogeneous layers, rather then a continuum. The process
is started with some initial values of the parameters at the top boundary of the ionosphere, and then the
boundary conditions are matched between adjacent layers, going downward until a lower boundary is reached.
The thin-layer method is iterative in a similar way to the full-wave method: the boundary conditions at the
lower boundary are used to correct the initial values, and the calculation process is repeated until convergence
is achieved to a desired accuracy.
Galejs (84) presents a thin-layer method where the surface impedance computed at a lower boundary is
used to correct the initial conditions. The field components are calculated for each layer, and from these the
wave impedances are derived. This procedure is used to calculate the frequency-dependent attenuation rate
for day and night conditions and to examine the effect of the anisotropy of the ionosphere.
A similar approach is presented by Altman and Cory (85,86), where the ionosphere is treated as a
collection of discrete layers and an iterative process is used to calculate ELF propagation parameters, each
iteration starting from the top of the ionosphere. This method is presented as a derivation from the thin-film
method used in optics.
Another thin-layer method is presented by Nagano et al. (87), where the calculated quantities are the
components of the field. Nagano et al. (87) compare the width of the layers required for their method with the
step size required in various full-wave methods and show that their thin-layer method produces comparable
results to full-wave methods, with much bigger step sizes. The complexity (computer time) of the two methods
is also compared, and it is shown that the thin-layer method is more efficient.
Lateral Variations in the Waveguide. Lateral variations in the earthionosphere waveguide may be
treated by the WKB approach, which assumes slow variation in the waveguide. However, this approach is of
PROPAGATION AT EXTREMELY LOW FREQUENCIES
17
little use in the ELF band, because the wavelengths involved are very large, with the result that in many cases
the ionospheric or ground variations are not slow compared to the wavelength.
An important approach to lateral variations in the waveguide is based on an integral equation describing
the lateral propagation [Field and Joiner (88)]. The integral equation is solved numerically in terms of the sine of
the eigenangle. This approach is based on the simplified assumptions that vertical variations in the ionosphere
are decoupled from lateral variations and that lateral variations are much slower than vertical variations. A
program that predicts vertical ELF fields in an inhomogeneous waveguide is described by Ferguson et al. (89).
This program uses the moment method to solve the integral equation numerically.
Another approach to the laterally varying waveguide is based on ray theory, where vertical inhomogeneities are assumed to be uncoupled from the lateral variations. The vertical inhomogeneities are usually
treated in a full-wave approach, and the lateral variations are treated using ray theory. Field et al. (78) use the
full-waveray method to analyze the effects of polar-cap absorption events on ELF propagation.
A calculation of the electric field, given the eigenvalues of the propagation, is described by Shellman (90).
The method is based on the knowledge of the sine of the waveguide eigenangles at elements of a mesh-point
array that extends around the earth. The lateral variation of the waveguide is described by the eigenvalues at
different locations.
Acknowledgments
The authors thank Martin Fullekrug
for the Schumann resonance data plotted in Fig. 2.
This work was supported in part by the Office of Naval Research through Grant N00014-92-J-1576.
BIBLIOGRAPHY
1. J. R. Wait Electromagnetic Waves in Stratified Media, revised ed. 1970. New York: Pergamon Press.
2. P. R. Bannister Simplified formulas for ELF propagation at shorter distances , Radio Sci. 21 (3): 529537, 1986.
3. A. P. Nickolaenko M. Hayakawa Natural electromagnetic pulses in the ELF range , Geophys. Res. Lett. 25 (16): 3103
3106, 1998.
4. P. R. Bannister ELF propagation update , IEEE J. Oceanic Eng. OE-9: 179188, 1984.
5. F. J. Kelly ELF/VLF/LF propagation and system design, Technical Report NRL 9028, Naval Research Laboratory, June
1987.
6. J. Villasennor
et al. Comparison of ELF/VLF generation modes in the ionosphere by the HIPAS heater array , Radio
Sci., 31 (1): 211226, 1996.
7. R. Barr P. Stubbe. ELF and VLF radiation from the polar electrojet antenna, Radio Sci., 19 (4): 11111122, 1984.
8. K. Papadopoulos et al. On the efficiency of ionospheric ELF generation, Radio Sci., 25 (6): 13111320, 1990.
9. P. Stubbe et al. Ionospheric modification experiments with the Troms Heating Facility, J. Atm. Terrestrial Phys., 47
(12): 11511163, 1985.
10. A. J. Ferraro et al. Measurements of extremely low frequency signals from modulation of the polar electrojet above
Fairbanks, Alaska, IEEE Trans. Antennas Propag., 37: 802805, 1989.
11. R. Barr et al. ELF and VLF signals radiated by the polar electrojet antenna: Experimental results, J. Geophys. Res.,
91 (A4): 44514459, 1986.
12. M. T. Rietveld et al. Ionospheric heater beam scanning: A new technique for ELF studies of the auroral ionosphere,
Radio Sci., 19 (4): 10691077, 1984.
13. R. Barr et al. Ionospheric heater bean scanning: A mobile source of ELF radiation. Radio Sci., 22 (6): 10731083,
1987.
14. J. Galejs Terrestrial Propagation of Long Electromagnetic Waves, New York: Pergamon, 1972.
15. E. L. Maxwell Atmospheric noise from 20 Hz to 30 kHz, Radio Sci., 2 (6): 637644, 1967.
18
PROPAGATION AT EXTREMELY LOW FREQUENCIES
16. E. L. Maxwell D. L. Stone Natural noise fields from 1 cps to 100 kc, IEEE Trans. Antennas Propag. AP-11 (3): 339343,
1963.
17. G. Gustafsson A. Egeland J. Aarons Audio-frequency electromagnetic radiation in the auroral zone, J. Geophys. Res. 65
(9): 27492758, 1960.
18. A. J. Smith P. J. Jenkins A survey of natural electromagnetic noise in the frequency range f = 110 kHz at Halley
station, Antarctica: 1. Radio atmospherics from lightning, J. Atm. SolarTerrestrial Phys., 60 (2): 263277, 1998.
19. J. R. Wait Earthionosphere cavity resonances and the propagation of ELF radio waves, Radio Sci., 69D (8): 10571070,
1965.
20. D. L. Jones Numerical computations of terrestrial ELF electromagnetic wave fields in the frequency domain, Radio
Sci. 5 (5): 803809, 1970.
21. D. B. Large J. R. Wait Theory of electromagnetic coupling phenomena in the earthionosphere cavity, J. Geophys. Res.
73 (13): 43354362, 1968.
22. M. Balser C. A. Wagner Observations of earthionosphere cavity resonances, Nature, 188: 638641, 1960.
23. M. Balser C. A. Wagner Diurnal power variations of the earthionosphere cavity modes and their relationship to
worldwide thunderstorm activity, J. Geophys. Res. 67 (2): 619625, 1962.
24. A. P. Nickolaenko L. M. Rabinowicz Study of the annual changes of global lightning distribution and frequency variations of the first Schumann resonance mode, J. Atm. Terrestrial Phys. 57 (11): 13451348, 1995.
25. A. P. Nickolaenko M. Hayakawa Y. Hobara Temporal variations of the global lightning activity deduced from the
Schumann resonance data, J. Atm. Terrestrial Phys. 58 (15): 16991709, 1996.
26. T. Ogawa Y. Tanaka M. Yasuhara Schumann resonances and worldwide thunderstorm activity, in S. C. Coroniti and J.
Hughes, (ed.), Planetary Electrodynamics, Vol. 2, New York: Gordon and Breach, 1969, Chap. V-7.
27. S. J. Heckman E. Williams B. Boldi Total global lightning inferred from Schumann resonance measurements, J. Geophys.
Res. 103 (D24): 31, 77531, 779, 1998.
28. C. Polk Relation of ELF noise and Schumann resonances to thunderstorm activity, in S. C. Coroniti and J. Hughes (ed.),
Planetary Electrodynamics, Vol. 2, New York: Gordon and Breach 1969, chap. V-6. pp. 5863.
29. J. Galejs Frequency variations of Schumann resonances, J. Geophys. Res. 75 (16): 32373251, 1970.
30. D. D. Sentman B. J. Fraser Simultaneous observations of Schumann resonances in California and Australia: Evidence
for intensity modulation by the local height of the D region, J. Geophys. Res. 96 (A9): 15,97315,984, 1991.
31. K. Sao et al. Experimental investigations of Schumann resonance frequencies, J. Atm. Terrestrial Phys. 35: 20472053,
1973.
32. E. R. Williams The Schumann resonance: A global tropical thermometer, Science 256: 11841187, 1992.
33. A. P. Nickolaenko Modern aspects of Schumann resonance studies, J. Atm. SolarTerrestrial Phys., 59 (7): 805816,
1997.
34. W. Harth Theory of low frequency wave propagation, in H. Volland (ed.), CRC Handbook of Atmospherics, Vol. II, CRC
Press, Boca Raton, FL, 1982, pp. 133202.
35. H. G. Booker F. Lefeuvre The relation between ionospheric profiles and ELF propagation in the earthionosphere
transmission line, J. Atm. Terrestrial Phys., 39: 12771292, 1977.
36. C. Greifinger P. Greifinger Approximate method for determining ELF eigenvalues in the earthionosphere waveguide,
Radio Sci., 13 (5): 831837, 1978.
37. R. J. Dinger W. D. Meyers J. R. Davis Experimental investigation of ambient electromagnetic noise from 1.0 to 4.0 kHz
in Italy and Norway, Radio Sci., 17 (1): 285302, 1982.
38. R. Barr Some new features of ELF attenuation, J. Atm. Terrestrial Phys. 34: 411420, 1972.
39. R. Barr The propagation of ELF and VLF radio waves beneath an inhomogeneous anisotropic ionosphere, J. Atm.
Terrestrial Phys. 33: 343353, 1971.
40. J. Galejs Near fields and antipodal fields in the terrestrial earthionosphere waveguide, Proc. IEE 116 (7): 11501158,
1969.
41. J. R. Wait Distortion of an ELF pulses after propagation through an antipode, J. Geophys. Res. 74 (11):29822986, 1969.
42. D. L. Jones Propagation of ELF pulses in the earthionosphere cavity and application to slow tail atmospherics, Radio
Sci. 5 (8,9): 11531162, 1970.
43. A. C. Fraser-Smith P. R. Bannister Reception of ELF signals at antipodal distances, Radio Sci. 33 (1): 8388, 1998.
44. J. E. Rogerson Airborne field strength measurements in the region of the NPM antipode, Radio Sci. 2 (6): 581587,
1967.
PROPAGATION AT EXTREMELY LOW FREQUENCIES
19
45. A. D. Richmond The ionospheric wind dynamo: Effects of its coupling with different atmospheric regions, in R. M.
Johnson and T. L. Killeen (eds.), The Upper Mesosphere and Lower Thermosphere, Washington: American Geophysical
Union, 1995.
46. M. H. Rees. Physics and Chemistry of the Upper Atmosphere, Cambridge University Press, 1989.
47. S. Kato Dynamics of the Upper Atmosphere, Dordrecht: Reidel, 1980.
48. S. I. Akasofu S. Chapman SolarTerrestrial Physics, Oxford: Clarendon, 1972.
49. J. A. Ratcliffe An Introduction to the Ionosphere and Magnetosphere, Cambridge University Press, 1972.
50. J. A. Ratcliffe Sun, Earth and Radio, McGraw-Hill, 1970.
51. H. Rishbeth O. K. Garriot Introduction to Ionospheric Physics, Academic Press, 1969.
52. R. C. Whitten I. G. Poppoff Physics of the Lower Ionosphere, Engelwood Cliffs, NJ: Prentice-Hall, 1965.
53. K. Rawer The Ionosphere: Its Significance for Geophysics and Radio Communications, New York: Ungar, 1957.
54. J. K. Hargreaves The SolarTerrestrial Environment, Cambridge Atmospheric and Space Science Series, Cambridge
University Press, 1992.
55. H. Volland Longwave sferics propagation within the atmospheric waveguide, in H. Volland (ed.), Handbook of Atmospheric Electrodynamics, Vol. II, Boca Raton, FL: CRC Press, 1995, Chap. 3, pp. 6593.
56. J. R. Wait K. P. Spies Characteristics of the earthionosphere waveguide for VLF radio waves, Technical Note 300,
National Bureau of Standards, 1964.
57. M. Mambo, et al. An improved method of estimating electron density profiles in the lower ionosphere with VLF reflection
coefficients, Electron. Commun. Japan, 67-B (7): 2837, 1984.
58. K. G. Budden The Propagation of Radio Waves, Cambridge University Press, 1985.
59. H. G. Hughes R. J. Gallenberger Propagation of extremely low-frequency (ELF) atmospherics over a mixed daynight
path, J. Atm. Terrestrial Phys. 36: 16431661, 1974.
60. J. R. Davis ELF propagation irregularities on northern and mid-latitude paths, in J. A. Holtet, (ed.), ELFVLF Radio
Wave Propagation Proc. NATO Advanced Study Institute, April 1727, 1974, Dordrecht: D. Reidel, 1974, pp. 263277.
61. V. G. Bezrodnyi et al. Reflection of superlong waves from the terminator in the earthionosphere waveguide, Radiophys.
Quantum Electron., 21 (11): 11001114, 1978.
62. E. C. Field R. G. Joiner Effects of lateral ionospheric gradients on ELF propagation, Radio Sci., 17 (3): 693700, 1982.
63. A. P. Nikolaenko Scattering of ELF radio waves on global inhomogeneities of the earthionosphere cavity, Radiophys.
Quantum Electron., 29 (1): 2632, 1986.
64. R. A. Pappert ELF scattering in the earthionosphere waveguide, Radio Sci., 24 (5): 629639, 1989.
65. A. P. Nickolaenko ELF radio wave propagation in a locally nonuniform earthionosphere cavity, Radio Sci., 29 (5):
11871199, 1994.
66. A. P. Nikolaenko Effects of a local inhomogeneity in the ionosphere on the propagation of ELF radio waves, Radiophys.
Quantum Electron., 27 (10): 856864, 1984.
67. K. I. Kostigov Theory of fluctuations of super-long radio waves caused by random deformations of the boundaries of the
earth surfaceionosphere waveguide channel, Radio Eng. Electron. Phys., 19 (8): 123126, 1974.
68. E. C. Field VLF and ELF propagation during sudden ionospheric disturbances, J. Geophys. Res., 75 (10): 927933, 1970.
69. J. R. Davis Localized nighttime D-region disturbances and ELF propagation, J. Atm. Terrestrial Phys., 38: 13091317,
1976.
70. B. Parady L. J. Cahill, Jr. ELF observations during the December 1971 storm, J. Geophys. Res., 78 (22): 47654770,
1973.
71. P. V. Vakulov et al. Spilling of charged particles from the radiation belts and variations in the intensity of the ELF
radiation in the recovery phase of the magnetic storm of December 1617, 1971, Cosmic Res., 12 (5): 645652, 1974.
72. M. S. Kovner et al. High-energy electron flux and ELF radiation flux at ionospheric heights during the magnetic storm
of December 16, 1971, Cosmic Res., 15 (3): 374379, 1977.
73. P. R. Bannister Localized ELF nocturnal propagation anomalies, Radio Sci., 17 (3): 627634, 1982.
74. R. A. Pappert Calculated effects of traveling sporadic E on nocturnal ELF propagation: Comparison with measurement,
Radio Sci., 20 (2): 229246, 1985.
75. R. A. Pappert Effects of a large patch of sporadic-E on night-time propagation at lower ELF, J. Atm. Terrestrial Phys.,
42, 417425, 1980.
76. R. A. Pappert W. F. Moler A theoretical study of ELF normal mode reflection and absorption produced by night-time
ionospheres, J. Atm. Terrestrial Phys., 40: 10311045, 1978.
20
PROPAGATION AT EXTREMELY LOW FREQUENCIES
77. J. R. Katan P. R. Bannister Summary of ELF propagation variations at mid and high latitudes during the November/December 1982 and February 1984 solar proton events, Radio Sci., 22 (1): 111124, 1987.
78. E. C. Field C. R. Warber R. G. Joiner Effects of the ionosphere on ELF signals during polar cap absorption event:
Comparison of theory and experiments, in H. Soicher, (ed.), AGARD Conf. Proc. No. 382: Propagation Effects on
Military Systems in the High Latitude Region, North Atlantic Treaty Organization, 1985, 8.3-1-8.3-10.
79. A. C. Fraser-Smith R. A. Helliwell ELF sferic occurrences in the antarctic during a solar proton event: Case study of
occurrences at Byrd station during the event of June 9, 1968, J. Geophys. Res., 85 (A5): 22962306, 1980.
80. C. Greifinger P. S. Greifinger Noniterative procedure for calculating ELF mode constants in the anisotropic earth
ionosphere waveguide, Radio Sci., 21 (6): 981990, 1986.
81. R. A. Pappert W. F. Moler Propagation theory and calculations at lower extremely low frequencies (ELF), IEEE Trans.
Commun., COM-22 (4): 438451, 1974.
82. E. C. Field Propagation of ELF waves under normal and naturally disturbed conditions, J. Geophys. Res., 74 (14):
36393650, 1969.
83. V. I. Aksenov M. V. Nazarova Numerical solution of the problem of transmission of ELF waves through the lower
ionosphere, Radio Eng. Electron. Phys., 17 (7): 10571064, 1972.
84. J. Galejs On the terrestrial propagation of ELF and VLF waves in the presence of a radial magnetic field, Radio Sci. J.
Res. NBS/USNC-URSI, 69D (5): 705720, 1965.
85. C. Altman H. Cory The simple thin-film optical method in electromagnetic wave propagation, Radio Sci., 4 (5): 449457,
1969.
86. C. Altman H. Cory The generalized thin-film optical method in electromagnetic wave propagation, Radio Sci., 4 (5):
459470, 1969.
87. I. Nagano M. Mambo G. Hutatsuishi Numerical calculation of electromagnetic waves in an anisotropic multilayered
medium, Radio Sci., 10 (6): 611617, 1975.
88. E. C. Field R. G. Joiner An integral-equation approach to long-wave propagation in a nonstratified earthionosphere
waveguide, Radio Sci., 14 (6): 10571068, 1979.
89. J. A. Ferguson L. R. Hitney R. A. Pappert A program to compute vertical electric ELF fields in a laterally inhomogeneous
earthionosphere waveguide, Technical Report NOCS TR 851, Naval Ocean Systems Center, San Diego, CA, 1982.
90. C. H. Shellman A model for propagation of ELF waves throughout the lateral extent of the inhomogeneous earth
ionosphere waveguide, Radio Sci., 24 (1): 3546, 1989.
DANA PORRAT
ANTONY C. FRASER-SMITH
Stanford University
SKY WAVE PROPAGATION AT LOW FREQUENCIES
SKY WAVE PROPAGATION
AT LOW FREQUENCIES
319
duced here. A discussion of Loran-C 100 kHz sky waves is
given in Ref. 10.
Radio waves generated by manufactured transmitters near
the surface of the earth or radio waves generated by such
natural causes as cloud-to-ground lightning strokes propagate
to great distances in a lateral direction between the surface
of the earth and the ionized region between approximately 60
km and 100 km above the earth. This region is known as the
lower ionosphere or the D and E regions of the ionosphere
(18). These radio waves are called low-frequency (LF) sky
waves in reference to this propagation phenomenon.
It is common practice to specify 30 kHz to 300 kHz as the
LF part of the radio spectrum. This is somewhat arbitrary,
and theoretical considerations described herein have been
found to be useful far above and below this range of frequencies. In fact, the LF sky wave propagation theory can be used
at the lower end of the VLF radio spectrum, and at the higher
end of the MF radio spectrum. Traditionally, the mode theory
is used at VLF (9), but the theoretical presentations herein
can provide an interesting alternative.
Although LF sky waves have been used since the earliest
days of radio science, vigorous study of these waves occurred
mainly between 1950 and 1970 (see Reading List). This research was in large measure a consequence of the development of precision pulsed ground-wave radio navigation and
positioning systems. While the ground wave is very stable
with time of day, season, and so on, the ionospheric waves
vary with time, and if such waves were not somehow sidestepped, the accuracy and precision of the navigation or positioning system would deteriorate. Thus, from an engineering
point of view, LF sky waves were a problem in developing
navigation systems such as Loran-C or Loran-D. One direct
consequence of this research was the development of the wave
hop theory of LF sky wave propagation, which will be intro-
EXPERIMENTS TO DETECT LOW-FREQUENCY SKY WAVES
An experimental pulse was radiated from a transmitter located in New York state during the 1953 preliminary tests of
the Loran-C radio navigation system (2). Figure 1(a,b,c)
shows oscillograms of the observed pulse at distances west of
the transmitter of 0, 1065, and 1381 km (or 0, 662, 858 statute miles), respectively. The electromagnetic (EM) field near
the transmitter is vertically polarized and consists of a pulse
that rises to a crest in 4 cycles, or 40 s. The observation at
1065 km was recorded at 12:55 A.M. EST and therefore includes reflections from the nighttime ionosphere at altitudes
as great as 90 km. At 1381 km the second pulse was recorded
at 4:00 A.M. EST and the shape reflects the changes that occur
between day and night. Finally, the daytime pulse is shown
at the distance 1381 km. Here at 9:55 A.M. a daytime sky
wave from the D region of the ionosphere dominates the
pulse.
It is not difficult to realize that the severe distortion of
the pulse is a consequence of multiple reflections from the
ionosphere. Thus, many discrete pulses may arrive at the receiver at different delayed points in local times t. Both the
different propagation times and phase distortions resulting
from reflection process cause constructive and destructive
phase-type interference between the cycles of each pulse.
ELEMENTARY THEORETICAL CONSIDERATIONS
A theory to explain these observed phenomina has been constructed by applying Maxwells equations to a model of the
ionosphere and the earth. Before describing this theory, let
Re E(t ,d) (V/m)
d0
100
t ( s)
(a)
d = 1381 km
Re E( ,d) (v/m)
Re E(t ,d) (V/m)
d = 1065 km
77 100
j=0
248
46
j=0
100
j=1
j=1
j=2
j=2
t ( s)
t ( s)
(b)
(c)
Figure 1. (a) Experimental LF pulse radiation
close to the transmitter distance, d 0, with a
characteristic frequency of 100 kHz. (By permission of IEEE, Ref. 2). (b) Pulse observed at
distance d 1065 km at 12:55 A.M. EST, illustrating nightime LF sky wave pulses, with
characteristic frequency at 100 kHz. (By permission of IEEE, Ref. 2). (c) Pulse observed at
a distance of 1381 km at 4:00 A.M. EST, illustrating early morning LF sky wave pulses. (By
permission of IEEE, Ref. 2.)
J. Webster (ed.), Wiley Encyclopedia of Electrical and Electronics Engineering. Copyright # 1999 John Wiley & Sons, Inc.
320
SKY WAVE PROPAGATION AT LOW FREQUENCIES
us use some simple intuition and a simple model for the propagation environment. Consider the relationship between time
and frequency:
E (t , d) =
1
2
exp(it) E (, d) f s () d
[1,1]
Tee
(1)
where E(t, d) is the propagated field at time t at a distance d.
Here the transformed field E(, d) depends on the frequency , and t t d/c, where c 0.299792458 m/ns, a
constant. E(, d), and f s() are transforms of the field and the
source, respectively. The observed pulse is E(t, d), and Re in
Fig. 1 denotes the real part (2).
The earliest pulse to arrive at the receiver travels the
shortest distance from the transmitter over the geodesic, d.
This is the ground wave pulse with the indexed order j 0.
Pulses are also reflected from the ionosphere, but these pulses
always arrive later in the local time (t) domain. The earliest
sky wave to arrive at the receiver at the greater distances
occurs during daylight hours. The arrival of the first hop skywave is between 30 s and 40 s later than the arrival of the
earliest precursor of the ground wave pulse. Thus, the natural
world allows only 30 s to 40 s of pure ground-wave pulse,
in daylight hours, for operation of Loran-C. This difficulty was
overcome by time-domain data sampling on the leading edge
of the pulse at a point (say less than 30 s to 40 s) where
pure ground wave pulse energy can be found.
Figure 2 is a diagrammatic representation of the LF sky
waves, valid in both the time and frequency domains of Eq.
(1) (2,11). The concept of a somewhat localized system of reflection coefficients, R and T, at the ground and at the ionosphere respectively is introduced there. For a spherical model
of the earth with a concentric ionosphere as depicted in Fig.
2(a,b), a coordinate system is used such that the distance
from the center of the earth to the surface of the ground is r
a and at the ionosphere is r a h, where r is the radial
distance form the center of the earth. The geodesic distance is
d a, where is the angle at the earths center. The lines
connecting the source or transmitter (S) with the observer or
receiver (O), via the various local reflecting regions both at
the ground and at the ionosphere, are called geometric-optical
rays (12) or wave hops (13), where the latter include diffraction around the curve of the earth. The ionosphere lower
boundary is located at r a h g. For r coordinates the
earths surface is r a. The earliest part of each sky wave
pulse arrives at a time Dj /c, where
D j = 2 j[(a + h) cos i, j a cos j ]
in which
i, j angle of incidence on the ionosphere
j angle of incidence on the earth
If the ionosphere were perfectly conducting (i.e., sharply
bounded and of very high conductivity), the reflection coefficient would be T 1. The composite reflection process with
j=1
[2,1]
Tem
[2,2]
TmeTee
Tee
j=2
Rm Re
[2,1]
(a)
[3,2]
[3,1]
[3,3]
Tme Tee T
Tme Tem
Tmm Tee
Tem
Tee
me T ee
j=3
Rm Re Rm Re
[3,1] Rm Re
[4,1]
Tem
Tee
[4,2]
Tme Tem
Tmm Tee
[3,2]
[4,3]
[4,4]
Tme Tem Tme Tem
Tee T
Tmm Tee Tmm Tee
ee
Tme Tme Tee Tee
Tme T
me
j=4
Rm Re
[4,1]
Rm Re Rm Re
[4,2]
Rm Rm R R
m m
Re Re R R
e e
[4,3]
(b)
TM
TE
Figure 2. (a) Diagrammatic representation of LF sky waves, depicting an indexing method for identification of a series of time-domain pulses traveling laterally between the ground and the ionosphere, j 1, 2, 3, . . .. Reflecting regions at the ionosphere are
indicated as ( j, k), j 1, 2, 3, . . ., k 1, 2, 3, . . .. (From Ref. 21)
(b) Diagrammatic representation of LF sky waves, depicting an indexing method for identification of a series of time-domain pulses
traveling laterally between the ground and the ionosphere, j 2, 4,
. . .. Reflecting regions at the ionosphere are indicated as ( j, k), j
1, 2, 3, . . ., k 1, 2, 3, . . .. (From Ref. 21.)
SKY WAVE PROPAGATION AT LOW FREQUENCIES
reflections at the ground, R, for any index j would be
C j = (1) j p0j Rej1
These equations can be generalized using matrix notation, for
vertically polarized transmitters and receivers:
(2)
where
Rj1
( j 1)th earth reflection coefficient
e
p0j ionosphereground curvature focusingdefocusing
factor.
If the ionosphere is simply imperfectly reflecting with a finite
conductivity, an ionospheric reflection coefficient T can be introduced:
C j = (Tee ) j p0j Rej1
Tee is the coefficient for the TM incident, TM reflected
wave.
Tmm is the coefficient for the TE incident, TE reflected
wave.
Tem is the TM-incidentTE-reflected conversion coefficient.
Tme is the TE-incidentTM-reflected conversion coefficient.
The coupling coefficients for transmission between vertically
polarized transmission and reception can now be written:
C1 = Tee
2
C2 = Re Tee
+ Rm Tem Tme
..
.
0
Re
Ge =
0 1
Tee Tem
T=
Tme Tmm
1
0
1 0
,
1=
Gm = p 0
0 Rm
0 1
xj
j1
j Cj
(Gm TGe ) Gm T = p0
yj zj
(5)
(6)
(7)
(8)
(3)
The subscript ee denotes vertical electric (TM, or transverse
magnetic) polarization at both the source and the receiver
(vertical means in the r direction).
As extensively discussed in the literature (refs. 1416, for
example), the reflection process at the ionosphere is complicated by the effects of the earths magnetic field on the reflection process. Thus, notwithstanding the fact that the excitation waves generated at the transmitter are pure vertical
polarization (TM) waves, such waves arriving at the receiver,
after reflecting from the ionosphere magnetoplasma, contain
horizontally polarized (TE) wave components. These components of the reflected waves are shown in Fig. 2(a,b) as
dashed lines. These dashed geometric-optical lines always
originate at the ionosphere magnetoplasma. The ground, for
most practical purposes, is considered to be isotropic.
The wave j 1 is independent of the TE waves if the receiving antenna is a vertical structure (i.e., receives only TM
waves). But the wave j 2 depends on the TE component,
because this component is converted into TM waves at the
second reflection point. Thus, in principle, one cannot escape
the TE waves that originate in the ionosphere by using vertically polarized transmitting and receiving antennas.
The following reflection coefficients for TM and TE waves
are defined as:
3
C3 = 2Re Rm Tee Tem Tme + R2e Tee
+ R2m Tmm Tem Tme
321
THEORY OF PROPAGATION OF
LOW-FREQUENCY SKY WAVES
Since Eq. (1) is independent of Maxwells equations, it is necessary only to construct a model for the propagation of the E
field in the frequency domain (2,11). The propagation model
usually employed to describe the LF sky waves in their natural environment is a finitely conducting spherical earth with
radius r a. This spherical earth is surrounded by a magnetoplasma, comprising electrons, ions, and neutral particles
with a superposed terrestrial magnetic field and a finite frequency of collision between particles. These particles go from
random into orbital motion when excited by the LF electromagnetic waves. Each particle has a finite collision frequency,
which tends to dampen the activity. Since the earths magnetic field changes the particle motion from linear to orbital,
and the direction of propagation of the EM wave may vary
with respect to the terrestrial magnetic field, the ionospheric
reflection process becomes anisotropic.
The detailed structure of the lower ionosphere between approximately 60 km and 100 km above the surface of the earth
is well documented in the literature (3,17). In the daytime, a
layer of such plasma placed at 60 km and concentric with the
earth usually serves as a simple model. This can be improved
by introducing a large number of such concentric shells of
plasma as a function of altitude, to take account of the variation of electron and ion density with altitude. Application of
Maxwells equations to this model results in a rigorous fullwave solution, if the ionosphere is assumed to be isotropic (1).
As is indicated in the discussion leading to Eq. (8) and in
Refs. 4, 1416, the curved, concentric, isotropic ionosphere reflection coefficient is normally replaced by a suitable planar
anisotropic reflection coefficient. This replacement involves
the magneto-ionic theory with a full-wave (plane-wave) reflection coefficient. The spherical-wave focusing effect of the
curve ionosphere can be retained in this process.
The LF radio wave field for a transmitter and receiver on
the surface of the earth separated by a distance d can be represented rigorously by a system of waves traveling in the radial (r) direction:
(4)
Er = A
n=0
G(, n)F(r, n)
(9)
322
SKY WAVE PROPAGATION AT LOW FREQUENCIES
GEOMETRIC SERIES REPRESENTATION
Here
A=
I0 L0 c
4k21 a4
Equation (10) can be expanded into a geometric series:
where I0L is the current moment of the source, 0 4
107 F/m, c 0.299792458 m/ns, and the wave number for
the earthionosphere space is k1 /c. We also define
k2
=
c
F (r, n) = (1 + R)(1 + p0 Tee ) 1 +
(p0 RTee )
j=1
(1) (2)
1a
1a
(13)
for
2 i 2
0
|p0 RTee | < 1
where 2 is the ground conductivity (s/m), 2f, 0
1/c20, 2 is the relative dielectric constant of the ground,
and f the frequency (Hz). Furthermore,
(14)
which converges absolutely. The propagated field of Eq. (9)
can now be written
Er = Er,0 +
Er, j
(15)
j=1
G(, n) = n(n + 1)(2n + 1)Pn (cos )
where at the surface of the ground the zero-order term is the
ground wave:
where P0(z) 1, P1(z) z, and
Pn+1 (z) =
2n + 1
1
z Pn (z)
P (z),
n+1
n + 1 n1
n = 1, 2, . . .
Er,0 = B
n=0
Finally,
(1) (2)
G(, n)1a
1a (1 + Re )
(16)
Here
(1) (2)
F (r, n) = 1a
1a
(1 + R)(1 + p0 Tee )
1 p0 RTee
B=
(10)
0 c I0 L
8 k21 a4
and Re is the ground reflection coefficient:
where
p0 =
(1) (2)
1a
1g
(2)
1a
1 + Re =
(11)
(1)
1g
2n + 1 (1,2)
(1,2)
n (z) n1
(z)
z
0(1,2) (z) = exp( iz)
(1,2)
n+1
(z) =
(12)
(1,2)
1
(z) = exp( iz)
(1,2)
(1,2)
means 1
(k1a). The earths reflection
The abbreviation 1a
coefficient for the spherical waves is
k1
ln 2a
k2
Re =
k
(2)
ln 1a
+ 1 ln 2a
k2
ln 1a
n (z) = 12 [n(1) (z) + n(2) (z)]
2i
(2)
(1) (2) [ ln (2) + (k /k
1a
1
2 ) ln 1a ]
1a
1a
where the logarithmic derivative is
(1,2) (z)
(1,2)
ln 1a =
n(1,2) (z)
with
n(1,2) (z) =
ln 1a
n (z)
=
n (z)
,
z=k 1 a
d
n (z)
n (z) =
dz
These spherical wave functions are given in Ref. 18.
Equation (10) is derived from this isotropic model by identification of Tee as the reflection coefficient of the lower boundary of this model. If Tee is calculated by the recursion process
described in Ref. 1, a rigorous solution of the problem for an
arbitrary variation of the electron and ion densities with altitude can be found.
d (1,2)
dz n
The quantity B contains I0L, the source dipole current moment, and I0L 1 A m determines the E-field amplitude.
A particular sky wave can now be written
Er, j = B
n=0
where ln is the logarithmic derivative defined by
z=k 1 a
(1) (2)
G(, n)1a
1a (1 + Re )2 p0 Rej1 Teej
(17)
The anisotropic sky waves can now be written using Eqs. (5)
(8):
F(r, n) = (1 + Re )
|1 + Gm T| (1) (2)
|1 Gm Ge T| 1a 1a
(18)
whereupon expansion of the determinate ratio in a geometric
series yields
(1) (2)
F (r, n) = 1a
1a (1 + Re )I + (I + Ge )
(Gm Ge T ) j1 Gm T
j=1
(19)
SKY WAVE PROPAGATION AT LOW FREQUENCIES
provided n v . This is called the Debye approximation;
it is used extensively in the literature, for example Refs. 9,
11, 13, 19, 24, and 25. The isotropic reflection from the ionosphere lower boundary located at altitude h above the earth,
or r g a h, is
with
I=
1
0
0
1
and so
Er, j =
n=0
(1) (2)
G(, n)1a
1a (1
+ Re )
p0j C j
!j
|I + Gm T|
= I +
Gm Ge T (I + Gm T )
|I Gm Ge T|
j=1
Here
3
(1,2)
v1/2
(z) =
z
2
where Hv(1,2)(z) is the well-known Hankel function (24).
The spherical reflection coefficients introduced in Fig. 2
and used in Eqs. (16), (17) can now be written
(1)
1a
where
D2a =
(m)
2a
(m)
2a
(1)
s
Tee
(25)
k1
D (2)
k3 3g 1g
s
Tee =
k
(1) + 1 D (1)
1g
k3 3g 1g
(26)
where K3 is an isotropic wave number representing a simple
model ionosphere reflector. The angle of incidence on the ionosphere and the earth shown in Fig. 2(ad) can now be identified:
i, j = sin1
v
,
k1 g
j = sin1
v
k1 a
(27)
Using the expansion given in Eq. (21) for an anisotropic ionosphere and integrating each term of the series along a suitable contour in the complex v plane again gives the ground
wave analog of Eq. (16) and a series of terms, each of which
is a particular sky wave analog of Eq. (17).
The reduction of Eq. (22) is detailed in Refs. 13, 19.
The wave hop series now becomes
Er =
f (v)(1 + Re ) dv
c
aj
cj
j=1
f (v)(1 + Re )2 p0j (a j Tee + c j Tme ) dv
bj
a j1
= G1m Ge T
dj
c j1
b j1
d j1
(28)
(29)
where
Hv(1,2) (z)
k1
D (1)
k2 2a 1a
Re =
k
(2) + 1 D
1a
(2)
k2 2a 1a
(2)
1g
(23)
where Pv1/2(cos ) is the Legendre function of complex order.
The function (23) is the complex-order analog to that given in
Eq. (9), and we have (12)
(1,2)
(1,2)
1a
= v1/2
(k1 a),
(2)
1g
(2)
1g
(21)
which is equivalent to the determinant in Eq. (19). This is
analogous to the determinant given by Eq. (7) in Ref. 19.
Equation (21) was central to the development of a widely
distributed computer program (22,23) based on the use of the
asymptotic computation methods for spherical wave functions
of complex, noninteger order n. This approach is based on
Ref. 13.
The j-series expansion discussed above is called by its author the wave hop series. In this approach Eq. (9) is rewritten
in the complex v plane where a suitable contour c is used (24):
|I + Gm T|
dv
(22)
Er =
f (v)(1 + Re )
|I
Gm Ge T|
c
v
(1) (2)
P
( cos )1a
1a
cos v v1/2
s
T = Tee
=
(20)
Since the ground wave (16) has been removed as a separate
entity, the summation of all the j-terms generalized by Eq.
(19) is a full-wave solution for the LF sky waves in the presence of an anisotropic ionosphere. The method for inserting
anisotropy is discussed in detail in Refs. 1, 13, 19, 20. Reference 21 describes a computer program to calculate this equation directly.
The ratio of the two determinants in Eq. (18) can be written
f (v) = iA
323
(24)
Gm
p0
a1 = 1
G1m =
c1 = 0
a j = Re (Tee a j1 + TmeC j1 )
for
j2
c j = Rm (Tem a j1 + Tmm c j1 )
for
j2
Notwithstanding the fact that only TM waves have been excited and only TM waves are received at the receiver, a TEtype ground reflection coefficient is required at the ground
as a result of the effects of the earths magnetic field on the
ionosphere. Thus, due to anisotropy, the TE ground reflection
coefficient is required, as found explicitly in Refs. 19, 23:
k2
D (1)
(2)
k1 2a 1a 1a
Rm =
(1)
k
(2) 1a
(2) + 2 D
1a
2a 1a
k1
v
k2 a
2
(1)
1a
(30)
324
SKY WAVE PROPAGATION AT LOW FREQUENCIES
Table 1. Coefficients of Quartic Equation for an
Electron Plasma
a L sin i cos a,
a 0 S 22
a1 2
hLhTaL
S2
s(s 2 h 2 )
a2 2 1
a3 2
X
Figure 3. Coordinate system for each reflecting region of the ionosphere.
Tee ,
Tmm ,
Tem ,
h 2L
h 2L a 2T
s
s2
S2 2
2
2
2
s h
s(s h )
s(s h 2 ) s(s 2 h 2 )
2
s 2 h 2L
s(s 2 h 2 )
1i ,
2N
H
,
2N
S 2 sin 2 i 1
h L h sin I,
h T h cos I
2N
N e2
0m
0 e Hm
gyrofrequency
m
THEORY OF REFLECTION FROM THE IONOSPHERE
There remains to be explained a reflection process that leads
to the four ionosphere reflection coefficients:
hLhTaL
a L sec 2 i
s(s 2 h 2 )
a4 1
Hm
a T sin i sin
a 21 h 2T
s2
s
s1
1
2
1 2
S2
2
2
s h
s s h
s(s 2 h 2 )
s(s 2 h 2 )
N and m are the electron number density and mass. H m is the earths magnetic
intensity; I is the madnetic declination.
Tme
The details leading to the mathematical theory of this reflection process are given in Refs. 4, 14, 15, and 16. Consider one
of the reflecting regions at the ionosphere depicted in Fig.
Coupling
oe
eo
Zp=
Waves
Np, p p
Zp-1
Np-1, p-1 p-1
Zn+1
Nn+1, n+1 n+1
Zn
Nn, n n
Z6
N6, 6 6
ir
ri
Z5
Figure 4. Magnetoplasma model showing
detailed structure of the system of waves.
The model is flexible, and the number of layers can be increased and the thickness decreased until convergence is obtained on electronion density profiles of the lower
ionosphere. (After Ref. 2, by permission of
IEEE.)
Upgoing (i)
Downgoing (r)
Coupling N5, 5 5
Z4
N4, 4 4
Z3
N3, 3 3
Z2
N2, 2 2
Z1
N1, 1 1
0 = 1
i
Ordinary (o) mode
Extraordinary (e) mode
Upgoing (i) and downgoing (r)
waves are coupled at each boundary
Hm
SKY WAVE PROPAGATION AT LOW FREQUENCIES
325
Table 2. Reflection and Transmission Coefficient Matrices
a11 a12 a13 a14 a15 a16
Tem
Tmm
aoe
aom
Tee
Tme
boe
bom
(1)
U eio
(1)
U mio
coe
com
(1)
eic
(1)
mie
doe
dom
b11 b12 b13 b14 b15 b16
c11 c12 c13 a14 c15 c16
d11 d12 d13 d14 d15 d16
(1)
U ero
(1)
U mro
b23 b24 b25 b26 b27 b28 b29 b2(10)
(1)
U ere
(1)
U mre
(2)
eio
(2)
U mio
(2)
eie
(2)
U mie
c23 c24 c25 c26 c27 c28 c29 c2(10)
d23 d24 d25 d26 d27 d28 d29 b2(10)
a37 a38 a39 a3(10) a3(11) a3(12) a3(13) a3(14)
(2)
(2)
U ero
U mro
b37 b38 b39 b3(10) b3(11) b3(12) b3(13) b3(14)
(2)
U ere
d37 d38 d39 d3(10) d3(11) d3(12) d3(13) d3(14)
(p1)
eio
(p1)
U mio
(p1)
eie
(p1)
U mie
ap(p4) ap(p9)
(p1)
(p1)
U ero
U mro
bp(p4) bp(p9)
(p1)
(p1)
U ere
U mre
cp(p4) cp(p9)
U (p)
U (p)
eio
mio
dp(p4) dp(p9)
U (p)
U (p)
mie
eie
(2a,b). The reflection process at one such region will now be
given in detail.
The particle statistics of the electronionneutral gas with
superposed electrodynamic and magnetostatic fields was first
treated in radio science between 1927 and 1931 (58). See
also Ref. 26, pp. 5999.
a 11 cos i
a 12 cos i sin a
(1)
a 13 Q 1o
b 11 sin i
b 12 cos i cos a
b 13 1
a 14 Q ie(1)
(1)
a 15 Q ro
(1)
a 16 Q ro
b 14 1
b 15 1
b 16 1
z 1 io(1)
c
a 24 a 14 exp i z 1 ie(1)
c
(1)
a 25 a 15 exp i z 1 ro
c
(1)
a 26 a 16 exp i z 1 re
c
a 27 Q io(2)
a 28 Q ie(2)
(2)
a 29 Q ro
a 37 a 27 exp i z 2 io(2)
c
.
..
a p( p9) Q 1e
The use of the full magneto-ionic theory to model LF sky
wave propagation is given in Ref. 14. A local coordinate system is set up at a particular reflecting region depicted in Fig.
2. This Cartesian xyz, coordinate system is shown in Fig. 3.
The terrestrial magnetic field vector is contained in the yz
plane. The EM wave propagates in the z direction, and the
Table 4. Elements of Matrix Equation
Table 3. Elements of Matrix Equation
(2)
U mre
c37 c38 c39 c3(10) c3(11) c3(12) c3(13) c3(14)
a 23 a 13 exp i
a23 a24 a25 a26 a27 a28 a29 a2(10)
z 1 io(1)
c
b 24 exp i z 1 ie(1)
c
(1)
b 25 exp i z 1 ro
c
(1)
b 26 exp i z 1 re
c
b 27 1
b 28 1
b 23 exp 1
b 29 1
b 37 exp i
.
..
b p( p9) 1
z 2 io(2)
c
c 11 cos i sin a
c 12 cos a
c 13 (a L P io(1) io(1))
c 14 (a L P ie(1) ie(1))
(1)
(1)
ro
)
c 15 (a L P ro
d 11 cos i cos a
d 12 sin a
d 13 ( io(1) Q io(1) a T P io(1))
d 14 ( ie(1) Q ie(1) a T P ie(1))
(1) (1)
(1)
d 15 ( ro
Q ro a T P ro
)
(1)
(1)
re
)
c 16 (aL P re
(1) (1)
(1)
d 16 ( re
Q re a T P re
)
c 23 c 13 exp
c 24 c 14 exp
c 25 c 15 exp
c 26 c 16 exp
i z 1 io(1)
c
i z 1 ie(1)
c
(1)
i z 1 ro
c
(1)
i z 1 re
c
c 27 (a L P 2io io(2))
c 28 (a L P ie(2) ie(2))
(2)
(2)
c 29 (a L P ro
ro
)
(2)
(2)
)
c 2 10 (a L P re re
c 37 c 27 exp i z 2 io(2)
.
c
..
c p( p9) (a L P (iep) (iep))
z 1 io(1)
c
d 24 d 14 exp i z 1 ie(1)
c
(1)
d 25 d 15 exp i z 1 ro
c
(1)
d 26 d 16 exp i z 1 re
c
d 23 d 13 exp i
d 27 ( io(2) Q io(2) a T P io(2))
d 28 ( ie(2) Q ie(2) a T P ie(2))
(2) (2)
(2)
d 29 ( ro
Q ro a T P ro
)
(2) (2)
(2)
d 2 10 ( ie Q ie a T P re
)
(2)
d 37
. d 27 exp i c z 2 io
..
d p( p9) ( (iep) Q (iep) a T P (iep))
326
SKY WAVE PROPAGATION AT LOW FREQUENCIES
Table 5. Elements of Matrix Equation
ness of each layer is decreased and the number of layers is
increased until a stable reflection coefficient set is obtained
for a particular electron number density, collision frequency,
and so on. Figure 4 depicts ordinary and extraordinary propagation components coupled at each boundary. Each plasma
slab, n 1, 2, 3, . . ., p, becomes smaller as the number of
slabs, p, is increased. The complex index of refraction for each
layer in the model is obtained from a simultaneous solution
of Maxwells equations and the equation of motion in the velocity V:
a oe cos i sin a
a om cos a
b oe cos i sin a
b om sin a
c oe cos a
c om cos i sin a
d oe sin a
d om cos i cos a
hT hL
s
A1 aL
1 a 2L 2 2
s(s 2 h 2 )
s h2
hL
hT
B 1 a L a T i 2
aT i 2
s h2
s h2
hL
s 2 h 2L
C 1 1 a 2s
aL aT i 2
s(s 2 h 2 )
s h2
hT
hLhT
D1 aT i 2
aL
s h2
s(s 2 h 2 )
s 2 h 2L
s
E 1 1 a 2s
1 a 2L 2 2
s(s 2 h 2 )
s h2
hT
hT
F1 aT i 2
aT i 2
s h2
s h2
A1 B1
C1 D1
P
Q
E1 F1
E1 F1
H
H
=0
t
E
E
H J 0
=0
t
E + 0
Here (n), P P (n), Q Q (n) for a particular slab (Fig. 4), and a s sin i ;
io, ie, ro, re refer to the four roots of Eq. (35).
wavefronts are contained in the xy plane. The directions relative to the magnetic field vector are as follows:
i is the angle of incidence.
a is the magnetic azimuth.
I is the magnetic dip angle.
i, j is now abbreviated to i.
(31)
V
dV
V + 0 eV
V H + eE
E=0
+ mV
dt
(32)
where 0 4 107 H/m, is the collision frequency, m is
the electronic mass, and e is the electronic charge. Reference
4 extends these equations to include ions, and Ref. 16 provides an extension for collision rate proportional to energy
(2729). The lower boundary of the model electron plasma
(Fig. 3), below which (z 0) the ionization is nil (N 0), is
taken as the xy plane. The region above the xy plane (z 0)
is characterized by an electron number density N, which,
along with the collision frequency, varies with altitude z. A
plane wave incident upon the ionosphere is assumed to be
E i | exp i t D
Ei = |E
c
(33)
Below the ionosphere, 0 1. Here
Figure 4 depicts a flexible model for the lower ionosphere.
The plasma electron density is divided into layers. The thick-
D = x sin i sin a + y sin i cos a + z cos i
107
Amplitude (V/m)
108
3
4
5
6
7
j=8
109
3
1010
5
4
Figure 5. Amplitude of the total E field as a function of distance along the surface of the earth, together with the individual wave hops, j 0 (the ground wave), 1, 2, 3, . . ., illustrating propagation in the presence of an ionosphere with infinite
conductivity. (From Ref. 21 by permission IEEE (2).)
j=0
j=1
6
j=2
7
8
1000 2000 3000 4000 5000 6000 7000 8000 9000 10,000
Distance (km)
1011
SKY WAVE PROPAGATION AT LOW FREQUENCIES
327
A wave transmitted into the ionosphere is assumed to have
the form
such as Refs. 21, 22. The complex index of refraction of a particular layer in the model now can be found:
E t | exp i t D
Et = |E
c
2 = 2 + sin2 i
(34)
Elimination of the vectors V and H results in a quartic equation:
D = x sin i sin a + y sin i cos a + z
a4 4 + a3 3 + a2 2 + a1 + a0 = 0
(35)
The detailed explicit expressions for the coefficients of this
quartic are given in Table 1 for an electron magnetoplasma.
The coefficients for an electronion magnetoplasma are given
in Ref. 4. The quartic equation is readily solved numerically
with various computer programs that exist in the literature,
120
The significance of the four roots of the quartic equation is
depicted in Fig. 4. There are two upgoing waves, called the
ordinary and the extraordinary upgoing wave. There are two
downgoing waves, the ordinary and the extraordinary downgoing wave. These four waves are coupled at each boundary
in the model and hence are modified by the change in electron
density on each side of the boundary.
The boundary of each stratum, or plasma slab, is introduced through the continuity of the tangential E and H field
components in Maxwells equations. This is accomplished by
equating the fields immediately above and immediately below
each boundary, which results in the matrix equation in Table 2.
Using Fig. 3, the following transmission and reflection coefficients can be recovered by a numerical solution of the matrix equation in Table 2:
Tee =
140
Tme =
160
U (n)eio =
4
5
Decibels
180
U (n)eie =
6
3
7
200
Ey r
Ey i
Ey r
Ex i
(n)
Eyio
Ey i
(n)
Eyie
Ey i
Tem =
Ex r
Ey
i
E
= xr
Ex i
Tmm
U (n)mio =
U (n)mie =
(n)
Eyio
Ex i
(n)
Eyie
Ex r
using the ratios Q Ex /Ey, P Ez /Ey, where (Fig. 2) n 1,
2, 3, . . ., p 1. The elements of the matrix equation in Table
2 are defined in Tables 3, 4, and 5 for an electron magnetoplasma.
8
2
9
220
(36)
10
240
11
1
SKY WAVE DIFFRACTION
The angle of incidence of the sky wave on the earth [Eq. (26)]
is in general complex:
260
j =
280
300
2000
4000
6000
8000
10,000
Distance (km)
Figure 6. Amplitude of the total E field as a function of distance
along the surface of the earth, together with the individual wave
hops, j 0 (the ground wave), 1, 2, 3, . . ., illustrating propagation
in the presence of a sharply bounded semiinfinite plasma slab placed
at the height h 65 km for a signal at 26.1 kHz and an assumed
ratio of electron number density to collision frequency equal to 9.375
(1). (From Ref. 1, by permission of AGU.)
v
k1 a
This is implied in the full-wave solutions given by Eqs. (16),
(17), (26). LF sky wave diffraction theory has been treated in
Refs. 25, 30, and 31. However, it can now be demonstrated
that the full-wave solution discussed herein implies sky wave
diffraction into the earths shadow region. The hypothetical
ionosphere reflection was set equal to that of a perfect reflector [Eq. (2)], and Eqs. (16), (17) were used to generate the
wave hop series shown in Fig. 5. Each wave hop in Fig. 5
j 0 (the ground wave) and j 1, 2, 3, . . .will ultimately
go beyond the geometric-optical horizon, and the attenuation
finally takes on a slope as a function of distance like the
ground wave. The conductivity of the ground is assumed to be
0.005 S/m with a dielectric constant of 15 relative to space.
328
SKY WAVE PROPAGATION AT LOW FREQUENCIES
107
j=0
Amplitude (V/m)
108
j=3
109
j=4
j=5
j=6
1010
j=2
j=7
j=1
j=8
j=0
Figure 7. Sample wave hop calculation using the full magneto-ionic theory for the model given in Table 2, for a propagation path over sea water with magnetic parameters and ionosphere electron density profiles given in Ref. 21.
The frequency is 26 kHz, and the ionosphere height h 65
km.
A more sophisticated model of the ionosphere is shown in
Fig. 6. Here a single sharply bounded electron plasma slab is
placed at 65 km and extends uniformly out to infinity with a
constant ratio of electron density to collision frequency of
9.375 (1). The frequency is again 26 kHz. The influence of a
finitely conducting ionosphere is here introduced into the
model. Again, each sky wave at great distance attenuates
with a slope parallel to that of the ground wave on this linear
distancedecibel (logarithmic) amplitude scale. A rather interesting set of standing waves as a function of distance appears at shorter distances on the higher-order wave hops. The
total field, which sums the wave hops together with the
ground wave, does not give any evidence of the existence of
these waves.
It is concluded from Fig. 6 that the inclusion of a finitely
conducting ionosphere is an important improvement in the
modeling technique. At this juncture one could still go further
with the full-wave solution and use concentric spherical shells
that follow known electron-densitycollision-frequency profiles such as given in Ref. 1. But this would not take into
account the effects of the earths magnetic field on the propagation. By following the procedures dictated by Eq. (20) or by
using the asymptotic methods used with Eq. (28), the effects
of a more realistic model can be generated as shown in Fig.
7. The electron density profile for the ionosphere is given in
Ref. 21.
It is interesting to note that the standing waves on the
individual wave hops are more highly damped as a function of
distance. This is quite reasonable, since the ionosphere lower
boundary is more diffuse (gradual) in this model. It therefore
seems quite justified to use this more sophisticated model.
1011
1000 2000 3000 4000 5000 6000 7000 8000 9000 10,000
Distance (km)
BIBLIOGRAPHY
1. J. R. Johler, Spherical wave theory for MF, LF, and VLF propagation, Radio Sci., 5: 14291443, 1970.
2. J. R. Johler, Propagation of the low-frequency radio signal, Proc.
IRE, 50 (4): 404427, 1962.
3. J. S. Belrose, L. A. Bound, and L. W. Hewitt, Ground based radio
wave propagation studies of the lower ionosphere, Proc. Conf.,
held 1115 April 1966, Defence Research Telecommunications
Establishment, Radio Physics Lab., Defense Research Board,
Dept. of National Defence, Shirley Bay, Ottawa, Canada, 1967.
4. J. R. Johler and L. A. Berry, On the effect of heavy ions on LF
propagation, with special reference to a nuclear environment,
Natl. Bur. Stand. Tech. Note 313, Washington: U.S. Govt. Printing Office, 1965.
5. D. R. Hartree, Proc. Cambridge Philos. Soc., 25: 27, 1927.
6. D. R. Hartree, The propagation of electro-magnetic waves in a
refracting medium in a magnetic field, Proc. Cambridge Philos.
Soc., 27: 143, 1931.
7. E. V. Appleton, J. Inst. Elec. Eng., 71: 642, 1932.
8. J. A. Ratcliffe, The Magneto-ionic Theory and Its Applications to
the Ionosphere, Cambridge, UK: Cambridge Univ. Press, 1959.
9. J. R. Wait, Electromagnetic Waves in Stratified Media, New York:
Pergamon, 1962.
10. G. Hefley, The Development of Loran-C Navigation and Timing,
NBS Monograph 129, Washington: U.S. Govt. Printing Office,
1972.
11. J. R. Johler, Propagation of an electromagnetic pulse from a nuclear burst, IEEE Trans. Antennas Propag., AP-15: 256263,
1967.
12. H. Bremmer, Terrestrial Radio Waves, New York: Elsevier, 1949.
13. L. A. Berry, Wave hop theory of long distance propagation of LF
radio waves, J. Res. Natl. Bur. Stand., 68D: 12751282, 1964.
SKY WAVE PROPAGATION AT MEDIUM AND HIGH FREQUENCIES
14. J. R. Johler and J. D. Harper, Jr., Reflection and transmission of
radio waves at a continuously stratified plasma with arbitrary
magnetic induction, J. Res. Natl. Bur. Stand., 66D: 8191, 1962.
15. J. R. Johler and J. D. Harper, Jr., On the effect of a solar disturbance on the LF ionosphere reflection process, 6th AGARD
(NATO) Ionospheric Research Meeting, 1518 May 1961, Naples,
Italy, New York: Pergamon, 1963.
16. J. R. Johler, On radio wave reflections at a continuously stratified
plasma with collisions proportional to energy and arbitrary magnetic induction, Proc. Int. Conf. Ionosphere, Imperial College, London, July, 1962, London: Inst. Phys., Phys. Soc., Chapman &
Hall, 1963.
17. R. M. Davis and L. A. Berry, A revised model of the electron
density in the lower ionosphere, Alexandria, VA: Defense Documentation Center, Cameron Station, 1977.
18. M. Abromowitz and I. A. Stegun, Handbook of Mathematical
Functions, NBS, AMS 55, Washington: U.S. Govt. Printing Office, 1964.
19. L. A. Berry, G. Gonzalez, and J. L. Lloyd, Wave hope series for
an anisotropic ionosphere, Radio Sci., 4: 10211027, 1969.
20. J. R. Johler, Zonal harmonics in low frequency terrestrial radio
wave propagation, Natl. Bur. Stand. Tech. Note 335, Washington:
Supt. of Documents, U.S. Govt. Printing Office, 1966.
21. J. R. Johler and C. Mellecker, Theoretical LF, VLF field calculations with spherical wave functions of integer order, ESSA Tech.
Rep. ERL 165ITS 106, Boulder, CO: U.S. Dept. of Commerce,
Inst. for Telecommunication Sciences (ITS), 1970.
22. L. A. Berry and J. E. Herman, A wave hop propagation program
for an anisotropic ionosphere, OT/ITS Res. Rep. 11, Boulder, CO:
U.S. Dept. of Commerce, Inst. for Telecommunication Sciences
(ITS), 1971.
23. J. K. Oliver, Jr., G. Gonzalez, and J. L. Lloyd, LFVLF propagation analysis computer program documentation, RADC-TR-68453, Rome, NY: Rome Air Development Center, 1968.
24. J. R. Johler and L. A. Berry, A Complete Mode Sum for LF, VLF,
ELF Terrestrial, Radio Wave Fields, NBS Monograph 78, Washington: U.S. Govt. Printing Office, 1964.
25. J. R. Wait, A diffraction theory for LF sky-wave propagation, J.
Geophys. Res., 66: 17131724, 1961.
26. K. Davies, Ionospheric Radio Propagation, NBS Monograph 80,
Washington: U.S. Govt. Printing Office, 1965.
27. A. V. Phelps, Propagation constants for electromagnetic waves in
weakly ionized, dry air, J. Appl. Phys., 21: 17231729, 1960.
28. P. Molmud, Langevin equation and the ac conductivity of nonmaxwellian plasmas, J. Phys. Res., 114: 2932, 1959.
29. R. B. Dingle, D. Arndt, and S. K. Roy, The integrals Ep(x) and
Dp(x) and their tabulation, Appl. Sci. Res., 6B: 155164, 1956.
30. L. A. Berry and M. C. Chrisman, Numerical values for the path
integrals for low and very low frequencies, NBS Tech. Note 319,
Washington: U.S. Govt. Printing Office, 1965.
31. O. E. H. Rydbeck, On the propagation of radio waves, Trans.
Chalmers Univ. 34, 1944.
Reading List
L. A. Berry and R. M. Jones, A time-varying electron density model
for LF/VLF propagation calculations, OT/ITSTM 3, Boulder, CO:
U.S. Dept. of Commerce, Inst. for Telecommunication Sciences
(ITS), 1970.
J. R. Johler, On the analysis of LF ionospheric radio propagation phenomena, J. Res. Natl. Bur. Stand Radio Propag., 65D: 5, 1961.
J. R. Johler, Theory of propagation of low frequency terrestrial radio
wavesmathematical techniques for the interpretation of D-region propagation studies, 1966 Ottawa Conf. Proc., Vol. 2, 1967,
pp. 399422. See also Ref. 3.
329
J. R. Wait, Terrestrial propagation of very low frequency radio waves,
J. Res. Natl. Bur. Stand. D Radio Propag., 64D: 183204, 1960.
J. RALPH JOHLER
Johler Associates
FRIIS FREE-SPACE TRANSMISSION FORMULA
pinging on the ground. Also, in this frequency range atmospheric effects are small. Thus the Friis formulation is directly applicable. In other situations the free-space equation
may be the starting formulation to which corrections for other
effects (e.g., atmospheric attenuation, attenuation due to the
ground) can be added. In fact, attenuation over a path is often
specified as the attenuation relative to that for the same distance in free space; the latter would be calculated by the
Friis formula.
STATEMENT OF THE FORMULA
For a free-space transmission path, the available power Pr at
the receiving antenna terminals is related to the power input
Pt at the transmitting antenna terminals by the ratio
Pr
A Ar
= 2t 2
Pt
R
(1)
Far Field
In this expression Ar denotes the effective area of the receiving antenna in the direction of the transmitter, At is the effective area of the transmitting antenna in the direction of the
receiver, is the wavelength, and R is the distance separating
the transmitting and receiving antenna, all measured in compatible units (e.g., both powers in watts, both areas in square
meters, and R in meters). An alternative form is
G Gr 2
Pr
= t 2 2
Pt
(4 ) R
The antennas must be in the far field (Fraunhofer region)
with respect to one another. This requires a sufficient distance so that the transmitted field at the receiving antenna
is a spherical wave which may be approximated as a plane
wave over the receiving antenna aperture. A frequently used
criterion is that the longest and shortest paths between the
two antenna apertures should not differ by more than 1/16
wavelength. Also the distance must be greater than five times
the sum of the largest transmitting antenna dimension and
the largest receiving antenna dimension, and it must exceed
1.6 wavelengths.
(2)
where Gt and Gr denote, respectively, the gain of the transmitting antenna in the direction of the receiver and the gain of
the receiving antenna in the direction of the transmitter, and
and R must be given in the same units. For convenience, a
decibel formulation is sometimes used. Such formulations are
obtained by setting c/f, where c is the free-space velocity
of light and f the frequency, converting quantities to the desired units (e.g., R from meters to kilometers) taking the common logarithm of both sides of the resulting equation, and
multiplying by 10. Algebraic and arithmetic manipulation
then yield expressions such as
Pr,dbW = Pt,dbW + Gt,dB + Gr,dB 20log10 Rkm
20log10 f MHz 32.4
Polarization
In this formulation it has been assumed that the polarization
of the wave at the receiving antenna is optimal for that antenna; otherwise a polarization-mismatch factor (in the decibel formula, a polarization-mismatch term) must be included.
DERIVATION
Equations (1) and (2) can be derived easily from basic physical principles. For a hypothetical lossless, isotropic transmitting antenna the power applied to the input terminals would
be spread uniformly over a sphere at radius R, giving a flux
density
(3)
In this example of the decibel form of the formula, the power
at the transmitting antenna terminals and the power available at the receiving antenna terminals are expressed in
dBW, (i.e., the power relative to 1 W expressed in decibels),
the distance is given in kilometers, and the frequency in megahertz. When other units are used, the form of the equation
remains the same, but the value of the constant term may
differ. In the decibel form of the formula, the antenna gains
are always specified in decibels.
Si =
Pt
4R2
(4)
For an actual antenna, by the definition of antenna gain, the
field in the receiver direction is obtained by
S = S i Gt =
Pt Gt
4R2
(5)
By definition of the effective area, the power available at the
receiving antenna terminals for optimum polarization is given
by
ASSUMPTIONS
Free Space
Pr = SAr =
The term free-space implies that environmental effects,
such as the effects of the ground and of the atmosphere, are
negligible. This is sometimes a very good approximation. For
example, in the case of a transmission from the earth surface
to a satellite appearing not too near the horizon in the frequency range 500 MHz to 10 GHz, the antenna will usually
be sufficiently directive to prevent substantial power from im-
Pt Gt Ar
4R2
(6)
Use of the relationship between gain and effective area
G=
4A
2
then leads to either Eq. (1) or Eq. (2).
1
J. Webster (ed.), Wiley Encyclopedia of Electrical and Electronics Engineering. Copyright # 1999 John Wiley & Sons, Inc.
(7)
FUEL CELL POWER PLANTS
HISTORICAL NOTE
The formulas in Eqs. (1) and (2) are named after Harald Trap
Friis, who published Eq. (1) in 1946 while he was a research
engineer at Bell Telephone Laboratories, Inc. (1). Mr. Friis
later became Director of Research in High Frequency and
Electronics at Bell Telephone Laboratories and was honored
with numerous medals and awards for his technical work and
his leadership.
BIBLIOGRAPHY
1. H. T. Friis, A note on a simple transmission formula, Proc. IRE
(subsequently Proc. IEEE ) 34: 254256, 1946.
Reading List
An interesting summary of derivations of the Friis formula from various perspectives can be found in D. C. Hogg, Fun with the Friis freespace transmission formula, IEEE Antennas Propag., Mag. 35 (4): 33
35, 1993.
CURT A. LEVIS
The Ohio State University
FUEL CELL. See HYDROGEN ENERGY SYSTEMS.
Você também pode gostar
- About Transmission Line and Satellite CommunicationDocumento20 páginasAbout Transmission Line and Satellite Communicationangelie sambayanAinda não há avaliações
- Radio Propagation and Antennas: A Non-Mathematical Treatment of Radio and AntennasNo EverandRadio Propagation and Antennas: A Non-Mathematical Treatment of Radio and AntennasNota: 5 de 5 estrelas5/5 (1)
- Assignment Comm3Documento27 páginasAssignment Comm3angelie sambayanAinda não há avaliações
- RMSA Project ReportDocumento97 páginasRMSA Project Reportprashanth kumarAinda não há avaliações
- Antenna 3Documento11 páginasAntenna 3oyianonymousAinda não há avaliações
- The Fast Track to Understanding Ham Radio PropagationNo EverandThe Fast Track to Understanding Ham Radio PropagationAinda não há avaliações
- Waves and AntennasDocumento16 páginasWaves and AntennasTony ClaytonAinda não há avaliações
- The Radio Amateur's Hand Book: A Complete, Authentic and Informative Work on Wireless Telegraphy and TelephonyNo EverandThe Radio Amateur's Hand Book: A Complete, Authentic and Informative Work on Wireless Telegraphy and TelephonyAinda não há avaliações
- Microstrip AntennaDocumento84 páginasMicrostrip Antennanishanth17492% (12)
- Electrodynamic Waves: Wireless Transmission Through Air, Metals, Water and the Human BodyNo EverandElectrodynamic Waves: Wireless Transmission Through Air, Metals, Water and the Human BodyAinda não há avaliações
- Antenna Theory HistoryDocumento77 páginasAntenna Theory HistoryDELIRI GARGAinda não há avaliações
- Radio Antennas and Propagation: Radio Engineering FundamentalsNo EverandRadio Antennas and Propagation: Radio Engineering FundamentalsAinda não há avaliações
- Annalen Der Physik Und ChemieDocumento35 páginasAnnalen Der Physik Und ChemieVinod KumarAinda não há avaliações
- Radio Broadcast Technology: Harsh VardhanDocumento11 páginasRadio Broadcast Technology: Harsh VardhanLaraAssiAinda não há avaliações
- Antenna Report Update 1.0Documento23 páginasAntenna Report Update 1.0Shambhavi KAinda não há avaliações
- Antenna (Radio)Documento24 páginasAntenna (Radio)sorin birouAinda não há avaliações
- Basics of Communication & Astronomy: APTMU Books, #1No EverandBasics of Communication & Astronomy: APTMU Books, #1Ainda não há avaliações
- Theory Antenna CppasteDocumento312 páginasTheory Antenna CppastekenAinda não há avaliações
- Antennas L01Documento15 páginasAntennas L01Domenico RizzoAinda não há avaliações
- Antenna BasicsDocumento125 páginasAntenna Basicssamadhan1187Ainda não há avaliações
- UntitledDocumento9 páginasUntitledGautham VenugopalAinda não há avaliações
- Radio ReceiversDocumento123 páginasRadio ReceiversThắng Phan Duy100% (2)
- Antena - L1 - Antenna Basics22 - 23Documento9 páginasAntena - L1 - Antenna Basics22 - 23مؤمل عبدالهاديAinda não há avaliações
- Ant TheoryDocumento45 páginasAnt TheoryAbood AliAinda não há avaliações
- Caveman RadioDocumento6 páginasCaveman RadioDonT_RNAinda não há avaliações
- Radio ReceiversDocumento106 páginasRadio ReceiversAgus Dwiaji Abi HaffafAinda não há avaliações
- Communication Antenna PDFDocumento9 páginasCommunication Antenna PDFJaison PhilipAinda não há avaliações
- Hydroacoustics and Underwater Acoustics: SessionsDocumento53 páginasHydroacoustics and Underwater Acoustics: Sessionsle_luan_166Ainda não há avaliações
- CHAPTER 14 Antennas and Wave PropagationDocumento4 páginasCHAPTER 14 Antennas and Wave PropagationPatrick GarciaAinda não há avaliações
- Whip AntennaDocumento4 páginasWhip AntennatechzonesAinda não há avaliações
- NiceDocumento7 páginasNicekisan bistaAinda não há avaliações
- AntennaDocumento56 páginasAntennaAbhinav SharmaAinda não há avaliações
- Antenna and Wave Propagation (EEE) - EEC708Documento5 páginasAntenna and Wave Propagation (EEE) - EEC708Mayur BundelAinda não há avaliações
- Antenn Chapter 1Documento39 páginasAntenn Chapter 1Chernet ThomasAinda não há avaliações
- Antenna Basics - Christ of Rohner With Contributuons From Franz Demmel Axel Stark Manfred Stecher (Rohde & Schwarz)Documento76 páginasAntenna Basics - Christ of Rohner With Contributuons From Franz Demmel Axel Stark Manfred Stecher (Rohde & Schwarz)abcd2690Ainda não há avaliações
- Antenna Basic TermsDocumento24 páginasAntenna Basic Termsmarc.ysak.13Ainda não há avaliações
- Antenna (Radio) : Jump ToDocumento29 páginasAntenna (Radio) : Jump ToAudacieux FilleAinda não há avaliações
- Antenna Basics: Frequency Is One of The Most Important Concepts in The Universe and To Antenna Theory, WhichDocumento39 páginasAntenna Basics: Frequency Is One of The Most Important Concepts in The Universe and To Antenna Theory, WhichCharbel RahmeAinda não há avaliações
- Ec2353 AwpDocumento137 páginasEc2353 AwpNirmalyaMazumdarAinda não há avaliações
- Balanis Antenna Theory A ReviewDocumento17 páginasBalanis Antenna Theory A ReviewGetahun TesfaAinda não há avaliações
- Introduction of Microwave Antenna and ApplicationsDocumento34 páginasIntroduction of Microwave Antenna and ApplicationsdineshAinda não há avaliações
- Antennas:: Subject: Electro Magnetic Field TheoryDocumento38 páginasAntennas:: Subject: Electro Magnetic Field Theorynadeem saeedAinda não há avaliações
- CNS-ST1.2Documento8 páginasCNS-ST1.2Achilles Aldave50% (2)
- Spark Gap TransmitterDocumento34 páginasSpark Gap TransmitterpepeAinda não há avaliações
- Fluxgate MagnetometerDocumento6 páginasFluxgate MagnetometerAleksey HblastoAinda não há avaliações
- Antenna Design, Other StyleDocumento51 páginasAntenna Design, Other Styleapi-26989621100% (3)
- Bandwidth Is Another Fundamental Antenna Parameter. Bandwidth Describes The RangeDocumento6 páginasBandwidth Is Another Fundamental Antenna Parameter. Bandwidth Describes The RangeOral RobertzAinda não há avaliações
- Electromagnetic Radiation Ron Bertrand PDFDocumento10 páginasElectromagnetic Radiation Ron Bertrand PDFMiller Edisson Hernández PeñaAinda não há avaliações
- Antenna Theory - A Comparative Study PDFDocumento6 páginasAntenna Theory - A Comparative Study PDFoldjanusAinda não há avaliações
- Review of Antenna Geometries and Arrangements. Wireless vs. Cable Communication Systems. The Radio-Frequency Spectrum.)Documento15 páginasReview of Antenna Geometries and Arrangements. Wireless vs. Cable Communication Systems. The Radio-Frequency Spectrum.)DanielAeiounAeioumAinda não há avaliações
- Antenna and Wave PropagationsDocumento146 páginasAntenna and Wave PropagationswiyetoAinda não há avaliações
- Antenna (Radio)Documento50 páginasAntenna (Radio)Radhika SethuAinda não há avaliações
- An Introduction To Antenna TheoryDocumento95 páginasAn Introduction To Antenna TheorytriplojotaAinda não há avaliações
- 08 Circuits & Systems. Fundamental Theory & ApplicationsDocumento715 páginas08 Circuits & Systems. Fundamental Theory & ApplicationsAngel CelestialAinda não há avaliações
- 05 Biomedical EngineeringDocumento357 páginas05 Biomedical EngineeringAngel Celestial100% (1)
- 07 Circuits & Systems. Analog & Digital Signal ProcessingDocumento1.041 páginas07 Circuits & Systems. Analog & Digital Signal ProcessingAngel CelestialAinda não há avaliações
- 01 Aerospace & Electronic SystemsDocumento380 páginas01 Aerospace & Electronic SystemsAngel Celestial100% (1)
- 03 Applied SuperconductivityDocumento522 páginas03 Applied SuperconductivityAngel CelestialAinda não há avaliações
- Ogino - Qi ZhengDocumento5 páginasOgino - Qi ZhengAngel CelestialAinda não há avaliações
- Tibetan AstrologyDocumento84 páginasTibetan AstrologyAngel CelestialAinda não há avaliações
- Hsia Calendar 1974 - 2020Documento931 páginasHsia Calendar 1974 - 2020Angel Celestial67% (3)
- Ogino - Mei Hua Xin YiDocumento11 páginasOgino - Mei Hua Xin YiAngel Celestial100% (1)
- Ogino - Qi Men Dun JiaDocumento16 páginasOgino - Qi Men Dun JiaAngel Celestial100% (7)
- Heluo WorkbooksDocumento12 páginasHeluo WorkbooksAngel Celestial100% (5)
- How To Make Salt Coin Cure PDFDocumento4 páginasHow To Make Salt Coin Cure PDFAngel CelestialAinda não há avaliações
- Scientific Evidences On How Cell Phones Affect HealthDocumento28 páginasScientific Evidences On How Cell Phones Affect HealthInfo_scribd100% (1)
- Electronics Lecture 2Documento28 páginasElectronics Lecture 2Sathish Kumar KarneAinda não há avaliações
- GSM Based Stepper Motor ControlDocumento81 páginasGSM Based Stepper Motor ControlEKo SaputroAinda não há avaliações
- En-Ieb Aen Oe1 1Documento52 páginasEn-Ieb Aen Oe1 1Serban MihaelaAinda não há avaliações
- HIFREQ: An OverviewDocumento7 páginasHIFREQ: An OverviewMohammad SeyediAinda não há avaliações
- Capacitor ShootoutDocumento17 páginasCapacitor ShootoutlookbooksAinda não há avaliações
- Bee Book by Gulzar AhmadDocumento337 páginasBee Book by Gulzar AhmadMegha Singh100% (1)
- Physics: Final Syllabus For NEET-UGDocumento8 páginasPhysics: Final Syllabus For NEET-UGVansh SainiAinda não há avaliações
- 4.5-V To 65-V Input, Compact Bias Supply With Power Stage Reference Design For Igbt/Sic Gate DriversDocumento44 páginas4.5-V To 65-V Input, Compact Bias Supply With Power Stage Reference Design For Igbt/Sic Gate DriversDebasish MishraAinda não há avaliações
- PC 23 SpecDocumento19 páginasPC 23 Speccht0% (1)
- Capacitance and Inductance of A Parallel Plate Transmission LineDocumento2 páginasCapacitance and Inductance of A Parallel Plate Transmission LineBabasrinivas GuduruAinda não há avaliações
- Design and Testing of A Three-Break Shunt Reactor SwitchingDocumento9 páginasDesign and Testing of A Three-Break Shunt Reactor SwitchingWardencasianAlanisAinda não há avaliações
- RS1117Documento10 páginasRS1117asismamenAinda não há avaliações
- 1 5ke68aDocumento7 páginas1 5ke68amohammadsajjad sedqiAinda não há avaliações
- Restricted Earth Fault Relay Application Within Shunt Capacitor Bank PDFDocumento115 páginasRestricted Earth Fault Relay Application Within Shunt Capacitor Bank PDFyasadik9Ainda não há avaliações
- EMF-Assignment Questions 0Documento6 páginasEMF-Assignment Questions 0Akula VeerrajuAinda não há avaliações
- Infineon-AN2009 10 Using The NTC-ApplicationNotes-V01 00-EnDocumento10 páginasInfineon-AN2009 10 Using The NTC-ApplicationNotes-V01 00-EnjuanAinda não há avaliações
- Lecture 1HV - Basics of High Voltage EngineeringDocumento34 páginasLecture 1HV - Basics of High Voltage EngineeringFaizan AhmadAinda não há avaliações
- Switching Overvoltages in Wind FarmsDocumento13 páginasSwitching Overvoltages in Wind FarmsMarko IstenicAinda não há avaliações
- Smoothing and Filtering The Power SupplyDocumento2 páginasSmoothing and Filtering The Power SupplyDiego García MedinaAinda não há avaliações
- F5 Elevator Drive Manual - V1.72Documento180 páginasF5 Elevator Drive Manual - V1.72Hongquan SuAinda não há avaliações
- CH 12Documento7 páginasCH 12Zhuo YaoAinda não há avaliações
- (IEEE Format) Design Consideration A Low-Cost Capacitive Sensor For Plantar Pressure Insole - R1Documento5 páginas(IEEE Format) Design Consideration A Low-Cost Capacitive Sensor For Plantar Pressure Insole - R1WAHYU DWI LESTARIAinda não há avaliações
- Basic Electrical EngineeringDocumento68 páginasBasic Electrical Engineeringahg#Ainda não há avaliações
- Paper 14Documento4 páginasPaper 14Kiran YaddanapudiAinda não há avaliações
- Asso CET 2014 SyllabusDocumento21 páginasAsso CET 2014 SyllabusJitendra ChogleAinda não há avaliações
- Single Phase Tefc Cage MotorsDocumento12 páginasSingle Phase Tefc Cage Motors1cvbnmAinda não há avaliações
- Career1737 ElectronicsDocumento6 páginasCareer1737 Electronicsbabu balochAinda não há avaliações
- Real-Time Co-Simulation Platform Using OPAL-RTDocumento5 páginasReal-Time Co-Simulation Platform Using OPAL-RTLuis Fernando Cortez PeñuelasAinda não há avaliações
- TB6560AHQ Datasheet en 20110118Documento34 páginasTB6560AHQ Datasheet en 20110118jhond1000Ainda não há avaliações
- EEX 107.02 Evaluating Insulation TestsDocumento107 páginasEEX 107.02 Evaluating Insulation TestsJellyn Base100% (1)
- Evaluation of Some Websites that Offer Virtual Phone Numbers for SMS Reception and Websites to Obtain Virtual Debit/Credit Cards for Online Accounts VerificationsNo EverandEvaluation of Some Websites that Offer Virtual Phone Numbers for SMS Reception and Websites to Obtain Virtual Debit/Credit Cards for Online Accounts VerificationsNota: 5 de 5 estrelas5/5 (1)
- Hacking: A Beginners Guide To Your First Computer Hack; Learn To Crack A Wireless Network, Basic Security Penetration Made Easy and Step By Step Kali LinuxNo EverandHacking: A Beginners Guide To Your First Computer Hack; Learn To Crack A Wireless Network, Basic Security Penetration Made Easy and Step By Step Kali LinuxNota: 4.5 de 5 estrelas4.5/5 (67)
- Microsoft Azure Infrastructure Services for Architects: Designing Cloud SolutionsNo EverandMicrosoft Azure Infrastructure Services for Architects: Designing Cloud SolutionsAinda não há avaliações
- Cybersecurity: A Simple Beginner’s Guide to Cybersecurity, Computer Networks and Protecting Oneself from Hacking in the Form of Phishing, Malware, Ransomware, and Social EngineeringNo EverandCybersecurity: A Simple Beginner’s Guide to Cybersecurity, Computer Networks and Protecting Oneself from Hacking in the Form of Phishing, Malware, Ransomware, and Social EngineeringNota: 5 de 5 estrelas5/5 (40)
- AWS Certified Solutions Architect Study Guide: Associate SAA-C02 ExamNo EverandAWS Certified Solutions Architect Study Guide: Associate SAA-C02 ExamAinda não há avaliações
- Microsoft Certified Azure Fundamentals Study Guide: Exam AZ-900No EverandMicrosoft Certified Azure Fundamentals Study Guide: Exam AZ-900Ainda não há avaliações
- Open Radio Access Network (O-RAN) Systems Architecture and DesignNo EverandOpen Radio Access Network (O-RAN) Systems Architecture and DesignAinda não há avaliações
- The Compete Ccna 200-301 Study Guide: Network Engineering EditionNo EverandThe Compete Ccna 200-301 Study Guide: Network Engineering EditionNota: 5 de 5 estrelas5/5 (4)
- Cybersecurity: The Beginner's Guide: A comprehensive guide to getting started in cybersecurityNo EverandCybersecurity: The Beginner's Guide: A comprehensive guide to getting started in cybersecurityNota: 5 de 5 estrelas5/5 (2)
- ITIL® 4 Create, Deliver and Support (CDS): Your companion to the ITIL 4 Managing Professional CDS certificationNo EverandITIL® 4 Create, Deliver and Support (CDS): Your companion to the ITIL 4 Managing Professional CDS certificationNota: 5 de 5 estrelas5/5 (2)
- Palo Alto Networks: The Ultimate Guide To Quickly Pass All The Exams And Getting Certified. Real Practice Test With Detailed Screenshots, Answers And ExplanationsNo EverandPalo Alto Networks: The Ultimate Guide To Quickly Pass All The Exams And Getting Certified. Real Practice Test With Detailed Screenshots, Answers And ExplanationsAinda não há avaliações
- Azure Networking: Command Line Mastery From Beginner To ArchitectNo EverandAzure Networking: Command Line Mastery From Beginner To ArchitectAinda não há avaliações
- Set Up Your Own IPsec VPN, OpenVPN and WireGuard Server: Build Your Own VPNNo EverandSet Up Your Own IPsec VPN, OpenVPN and WireGuard Server: Build Your Own VPNNota: 5 de 5 estrelas5/5 (1)
- ITIL 4 : Drive Stakeholder Value: Reference and study guideNo EverandITIL 4 : Drive Stakeholder Value: Reference and study guideAinda não há avaliações
- Digital Health: Mobile and Wearable Devices for Participatory Health ApplicationsNo EverandDigital Health: Mobile and Wearable Devices for Participatory Health ApplicationsShabbir Syed-AbdulAinda não há avaliações
- AWS Certified Cloud Practitioner Study Guide: CLF-C01 ExamNo EverandAWS Certified Cloud Practitioner Study Guide: CLF-C01 ExamNota: 5 de 5 estrelas5/5 (1)
