Laboratory exercise TFE4165 Applied Photonics

Version: September 6, 2007

Dispersion and Simulation of Fiber Optic Link

Room: A478 (Koherentlab)
Supervisor: ystein Mary, room A479, oystein.maroy@iet.ntnu.no

Motivation

The objective of this exercise is to study how chromatic dispersion affects the
performance of an optical high-speed network. We will use the simulation tool
OptSim to calculate common performance parameters in a basic fiber optic link. Thus,
we will also learn how to characterize a transmission system.


Before the Lab

Read through this text and get an overview of the exercise. A short introduction to
OptSim will be given in the lab.


Introduction

In an optical communication system we use the intensity of the light to represent the
digital data bits 1 and 0. Normally, light on means 1 and light off is 0. The data
rate or bit rate is the speed at which these bits are transmitted. Clearly, the temporal
width of the light pulses we use to represent the bits must not exceed the bit time
interval.

If we look at the wavelength spectrum of an optical pulse coming from a laser source
(Fig. 1), we see that the laser not only emits light at one wavelength but emits a small
continuous spectrum around a center wavelength !
0
. Unfortunately, the different
spectral components propagate through the fiber at different velocities due to a
wavelength dependent refractive index. Thus, the pulse will spread in time. This is
what we call chromatic dispersion. After propagating a certain length, the pulses will
exceed the bit time interval and we get intersymbol interference, as illustrated in Fig.
2.







Dispersion

The dispersion coefficient D(!) is a parameter describing the relation between the
initial spectral width "
w
of the pulse and the temporal pulse width "
t
at the fiber output
due to chromatic dispersion. Its unit is ps/(nm-km). This parameter tells us how many
picoseconds the pulse broadens per kilometer of fiber per nanometer of pulse spectral


Fig. 1: Wavelength spectrum of a laser pulse. !
0
is the center wavelength. The spectral
bandwidth "
w
is often denoted the full pulse width at half maximum value.


Fig. 2: Two pulses before and after propagation through fiber. The pulses are spread in time and will
eventually overlap (intersymbol interference).
width [1]:

(1)

L is the length of the fiber. An empirical formula used to estimate D(!) is [2]:

(2)

!
0
is the reference wavelength, i.e. the wavelength where the dispersion coefficient is
0, which is ~1312 nm for a standard single-mode fiber. S
0
is the dispersion slope,
which tells us how much the dispersion changes with wavelength. These parameters
are available from fiber specification sheets. OptSim uses Equation (2) to calculate
chromatic dispersion.

At common communication wavelengths around 1550 nm, the dispersion coefficient
is positive, which means that shorter-wavelength components travel faster than the
longer components. However, it is possible to produce fibers with customized
coefficient values by altering the material composition and waveguide design.


The Fiber Optic Link

We will consider a basic fiber optic link with a 1550 nm laser source, a standard
single-mode fiber and a receiver, as shown in Fig. 3. There are also two attenuators
called Normalizer and Attenuator for adjusting the launch power and the power
onto the receiver. Two monitors display the accumulated dispersion in the fiber and
the power at the receiver.



A PRBS generator is a common tool for characterizing network performance. PRBS is
short for Pseudo-Random Bit Sequence and is a sequence of 2
N
-1 data bits with a


Fig. 3: Basic fiber optic link with test and monitoring components.
random pattern of 1s and 0s. This sequence is transmitted repeatedly through the
system and a bit-error ratio (BER) tester at the receiver counts the number of
erroneously detected bits. The bit-error ratio is then the probability of detecting a 1
when a 0 is transmitted, and vice versa.

OptSim does not count the number of errors, as one would do when measuring BER
in a lab, but estimates the BER from the eye diagram in order to decrease the
simulation time. We will not delve into the details here but refer the interested student
to the OptSim manual [3]

In Fig. 3 the logical PRBS signal is transferred to an electrical component that
controls a direct modulated laser. Thus, the data signal is converted into an optical
signal where no light means logical 0 and presence of light means 1. The BER tester
at the receiver end is synchronized to the PRBS generator.

A receiver needs a minimum optical input power in order to keep BER below a given
maximum value [4]. By coupling the light from the laser directly into the receiver
(except for a variable attenuator between the laser and the receiver), we can measure
BER versus power. This is a back-to-back measurement, which is the reference to
which we compare all other measurements. The leftmost graph in Fig. 4 shows a
back-to-back measurement. Inserting fibers, amplifiers and other components into the
transmission path will introduce impairments like noise and dispersion, and we need
to increase the power in order to maintain the same BER. The value of this power
increase is termed the power penalty and is often defined at BER=10
-9
. The power
required to achieve BER=10
-9
is called the sensitivity. The second plot in Fig. 4 is the


Fig. 4: BER curves used to characterize an optical link. The leftmost line is a
back-to-back measurement while the rightmost line shows the BER for the total
system
BER of the complete system, and in this case there is a power penalty of 3 dB.




The last monitor in the system, Eye_Diagram, is an oscilloscope that displays a
superposition of many bits. This produces an eye like diagram (see Fig. 5) in which
the eye opening is a measure of the degradation of the link. An open eye with sharp
lines means good performance, while noise and intersymbol interference appears as
spreading of the rails.


Exercises

We will in the following exercises study what impact chromatic dispersion has on an
optical transmission system with a bit rate of 10 Gbps per channel. A widely deployed
bit rate in commercial systems today is 2.5 Gbps, with 10 Gbps as the next step up.
We will use a PRBS sequence with N=9. S
0
of a standard fiber is ~ 0.090 ps/(nm
2
km).


A: Back-to-back

First of all we need to characterize our system without any components between the
transmitter and the receiver, i.e. a back-to-back measurement. We will calculate BER
versus received power and use this to evaluate performance of our transmission
system later on.

1. Open the back2back.moml file and look at the different network components. By
right-clicking the blocks, one can access and set the properties of the components
or view the simulation results. We will not go into details of each component here,


Fig. 5: Eye diagram as calculated by OptSim. The top line is the signal
level of 1s, while the bottom line is the signal level of 0s.
so most of the properties have already been set. There is a 1 m long fiber between
the transmitter and the receiver.
2. Run a simulation by pushing the F5 key and clicking OK in the dialog box that
opens. You can watch the status of the simulation in the lower left corner of the
OptSim window.
3. From the eye diagram (right-click Eye_Diagram and choose View Results) we
can see that there is a distinct difference between the zero level and the one level.
The BER calculation (right-click BER-Tester, choose View Results and
double-clik the name of the text file) shows error-free transmission, which is as
expected.
4. Start the back-to-back simulation by pushing the F4 key and clicking OK in the
dialog box. OptSim will now vary the attenuation of the attenuator in front of the
receiver and calculate the BER.
5. Use the values of Optical_Monitor and BER-Tester to plot BER as a function
of received power. What is the sensitivity of the receiver (i.e. the received power
giving BER = 10
-9
)?


B: Dispersion

1. Open the file called fiberdispersion.moml. We now have a 100 km long fiber as
our transmission line.
2. Push F5 and run a simulation. Look at the eye diagram. We can clearly see the
effect of dispersion as the bits are starting to interfere with each other and close the
eye. But still the BER is low.
3. Push F4 and run the simulation. As before, use Optical_Monitor and
BER_Tester to find BER as a function of received power. Plot BER in the same
chart as the back-to-back graph in exercise A.
4. What is the sensitivity of the system? What is the penalty?
5. What is the maximum transmission length, given that BER should be better than
10
-9
?
6. Change the bit rate to 2.5 Gbps (right-click PRBS_Generator, choose
Properties and under General change bitrate from 10e9 to 2.5e9.
7. Repeat step 3-5 and compare the results.


C: Dispersion compensation

1. Open the file called compensation.moml. A second fiber called DCF is added
to the link. This is a dispersion compensating fiber designed to have a large,
negative dispersion coefficient. The length of the standard fiber is now 80 km
while the DC fiber is 20 km long.
2. Repeat steps 2 through 4 from exercise B. Compare the results. You can see how
the dispersion is accumulated and compensated by double-clicking
Dispersion_Monitor and then double-clicking the dispersion plot from the list.
3. What is the maximum transmission length in this case?


Discussion

The bandwidth "
w
of the 10 Gbps data signal is approximately 0.07 nm. Find "
t
and
compare to the bit time interval. Are the simulation results OK?


References

[1] B. E. A. Saleh, M. C. Teich: Fundamentals of Photonics, John Wiley & Sons,
Inc., 1991.
[2] J. M. Senior: Optical Fiber Communications Principles and Practice, Prentice
Hall, second edition, 1992.
[3] OptSim 4.0 Models Reference Volume II Block Mode, RSoft Design Group,
2004.
[4] G. P. Agrawal: Fiber-Optic Communication Systems, John Wiley & Sons, Inc.,
second edition, 1997.