MULTIBAND AND SILICON INTEGRATED ANTENNAS FOR WIRELESS SENSOR NETWORKS

Multiband and Silicon Integrated Antennas for Wireless Sensor Networks

Ph.D. Dissertation Joan Gemio Valero

J. Gemio
Supervisor: Josep Parrn Granados Department of Telecommunications and Systems Engineering July, 2011

Multiband and Silicon Integrated Antennas for Wireless Sensor Networks

by

Joan Gemio Valero

Ph.D. Dissertation

Supervisor: Josep Parrn Granados Antenna and Microwave Systems Group Department of Telecommunications and Systems Engineering

Barcelona, July 2011

Abstract

Wireless technologies have triggered an important revolution in the world of telecommunication networks and services. New systems, such as the latest generation mobile phones or wireless LANs, have appeared being enthusiastically accepted by people. In particular, applications related to remote sensor networks are growing rapidly (for temperature monitoring, the detection of free parking spaces, etc.). This thesis is focused on antenna design for distributed wireless sensor networks. The remote sensors we are considering are connected to a central unit using a wireless link in the 5.8 GHz ISM band and the central unit transmits the collected information to the Internet via a WLAN connection. In the central unit multiband antennas are necessary to cover several frequency bands with a single radiating element. Two solutions to obtain this multiband performance are proposed: fractal-based ground planes and resonator loaded monopoles. Several novel antennas that easily meet the system requirements are manufactured and measured. For remote sensors small dimensions and low costs are desired. The use of silicon integrated antennas fabricated with standard CMOS technology is a good choice to achieve these requirements. First, the main problems of antenna integration are discussed and, then a possible solution to improve the antenna eciency is presented together with a study of the eects of other elements integrated in the same chip with the antenna. The measurements of the manufactured prototypes demonstrate that antenna integration at 5.8 GHz is feasible, and that system ranges of some meters can be obtained even when the antenna is fabricated together with other elements integrated in the same chip. The possibility of having some of the sensors implanted inside the body will be a reality in the near future, for this reason the particular case of implanted antennas is also analyzed. The last part of this thesis is devoted to study the body eects on antenna performance and evaluate the additional propagation losses introduced by the body tissues.

Agra ments

En aquesta vida tot t e un inici i un nal. Ara ha arribat un dels moments m es esperats daquesta tesis, el moment descriure els agra ments, fet que indica que aquest treball ha arribat a la seva culminaci o. A m es es locasi o perfecte per donar les gr` acies a totes aquelles persones que mhan acompanyat durant aquests 4 anys. Gr` acies Josep (director daquesta tesis) per donar-me loportunitat de realitzar aquest treball i per poder gaudir de la teva ajuda sempre que lhe necessitat. Gr` acies tamb e als altres membres del grup AMS (Pedro, Oscar, Gary), per totes les aportacions que heu fet durant aquest per ode. Ram on, tot i que ja no est` as en el grup, no moblido pas de tu. Jordi Soler, et vull agrair tots els comentaris que has anat fent per millorar el treball realitzat. Felix i Jordi, gr` acies per totes les antenes fabricades. Gr` acies a tots els professors de la ETSE, ja que sense el que ells em van ensenyar no hauria pogut fer aquesta tesis. Tamb e vull agrair lajuda rebuda per part de membres daltres institucions: a Jose Maria Gonz alez Arbes u (UPC) per ajudar-me a fer les mesures dels diagrames de radiaci o i a Jordi Sacrist an (CNM) i Toni Baldi (CNM), per fer possible que pogu es fabricar les antenes integrades en silici. Gr` acies a tots els company del despatx (M onica, Jordi, Eden, Rafa, Mercedes, Verd u) i als que no esteu en el despatx (Jose, Toni) per haver fet m es entretinguts els dies que ens hem passat aqu tancats treballant. Algun@s de vosaltres heu estat molt m es que simples company@s de feina, i espero que aquesta amistat perduri en el futur. A tota la gent que fora de la universitat mha donat suport i mha ajudat a gaudir daquest per ode, tamb e els vull dir gr` acies. Gr` acies Monica, Oscar, Jordi, Laia, i gr` acies tamb e a tots els del Patinet (Anna, August, Canari, David, Jaume, Jordi, ......) per haver compartit amb mi molts caps de setmana, ja sigui sobre dues o quatre rodes. Pare, mare i avis, a vosaltres us deixo pel nal. Vosaltres em vareu motivar a iniciar aquesta tesis i sempre heu estat presents en aquest llarg cam . Heu suportat el meu mal humor quan les coses no anaven com jo volia i sempre heu estat all` a, animant-me a tirar endavant i a arribar al nal daquest cam que vaig iniciar ja fa un bon temps. Moltes gr` acies per tot, i recordeu que aquesta tesis va dedicada a vosaltres !!!! Joan Gemio Valero, Juliol 2011

This work has been supported by the Spanish Ministry of Science and Innovation through grants TEC2006-13248-C04-01 and TEC2009-13897-C03-02, and by the Spanish Ministry of Education through the FPU program. Silicon integrated antennas were manufactured in the integrated clean room for micro and nano manufacturing of Instituto de Microelectr onica de Barcelona - Centro Nacional de Microelectr onica (IMB-CNM) supported by the Spanish Ministry of Science and Innovation through the GICSERV program ref. 2009-NGG 151 and ref. 2010-NGG 183. Radiation pattern measurements were carried out at the anechoic chamber of the Antenna Lab group of the Universitat Politecnica de Catalunya.

Contents

1 Introduction 1.1 Wireless communication systems: New trends and use of the ISM bands . . 1.1.1 1.1.2 1.2 New trends . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Use of the ISM bands . . . . . . . . . . . . . . . . . . . . . . . . . .

1 1 1 2 4 4 5 7 8 11

Motivation and focus of this thesis . . . . . . . . . . . . . . . . . . . . . . . 1.2.1 1.2.2 Thesis motivation and system description . . . . . . . . . . . . . . . Focus of the thesis and objectives . . . . . . . . . . . . . . . . . . . .

1.3 1.4

Thesis organization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . List of Authors Contributions . . . . . . . . . . . . . . . . . . . . . . . . .

2 Multiband antennas for the central unit 2.1 2.2

Introduction to multiband antennas . . . . . . . . . . . . . . . . . . . . . . 11 Multiband antennas inspired by fractal geometries . . . . . . . . . . . . . . 15 2.2.1 2.2.2 2.2.3 Introduction to the use of fractal shapes in antenna design . . . . . 16

The Sierpinski monopole . . . . . . . . . . . . . . . . . . . . . . . . . 19 Use of the Sierpinski gasket in the ground plane . . . . . . . . . . . 26 . . . . . . . . . . . . . . . 37

2.3

Resonator loaded multiband monopole antennas 2.3.1 2.3.2 2.3.3 2.3.4

Introduction to resonator loaded antennas . . . . . . . . . . . . . . . 37 Dual band resonator loaded monopole antenna . . . . . . . . . . . . 40 Triple band SRR loaded monopole antenna . . . . . . . . . . . . . . 46 Triple band resonator loaded monopole antenna based on the use of an asymmetric structure . . . . . . . . . . . . . . . . . . . . . . . . . 52 i

ii 2.4

Contents Chapter conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59 63

3 Small silicon integrated antennas for the remote sensors 3.1 3.2 3.3

Introduction to silicon integrated antennas . . . . . . . . . . . . . . . . . . . 63 Prototype antennas fabrication and measurement . . . . . . . . . . . . . . . 68 Miniaturization techniques applied to low resistivity silicon integrated antennas . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74 3.3.1 3.3.2 Antenna designs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74 Miniaturization study: Measurements and simulation results . . . . . 77

3.4

Improving silicon integrated antennas by substrate micromachining . . . . . 81 3.4.1 3.4.2 3.4.3 Antenna designs, denition of the etching areas and fabrication process 81 Performance improvement due to the micromachining process: Measurements and simulation results . . . . . . . . . . . . . . . . . . 83 Antenna performance improvement due to micromachining in a free space scenario . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87

3.5

Antenna performance degradation due to the presence of coils . . . . . . . . 88 3.5.1 3.5.2 3.5.3 Performance degradation due to the coil eects: measurements and simulation results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89 Performance degradation due to the coil eects in a free space scenario 91 Eects of the performance degradation in a real system link budget 92

3.6

Chapter conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93 95

4 Human body eects on antennas performance 4.1

Introduction to the human body eects . . . . . . . . . . . . . . . . . . . . 95 4.1.1 4.1.2 Antennas in the vicinity of the human body . . . . . . . . . . . . . . 95 Antennas implanted inside the human body . . . . . . . . . . . . . . 98

4.2

Human body models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103 4.2.1 4.2.2 Simulations with human body models . . . . . . . . . . . . . . . . . 104 Measurements using human body phantoms . . . . . . . . . . . . . . 109

4.3

Study of the propagation losses introduced by body tissues . . . . . . . . . 110 4.3.1 4.3.2 4.3.3 Free space wireless link . . . . . . . . . . . . . . . . . . . . . . . . . 110 One layer model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111 Three layer model . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113

Contents 4.3.4 4.3.5 4.4 4.5

iii Eects of the layers thickness . . . . . . . . . . . . . . . . . . . . . . 117 Eects of a coating layer . . . . . . . . . . . . . . . . . . . . . . . . . 118

In-body antenna measurements with body tissue simulating liquids . . . . . 119 Chapter conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 124 127

5 Conclusions and future work 5.1 5.2

Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 127 Future work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 130 133

References

List of Figures

1.1

Sensor network with a central unit that collects the sensors information and sends it to the Internet using a WLAN connection. The central unit and the remote sensors are connected using a wireless link working in the 2.45 GHz or 5.8 GHz ISM bands. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

2.1

Classic broadband and multiband antennas. (a) Helix antenna. (b) Biconical antenna. (c) Log-periodic antenna. (d) Spiral antenna. . . . . . . . . . . . . 12 Multiband antennas presented in recent years. (a) Monopole with stub tuning [28]. (b) Antenna based on dielectric resonators [32]. (c) PIFA Antenna [35]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14 Top-loaded triangular monopole with dierent load numbers. (a) Antenna geometry. (b) Antenna input reactance. . . . . . . . . . . . . . . . . . . . . 15 Modied top-loaded triangular monopole. The width w has been increased to improve the matching levels. (a) Fabricated prototype. h1 = 50 mm, h2 = 41.7 mm, h3 = 33.3 mm, hl = 25 mm, s = 2 mm, w = 25 mm. (b) Measured and simulated input reection coecient. . . . . . . . . . . . . . . 16 An example of fractal shape in nature: Cauliower [38]. . . . . . . . . . . . 17 Generation of dierent fractal shapes using an iterative process. (a) Sierpinski gasket. (b) Sierpinski carpet. (c) Koch curve. . . . . . . . . . . . 18 Sierpinski monopole antenna with dierent numbers of iterations. h1 = 100 mm, = 60o and hn+1 = hn /2. . . . . . . . . . . . . . . . . . . . 20 Normalized current distributions at the odd resonances of a Sierpinski monopole with 4 fractal iterations. . . . . . . . . . . . . . . . . . . . . . . . 21 Radiation patterns at the odd resonances of the Sierpinski monopole with dierent are angles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23 v

2.2

2.3

2.4

2.5 2.6

2.7

2.8

2.9

vi

List of Figures 2.10 Sierpinski monopole perturbations to control the spacing between resonances: (a) The scale factor sf = hn /hn+1 can be changed (b) or dierent Mod-p structures can be used. . . . . . . . . . . . . . . . . . . . . 24 2.11 Top-loaded Sierpinski monopole. h11 = 300 mm, h21 = 150 mm, h12 = 200 mm, h22 = 100 mm, = 60o and t = 2 mm. . . . . . . . . . . . 24 2.12 Input reactance of an asymmetric top-loaded Sierpinski monopole (gure 2.11). The input reactance of the two isolated halves and the reactance of the parallel of the two halves impedance are also plotted. . . . . . . . . . . 25 2.13 Input parameters of an asymmetric top-loaded Sierpinski monopole (scaled prototype with h11 = 90 mm). The reference impedance for the S11 parameter calculation is 170 instead of 50 . . . . . . . . . . . . . . . . . 26 2.14 Radiation patterns at the even resonances of an asymmetric top-loaded Sierpinski monopole (gure 2.11). Black dotted lines are used for the copolar component; red dashed lines are used for the crosspolar component. . . . . 27 2.15 Triangular monopole antenna on a nite dimensions ground plane. (a) Solid ground plane. (b) Fractal-based ground plane after 1 iteration. (c) Fractalbased ground plane after 2 iterations. (d) Fractal-based ground plane after 3 iterations. The triangular monopole height is 50 mm and the ground plane dimensions are 200 mm x 200 mm. . . . . . . . . . . . . . . . . . . . . . . . 28 2.16 Input parameters of a triangular monopole on a fractal-based ground plane with dierent number of iterations (gure 2.15). . . . . . . . . . . . . . . . . 29 2.17 Triangular monopole on a fractal-based ground plane. (a) Monopole element. (b) Modied ground plane inspired by the Sierpinski gasket. . . . . 30 2.18 Antenna input parameters. (a) Triangular monopole on a solid ground plane. (b) Triangular monopole on a fractal-based ground plane. . . . . . . 31 2.19 Simulated normalized radiation patterns. Black dotted lines are used for the copolar component, red dashed lines are used for the crosspolar component. (a) Triangular monopole on a solid ground plane. (b) Triangular monopole on a fractal-based ground plane. . . . . . . . . . . . . . . . . . . . . . . . . 33 2.20 Current distributions in the ground plane. The area with higher current density levels (active area) has been marked with a dashed circle. . . . . . . 34 2.21 The monopole element is a two iteration Sierpinski gasket with dierent are angles . Antennas were fabricated on a low cost FR4 substrate ( r = 4.5, tan = 0.02). hm = 50 mm. . . . . . . . . . . . . . . . . . . . . . . . . . . . 34 2.22 Measured input parameters of a Sierpinski monopole with dierent are angles on a fractal-based ground plane. . . . . . . . . . . . . . . . . . . . 35 2.23 A small gap g between the ground plane and the monopole can be used to optimize the antenna performance and improve the matching levels. . . . . 36

List of Figures

vii

2.24 Measured input parameters of a Sierpinski monopole with = 75o mounted on a fractal-based ground plane with dierent gap sizes g (gure 2.23). . . . 36 2.25 Simulated normalized radiation patterns. Black dotted lines are used for the copolar component, red dashed lines are used for the crosspolar component. (a) Sierpinski monopole on a solid ground plane. (b) Sierpinski monopole on a fractal-based ground plane. . . . . . . . . . . . . . . . . . . . . . . . . 38 2.26 Schematic representation of a trap dipole antenna [66]. . . . . . . . . . . . . 39 2.27 Monopole antenna loaded with two double layer magnetic resonators DLCLL [67] of identical dimensions fabricated on Rogers RO4003C [68] double sided 0.508 mm substrate ( r = 3.38, tan = 0.0021). . . . . . . . . . . . . 40 2.28 Parametric study of the variation of the input reection coecient S11 with respect to (a) the monopole length LM , (b) the resonator width RW , (c) the length of the resonator capacitor strip RCL . . . . . . . . . . . . . . . . . 41 2.29 Parametric study of the variation of the input reection coecient S11 with respect to (a) the monopole width WM and (b) the resonator oset position RO . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42 2.30 Simulated and measured S11 parameters of the prototype antenna with the dimensions shown in table 2.3. . . . . . . . . . . . . . . . . . . . . . . . . . 44 2.31 Simulated current distributions at 2.45 GHz, 4.9 GHz and 5.875 GHz. Each image is individually normalized and displayed on a 30 dB scale. . . . . . . 44 2.32 Simulated radiation patterns of the prototype antenna whose dimensions are shown in table 2.3. Black dotted lines are used for the copolar components. Crosspolar components are too low to be plotted with the current axis conguration. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45 2.33 Radiation eciency and total eciency of the prototype whose dimensions are shown in table 2.3. Eciency measurements were carried out using the method presented in [69]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46 2.34 SRR geometry and impinging plane wave for computation of the monostatic radar cross section. The SRR is supported on Rogers RO4003C 0.508 mm substrate (r = 3.38, tan = 0.0027). L1 = 8.2 mm, L2 = 6.6 mm, w1 = w2 = 0.5 mm, g1 = g2 = 1.5 mm, t = 0.3 mm. . . . . . . . . . . . . . 47 2.35 Computed monostatic radar cross section in free space for the SRR of gure 2.34. Solid line: stand alone outer open ring. Dashed line: stand alone inner open ring. Dotted line: SRR. . . . . . . . . . . . . . . . . . . . . . . . . . . 48 2.36 Simulated current distributions. Each image is individually normalized and displayed on a 20 dB scale. Arrows are scaled according to the normalized current levels. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48

viii

List of Figures 2.37 SRR loaded monopole. The structure was fabricated on Rogers RO4003C 0.508 mm substrate (r = 3.38, tan = 0.0027). Lm = 22 mm, wm = 6 mm, L0 = 1 mm, w0 = 1 mm, h = 2.5 mm, s = 0.3 mm. SRRs dimensions are the same as in gure 2.34. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49 2.38 Simulations comparing the stand alone monopole and the SRR loaded monopole. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50 2.39 Simulated radiation eciency of the stand alone monopole and the SRR loaded monopole. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51 2.40 SRR loaded monopole prototype antenna. (a) Prototype antenna mounted on a large ground plane. (b) Simulated and measured S11 parameters. (c) Simulated and measured radiation eciency. . . . . . . . . . . . . . . . . . 51 2.41 Measured normalized radiation patterns of the SRR loaded monopole. Each cut has been normalized with reference to its maximum. Black dotted lines are used for the copolar component; red dashed lines are used for the crosspolar component. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53 2.42 Fabricated prototype. In the rear side a conductor strip is used to reduce the resonance frequency of the large resonator. . . . . . . . . . . . . . . . . 54 2.43 Parametric study of the variation of the S11 parameter with respect to (a) the length of the monopole LM , (b) the width of the monopole WM , (c) (d) the height of the resonators HR1 , HR2 , (e) the length of the rear side strip LS and (f) the width of the resonators WR . . . . . . . . . . . . . . . . . . . 55 2.44 Simulated current distributions at the S11 minimums. Each image is individually normalized and displayed on a 30 dB scale. . . . . . . . . . . . 56 2.45 Measured and simulated S11 parameter of the asymmetric loaded monopole antenna. Blue solid line shows simulated results, and red dashed line shows measured results. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57 2.46 Measured and simulated results for the radiation eciency and the total eciency. Total eciency takes into account the mismatch loss. . . . . . . . 58 2.47 Measured radiation patterns of the asymmetric loaded monopole in the three operating bands. Each cut has been normalized with reference to its maximum. Black dotted lines are used for the copolar component; red dashed lines are used for the crosspolar component. . . . . . . . . . . . . . . 58 3.1 Example of intra-chip communication system. (a) Conceptual system illustration of intra-chip wireless interconnected system for clock signal distribution. (b) 0.18 m clock transmitter and receiver with zigzag dipole antennas [78]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64

List of Figures 3.2

ix

Examples of integrated antennas operating at frequencies higher than 60 GHz. (a) Array antenna fabricated on high resistivity silicon presented by Buechler in 1986 [79]. (b) Folded slot antenna designed at 79 GHz [80]. (c) and (d) 60 GHz on-chip Yagi antennas [81, 82]. . . . . . . . . . . . . . . 65 Integrated antenna presented by Lin in [83]. (a) Layout of the 2 mm zigzag dipole. (b) Antenna measurement setup with a signal generator and a spectrum analyzer. (c) Antenna pair gain as a function of the distance in three dierent environments at 24 GHz. The inset shows the 2 mm zigzag dipole antenna radiation pattern. . . . . . . . . . . . . . . . . . . . . . . . . 66 Fully integrated transmitter with an on-chip dipole antenna at 24 GHz [91]. (a) Photomicrograph of the fully integrated transmitter and frequency synthesizer. (b) Reception of the signal generated by a transmitter IC with an on-chip antenna using a 20 dBi gain horn antenna. . . . . . . . . . . . . 67 Antenna pair gain as a function of the distance of a 5.8 GHz monopole and a 24 GHz dipole antenna. The antennas are located 52 cm and 5 mm above the ground [75]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67 Simplied metal/dielectric layer structure of the fabricated prototypes. It is equivalent to a standard 0.18 m RF CMOS technology. . . . . . . . . . 69 Antenna input parameters and radiation eciency with and without the SiO2 layer. The silicon conductivity is = 5 S/m. The antenna is a meandered dipole and the chip dimensions are 5.25 mm x 1.9 mm. . . . . . 70 Eects of the silicon conductivity in antenna parameters. Input reection coecient, input impedance and eciency are shown. The antenna is a meandered dipole and chip dimensions are 5.25 mm x 1.9 mm. . . . . . . . 71 Silicon antennas measurement setup. (a) A 1 cm PVC support was used to separate the antenna from the metallic chuck. (b) The dierential dipole antennas were fed with a Picoprobe GSGSG probe. . . . . . . . . . . . . . . 72

3.3

3.4

3.5

3.6 3.7

3.8

3.9

3.10 Propagation measurement setup used for the gain measurement of the fabricated prototypes. Gain measurements were carried out in the 5.8 GHz ISM band. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73 3.11 Dierent components used for the propagation measurements. (a) Signal generator with a balun. (b) Receiving patch antenna fabricated on Rogers RO4003C substrate (r = 3.38, tan = 0.0027). . . . . . . . . . . . . . . . . 74 3.12 Input parameters of a meandered dipole antenna fabricated on silicon. Solid line shows the simulated results. Dashed lines show the measured results of the same antenna fabricated on dierent wafers. A 100 impedance was used as dierential reference impedance. . . . . . . . . . . . . . . . . . . . . 75 3.13 Topologies included in the miniaturization study. . . . . . . . . . . . . . . . 76 3.14 Measured and simulated results of a meandered dipole (gure 3.13) with N = 8 meanders, conductor width = 100 m and L1 = 9.2 mm. . . . . . . 76

List of Figures 3.15 Antennas with dierent miniaturization techniques. The chip dimensions are 12.5 mm x 2 mm. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77 3.16 Results of the miniaturization study in the probe station scenario. (a) Measured miniaturization factor (M F ). (b) Simulated total eciency of the dierent topologies. N is the number of meanders. The dimensions of the other variables (dened in gure 3.13) are in mm. The reference antenna for the M F calculation is a simple at dipole with Ff latdipole = 68 GHzmm. 78 3.17 Results of the miniaturization study in a free space scenario. (a) Simulated miniaturization factor (M F ) (b) Simulated total eciency of the dierent topologies. N is the number of meanders. The dimensions of the other variables (dened in gure 3.13) are in mm. The reference antenna for the M F calculation is a simple at dipole with Ff latdipole = 89 GHzmm. . . . 79 3.18 Results of the miniaturization study in the probe station scenario when antennas are fabricated on a low-loss substrate (RO3010, r = 10.2, tan = 0.0023). (a) Simulated miniaturization factor (M F ). (b) Simulated radiation eciency of the dierent topologies. N is the number of meanders. The dimensions of the other variables (dened in gure 3.13) are in mm. The reference antenna for the M F calculation is a simple at dipole with Ff latdipole = 75.5 GHzmm. . . . . . . . . . . . . . . . . . . . . 80 3.19 Design of the fabricated prototype antennas. (a) Image of the fabricated meandered dipole antenna. (b) Schematic representation of the etching patterns used for the micromachining process. Black rectangles indicate the areas where silicon is etched. . . . . . . . . . . . . . . . . . . . . . . . . 82 3.20 Prototype antennas with micromachining. (a) Top view of one of the antennas. (b) Rear view of some of the antennas. . . . . . . . . . . . . . . . 83 3.21 Simulated input reection coecient. (a) Dierent etching congurations are applied. The etching depth is 450 m. (b) Dierent etching depths (m) are applied. The M2 pattern is used for the micromachining process. The wafers thickness is 500 m. WM = Without Micromachining. . . . . . . . . 84 3.22 Simulated radiation parameters when micromachining techniques are applied. (a) Radiation eciency as a function of frequency. The etching depth is 450 m. (b) and (c) radiation eciency and gain improvement as a function of the etching depth. The operating frequency is 5.8 GHz and the wafers thickness is 500 m. The reference gain for the gain improvement plot is the gain of the antenna without micromachining (18.95 dB). WM = Without Micromachining. . . . . . . . . . . . . . . . . . . . . . . . . 85 3.23 Simulated gain improvement as a function of the etched area position. (a) Pattern used in the micromachining process. (b) Gain improvement plot. The reference gain is the gain of the antenna without micromachining (18.95 dB). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87

List of Figures

xi

3.24 Performance improvement due to micromachining in a free space scenario. (a) Simulated input reection coecient. The etching depth is 450 m. (b) Gain improvement as a function of the etching depth. The operating frequency is 5.8 GHz and the wafers thickness is 500 m. WM = Without Micromachining. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88 3.25 Fabricated prototype antenna for the evaluation of the coil eects on the RF link. The chip dimensions are 3.8 mm x 5.2 mm. . . . . . . . . . . . . . 89 3.26 Prototype antennas. The chip dimensions are 3.8 mm x 5.2 mm. The distances d between the RF antenna and the coil are 320 m, 570 m, 820 m, 1.07 mm, 1.32 mm and 1.57 mm. . . . . . . . . . . . . . . . . . . . 89 3.27 Measured input reection coecient of an integrated antenna fabricated together with a coil for an inductive link (gure 3.25). The distance between the antenna and the coil is d = 320 m. . . . . . . . . . . . . . . . . . . . . 90 3.28 Radiation parameters as a function of the distance d between the RF antenna and the coil. (a) Antenna gain. The simulated gain without the coil is 17.7 dB. (b) Gain degradation. The reference gain is the gain when the coil is at d = 1.57 mm. Operating frequency f = 5.8 GHz. . . . . . . . 90 3.29 Simulated gain degradation in a free space scenario as a function of the distance d between the RF antenna and the coil. The reference gain (25.5 dB) is the gain when the coil is at d = 1.57 mm. Operating frequency f = 5.8 GHz. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92 4.1 4.2 Recent implantable cardiac debrillators provide in-home monitoring via wireless base stations that relay data to doctors [136]. . . . . . . . . . . . . 99 Medtronic oers a wide catalog of solutions for the treatment of chronic diseases [20]. (a) Real time glucose monitoring system. (b) Implantable cardioverter debrillator with wireless telemetry system. (c) Implantable cardiac monitor. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100 Antennas for implantable devices. (a) Microstrip antenna. [139]. (b) PIFA antenna. [140]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101 Examples of antennas operating in the 2.45 GHz ISM band. (a) PIFA antenna for an intracranial pressure monitoring system [104]. (b) Cavity slot antenna to be implanted in the patient arm [148]. (c) Dual band antenna for the MICS and ISM bands [144]. . . . . . . . . . . . . . . . . . . . . . . . 102 One layer model used by K. Ito for the antenna simulation inside the human body [148]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104 The human body is an inhomogeneous structure. (a) MRI image of the thesis author. (b) Cut of the FEKO inhomogeneous body model. The simulation model takes into account most real body details. . . . . . . . . . 106

4.3 4.4

4.5 4.6

xii 4.7

List of Figures Examples of human body models found in literature. (a) 3D human body in dierent positions [164]. (b) Cut of an inhomogeneous body model [168]. (c) 3D inhomogeneous head model [139]. . . . . . . . . . . . . . . . . . . . . 107 Radiation pattern of a dipole antenna in front of a human body at a distance of 10 cm. The inhomogeneous FEM model was used, but the same results were obtained with the homogeneous one. A 30 dB dynamic range has been used for the 3D far eld plot. . . . . . . . . . . . . . . . . . . . . . . . . . . 108 Dasy 5 measurement setup. It is designed for SAR measurements, but it could be also used for the measurement of implanted antennas [166]. . . . . 109

4.8

4.9

4.10 One layer scenario. The transmitting and receiving antennas are half wavelength resonant dipoles. The equivalent tissue properties are shown in table 4.2. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112 4.11 Simulation results for the one layer model (gure 4.10). (a) Antenna pair gain. (b) Implanted dipole length to keep the antenna resonant at the desired frequency. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112 4.12 Three layer (skin - fat - muscle) model simulation scenario. As in the previous scenario, the transmitting and receiving antennas are half wavelength resonant dipoles. The body tissues properties are shown in table 4.3. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113 4.13 Simulation results with the three layer model (gure 4.12). Yellow lines show the theoretical results. (a) Antenna pair gain. (b) Implanted antenna gain. (c) Air propagation gain as a function of the implanted antenna depth. (d) Air propagation gain as a function of the distance to the external antenna. (e) Implanted dipole length to keep the antenna resonant at the desired frequency. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115 4.14 Coverage maps showing the antenna pair gain (Ga) at dierent frequencies as a function of the implanted antenna depth and the distance to the external antenna. Coverage maps are useful to predict the system performance as a function of the frequency band and the antennas position. 116 4.15 Eects of the fat layer thickness. Yellow lines show the results obtained with the one layer model. Operating frequency f = 915 MHz. (a)Implanted antenna gain. (b) Implanted dipole length to keep the antenna resonant at the desired frequency. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 117 4.16 Eects of the skin layer thickness. Yellow lines show the results obtained with the one layer model. Operating frequency f = 915 MHz. (a) Implanted antenna gain. (b) Implanted dipole length to keep the antenna resonant at the desired frequency. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 118 4.17 Eects of a coating layer on the antenna performance. Operating frequency f = 2450 MHz. (a) Implanted antenna gain. (b) Implanted dipole length to keep the antenna resonant at the desired frequency. . . . . . . . . . . . . . . 119

List of Figures

xiii

4.18 Microstrip patch antenna with the feeding point at the edge of the patch to improve the matching levels when it is immersed in liquid. (a) Patch antenna without superstrate. (b) Patch antenna with superstrate. . . . . . . 121 4.19 Measurement setup. The antenna under test was immersed in liquid. The depth can be easily changed by adjusting the liquid level. A ruler was used to measure the antenna depth. . . . . . . . . . . . . . . . . . . . . . . . . . 121 4.20 Input reection coecient of a microstrip patch with superstrate (gure 4.18(b)) when the antenna depth is 8 mm. The measurement setup is shown in gure 4.19. (a) Antenna in distilled water. (b) Antenna in the SPEAG body tissue simulating liquid MSL 2450 [166]. . . . . . . . . . . . . . . . . 122 4.21 Measurement setup used to carry out propagation measurements. The receiving antenna was a microstrip patch with superstrate designed to operate in the desired frequency band (2.45 GHz ISM band) in free space. The distance between both antennas was 1.25 m. The implanted antenna depth was modied by changing the liquid level. . . . . . . . . . . . . . . . 123 4.22 Antenna gain in the normal direction of the microstrip patch with superstrate (gure 4.18(b)) as a function of the depth. The measurement frequency is near the 2.45 GHz ISM band. (a) Antenna in distilled water. (b) Antenna in the SPEAG body tissue simulating liquid MSL 2450 [166]. . 124

List of Tables

1.1

Frequency bands for ISM applications dened by the Federal Communications Commission (FCC) [12]. 1 The use of these frequency bands for ISM applications shall be subject to special authorization. 2 Only in Region 2. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

2.1 2.2 2.3 2.4 2.5

Measured operating bands of the triangular monopole on the dierent ground plane congurations. . . . . . . . . . . . . . . . . . . . . . . . . . . . 32 Measured operating bands of the Sierpinski monopole on the dierent ground plane congurations. . . . . . . . . . . . . . . . . . . . . . . . . . . . 37 Dimensions of the dual band WLAN monopole antenna. . . . . . . . . . . . 43 Measured operating bands of the dual band resonator loaded monopole antenna (gure 2.27) with the dimensions of table 2.3. . . . . . . . . . . . . 43 Measured operating bands of the triple band SRR loaded monopole antenna (gure 2.40(a)). The second and the third bands have been combined obtaining a broadband matching frequency range from 3 GHz to 6.2 GHz. . 52 Dimensions of the asymmetric prototype antenna. . . . . . . . . . . . . . . 57 Measured operating bands of the triple band asymmetric resonator loaded monopole antenna (gure 2.42). . . . . . . . . . . . . . . . . . . . . . . . . . 57 Radiation eciency and additional losses due to miniaturization of a 100 m meandered dipole when the number of meanders is increased. . . . . . . . . 80 Measured and simulated gain improvement for the dierent etching patterns. The last column shows the etched area. The etching depth is 450 m and the wafers thickness is 500 m. . . . . . . . . . . . . . . . . . . 86 Measured gain degradation and system range reduction as a function of the distance d between the RF antenna and the coil. . . . . . . . . . . . . . . . 91 System parameters used for the link budget. . . . . . . . . . . . . . . . . . . 92 xv

2.6 2.7

3.1 3.2

3.3 3.4

xvi 3.5 4.1 4.2 4.3 4.4

List of Tables System range for dierent chip congurations. . . . . . . . . . . . . . . . . . 93 SAR: Whole body and maximum local limits [125128]. . . . . . . . . . . . 98 Equivalent tissue dielectric parameters for head and body models [154]. . . 105 Electrical properties of body tissues in dierent ISM frequency bands [161]. 105 Computational requirements of FEKO body models. . . . . . . . . . . . . . 107

Chapter

Introduction
1.1 Wireless communication systems: New trends and use of the ISM bands
New trends

1.1.1

Over recent years, wireless communication systems have spread widely and new communication services have been enthusiastically adopted by people around the world. Mobile communication systems have had a massive growth, rstly with voice services, and recently, with the latest generation of mobile data services that allow for high data transmission rates. Nowadays, it is quite common to nd people reading their electronic mail, reading the last news in an online newspaper or updating their Facebook page in the train during their journey to the oce. The proliferation of computers and wireless communications has evolved into an era of wireless networking. Wireless Local Area Networks (WLANs) have replaced the traditional LANs in a huge number of applications because WLANs introduce benets such as improved exibility, mobility and ease of installation. Furthermore, the most modern commercial devices [14] are starting to use the latest generation mobile phone systems together with the WLAN systems to improve the user experience, providing access to the network inside and outside the oce and taking advantage of the higher performance of WLANs when the user is inside. Mobile phones and WLANs are probably the most popular wireless systems, but over recent years the use of other wireless devices has also increased. For instance remote sensor networks can be used to detect the free spaces in a parking lot and send this information to 1

1.1. Wireless communication systems: New trends and use of the ISM bands

the people that are looking for parking space [57]; or to precisely control the temperature at dierent points of a room or a building and adjust the air conditioner according to the sensors measurements [8]. In sensor networks a group of sensors are spatially distributed and, using a wireless link, they transmit the information collected to a central unit which can process it or retransmit it around the world over the Internet. In these applications sensor size and cost are key factors to ensure system viability. For this reason, the integration of the RF front-end together with the sensor elements can help obtain really small and cheap devices. Due to the increasing use of these systems, multiple commercial transceiver modules for these applications can be found [9]. However, the radiating element is usually an external antenna that increases the system size. Furthermore, costs are also increased by the need for a physical connection between the antenna and the integrated front-end. The integration of the whole system (including the antenna and the sensor) in a single chip can be advantageous and avoid the need for external connections. Another eld where the use of wireless devices is increasing is in medical applications, which can be understood as a specic case of sensor networks. The aim of these systems is to improve the life quality of patients, improving their mobility during treatment and the subsequent monitoring period. These systems can be also extremely useful for the treating of chronic diseases. As an example, wireless implantable glucose monitoring systems can improve the life and mobility of people with diabetes related problems. Moreover, these systems could also increase safety by alerting the emergency services when a problem is detected.

1.1.2

Use of the ISM bands

In recent years there has been a signicant increase in the number of systems and devices that operate in the Industrial, Scientic and Medical (ISM) applications frequency bands. According to the Code of Federal Regulations (CFR) of the Federal Communications Commission (FCC) [10] and the International Telecommunications Union (ITU) regulations [11], the ISM bands are for operation of equipment or appliances designed to generate and use locally RF energy for industrial, scientic, medical, domestic or similar purposes. Typical ISM applications are the production of physical, biological, or chemical eects such as heating, ionization of gases, mechanical vibrations, hair removal or acceleration of charged particles. Specic examples of the previous applications are domestic microwave ovens, jewelry cleaners for home use or ultrasonic humidiers. Table 1.1 shows a summary of the frequency bands dened by the FCC for ISM

Chapter 1. Introduction Table 1.1: Frequency bands for ISM applications dened by the Federal Communications Commission (FCC) [12]. 1 The use of these frequency bands for ISM applications shall be subject to special authorization. 2 Only in Region 2. Frequency Range 13553 - 13567 kHz 26957 - 27283 kHz 40.66 - 40.70 MHz 433.05 - 434.79 MHz 902 - 928 MHz 2400 - 2500 MHz 5725 - 5875 MHz 24 - 24.25 GHz 61 - 61.5 GHz 122 - 123 GHz 244 - 246 GHz Center frequency 13560 kHz 27120 kHz 40.68 MHz 433.92 MHz1 915 MHz2 2450 MHz 5800 MHz 24.125 GHz 61.25 GHz1 122.5 GHz1 245 GHz1

applications, also known as ISM frequency bands. The previous denition states that ISM bands were not originally specied for telecommunication services, but the FCC mentions that these bands can also be used for this kind of services [13]. However, the radio communication services operating within these frequency bands must accept harmful interference which may be caused by the other ISM applications. The FCC has several rules and regulations that must be satised by these communication systems to ensure their correct operation and avoid undesirable interferences to other systems [13]. Multiple license free communication systems that use some of the ISM bands have appeared in the last years. WLANs under the IEEE (Institute of Electrical and Electronic Engineers) 802.11b standard [14], one of the most widespread standards, use the 2.45 GHz ISM band for data transmission. Other examples can be found in Bluetooth systems [15], a wireless technology standard for exchanging data over short distances from xed and mobile devices, or Zigbee systems [16]. Apart from these communication systems, dierent examples where the ISM bands are used in sensor networks can be also found in literature [17, 18]. Implantable devices could also operate in the ISM bands. Originally, the Medical Implantable Communications Services (MICS) 402 - 405 MHz frequency band was dened for these devices [19]. Some of the rst examples of communication systems in implantable devices can be found in wireless links for pacemakers in order to monitor the pacemaker state [20]. Pacemakers are big devices, but in new applications, like implantable glucose monitoring systems, smaller devices are necessary. For that reason, the 2.45 GHz and

1.2. Motivation and focus of this thesis

Internet

WLAN ACCES POINT

REMOTE SENSOR

REMOTE SENSOR REMOTE SENSOR

CENTRAL UNIT
REMOTE SENSOR REMOTE SENSOR REMOTE SENSOR REMOTE SENSOR

WLAN LINK

ISM BAND WIRELESS LINK

Figure 1.1: Sensor network with a central unit that collects the sensors information and sends it to the Internet using a WLAN connection. The central unit and the remote sensors are connected using a wireless link working in the 2.45 GHz or 5.8 GHz ISM bands.

5.8 GHz ISM bands could be considered to reduce the system dimensions.

1.2
1.2.1

Motivation and focus of this thesis

Thesis motivation and system description

This thesis is focused on the antenna design for a wireless remote sensor network with a central unit connected to the Internet through a WLAN access point (gure 1.1). According to gure 1.1 all the sensors will be connected to a central unit and this unit will be connected to the Internet through a WLAN access point. Using the WLAN connection the central unit will be able to send the sensors information to anyplace around the world. In order to assure maximum coverage and compatibility with the latest standards, the central unit will use the 2.45 GHz and the 5 GHz bands dened by the 802.11 standards. With this dual band operation, the typical 802.11 a/b/g standards can be used, and furthermore the system will be also compatible with the most modern versions of the 802.11 standards, such as the 802.11n. The remote sensors and the central unit will be connected through a wireless link using the ISM bands. To achieve small devices, the 5.8 GHz frequency band is advisable, but in some applications, such as implantable sensors, lower frequencies like the 2.45 GHz ISM

Chapter 1. Introduction

band can be desirable. Therefore the central unit must be able to simultaneously receive information from sensors operating in both bands. The use of a central unit that collects the information instead of having the sensors directly connected to the Internet will allow for smaller and simpler sensors. Moreover, the distance between the sensors and the central unit will be smaller than the distance between the sensors and the WLAN access point, thus allowing for low power sensors that transmit tiny power levels, reducing power consumption and increasing the sensors battery life. A particular case of the previous system is when one of the remote sensors is implanted inside the human body. An example of application with implanted sensors could be in a hospital, where they could be used to monitor the patients health status. Sensors would be connected to the central unit that could be a small receiver located on the patient bed or table. This central unit would be connected to the hospital network using the nearest WLAN access point, allowing the authorized doctors and nurses to access the patient information from any computer in the hospital. It is true that the central unit could be connected with a cable to the hospital network, but a WLAN connection has advantages, like the fact that the patient could be moved from one room to another (for instance for a special diagnostic operation) while keeping open the connection to the hospital network. Multiple small variations could be introduced to the described system in order to modify its functionality. For instance a portable central unit with a Bluetooth connection could use the patient mobile phone to send the collected information to the Internet when away from a WLAN access point. Another option could be to use the new WiMAX (IEEE 802.16 [14]) systems to transmit the information to the Internet.

1.2.2

Focus of the thesis and objectives

The aim of this thesis is the design of antenna solutions to cover the needs of the sensor network presented in the previous section. The requirements of the central unit antenna and the sensors antenna are quite dierent; therefore both problems will be analyzed separately. According to the above system description, the central unit must be able to use the 2.45 GHz and 5 GHz bands dened by the 802.11 standards, as well as the 5.8 GHz ISM band. One possible option could be to use a dierent antenna for each band; but this is not an optimal solution. Multiband radiators that can give service to several bands with a single radiating element are desirable. For this reason an important part of the work

1.2. Motivation and focus of this thesis

presented in this thesis will be devoted to the design of multiband antennas to try to optimize performance while keeping to reduced dimensions. Among the possible solutions to achieve this multiband functionality, we will consider fractal geometries and resonator loaded structures. Our study will be focused on monopole antennas because, as can be seen in gure 1.1, the central unit will be similar to an ADSL router and monopole antennas are typically used in this kind of devices. In remote sensors, small and cheap antennas are necessary. The integration of the RF front-end together with the sensors can be of interest to obtain small devices. However, the radiating element is usually an o-chip antenna thus increasing the system size and cost due to the need for a physical connection between the antenna and the RF front-end. A good choice to obtain really small systems is the use of system-on-chip solutions [21], where antenna and circuits are integrated together in the same chip. For this reason one chapter of this thesis will focus on the design of small antennas fabricated with standard CMOS silicon technologies. Furthermore, antenna performance degradation due to the presence of other components will be also analyzed. The special case of a sensor implanted inside the human body will be considered. In implantable devices the human body eects on the antenna performance must be taken into account. Human tissues near the antenna will produce losses that will degrade the antenna performance; this being one of the most important limiting factors of the systems using implantable devices. Therefore, the three main objectives of this thesis can be summarized as:

Design of monopole multiband antennas for the central unit.

Design of silicon integrated antennas for remote sensors.

Study of the body eects when one of the sensors is implanted inside the human body.

Finally, we want to point out that the purpose of this thesis is not only to give a particular solution for a specic problem but to carry out a thorough study of the proposed topologies and its operating principles as well. Design guidelines will be given to assure that the topologies and results presented can also be applied in other systems with similar requirements.

Chapter 1. Introduction

1.3

Thesis organization

This thesis is divided into ve chapters to gradually cover the aforementioned objectives. Chapter 1, is the chapter that you are reading at this moment. It gives a brief introduction to the modern wireless communication systems and to the ISM bands. This chapter also presents our motivation and goals, showing the system architecture that has inspired and motivated this thesis. The results obtained during this thesis have permitted the author to publish and contribute to several papers which are detailed at the end of this chapter. To satisfy the central unit requirements, high performance antennas with multiband behavior are necessary. For this reason, chapter 2 will focus on the design of multiband antennas covering the WLAN and ISM bands. Firstly a briey introduction to multiband antennas will be given. Then, the chapter is divided in two sections. The rst section analyzes multiband antennas inspired by fractal shapes. Some novel designs where fractals are used in the ground plane will be shown, and we will take advantage of the combination of the fractal shape in the ground plane and in the monopole element to improve antenna performance. In the second section, multiband behavior will be obtained using monopole structures loaded with resonators. Dierent novel topologies will be presented with a thorough analysis of their operating principles. Firstly the monopole will be loaded with a simple resonator and, next, it will be loaded with a dual resonator to introduce more operating bands. Finally, asymmetric structures will be also considered to improve multiband performance. Chapter 3 will focus on the design of antennas for remote sensors. This chapter will analyze the design, fabrication and measurement of miniature integrated antennas fabricated with standard CMOS technologies. Firstly the performance oered by classical miniaturization technique when they are applied to antennas fabricated on a silicon substrate will be analyzed. Next, micromachining techniques will be suggested as a way to improve antenna performance and nally, the degradation introduced by other elements when they are fabricated together with the antenna will be studied. In chapter 4 implantable antennas will be analyzed, focusing on the study of the propagation losses introduced by the human body. Furthermore, some of the results will be tested with measurements carried out using tissue simulating liquids. Finally, chapter 5 includes the conclusions of this thesis and some guidelines for future work.

1.4. List of Authors Contributions

1.4

List of Authors Contributions

This section shows the contributions of the author to international journals and conferences during the period of development of this thesis. A distinction is made between those contributions related to the work of this thesis and those that are the result of the collaboration with other researchers but not directly related to the thesis content.

Journal Papers J. Gemio, G. Junkin, J. Parr on, R. Villarino, Resonator-Loaded Dual-Band Monopole for Universal WLAN, IEEE Antennas and Wireless Propagation Letters, Vol. 8, 2009, pp. 736 - 739. J. Gemio, J. Parr on, J. Soler, Dual Band Antenna with Fractal-Based Ground Plane for WLAN Applications, IEEE Antennas and Wireless Propagation Letters, Vol. 8, 2009, pp. 748 - 751. J. Gemio, J. Parr on, P. de Paco, G. Junkin, J. Marin, O. Men endez, A Split-Ring-Resonator Loaded Monopole for Triple Band Applications, Journal of Electromagnetic Waves and Applications, Vol. 24, 2010, pp. 241 - 250. J. Gemio, J. Parr on, and J. Soler, Human Body Eects on Implantable Antennas for ISM Bands Applications: Models Comparison and Propagation Losses Study, Progress in Electromagnetics Research, Vol. 110, 2010, pp. 437 - 452. J. Gemio, J. Parr on, P. de Paco, J. Sacristan and A. Baldi, Improving Silicon Integrated Antennas by Substrate Micromachining: A Study of Etching Patterns, Progress in Electromagnetics Research, Vol. 117, 2011, pp. 365 - 378.

Conference Papers J. Soler, J. Parr on, J. Gemio, New Multiband Asymmetric Top-Loaded Mod-p Antenna for Terrestrial and Satellite Communications Services, IEEE International Symposium on Antennas and Propagation, San Diego, 2008. J. Gemio, J. Parr on, J. Soler, Antena Dual con Plano de Masa Fractal para Aplicaciones WLAN, XXIII Simposium Nacional URSI, Madrid, 2008. nola de Antenas (RESA) a Best Paper Award: Premio de la Red Espa la mejor presentaci on de antenas.

Chapter 1. Introduction

J. Gemio, J. Parr on, J. Soler, Multiband Antenna for WLAN Applications Using a Fractal-Based Ground Plane, 3rd European Conference on Antennas and Propagation, Berlin, 2009. J. Gemio, J. Parr on, G. Junkin, Monopolo Dual Cargado con Resonadores Magn eticos para Aplicaciones WLAN, XXIV Simposium Nacional URSI, Santander, 2009. J. Gemio, J. Parr on, J. Sacristan, F. Vila, F. Serra-Graells, A. Baldi, P. de Paco, Miniaturization Techniques Applied to Standard CMOS Technology Integrated Antennas, 4th European Conference on Antennas and Propagation, Barcelona, 2010. J. Gemio, J. Parr on, G. Junkin, Resonator-Loaded Monopole Antenna with Triple Band Performance for WLAN/WiMAX Applications, 4th European Conference on Antennas and Propagation, Barcelona, 2010. J. Gemio, J. Parr on, J. Sacristan, Performance Degradation in Silicon Integrated Antennas due to Coils for Inductive Energy Coupling, 5th European Conference on Antennas and Propagation, Rome, 2011.

Other journal papers O. Men endez, P. de Paco, J. Gemio, J. Verd u, E. Corrales, Methodology for Designing Microwave Acoustic Filters with Butterworth / Chebyshev Response, International Journal of Microwave and Wireless Technologies, Vol. 1, 2009 , pp. 11 18. O. Men endez, P. de Paco, J. Gemio, General Equations for the Design and Evaluation of a Stack of Two Bulk Acoustic Wave Resonators, Microwave and Optical Technology Letters, Vol. 51, 2009, pp. 752 - 755. M. Ram rez, J. Parr on, J. M. Gonz alez-Arbes u, J. Gemio, Concentric AnnularRing Microstrip Antenna with Circular Polarization, IEEE Antennas and Wireless Propagation Letters, Vol. 10, 2011, pp. 517 - 519.

Other conference papers J. Gemio, O. Men endez, P. de Paco, Performance Evaluation and Design of Stacked Crystal Filter, European Microwave Week 2008, Amsterdam, 2008.

10

1.4. List of Authors Contributions A. Arriola, G. Sasiain, J. I. Sancho, J. Parr on, J. Gemio, R. Villarino, Compact High Impedance Surface Based on Interdigital Capacitors, 4th European Conference on Antennas and Propagation, Barcelona, 2010.

Chapter

Multiband antennas for the central unit

This chapter will focus on the design of multiband antennas for the central unit of the remote sensor network. As it has been said in the introduction, multiband performance is necessary to serve the ISM and WLAN frequency bands with a single radiating element. Firstly a brief introduction to the concept of multiband antennas will be given and previous works related to multiband radiators will be shown. Next, two options to solve our problem will be analyzed: fractal shapes and resonator loaded structures. Some novel antenna designs will be presented for each option. Although this chapter is focused on monopole congurations, the same ideas could also be applied in the design of other kinds of antennas, such as dipoles or PIFA antennas.

2.1

Introduction to multiband antennas

Some of the classic antennas that can be found in literature, like straight dipoles or horns, have a behavior highly dependent on frequency [2224], which means that they only work properly at the design frequency. However, it would be desirable to have antennas that could work simultaneously at multiple frequencies. For this reason multifrequency antennas are required. There are two types of multifrequency antennas: multiband and broadband antennas. While a multiband characteristic is related to a discrete set of frequency bands where the antenna parameters are similar [25], broadband behavior is associated with an antenna whose parameters are similar within one large frequency band [22]. 11

12

2.1. Introduction to multiband antennas

(a)

(b)

Rn+1 Rn

(c)

(d)

Figure 2.1: Classic broadband and multiband antennas. (a) Helix antenna. (b) Biconical antenna. (c) Log-periodic antenna. (d) Spiral antenna.

Dierent basic topologies have been used to obtain multiband or broadband behavior. Some well known antennas with broadband behavior are the helix (gure 2.1(a)) and the biconical antennas (gure 2.1(b)). To simplify the fabrication process, some geometrical approximation to the biconical antenna, like the triangular monopole or the bow tie dipole, can also be used [22]. The self scalable principle of antennas states that the input and radiation parameters of an antenna are invariant to a change of its physical dimensions if a similar change is also made in the operating frequency [22]. Based on this idea, DuHamel [26] introduced the concept of log-periodic antennas. An example of a real implementation can be seen in

Chapter 2. Multiband antennas for the central unit gure 2.1(c), where the geometric ratio ( ) is dened as = fm+1 Rn = Rn+1 fm

13

(2.1)

being fm and fm+1 the frequency of two consecutive operating bands of the antenna. In this kind of antennas, the geometric ratio ( ) observed in the geometry also denes the antenna log-periodic behavior in frequency. Based on the same idea, in 1957 Rumsey [27] introduced the frequency independent antennas. He said that the behavior of an antenna that could be fully described by angles would be frequency independent. This is a theoretical concept, nevertheless there are some real implementations, like the spiral antenna of gure 2.1(d), that come close to this idea. These are just a few examples of classical topologies used for broadband and multiband antennas. They were presented half century ago, but they are still in use, and some of them have been the base for the development of modern multiband antennas, such as the fractal antennas that will be analyzed later in this chapter. Over the past few years an important eort has been put into the design of new multiband antennas for multiple system coverage in order to satisfy the needs of the market, which demands high performance small radiators with multiband behavior. Multiple examples of multiband antennas can be found, with designs that range from those based on monopoles with stub tuning [28, 29], sleeve dipoles [30] or multi-branch [31], to those based on dielectric resonators [32, 33]. Recently, the use of PIFA antennas [3436] for mobile and portable terminals has been increasing due to their compact size and ease of integration inside the mobile terminal. Figure 2.2 shows some examples of the aforementioned solutions. Top-loading techniques are another interesting option to achieve multiband behavior. These techniques have been typically used to reduce antenna dimensions [22]. For instance the introduction of a top capacitive load in a monopole antenna increases its electrical length, reducing the operating frequency without increasing the monopole height. However, top-loading techniques can also be applied to some antenna geometries, such as a triangular monopole or a Sierpinski monopole, in order to obtain multiband radiators [37]. From gure 2.3 it is seen that when loads are introduced to the antenna geometry, new current paths are created thus adding new resonances to the antenna response. It is advisable to use these antennas at even resonances because relative changes in the input resistance are smaller than at odd resonances, but these resistance values dier from 50 and hence the matching levels are poor. As was suggested in [37] the input resistance can be adjusted by modifying the antenna geometry. Figure 2.4 shows a modied version

14

2.1. Introduction to multiband antennas

(a)

(b)

(c)

Figure 2.2: Multiband antennas presented in recent years. (a) Monopole with stub tuning [28]. (b) Antenna based on dielectric resonators [32]. (c) PIFA Antenna [35].

Chapter 2. Multiband antennas for the central unit

Top Loaded Monopole (1 load)
t a hL h1 h2 a hL h1 h2 hL a h1 h3

15
Top Loaded Monopole (3 loads)

Top Loaded Monopole (2 loads)

(a)

1 Load 200 100

Xin (W)

2 Loads

3 Loads

0 -100 -200 0 1 2 3 4 5 6 f (GHz)

(b)

10

Figure 2.3: Top-loaded triangular monopole with dierent load numbers. (a) Antenna geometry. (b) Antenna input reactance.

of the triangular monopole antenna where the width w has been increased to reduce input resistance at even resonances and improve matching levels. Input reection coecients better than -10 dB are achieved in the four antenna matching bands. It is important to point out that, in the framework of this thesis, similar input parameters in all the bands will not be the only design requirement for our multiband antennas. Similar radiation patterns will also be necessary.

2.2

Multiband antennas inspired by fractal geometries

In this section, it will be shown that fractal shapes can improve multiband antenna performance when they are used either in the monopole element or in the ground plane generation process. Several antenna designs suitable for the central unit of the wireless sensor network will be presented.

16

2.2. Multiband antennas inspired by fractal geometries

s
h2 hL h1 h3

w
(a)

Simulated
0 -5

Measured

S11 (dB)

-10 -15 -20 0.5 1 1.5 2 2.5 3 3.5 f (GHz)

(b)

4.5

5.5

Figure 2.4: Modied top-loaded triangular monopole. The width w has been increased to improve the matching levels. (a) Fabricated prototype. h1 = 50 mm, h2 = 41.7 mm, h3 = 33.3 mm, hl = 25 mm, s = 2 mm, w = 25 mm. (b) Measured and simulated input reection coecient.

2.2.1

Introduction to the use of fractal shapes in antenna design

The denition of fractal is not a clear concept. The most important properties that characterize a fractal are its fractal dimension and its self similarity property [3841]. Mandelbrot gave the following denition: A fractal is by denition a set for which the Hausdor dimension strictly exceeds the topological dimension, which he later replaced with: A fractal is a shape made of parts similar to the whole in some way [3941]. In [39] it is said that the condition that must be accomplished to dene an object as a fractal is that its fractal dimension exceeds its topological dimension. One of the main ideas behind the fractal concept is that most fractal objects have self similarity at dierent scale levels, which means that a fractal object is composed of small but similar copies of itself. An example of a self similar fractal shape in nature is the

Chapter 2. Multiband antennas for the central unit

17

cauliower (gure 2.5). If it is analyzed at dierent scale levels, the same shape is observed. In nature, other examples of fractal shapes can be found in trees, leaves, snowakes or the coastline, to mention just a few examples. Fractal shapes can be generated with an iterative process where each iteration is created using multiple rotated and/or scaled copies of the previous iteration. As can be seen in gure 2.6(a), the Sierpinski gasket generation process starts with a solid triangle. In each iteration, the new geometry is generated using three copies scaled by a factor of 2 of the previous iteration. Figure 2.6(a) shows the evolution of the geometry when the number of iterations is increased. The fractal shape is the geometry obtained after an innite number of iterations. The generation process of other fractals, like the Sierpinski carpet or the Koch curve, is also shown in gure 2.6. Strictly speaking, fractal shapes can not be physically built since they are a mathematical abstraction which includes an innite number of elements. However, some real structures inspired by these ideal fractals are typically used. These real implementations are usually known as truncated fractals, prefractals or fractals with a nite number of iterations [39]. The use of fractal shapes is widespread in antenna design. One interesting property of some fractals, like the Koch or the Hilbert curves, is the space lling property which says that they can have an innite length and be tted in a nite area. This property is useful to reduce the antenna dimensions [42, 43], because when the number of fractal iterations is increased, the total antenna length is also increased, reducing the resonance frequency

Figure 2.5: An example of fractal shape in nature: Cauliower [38].

18

2.2. Multiband antennas inspired by fractal geometries

without a signicant change in the area. The self similarity property of some fractals makes these structures very suitable for multiband antennas. One of the rst reported examples of a multiband fractal-based antenna was the Sierpinski antenna shown in [44, 45], where the Sierpinski gasket was used to obtain a multiband monopole. Since these initial references, further works have been presented in order to improve the Sierpinski monopole performance and have more control in the allocation of the operating bands [4653]. Most of the aforementioned examples are based on dipole or monopole congurations; however similar ideas can be applied to other types of antenna. For instance in [5456] the Sierpinski gasket (or a modied version of it) is used in the design of multiband patch antennas.

Initiator

1 iteration

2 iterations

3 iterations

(a)

(b)

(c)

Figure 2.6: Generation of dierent fractal shapes using an iterative process. (a) Sierpinski gasket. (b) Sierpinski carpet. (c) Koch curve.

Chapter 2. Multiband antennas for the central unit

19

2.2.2

The Sierpinski monopole

In this section the performance of the Sierpinski monopole will be further analyzed because some of the new antennas presented in this thesis will be based on this structure. In the Sierpinski monopole the number of fractal iterations is directly related to the antenna multiband behavior. When additional fractal iterations are applied to the geometry, new resonances are introduced to its response. According to gure 2.7 when the number of iterations is equal to two, three pairs of resonances can be clearly seen at low frequencies (0 - 6 GHz). When the number of iterations is increased to three, a new pair of resonances appears at higher frequencies (8 - 12 GHz). This tendency goes on as more fractal iterations are included in the antenna geometry. Good matching levels are obtained at the odd resonances when the reference impedance is 50 , therefore the position of the operating bands will be mainly dened by the odd resonances of the structure. Moreover, if radiation patterns at the odd resonances are analyzed, acceptable similarity between them is obtained, and hence the Sierpinski monopole can be considered a multiband radiator [45]. The Sierpinski monopole multiband behavior is related to the self similarity property at dierent scale levels of this geometry and it can be easily understood analyzing the current distributions of the antenna (gure 2.8). When the frequency increases, currents are concentrated in a smaller part of the antenna, thus the antenna active area (area where currents are important) scales with frequency. If we zoom in on the active area, the shape of the current distributions at the dierent resonances is the same but scaled. This similarity in the current distributions allows for the antenna multiband behavior, oering similar input and radiation parameters in multiple operating bands. In the particular case of gure 2.8 the antenna will present a frequency log-periodic behavior, with a period equal to 2, due to the fact that the geometry has been created with a scale factor of 2. It is important to point out that the rst resonance behavior will be slightly dierent due to the truncation eects related to the nite dimensions of the real antenna [44, 45]. The antenna of gure 2.7 is the basic Sierpinski monopole, however multiple perturbations can be applied to the original geometry to change and adjust the position of the bands, the periodicity among them and the radiation patterns shape. In the Sierpinski monopole the resonance frequencies are associated to the monopole edge length rather than to its height [48]. For a xed height, when the are angle is increased, the edge length is also increased and the resonances will be shifted to lower frequencies. Another important factor is that the are angle also modies the radiation pattern shape. According to gure 2.9 when the are angle = 90o the monopole oers a bidirectional radiation

20

2.2. Multiband antennas inspired by fractal geometries

h1 h2

h1 h2

2 iterations

h3

3 iterations

h4

h3

z x

h1 h2

4 iterations

h4

h3

(a)

2 iterations
0

3 iterations

4 iterations

S11 (dB)

-10 -20 0 300

10

12

14

16

18

Rin (W)

150 0 150

10

12

14

16

18

Xin (W)

0 -150 0

8 10 f (GHz)
(b)

12

14

16

18

Figure 2.7: Sierpinski monopole antenna with dierent numbers of iterations. h1 = 100 mm, = 60o and hn+1 = hn /2.

Chapter 2. Multiband antennas for the central unit

21

Zoom x1
f3

Zoom x2
f5

f7

Zoom x4

Zoom x8
f9

0 dB

-20 dB

Figure 2.8: Normalized current distributions at the odd resonances of a Sierpinski monopole with 4 fractal iterations.

22

2.2. Multiband antennas inspired by fractal geometries

patterns, whereas if this angle is increased to = 110o radiation patterns become quaddirectional. Therefore, depending on the application, the are angle can be changed to adapt the radiation patterns to the specic application requirements. Flare angle and antenna height can control, the radiation patterns and the operating bands position while keeping the same frequency spacing. The frequency spacing can be adjusted by changing the scale factor of the fractal geometry (gure 2.10(a)) [46]. Another option is the structure known as Mod-p (gure 2.10(b)), where multiple bands and subbands can be dened [49, 50]. As is shown in gure 2.11, the Sierpinski monopole can be combined with the aforementioned top-loading techniques with the aim of improving its multiband behavior [57]. Top-loading resonances are added at low frequencies and can be used to adjust the frequency spacing between the two rst bands which does not agree with the scale factor because of the truncation eect of the fractal geometry. At high frequencies the toploading eects will be less important because the antenna currents self scale with frequency approaching the antenna feeding point. Furthermore asymmetric structures can increase the design exibility, due to the fact that the whole structure resonances will combine resonances of both halves. As an example, gure 2.11 shows an asymmetric antenna based on two halves of top-loaded Sierpinski monopoles of dierent heights [57]. The asymmetric antenna reactance and the reactance of the isolated halves are plotted in gure 2.12. At low frequencies the input impedance of the asymmetric structure is equal to the parallel of the impedance of both isolated halves. For this reason the odd resonances of the asymmetric structure approximately coincide with the odd resonances of the two separated halves. The relative variations in the real part of the impedance are generally smaller in even resonances than in odd resonances. Therefore to obtain multiband behavior the use of even resonances is advisable because it will allow us to obtain good matching levels simultaneously in the dierent bands if the average value of the input resistance at these resonances is used as reference impedance. Good antenna matching is achieved when the reference impedance is 170 (gure 2.13). According to gure 2.14, radiation patterns at even resonances are not equal in all the bands, but it could be useful because two dierent patterns are obtained. At the rst three resonances the typical monopole-like pattern is obtained. This pattern could be suitable for terrestrial communications. At the fourth and fth even resonances, the radiation levels for higher elevation angles increase, and this is useful for satellite communications. These results open the possibility of using the symmetric antenna as a solution for terrestrial

Chapter 2. Multiband antennas for the central unit

23

a = 90

f1

f3 0 dB

f5

f7

-30 dB

f9

a = 110

f1

f3 0 dB

f5

f7

-30 dB

f9

Figure 2.9: Radiation patterns at the odd resonances of the Sierpinski monopole with dierent are angles .

24

2.2. Multiband antennas inspired by fractal geometries

a h1 h2

h1 h2

sf = 5/3

h3

sf = 2
(a)

h3

h1 a

h1

Mod-3
(b)

Mod-4

Figure 2.10: Sierpinski monopole perturbations to control the spacing between resonances: (a) The scale factor sf = hn /hn+1 can be changed (b) or dierent Mod-p structures can be used.

Symmetric top-loaded Sierpinski monopole

t
a

Asymmetric top-loaded Sierpinski monopole

t
a

h1 h2

h11 h12 h21 h22

Figure 2.11: Top-loaded Sierpinski monopole. h11 = 300 mm, h21 = 150 mm, h12 = 200 mm, h22 = 100 mm, = 60o and t = 2 mm.

Chapter 2. Multiband antennas for the central unit

Asymmetric Monopole Big Half Impedance Parallel 300 150
Xin (W)

25
Small Half

0 -150 -300 0 1 2 3 4 f (GHz) 5 6 7 8

300 150
Xin (W)

0 -150 -300 0.1 0.2 0.3 0.4 0.5 0.6 f (GHz) 0.7 0.8 0.9 1

Figure 2.12: Input reactance of an asymmetric top-loaded Sierpinski monopole (gure 2.11). The input reactance of the two isolated halves and the reactance of the parallel of the two halves impedance are also plotted.

and satellite services with a single radiating element. In the sensor network these radiation patterns could also be of interest. For instance in an outdoor sensor network where there is no WLAN or mobile phone system coverage, the low frequency bands could be used to collect information from the sensors, and this information could be transmitted to the Internet using a satellite communication system operating in high frequency bands. In this scenario, the satellite communication system would replace the WLAN connection of the original system. Another possible application could be in indoor scenarios where dierent frequency bands could be used to receive information from sensors at dierent elevation angles. However, in order to use this antenna in the sensor network, the antenna geometry should be adjusted to correctly allocate the operating bands according to the system requirements.

26

2.2. Multiband antennas inspired by fractal geometries

Simulated 0
S11 (dB)

Measured

-10 -20 0 300

0.5

1.5

2.5

3.5

Rin (W)

150 0 200

0.5

1.5

2.5

3.5

Xin (W)

0 -200 0

0.5

1.5

2 f (GHz)

2.5

3.5

Figure 2.13: Input parameters of an asymmetric top-loaded Sierpinski monopole (scaled prototype with h11 = 90 mm). The reference impedance for the S11 parameter calculation is 170 instead of 50 .

2.2.3

Use of the Sierpinski gasket in the ground plane

In the previous section, and in most of the previous works presented in literature, the Sierpinski gasket is used in dipoles or monopoles with solid ground planes. In this section the fractal shape will be used in a dierent way, as part of the antenna ground plane. If not properly taken into consideration, a ground plane with nite dimensions can introduce undesirable eects and degrade antenna performance. Nevertheless, it will be shown that we can make good use of the nite dimensions ground planes eects to improve antenna performance. In particular we will use a nite ground plane inspired by the Sierpinski gasket to improve the multiband behavior of the whole radiating structure. The ground plane resonances will be used to reinforce the monopole resonances, improve the matching levels and bandwidths, and even introduce new operating bands. Several examples that use the ground plane eects to improve antenna performance can be found in literature [35, 5863], but in these references the ground plane is not based on fractal shapes, which is one of the main novelties presented in this thesis.

Chapter 2. Multiband antennas for the central unit

27

f =0
-30 0 0 30 60
-20

f =90
-30 -60
-20

q=90
330 30 60 300 270 240 90 210 180 150 0 0
-10 -20

0 0

30 60 90 120

f2

-10

-10

-60

-90

90

-90

f =0
-30 0 0 30 60
-20

f =90
-30 -60
-20

q=90
330 30 60 300 270 240 90 210 180 150
0

0 0

30 60 90 120

f4

-10

-10

-10 -20

-60

-90

90

-90

f =0
-30 0 0 30 60 -60 -30
-10

f =90
0 0 30 60 300 270 240 90
-10 -20

q=90
330
-10 -20

0 0 30 60 90 120

f6

-60

-20

-90

90

-90

210

180

150

f =0
-30 0 0 30 60
-20

f =90
-30 -60
-20

q=90
330 30 60 300 270 240 90 210 180 150 0 0
-20

0 0

30 60 90 120

-10

f8

-10

-10

-60

-90

90

-90

f =0
-30 0 0 30 60
-20

f =90
-30 -60
-20

q=90
330 30 60 300 270 240 90 210 180 150 0 0
-10 -20

0 0

30 60 90 120

f10

-10

-10

-60

-90

90

-90

Figure 2.14: Radiation patterns at the even resonances of an asymmetric toploaded Sierpinski monopole (gure 2.11). Black dotted lines are used for the copolar component; red dashed lines are used for the crosspolar component.

28

2.2. Multiband antennas inspired by fractal geometries

(a)

(b)

(c)

(d)

Figure 2.15: Triangular monopole antenna on a nite dimensions ground plane. (a) Solid ground plane. (b) Fractal-based ground plane after 1 iteration. (c) Fractalbased ground plane after 2 iterations. (d) Fractal-based ground plane after 3 iterations. The triangular monopole height is 50 mm and the ground plane dimensions are 200 mm x 200 mm.

Eects of using a fractal-based ground plane The fractal-based ground plane is created using four copies of a Sierpinski gasket rotated 90o to each other. In the ground plane generation process a dierent number of fractal iterations can be used obtaining the geometries shown in gure 2.15. In this gure it can be seen that the generated ground plane keeps the self similarity property of the original Sierpinski gasket, and hence it is predictable that the ground plane will also exhibit multiband behavior with multiple controlled resonances. Figure 2.16 shows that new resonances are introduced when the number of ground plane fractal iterations is increased. This is because the whole structure resonances are a combination of the monopole element resonances and the ground plane resonances. Moreover, the position of the ground plane resonances is similar to the position of the resonances obtained if the Sierpinski gasket used in the ground plane generation process is used as monopole element. Only a small frequency shift is observed due to the mutual loading eects between the dierent copies of the Sierpinski gaskets that make up the ground plane. Furthermore, good matching levels are obtained in the odd resonance when

Chapter 2. Multiband antennas for the central unit

Finite 0
S11 (dB)

29
2 iter 3iter

1 iter

-10 -20 0 300

Rin (W)

150 0 200

Xin (W)

0 -200 0

f (GHz)

Figure 2.16: Input parameters of a triangular monopole on a fractal-based ground plane with dierent number of iterations (gure 2.15). the reference impedance is 50 . It is important to point out that the aforementioned perturbations applied to the Sierpinski monopole, like the use of dierent scale factors, can also be applied to the ground plane in order to adjust the position of its resonances.

Dual band triangular monopole antenna on a fractal-based ground plane In this section an equilateral triangular monopole will be mounted on a fractal-based ground plane (gure 2.17) to obtain a multiband antenna that covers the needs of the central unit. The design process is as follows. The triangular monopole height hm is set in order to obtain 50 matching in the lower part of the 5 GHz band when it is on a solid ground plane. In the generation of the Sierpinski gasket for the ground plane, two fractal iterations with a scale factor of 2 were initially used, with the objective of obtaining three operating bands with frequency spacing around 2. However, the spacing between the two lower bands is larger than 2 due to the truncation eects of the real nite ground plane. Moreover, the rst band is not well matched to 50 , therefore, this band will not be considered in our design, setting h1 = 2h2 just to avoid the truncation eect (mismatch) for the second

30

2.2. Multiband antennas inspired by fractal geometries

h1 h2 h3
FR4 Substrate hm
z x

t
Ground plane
(a)

y x

Feeding
(b)

Figure 2.17: Triangular monopole on a fractal-based ground plane. (a) Monopole element. (b) Modied ground plane inspired by the Sierpinski gasket. band. The heights h2 and h3 have to be adjusted to allocate the 2.45 GHz and 5 GHz WLAN bands. For this reason the scale factor of the second iteration is increased to 2.36 to obtain the desired frequency spacing. The nal design (gure 2.17) has the following dimensions: hm = 30 mm, h1 = 52 mm, h2 = 26 mm, and h3 = 11 mm. These dimensions take into account the eects of the low cost FR4 1.55 mm substrate ( to support the antenna metallic parts. In order to demonstrate the improved performance of the antenna with the modied ground plane, it will be compared with the results obtained when the triangular monopole is mounted on a solid ground plane with the same dimensions. The input impedance and return losses of both congurations are plotted in gure 2.18, and table 2.1 summarizes the operating bands and their bandwidth, taking into account that an input reection coecient better than -10 dB is desired. For the triangular monopole on the solid ground plane there are two possible operating bands. However, the matching levels of the rst one are poor and therefore it may not be useful. The second one, with a center frequency of 5.1 GHz, is well matched and has a relative bandwidth of 17.2 %. In the case of the fractal-based ground plane, new resonances appear due to the fractal shape, obtaining good 50 matching at 2.475 GHz, with a relative bandwidth of 6.3 %. Moreover, the 5 GHz band has been improved, increasing its relative bandwidth to 28.6 %. This is due to the fact that dimensions have been adjusted to obtain the third resonance introduced by the ground plane at a frequency slightly higher than the monopole resonance. The combination of both resonances is responsible for the
r

= 4.5, tan = 0.02) used

Chapter 2. Multiband antennas for the central unit

31

Simulated 0
S11 (dB)

Measured

-10 -20 1 300

Rin (W)

150 0 200

Xin (W)

0 -200 1

4 f (GHz)
(a)

0
S11 (dB)

-10 -20 1 300

Rin (W)

150 0 200

Xin (W)

0 -200 1

4 f (GHz)
(b)

Figure 2.18: Antenna input parameters. (a) Triangular monopole on a solid ground plane. (b) Triangular monopole on a fractal-based ground plane.

32

2.2. Multiband antennas inspired by fractal geometries

Table 2.1: Measured operating bands of the triangular monopole on the dierent ground plane congurations.

Operating band (GHz) 4.675 - 5.556

Solid ground plane Center frequency (GHz) 5.1

Relative bandwidth(%) 17.2

Fractal-based ground plane Operating band (GHz) Center frequency (GHz) Relative bandwidth(%) 2.397 - 2.552 2.475 6.3 4.645 - 6.194 5.4 28.6

bandwidth increase. The simulated copolar and crosspolar components of both congurations are shown in gure 2.19. Good similarity between the two bands and reduced crosspolar levels are obtained. The radiation patterns keep the same properties when the solid ground plane is replaced by the fractal-based ground plane, proving that the fractal ground plane allows the introduction of new operating bands without signicantly degrading the radiation parameters of the original monopole antenna. The simulated directivity with the solid ground plane is 3.4 dB for the rst band and 4.6 dB for the second one, whereas the directivity with the fractal-based ground plane is 4.1 dB for the rst one and 5.1 dB for the latter. Back radiation could be considered signicant. However, this is not due to the use of a fractal-based ground plane but to the small ground plane dimensions, because similar back radiation levels are obtained with both ground plane congurations. These back radiation levels could be improved by using larger ground planes. The simulated radiation eciency is over 90% in the frequency range under consideration for both cases. Figure 2.20 shows the current distributions in the fractal-based ground plane in both operating bands. It can be seen as frequency increases, currents are located in a smaller region of the ground plane. Due to the self similarity property of the fractal geometry, current distributions at the dierent bands are similar with dierent scale factors, and this allows us to obtain the multiband performance of the antenna. This antenna oers good matching levels, good similarity in radiation patterns and good eciency levels in both frequency bands, and thus it could be perfectly well used as antenna for the central unit.

Chapter 2. Multiband antennas for the central unit

33

2.475 GHz
f =90
330 300 270 240 210 180 150 0 0 30 60 300 -10 -20 90 270 120 240

5.1 GHz
f =90
330 0 0 30 60 90 120 210 180 150 300 270 240 -10 -20

2.475 GHz
f =90
330 0 0
30

5.4 GHz
f =90
330 60 300 0 0 30 60 90 120 210 180 150 -10 -20 90 270 120 240

-10 -20

210

180

150

f =0
330 300 270 240 210 180 0

f =0
30 60 300 330 0

f =0
30 60 90 120 300 270 240 210 180 330 0

f =0
30 60 300 330 0

30 60 90 120

-10 -20

-10 -20

-10 -20

-10 -20

90 270 120 150 240 210 180 150

90 270 120 150 240 210 180 150

q =90
330 300 270 240 210 180 150 0 0 30 60 300 330 -10 -20 90 270 120 240 210

q =90
0 0 30 60 90 120 180 150 300 270 240 210 330 -10 -20

q =90
0 0 30 60 300 330 -10 -20 90 270 120 180 150 240 210

q =90
0 0 30 60 90 120 180 150 -10 -20

(a)

(b)

Figure 2.19: Simulated normalized radiation patterns. Black dotted lines are used for the copolar component, red dashed lines are used for the crosspolar component. (a) Triangular monopole on a solid ground plane. (b) Triangular monopole on a fractalbased ground plane.

Dual band Sierpinski monopole antenna on a fractal-based ground plane In the previous section the fractal shape was only used in the ground plane in order to introduce new operating bands to a triangular monopole. In this section, the Sierpinski gasket will be used in both elements to increase design exibility. The fractal-based ground plane will reinforce the resonances of a Sierpinski monopole improving the matching levels and bandwidths. The ground plane is similar to the one presented in gure 2.17(b), but dimensions are slightly dierent with h1 = 50 mm and a scale factor equal to two. The monopole is based on a Sierpinski gasket with the same scale factor. As can be seen in gure 2.21, several monopoles with various are angles were fabricated.

34

2.2. Multiband antennas inspired by fractal geometries

20log(J/Jmax)
0dB -20dB -40dB

2.475 Ghz

5.4 Ghz

Figure 2.20: Current distributions in the ground plane. The area with higher current density levels (active area) has been marked with a dashed circle.

hm

a=70

a=75

hm

hm
z

a=80

a=90

hm

Figure 2.21: The monopole element is a two iteration Sierpinski gasket with dierent are angles . Antennas were fabricated on a low cost FR4 substrate ( r = 4.5, tan = 0.02). hm = 50 mm.

Chapter 2. Multiband antennas for the central unit

a = 70 0
S11 (dB)

35
a = 80 a = 90

a = 75

-10 -20 1 400

Rin (W)

200 0 200

Xin (W)

0 -200 1

4 f (GHz)

Figure 2.22: Measured input parameters of a Sierpinski monopole with dierent are angles on a fractal-based ground plane.

If the Sierpinski gasket used as monopole and the Sierpinski gasket used in the ground plane generation process are the same, both will have approximately the same resonance frequencies, reinforcing the resonances of the whole structure. However, the antenna performance can be improved if the monopole is modied in order to separate the resonances of both elements and increase the available bandwidth. Antenna measurements of gure 2.22 show that double resonances appear near the operating bands when the monopole are angle is not 90o . Since the are angle of the monopole is dierent from the are angle of the Sierpinski gasket used in the ground plane, resonance frequencies of both elements are not the same, creating double resonances in the response of the whole antenna that help to improve the bandwidths. Similar results could be obtained by changing the monopole height hm instead of the are angle . As the are angle decreases, monopole resonance frequencies slightly increase and the frequency separation between resonances also increases improving the bandwidth. The problem is that if the monopole are angle is smaller than = 75o , the monopole resonances and the fractal-based ground plane resonances are separated too far and the rst band (2.45 GHz) splits into two bands with a reduced bandwidth.

36

2.2. Multiband antennas inspired by fractal geometries

hm

Monopole

Ground plane

Feeding

Figure 2.23: A small gap g between the ground plane and the monopole can be used to optimize the antenna performance and improve the matching levels.
g = 0 mm 0 -5
S11 (dB)

g = 0.6 mm

g = 1.4 mm

g = 3 mm

-10 -15 -20 1 2 3 4 f (GHz) 5 6 7

Figure 2.24: Measured input parameters of a Sierpinski monopole with = 75o mounted on a fractal-based ground plane with dierent gap sizes g (gure 2.23). Once the are angle is set to = 75o , a ne tuning process to optimize the matching levels can be carried out by adjusting the height of the monopole (g ) above the ground plane (gure 2.23). Measured return losses for dierent values of g are plotted in gure 2.24. As the height g increases, the insertion losses at the rst band decrease while the second band is slightly degraded. The optimum choice of g in order to maximize the S11 < 10 dB bandwidth in both bands is g = 1.4 mm. With this conguration the operating bands are 2.37 GHz - 2.74 GHz and 4.64 GHz - 5.9 GHz, with relative bandwidths of 14.5 % and 23.9 % respectively. Table 2.2 summarizes the information about the operating bands with and without the fractal-based ground plane. If we compare these results with those presented in the previous section where a triangular monopole was used, we can see that the use of the fractal shape simultaneously in the monopole and in the ground plane allows us to increase the rst band bandwidth without signicantly degrading the second band.

Chapter 2. Multiband antennas for the central unit

37

Table 2.2: Measured operating bands of the Sierpinski monopole on the dierent ground plane congurations. Solid ground plane Center frequency (GHz) 2.65 5

Operating band (GHz) 2.58 - 2.73 4.75 - 5.23

Relative bandwidth(%) 5.6 9.6

Fractal-based ground plane Operating band (GHz) Center frequency (GHz) Relative bandwidth(%) 2.37 - 2.74 2.55 14.5 4.64 - 5.90 5.27 23.9 Simulated copolar and crosspolar components of the radiation patterns for the optimized design are shown in gure 2.25(b). For comparative purposes, gure 2.25(a) shows the radiation patterns obtained when the same Sierpinski monopole is mounted on a solid ground plane with the same dimensions as the fractal ground plane. By comparing gures 2.25(a) and 2.25(b) it can be concluded that the fractal ground plane does not introduce important changes in the radiation patterns and that low crosspolar levels are obtained. The simulated directivity is around 4.2 dB for the rst band and 5.8 dB for the second. Although low cost FR4 dielectric losses are high, simulated eciency levels are over 70 % in both bands. As in the previous section, the presented antenna oers good multiband performance, and it could be used for the central unit.

2.3

Resonator loaded multiband monopole antennas

In this section, a new kind of antenna suitable for the central unit will be analyzed: monopoles loaded with resonators. The resonators eects will be used to improve the antenna multiband behavior.

2.3.1

Introduction to resonator loaded antennas

Recently, it has been demonstrated that loading a dipole/monopole with particles can also facilitate dual band operation [64, 65]. These particles are resonators that couple via the magnetic eld produced by currents owing along the dipole. In eect, such resonators load the dipole element at specically chosen locations with a reactance that is determined by the resonant properties of the particular particle geometry. The eects of these resonators

38
2.45 GHz
f =90
330 300 270 240 210 180 150 0 0 30 60 90 120 300 270 240 210 180 330 -10 -20

2.3. Resonator loaded multiband monopole antennas

5.4 GHz
f =90
0 0 30 60 90 120 150 300 270 240 210 180 150 330 -10 -20

2.45 GHz
f =90
0 0 30 60 90 120 300 270 240 210 330 -10 -20

5.4 GHz
f =90
0 0 30 60 90 120 180 150 -10 -20

f =0
330 300 270 240 210 180 0 150 0 0 0 -10 -20 90 120 270 240 210 30 60 300 330

f =0
0 0 -10 -20 90 120 180 0 150 0 270 240 210 30 60 300 330

f =0
0 0 -10 -20 90 120 180 0 150 0 270 240 210 30 60 300 330

f =0
0 0 -10 -20 90 120 180 0 150 0 30 60

q =90
330 300 270 240 210 180 150 30 60 90 120 300 270 240 210 330 -10 -20

q =90
30 60 90 120 180 150 300 270 240 210 330 -10 -20

q =90
30 60 90 120 180 150 300 270 240 210 330 -10 -20

q =90
30 60 90 120 180 150 -10 -20

(a)

(b)

Figure 2.25: Simulated normalized radiation patterns. Black dotted lines are used for the copolar component, red dashed lines are used for the crosspolar component. (a) Sierpinski monopole on a solid ground plane. (b) Sierpinski monopole on a fractalbased ground plane.

can introduce new matching bands to the antenna. Amateur radio enthusiasts have used a similar principle in trap antennas operating in HF bands for many years [66]. For operating in a number of bands, such as those between 3.5 MHz and 30 MHz, it is impractical for most amateurs to use a separate antenna for each band, for this reason multiband antennas are also necessary in this frequency range. As an example, in [66] a trap antenna with the structure shown in gure 2.26 for the 3.5 MHz and 7 MHz bands is presented. A trap in an antenna can perform two dierent functions depending on whether or not it is resonant at the operating frequency. A typical use is a resonant parallel trap. Let us assume that each LC combination is resonant in the 7 MHz band. Because of this parallel resonance, the trap presents high impedance at this point of the antenna. The electrical

Chapter 2. Multiband antennas for the central unit

L B A A L B

39

Figure 2.26: Schematic representation of a trap dipole antenna [66].

eect at 7 MHz is that the trap behaves as an insulator, an open circuit that cuts the antenna at this point, separating the B sections from the antenna. The result is that we have an antenna system that is resonant in the 7 MHz band, taking into account that each section A represents /4 at this frequency. One of the problems of this conguration is that the trap losses at the resonance frequency can reduce the antenna eciency. At the other operating band, where the trap is not resonant, the LC resonators acts as an electric load. If the operating frequency is below the trap resonance, the trap acts as an inductor; if above, as a capacitor. Inductive loading will electrically lengthen the antenna, and capacitive loading will electrically shorten the antenna. In the example we are considering, the traps behave as inductors when operation takes place at 3.5 MHz, electrically lengthening the antenna. This means that the total length of sections A and B may be something less than a physical /4 at 3.5 MHz. Thus, we have a dual band antenna that is shorter than a resonant dipole in the lower frequency band. In the previous example, traps were resonant at one of the two frequencies of operation. However, the same band coverage can be obtained with resonant traps at a frequency between both operating bands, for instance, 5 MHz. With proper selection of the LC ratio and the dimensions of A and B sections, the traps will act to electrically shorten the antenna at 7 MHz and electrically lengthen it at 3.5 MHz. Thus, an antenna that is intermediate in physical length between a resonant dipole at 3.5 MHz and a resonant dipole at 7 MHz can cover both bands, even though the trap is not resonant at these frequencies. Note that these traps have less RF current owing in the trap components in the operating bands, and hence trap losses are lower. The use of a trap resonating at a frequency quite apart from the antenna operating bands is also known as o-resonance loading.

40

2.3. Resonator loaded multiband monopole antennas Based on these concepts and operating principles, dierent antenna topologies with

multiband behavior suitable for the central unit of the sensor network will be presented and we will also give some guidelines for the antenna design process. In the frequency bands used by the sensor network, traps will be implemented in a distributed way instead of using discrete components, introducing some advantages such as the fact that the additional physical path created by the resonators can be used to tune the position of the straight monopole resonances.

2.3.2

Dual band resonator loaded monopole antenna

In this section the o-resonance loading principle will be employed in a design that avoids resonator losses using magnetic resonators whose natural frequency lies between the lower and upper WLAN bands.

Antenna design The basic geometry of the antenna consists of two identical capacitively loaded magnetic resonators attached to a uniform width monopole (gure 2.27). The magnetic resonators
WM Front Rear

RSW RH RCG LM RCL RO RVSW RW RCSW

Figure 2.27: Monopole antenna loaded with two double layer magnetic resonators DL-CLL [67] of identical dimensions fabricated on Rogers RO4003C [68] double sided 0.508 mm substrate ( r = 3.38, tan = 0.0021).

Chapter 2. Multiband antennas for the central unit are based on a double layer capacitive loaded loop (DL-CLL) [67].

41

The monopole length LM aects both operating bands (gure 2.28(a)), so the design begins with LM approximately equal to a quarter wavelength at 2.45 GHz. A small adjustment of this parameter must then be carried out to ne tune the rst band position. According to gure 2.28(b) the resonator width RW simultaneously shifts the second and the third S11 minimums without signicantly aecting the rst. The second S11 minimum shifts because the resonator inductance changes with RW . The third S11 minimum position is also aected by the resonator width because, as will be shown in gure 2.31, at frequencies near this S11 minimum currents tend to follow the perimeter
0
S11 (dB)

-5 -10 -15 -20 2

LM = 22 mm LM = 25 mm LM = 28 mm

f (GHz)
(a)

0
S11 (dB)

-5 -10 -15 -20 2

RW = 4.0 mm RW = 4.5 mm RW = 5.0 mm

f (GHz)
(b)

0
S11 (dB)

-5 -10 -15 -20 2

RCL= 3.0 mm RCL= 3.5 mm RCL= 4.0 mm

f (GHz)
(c)

Figure 2.28: Parametric study of the variation of the input reection coecient S11 with respect to (a) the monopole length LM , (b) the resonator width RW , (c) the length of the resonator capacitor strip RCL .

42

2.3. Resonator loaded multiband monopole antennas

of the structure. Consequently, the resonator external width RW is chosen based on the wavelength 2 corresponding to the upper frequency of the second band dened by the 802.11 standard (5.8 GHz) such that LM + 2RW 32 4 (2.2)

The second S11 minimum is mainly controlled by the resonance frequency of the resonator (gure 2.28(c)). The resonator dimensions are set to obtain a resonance frequency lower than 4.9 GHz. This can be achieved by changing the resonator height RH (modifying the inductance), the loop gap RCG , and the rear strip dimensions RCL , RCSW (modifying the capacitance). The separation between the second and the third S11 minimums aects the second band bandwidth and matching levels. A clear trade-o between both parameters exists: when the separation is increased, bandwidth is improved, but center band matching is degraded. The resonator quality factor will also aect the bandwidth of the impedance match over the 5 GHz band. The bandwidth can be increased by decreasing its capacitance and increasing the resonator height RH in order to increase the inductance. The resonator loop strip width RSW is chosen as 0.4 mm based on fabrication considerations.
0
S11 (dB)

-5 -10 -15 -20 2

WM = 2.0 mm WM = 3.0 mm WM = 4.0 mm

f (GHz)
(a)

0
S11 (dB)

-5 -10 -15 -20 2

RO = 11.25 mm RO = 9.75 mm RO = 8.25 mm

f (GHz)
(b)

Figure 2.29: Parametric study of the variation of the input reection coecient S11 with respect to (a) the monopole width WM and (b) the resonator oset position RO .

Chapter 2. Multiband antennas for the central unit

43

The match over both operating bands can be optimized by adjusting the monopole width WM and the resonator position oset RO (gure 2.29(a) and 2.29(b)). Furthermore, the oset parameter controls the resonator loading position, and the adjustment of this parameter allows for the separation of the minimum eciency point, due to resonator losses at the resonance frequency, from the match band (o-resonance loading principle). After the initial design, a ne tuning process of the aforementioned parameters is necessary to allocate the operating bands according to the system requirements.

Prototype antenna: Fabrication and measurement Figure 2.27 shows the prototype antenna fabricated on Rogers RO4003C [68] 0.508 mm substrate (
r

= 3.38, tan = 0.0021) using a LPKF ProtoMat S62 circuit board plotter.

Prototype dimensions are shown in table 2.3. The capacitor formed by overlapping strips on both faces is the most tolerance critical part of the fabrication process. The alignment tolerance was relaxed by making the front strip twice as wide as the rear strip. The prototype presented here is the second of two that were built on double-sided substrate, the rst one covered 802.11 a/b/g/h WLAN bands and the second one, the universal 802.11 a/b/g/h/j WLAN bands that include Japan. The input reection coecient of the fabricated prototype is plotted in gure 2.30. The available relative bandwidths (table 2.4) are enough to cover the WLAN and ISM bands requirements. In the 2.45 GHz band the relative bandwidth is 11.7 %. In the 5 GHz band, the combination of the second resonance of the straight monopole near 6 GHz and Table 2.3: Dimensions of the dual band WLAN monopole antenna. LM 25.5 mm RV SW 0.9 mm WM 3 mm RCG 0.5 mm RSW 0.4 mm RO 9.75 mm RW 4.68 mm RCL 3.6 mm RH 8 mm RCSW 0.45 mm

Table 2.4: Measured operating bands of the dual band resonator loaded monopole antenna (gure 2.27) with the dimensions of table 2.3. Resonator loaded monopole antenna Operating band (GHz) Center frequency (GHz) Relative bandwidth(%) 2.26 - 2.54 2.4 11.7 4.74 - 6.12 5.4 25.4

44

2.3. Resonator loaded multiband monopole antennas

Simulated 0
S11 (dB)

Measured

-5 -10 -15 -20 2 3 4 5 6 7

f (GHz)

Figure 2.30: Simulated and measured S11 parameters of the prototype antenna with the dimensions shown in table 2.3. the S11 minimum introduced by the resonator eects near 5 GHz, is used to improve the available bandwidth, achieving a relative bandwidth of over 25 %. The normalized current distributions (gure 2.31) show a /4 standing wave at 2.45 GHz where currents in the resonator arms are practically zero, a /4 standing wave at 4.9 GHz where the resonator eectively isolates the top half of the monopole, and a 3/4 standing wave at 5.875 GHz over a path that includes the monopole top and bottom strips and the resonator arms. This explains why the rst S11 minimum position is mainly dened by the monopole length, the second S11 is mainly dened by the properties of the

0 dB

- 30 dB

2.45 GHz

4.9 GHz

5.875 GHz

Figure 2.31: Simulated current distributions at 2.45 GHz, 4.9 GHz and 5.875 GHz. Each image is individually normalized and displayed on a 30 dB scale.

Chapter 2. Multiband antennas for the central unit

f =0
-30 0 30
-10

45
f =90 q=90
330 30
-10

-30 60 -60

2.45 GHz

300 60 270 90 240 210

-10 -20

30 60 90 120

-60

-20

-20

-90

90

-90

180 0 0

150

f =0
-30 0 30
-10

f =90
-30 60 -60 0 30
-10 -20

q=90
330 300 60 270 90 240 210 180 0 0 150
-10 -20

30 60 90 120

4.9 GHz

-60

-20

-90

90

-90

f =0
-30 0 30
-10

f =90
-30 60 -60 0 30
-10 -20

q=90
330 300 60 270 90 240 210 180 150
-10 -20

30 60 90 120

5.8 GHz

-60

-20

-90

90

-90

Figure 2.32: Simulated radiation patterns of the prototype antenna whose dimensions are shown in table 2.3. Black dotted lines are used for the copolar components. Crosspolar components are too low to be plotted with the current axis conguration.

monopole and the resonator, and the third S11 minimum is mainly dened by the length of the external perimeter of the structure (including the resonator arms). Figure 2.32 shows the radiation patterns.There is little variation in the shape of the 2.45 GHz radiation patterns, so only one is shown in this frequency band. The patterns at 4.9 GHz are characteristic of a quarter wavelength monopole, while, as expected, those at 5.8 GHz show a three quarter wavelength monopole pattern. The H-plane radiation patterns are essentially omnidirectional with less than 1 dB variation over both bands. Due to the antenna symmetry, the simulated maximum crosspolar levels are exceptionally low. The gain ranges from a maximum of 7.1 dB to a minimum of 5.8 dB. The radiation and total eciencies of the fabricated prototype are plotted in gure 2.33. Measurements were carried out using the method presented in [69] and FEKO [70] was used for the simulations. The minimum eciency occurs at the resonator resonance of 4.25 GHz, some 650 MHz below the band edge frequency of 4.9 GHz. The in-band radiation eciency is practically unaected by the resonators thanks to the use of the o-resonance loading principle where the resonator resonance is out of the antenna operating bands. Measured

46

2.3. Resonator loaded multiband monopole antennas

Simulated Radiation Efficiency Measured Radiation Efficiency 100
Efficiency (%)

Simulated Total Efficiency Measured Total Efficiency

80 60 40 20 0 2 3 4 5 6 7

f (GHz)

Figure 2.33: Radiation eciency and total eciency of the prototype whose dimensions are shown in table 2.3. Eciency measurements were carried out using the method presented in [69]. total eciency is higher than 85 % in both operating bands. For technical reasons, Method of Moments - Finite Element Method (MoM-FEM) FEKO simulations include dielectric losses but exclude copper losses, resulting in optimistic in-band radiation eciency and somewhat larger errors out-of-band near the resonator resonance where copper losses are particularly important. Interestingly, the measured eciencies conrm the overall FEKO predictions, which are especially useful for determining that the frequency of the minimum radiation eciency is out-of-band. The novel antenna analyzed in this section is also suitable for the central unit. Furthermore it is important to point out that its dimensions could be reduced loading the monopole with only one resonator, however with this asymmetric conguration the antenna performance is slightly degraded and crosspolar levels are increased.

2.3.3

Triple band SRR loaded monopole antenna

In the previous section a resonator loaded monopole has been presented in order to obtain a dual band radiator. Now, we will present a resonator loaded monopole that takes advantage of the dual band behavior of the Split Ring Resonator (SRR) to achieve triple band performance. The SRR has already been used to load a dipole and obtain dual band behavior [64, 65]. In [65] the lower band was dened by the dipole resonance whereas the upper band was based on the rst SRR resonance. However, the SRR also exhibits an upper resonance which has led recent papers to propose these particles as basic building blocks for dual

Chapter 2. Multiband antennas for the central unit

47

band lters [71, 72]. Therefore we will present a structure that combines the dual band SRR with the rst resonance of a monopole to obtain triple band performance with good radiation eciency and similar radiation patterns in all the three bands. In this design, the new frequency band will be used to cover one band of the 802.15 standard (3.244 4.742 GHz).

SRR analysis Conventional SRRs are made up of two concentric open rings with their slits on opposite sides. To evaluate the resonance frequency of the isolated rings and the SRR in free space we followed the procedure presented in [73] and calculated the monostatic radar cross section of the structure when a plane wave impinges as shown in gure 2.34. Because of the coupling between rings, the SRR has two resonances (gure 2.35): one below and one above the individual resonances of the isolated rings. The SRR resonances depend on the length of the open loops and the gap between rings which controls the strength of the coupling. More details about the SRR design can be found in [71]. The simulated SRR current distributions (gure 2.36) agree with those in [71]: in the lower resonance the induced circulating current generates a magnetic dipolar moment aligned with the impinging magnetic eld, whereas an electric dipolar moment aligned with the impinging electric eld is created in the upper resonance.

L2 w1 w2 E H t L1
Figure 2.34: SRR geometry and impinging plane wave for computation of the monostatic radar cross section. The SRR is supported on Rogers RO4003C 0.508 mm substrate (r = 3.38, tan = 0.0027). L1 = 8.2 mm, L2 = 6.6 mm, w1 = w2 = 0.5 mm, g1 = g2 = 1.5 mm, t = 0.3 mm.

L1

g2

m g3 g1

L2 k

48

2.3. Resonator loaded multiband monopole antennas

Outer open ring -10
RCS (dBm )

Inner open ring

SRR

-20 -30 -40 -50 -60 2 3 4 5 6 7

f (GHz)

Figure 2.35: Computed monostatic radar cross section in free space for the SRR of gure 2.34. Solid line: stand alone outer open ring. Dashed line: stand alone inner open ring. Dotted line: SRR.
0 dB -20 dB

Lower Resonance

Upper Resonance

Figure 2.36: Simulated current distributions. Each image is individually normalized and displayed on a 20 dB scale. Arrows are scaled according to the normalized current levels.

Antenna design The SRR loaded monopole antenna geometry consists of a monopole loaded with two identical SRRs (gure 2.37). The triple band performance could be achieved with a single SRR but a symmetric structure will result in highly symmetric radiation patterns with low crosspolar levels. The main drawback of the symmetric structure is an increase in the antenna width. The design begins by considering the middle band (3.244 - 4.742 GHz) and adjusting the monopole parameters Lm , wm , L0 and w0 to obtain impedance matching in a large

Chapter 2. Multiband antennas for the central unit

49

wm
z

Lm

L0

w0

Figure 2.37: SRR loaded monopole. The structure was fabricated on Rogers RO4003C 0.508 mm substrate (r = 3.38, tan = 0.0027). Lm = 22 mm, wm = 6 mm, L0 = 1 mm, w0 = 1 mm, h = 2.5 mm, s = 0.3 mm. SRRs dimensions are the same as in gure 2.34.

bandwidth centered just below the limit of this band, because according to gure 2.38, the presence of the SRRs will introduce an upward frequency shift in the monopole matching band. Using the values of gure 2.37, about 1 GHz of bandwidth centered at 3 GHz is obtained. The SRRs are designed to resonate just above the upper limits of the 802.11 standard lower (2.4 - 2.5 GHz) and upper (4.9 - 5.875 GHz) bands. The SRR shown in gure 2.34, which resonates at approximately 2.8 GHz and 6.2 GHz (gure 2.35), is used. The orientation of the SRRs with respect to the monopole is selected in order to obtain maximum coupling. Other orientations do not provide enough coupling to obtain good matching levels over all the operating bands. Figure 2.38 shows that when loading the monopole with the SRRs, two peaks appear at the SRRs resonances in both S11 and input impedance. Just before these peaks, two minimums in S11 appear used to allocate the lower and upper bands of the universal 802.11 standard. The mid-band response shifts higher in frequency when the SRRs are introduced. The return losses and the position of the minimums can be slightly adjusted with the

50

2.3. Resonator loaded multiband monopole antennas

Monopole 0
S11 (dB)

SRR loaded monopole

-10 -20 2 200

Rin (W)

100 0 200

Xin (W)

-200 2

f (GHz)

Figure 2.38: Simulations comparing the stand alone monopole and the SRR loaded monopole. parameters s and h dened in 2.37. The coupling between SRRs and monopole reduces with increased separation s, resulting in smaller peaks in the impedance and less shift in the middle band. The SRRs to ground plane separation h sets the SRRs load position on the monopole, with larger values of h having a less signicant loading eect on the input impedance. A ne tuning of these parameters gives the values of gure 2.37 and the results of gure 2.38 that meet the requirements. The simulated antenna radiation eciency is plotted in gure 2.39. The eciency minimums introduced by the resonators losses are out of band and coincide with the S11 peaks, as a result the radiation eciency in the operating bands is not expected to be signicantly degraded by the high losses of the resonators at their resonance frequencies. Prototype antenna: Fabrication and measurement Figure 2.40(a) shows the fabricated prototype mounted on a large ground plane. The measured S11 parameters agree well with simulations (gure 2.40(b)). However the upper SRRs resonance is slightly shifted in frequency and its eects on the monopole input impedance are weaker than the simulation predictions. This may be due to tolerances

Chapter 2. Multiband antennas for the central unit

51

in the fabrication process. Anyway, according to table 2.5 a clear matching band can be observed at 2.45 GHz, and a broadband matching frequency range is obtained from 3 GHz to 6.2 GHz, covering the requirements of the ISM bands and the 802.11 and 802.15 standards.

Monopole
Radiation Efficiency (%)

SRR loaded monopole

100 80 60 40 20 0 2 3 4 f (GHz) 5 6 7

Figure 2.39: Simulated radiation eciency of the stand alone monopole and the SRR loaded monopole.

(a)

Simulated
Radiation Efficiency (%)

Measured
100 80 60 40 20 0 2 3 4
(c)

0
S11 (dB)

-5 -10 -15 -20 2 3 4 5 f (GHz) 6 7

5 f (GHz)

(b)

Figure 2.40: SRR loaded monopole prototype antenna. (a) Prototype antenna mounted on a large ground plane. (b) Simulated and measured S11 parameters. (c) Simulated and measured radiation eciency.

52

2.3. Resonator loaded multiband monopole antennas

Table 2.5: Measured operating bands of the triple band SRR loaded monopole antenna (gure 2.40(a)). The second and the third bands have been combined obtaining a broadband matching frequency range from 3 GHz to 6.2 GHz.

SRR loaded monopole antenna Operating band (GHz) Center frequency (GHz) Relative bandwidth(%) 2.35 - 2.5 2.43 6.2 3.06 - 6.25 4.65 68.5 The radiation eciency (gure 2.40(c)) was measured using the method presented in [69]. By virtue of the o-resonance loading principle, the in-band radiation eciency is practically unaected by the SRRs losses, resulting in eciency levels higher than 85% over all the operating bands. These measurements conrm that, although FEKO radiation eciency simulation is optimistic, it predicts the overall behavior of the structure and the eciency minimums position. At frequencies lower than 2 GHz and frequencies higher than 6.4 GHz eciency measurements are distorted due to the limitations of the measurement method [69]. Figure 2.41 shows the measured copolar and crosspolar components of the normalized radiation pattern cuts in the main planes ( = 0 , = 90 and = 90 ) at 2.45 GHz, 3.5 GHz and 5.8 GHz. The E -plane cuts are quite similar for all the frequencies and the H -plane cuts are essentially omnidirectional. Furthermore, due to the antenna symmetry, measured crosspolar levels are low, normally less than 20 dB. The measured gain values range from 5 dB to 6 dB over all the bands, and are in agreement with the measured eciency values (gure 2.40(c)). These results demonstrate that the antenna presented in this section oers good triple band performance and could be used for the central unit, taking into account that the central unit could also use the 3.2 - 4.7 GHz frequency band of the 802.15 standard to transmit information.

2.3.4

Triple band resonator loaded monopole antenna based on the use of an asymmetric structure

As has been mentioned with fractal geometries, asymmetric structures are a good option to increase design exibility in multiband antennas. In this section a novel antenna design that uses two dierent loading resonators to obtain a triple band asymmetric monopole will be shown. The use of an asymmetric structure will introduce a new degree of freedom in the design process and with the o-resonance loading principle the resonators losses in

Chapter 2. Multiband antennas for the central unit

q=90
330 30 60 300 270 240 -90 90 -90 90 210 180 150 0
0

53

f =0

2.45 GHz

-30 -60

f =90
30 60 -60 -30 0 0
-10 -20

-10 -20

30 60 90 120

-10 -20

f =0

3.5 GHz

-30 -60

f =90
30 60 -60 -30 0 0
-10 -20

q=90
330 30 300 60 270 240 0
-10 -20

0
-10 -20

30 60 90 120

-90

90

-90

90

210

f =0

f =90
30 60 -60 -30 0 0
-10

q=90
330 30 300 60 270 240 0 0
-10 -20

180

150

5.8 GHz
-60

-30

0
-10 -20

30 60 90 120

-20

-90

90

-90

90

210

180

150

Figure 2.41: Measured normalized radiation patterns of the SRR loaded monopole. Each cut has been normalized with reference to its maximum. Black dotted lines are used for the copolar component; red dashed lines are used for the crosspolar component.

the matching bands will be avoided.

Antenna design The antenna geometry consists of an asymmetric structure based on a monopole loaded with two resonant particles with dierent resonance frequency (gure 2.42). Figure 2.43 shows the existence of three matching bands around 2.45 GHz, 3.5 GHz and 5.4 GHz. The rst S11 minimum is used to cover the lower band of the 802.11 standard (b / g) and the 2.45 GHz ISM band. As can be seen in gure 2.44 currents have the typical distribution of a /4 monopole with signicant currents in the resonators, especially in the largest one. Therefore this band corresponds to the rst resonance of the straight monopole, taking into account that it is slightly frequency-shifted due to the resonator loading eects. With the second S11 minimum, the 3.5 GHz band of WiMAX can be covered. This minimum basically appears due to the large resonator eects (it resonates at a frequency

54
Front WM

2.3. Resonator loaded multiband monopole antennas

Rear

WR

WR LM

HR1

GR WS

GR t

HR2

LS

WS

Figure 2.42: Fabricated prototype. In the rear side a conductor strip is used to reduce the resonance frequency of the large resonator.

around 2.9 GHz) and its position can be controlled adjusting the resonance frequency of this resonator. In the design of this resonator, a conductor strip in the rear side of the substrate is used to increase its capacitance and decrease the resonance frequency without a signicant increase in the resonator dimensions. In this frequency band currents in the small resonator are also signicant (gure 2.44), and hence it will be slightly frequencyshifted due to the eects of this resonator. Finally, the upper band of the international 802.11 standard (a / h / j) and the 5.8 GHz ISM band are covered by combining the third and fourth S11 minimums. According to gure 2.44, the third S11 minimum is mainly due to the minor resonator eects (it resonates at a frequency around 4.5 GHz) and its position can be controlled adjusting the dimensions of this resonator. The fourth S11 minimum is due to the monopole second resonance. However, as in the loaded monopole of section 2.3.2, at this frequency currents tend to follow the perimeter of the structure. Therefore, this S11 minimum is aected by the additional path introduced by the resonator arms. For this reason, the position of this minimum can be controlled adjusting the resonators width WR . Tuning the separation between both S11 minimums the available bandwidth of this band can be optimized, taking into account that a clear trade-o exists between the available bandwidth and the matching levels at the center of the band. Figure 2.43 shows a parametric analysis of the variations of S11 parameter with respect to the most important design parameters of gure 2.42. The monopole length modies the position of all the matching bands (gure 2.43(a)), while the monopole width does not signicantly change the position of the bands, although it aects the overall matching levels (gure 2.43(b)).

Chapter 2. Multiband antennas for the central unit

LM=20mm LM=23mm LM=26mm WM=4.0mm WM=6.0mm WM=8.0mm

55

S11 (dB)

-10 -15 -20 2 3 4

S11 (dB)
5 6 7

-5

-5 -10 -15 -20 2 3 4

f (GHz)

f (GHz)

(a)
HR1=9.5mm HR1=11.5mm HR1=13.5mm HR2=7.0mm

(b)
HR2=9.0mm HR2=11.0mm

S11 (dB)

S11 (dB)

-5 -10 -15 -20 2 3 4

-5 -10 -15 -20 2 3 4

f (GHz)

f (GHz)

(c)
LS=4.0mm LS=6.0mm LS=8.0mm WR=4.0mm

(d)
WR=6.0mm WR=8.0mm

S11 (dB)

S11 (dB)

-5 -10 -15 -20 2 3 4

-5 -10 -15 -20 2 3 4

f (GHz)

f (GHz)

(e)

(f)

Figure 2.43: Parametric study of the variation of the S11 parameter with respect to (a) the length of the monopole LM , (b) the width of the monopole WM , (c) (d) the height of the resonators HR1 , HR2 , (e) the length of the rear side strip LS and (f ) the width of the resonators WR . The resonance frequency of the large resonator lies between the rst and the second S11 minimums: for this reason when this resonator is modied, these two minimums will be aected. This is proved in gure 2.43(c) and gure 2.43(e), where the resonance frequency is changed by modifying the inductance and the capacitance respectively. The resonance frequency of the small resonator lies between the second and the third S11 minimums, therefore the modications applied to this resonator will aect the properties of both minimums(gures 2.43(d)). Finally, when the resonator width is changed all the S11 minimums are moved (gure 2.43(f)). This is logical, since this width will change the resonance frequency of both

56
0 dB

2.3. Resonator loaded multiband monopole antennas

-30 dB

f = 2.45 GHz

f = 3.45 Ghz

f = 4.85 GHz

f = 6.05 GHz

Figure 2.44: Simulated current distributions at the S11 minimums. Each image is individually normalized and displayed on a 30 dB scale. resonators, and as it has been shown before, this will change the position of the rst three S11 minimums. With reference to the fourth S11 minimum, the additional path added by the resonators arms will also change, and as a result, this minimum will be frequency shifted.

Prototype antenna: Fabrication and measurement Taking into account the aforementioned trends, a monopole antenna for

ISM / WLAN / WiMAX services was designed. The whole monopole dimensions are 20 mm x 25 mm and it was fabricated on Rogers RO4003C substrate ( tan = 0.0021). Table 2.6 shows the dimensions of the prototype antenna. In gure 2.45 three matching bands at 2.45 GHz, 3.5 GHz and 5.4 GHz are observed, with relative bandwidths of 14.7 %, 18.9 % and 23.2 % respectively, enough to cover the ISM, WLAN (802.11 a/b/g/h/j) and WiMAX requirements. More detailed information about the operating bands can be found in table 2.7.
r

= 3.38,

Chapter 2. Multiband antennas for the central unit

Simulated 0 -5
S11 (dB)

57
Measured

-10 -15 -20 2 3 4 5 6 7

f (GHz)

Figure 2.45: Measured and simulated S11 parameter of the asymmetric loaded monopole antenna. Blue solid line shows simulated results, and red dashed line shows measured results. Table 2.6: Dimensions of the asymmetric prototype antenna. LM 23 mm WM 5.8 mm LS 6.1 mm WS 0.9 mm GR 0.5 mm

t 0.5 mm

HR 1 11.5 mm

HR2 9 mm

WR 6.3 mm

Table 2.7: Measured operating bands of the triple band asymmetric resonator loaded monopole antenna (gure 2.42). Asymmetric resonator loaded monopole antenna Operating band (GHz) Center frequency (GHz) Relative bandwidth(%) 2.2 - 2.55 2.375 14.7 3.26 - 3.94 3.6 18.9 4.83 - 6.10 5.46 23.2 The measured and simulated position of the eciency minimums agree well (gure 2.46). These eciency minimums are related to the resonance of the resonators and again, thanks to the use of the o-resonance loading principle, they are out of the antenna operating bands, and therefore do not degrade the in-band total eciency. The measured total eciency is higher than 80 % in all the three bands. The measured radiation patterns of the proposed design are plotted in gure 2.47. Eplane radiation patterns are similar in all bands and have the typical shape expected from a quarter wavelength monopole. H plane radiation pattern cuts are essentially omnidirectional in all the bands. Crosspolar levels will be slightly degraded due to the antenna asymmetry, however in the radiation pattern cuts shown in gure 2.47, crosspolar

58

2.3. Resonator loaded multiband monopole antennas

Simulated Radiation Efficiency Measured Radiation Efficiency 100

Efficiency (%)

Simulated Total Efficiency Measured Total Efficiency

80 60 40 20 0 2 3 4 5 6 7

f (GHz)

Figure 2.46: Measured and simulated results for the radiation eciency and the total eciency. Total eciency takes into account the mismatch loss.

f =0
-30 0 0 30 60 -60 -30

f =90
0 0 30 60
-10 -20

2.45 GHz

-10

-60

-20

-90

90

-90

90

f =0

f =90
30 60 -60 -30 0 0
-10 -20

3.5 GHz

-30 -60

0
-10 -20

30 60

-90

90

-90

90

f =0

f =90
30 60 -60 -30 0 0
-10 -20

5.4 GHz

-30 -60

0
-10 -20

30 60

-90

90

-90

90

Figure 2.47: Measured radiation patterns of the asymmetric loaded monopole in the three operating bands. Each cut has been normalized with reference to its maximum. Black dotted lines are used for the copolar component; red dashed lines are used for the crosspolar component.

Chapter 2. Multiband antennas for the central unit

59

levels are below -15 dB in the directions of maximum radiation. The measured gain values range from 5 dB to 6 dB over all the bands. Measurements and simulations show that the proposed antenna oers good matching levels, measured total eciency levels of over 80 % and similar radiation patterns along all the three bands, therefore it is also suitable for the central unit. Furthermore, with this novel antenna the central unit capabilities could be improved using a WiMAX connection to send the collected information to the Internet without the need for any additional radiating element.

2.4

Chapter conclusions

In the central unit, multiband radiators are desirable to be able to receive information from sensors using the 2.45 GHz and 5.8 GHz ISM bands, and simultaneously send this information to the Internet using the 2.45 GHz and 5 GHz WLAN bands with a single radiating element. The solutions presented in this chapter are based on: Fractal shapes. Structures loaded with resonators. The most commonly used fractal shape for the design of multiband antennas is the Sierpinski gasket due to its self similarity property at dierent scale levels. Fractal shapes have usually been utilized in the monopole element assuming that it is on a large ground plane, however they can be also applied to the ground plane. As an example a fractal-based ground plane inspired by the Sierpinski gasket has been presented. The modied ground plane adds new resonances that can be combined with the monopole element resonances to improve the antenna multiband performance. As proof of concept, two novel antenna designs have been built and measured. The rst one is a dual band antenna based on a triangular monopole mounted on a fractal ground plane. The novel ground plane has allowed us to add new operating bands to the triangular monopole without degrading radiation patterns and eciency. Through current distribution analysis, it has been shown that currents in the ground plane scale with frequency, and thanks to the self-similarity property of fractal shapes, the current distributions in the dierent frequency bands are the same but at dierent scale levels. This is the main reason why this kind of ground plane oers good log-periodic multiband performance.

60

2.4. Chapter conclusions A Sierpinski monopole on a fractal-based ground plane has been also presented. The

use of the fractal shape in both elements gives more exibility to the design process. When the Sierpinski gaskets used in both elements are equal, the ground plane reinforces the monopole resonances, while if they are slightly dierent, resonance frequencies are not the same and the available bandwidths of the whole structure can be increased. Adjusting the height of the monopole above the ground plane, a ne tuning process can be carried out and the overall antenna performance can be improved. Comparing the rst antenna design with this second, the combination of the fractal shape in both elements improves the available bandwidth in the 2.45 GHz band. These two multiband antennas cover the 2.45 GHz and 5.8 GHz ISM bands, and the 2.45 GHz and 5 GHz WLAN bands, and hence they can be perfectly used as radiating element for the sensor network central unit. The second solution proposed in this chapter to obtain multiband behavior is resonator loaded monopoles. In order to improve the antenna performance it is advisable to use the o-resonance loading principle. This means that the resonators are designed to allocate their resonances out of the antenna operating bands. With this technique new matching bands are introduced without signicantly degrading the in-band radiation eciency. Based on this principle a simple design procedure for a dual-band ISM / WLAN monopole antenna based on capacitively loaded magnetic resonators has been presented. A fabricated prototype that easily meets the specications was demonstrated, showing good matching, high eciency levels and similar radiation patterns in both bands. This prototype antenna covers the ISM and WLAN bands, and hence it is also suitable for the central unit. The same concept can be also applied to introduce more operating bands. As an example, two novel triple band antennas have been presented. In the rst, the triple band performance is achieved using dual band SRRs. The lower and upper bands positions are essentially controlled by the SRRs dual band behavior whereas the monopole dimensions dene the middle band. A prototype that covers the universal 802.11 standard, the 5.8 GHz ISM bands and one band of the 802.15 standard (3.24 - 4.74 GHz) with radiation eciencies higher than 85% was fabricated and measured. In the second example, the triple band performance is achieved using simple resonators in an asymmetric structure. The asymmetric structure allows for the addition of new operating bands, and the position of these bands can be easily controlled by the design parameters of the loading resonators. An asymmetric prototype antenna for the ISM, 802.11 standard and 3.5 GHz WiMAX bands was fabricated and measured, proving the good performance oered by this topology. Moreover, the eects of two dierent dual band

Chapter 2. Multiband antennas for the central unit

61

resonators could be combined in an asymmetric structure to further improve the antenna multiband behavior. These two antennas cover the ISM and WLAN bands, and therefore are suitable for the central unit. Furthermore the third band could be used to improve the system performance with a WiMAX connection for those places away from any WLAN access point. In this chapter several novel antenna designs suitable for the central unit of the sensor network have been presented. In a real application, if we can dene the ground plane geometry, the solutions based on fractal ground planes can be applied. With these solutions, the ground plane area could be covered in a transparent plastic, and the antenna could even be seen as a decorative element. However, if access to the ground plane design is not available, the solutions based on resonator loaded monopoles must be used. Both solutions could be also combined to improve performance.

Chapter

Small silicon integrated antennas for the remote sensors

This chapter of the thesis will focus on the design of antennas for the remote sensors. As has been said in the introduction, sensors dimensions and cost are key factors to ensure system viability, therefore small and cheap sensors are desirable. One possible solution to achieve these requirements is the use of integrated silicon antennas fabricated on standard CMOS technology silicon substrates, where the antenna, the electronic circuits and the sensor are made together in the same chip, obtaining a system-on-chip solution. Firstly the performance oered by dierent miniaturization techniques when they are applied to antennas built on a low resistivity substrate will be analyzed. Next, micromachining techniques will be presented as a solution to improve antenna eciency, and nally the performance degradation due to the presence of parasitic elements will be studied.

3.1

Introduction to silicon integrated antennas

Packaging and assembly of radio front-ends, especially at high frequencies, is currently a signicant cost of the total product. Hence, monolithic integration of all the high frequency components on a single semiconductor chip has the potential of lowering the price of transceiver units to acceptable levels for mass market applications [74]. The fabrication of the antenna and the electronic circuits in the same chip reduces assembly costs and the system dimensions [75]. Despite their potential cost and integration benets, the use of onchip antennas has not yet become widespread. One of the main reasons is that the antenna is one of the most dicult RF-components to integrate on-chip with active circuits. 63

64

3.1. Introduction to silicon integrated antennas

(a)

(b)

Figure 3.1: Example of intra-chip communication system. (a) Conceptual system illustration of intra-chip wireless interconnected system for clock signal distribution. (b) 0.18 m clock transmitter and receiver with zigzag dipole antennas [78].

In order to reduce fabrication costs and obtain real system on-chip devices, it is also desirable to make the antenna fabrication process compatible with standard CMOS technologies using a silicon substrate not optimized for antennas production. The low bulk resistivity silicon of the wafers normally used in standard CMOS processes for the fabrication of active devices decreases the antenna radiation eciency drastically. Furthermore this eciency degradation is emphasized when miniaturization techniques are applied to reduce antenna dimensions [76]. All these problems mean that integrated antennas design is not an easy process. In literature, several examples of silicon integrated antennas for intra-chip and interchip communications can be found. As an example of intra-chip communications, in [77, 78] integrated antennas are used for the distribution of the clock signal to minimize the delay (gure 3.1). In a similar way, integrated antennas could be also used to transmit the clock signal among dierent chips. Some of the rst examples of antennas for inter-chip communications and for communications with an external receiver were presented in [79], where Buechler showed some antennas built on silicon substrates (gure 3.2(a)). In [80], the folded slot of gure 3.2(b) is proposed as basic radiating element for automotive short range radar applications in the 79 GHz band. At the millimeter wave frequency range wavelength is small and therefore complex antennas can be integrated in a reduced area. In [81] a 1.3 mm quasi-Yagi antenna with one driver, two directors and a truncated ground plane that acts as a reector is presented

Chapter 3. Small silicon integrated antennas for the remote sensors

65

(a)

(b)

(c)

(d)

Figure 3.2: Examples of integrated antennas operating at frequencies higher than 60 GHz. (a) Array antenna fabricated on high resistivity silicon presented by Buechler in 1986 [79]. (b) Folded slot antenna designed at 79 GHz [80]. (c) and (d) 60 GHz on-chip Yagi antennas [81, 82].

(gure 3.2(c)). Another example can be found in [82], where a director element, a driven element and a reector element are used on the design of a Yagi antenna with dimensions of 1.1 mm x 0.95 mm (gure 3.2(d)). The aforementioned integrated antenna examples work in the frequency bands of 60 GHz or higher. In [75, 83, 84] Lin studies the performance of integrated antennas in the 24 GHz ISM band. These antennas are fabricated using the conductor and dielectric layers available in mainstream silicon technologies to keep the integration of on-chip antennas viable and inexpensive [75]. Figure 3.3 shows the layout of the zigzag dipole antenna presented by Lin in [83], a schematic representation of the measurement setup and the measured antenna pair gain for dierent environments. The antenna pair gain includes the gain of the transmitting and receiving antennas and the propagation losses. At a distance of 1 m, antenna pair gains of less than -80 dB have been measured. In spite of these low values, it can be seen that the zigzag dipole increases the antenna pair gain by 35 dB with respect to the situation

66

3.1. Introduction to silicon integrated antennas

(a)

(b)

(c)

Figure 3.3: Integrated antenna presented by Lin in [83]. (a) Layout of the 2 mm zigzag dipole. (b) Antenna measurement setup with a signal generator and a spectrum analyzer. (c) Antenna pair gain as a function of the distance in three dierent environments at 24 GHz. The inset shows the 2 mm zigzag dipole antenna radiation pattern.

where the probes are the only radiating elements. Other examples of silicon integrated antennas operating at frequencies higher than 10 GHz can be found in [8589], where several techniques and topologies, such as IFA structures, micromachining or dielectric resonators, are used in order to improve the antenna performance. The previous references are focused on the antenna design, but there are also some references where the whole system is analyzed. For instance in [90, 91] a fully integrated transmitter with an on-chip dipole antenna is presented, showing that complete system integration with high system ranges is possible (gure 3.4). In some applications, such as in implantable devices, frequencies lower than 10 GHz are desirable to reduce the additional propagation losses introduced by body tissues [92, 93]. At lower frequency bands the radiating element dimensions become a problem because antennas are larger, therefore miniaturization techniques must be applied [76] to

Chapter 3. Small silicon integrated antennas for the remote sensors

67

(a)

(b)

Figure 3.4: Fully integrated transmitter with an on-chip dipole antenna at 24 GHz [91]. (a) Photomicrograph of the fully integrated transmitter and frequency synthesizer. (b) Reception of the signal generated by a transmitter IC with an on-chip antenna using a 20 dBi gain horn antenna.

allow antenna integration without an excessive enlargement of the chip dimensions. Some examples of integrated antennas operating at frequencies lower than 10 GHz can be found in [75, 76, 94]. For instance, in [75] Lin presented the gain plot of gure 3.5, showing that the antenna pair gain at 5.8 GHZ is signicantly better than at 24 GHz.

5.8 GHz monopole

24 GHz dipole

Figure 3.5: Antenna pair gain as a function of the distance of a 5.8 GHz monopole and a 24 GHz dipole antenna. The antennas are located 52 cm and 5 mm above the ground [75].

68

3.2. Prototype antennas fabrication and measurement

3.2

Prototype antennas fabrication and measurement

The objective of this chapter is the design, fabrication and measurement of standard CMOS silicon integrated antennas operating in the 5.8 GHz ISM band. A thorough study of the fabrication process is not included among the objectives of this thesis. Nevertheless some basic knowledge of the fabrication technology is necessary due to the fact that the resulting dielectric/metal layers structure and the materials properties will critically aect antenna performance and the design process. In order to make integration of an on-chip antenna viable, the fabrication process must be inexpensive. This means that the antenna must be fabricated using the conductor and dielectric layers available in standard CMOS technologies and a silicon substrate compatible with these technologies. The prototype antennas shown in this thesis were made in a dedicated process in the clean room of the Centro Nacional de Microelectronica - Instituto de Microelectronica de Barcelona (CNM-IMB, [95]). The process was intended to yield a metal/dielectric structure equivalent to a standard 0.18 m RF CMOS, that is, a top metal of 2 m thick Aluminum (Al) over 6 m of SiO2 (gure 3.6). The substrate was low resistivity silicon with a theoretical resistivity of 20 Ohmcm. Briey, the main steps of the fabrication process were: 1. Start with a 4 wafer of P doped silicon with a resistivity of 20 Ohmcm. 2. Growth of 2 m of SiO2 . 3. Additional 4 m-thick deposition of SiO2 . 4. 2 m thick deposition of Al. 5. Al patterning with standard photolithography and etching process. For the antennas fabricated during this thesis, two chip sizes were used. Firstly, the chip dimensions were 12.5 mm x 2 mm. Then, they were reduced to 5.25 mm x 1.9 mm to obtain smaller devices. Since the available area is very small, antenna miniaturization techniques will be necessary. According to gure 3.6, the antenna is at a distance of 6 m from the silicon substrate, and hence the silicon eects on the antenna performance will be important. The relative permittivity of silicon is high,
r

= 11.9, so it reduces the antenna dimensions. The problem

Chapter 3. Small silicon integrated antennas for the remote sensors

Aluminium (2mm)

69

SiO2

er = 3.9

6 mm

Silicon Substrate er = 11.9

500 mm

Figure 3.6: Simplied metal/dielectric layer structure of the fabricated prototypes. It is equivalent to a standard 0.18 m RF CMOS technology.

is that the low resistivity of silicon signicantly degrades the antenna radiation eciency and gain. The SiO2 layer helps to isolate the antenna from the silicon substrate, slightly improving antenna gain. A comparison of the simulated antenna input parameters and radiation eciency with and without the SiO2 layer can be found in gure 3.7, showing as in the case of using the SiO2 layer losses are slightly lower and the impedance curves are less smooth. Numerical simulations must take into account the dielectric layers structure of the fabricated prototypes (gure 3.6). When innite dielectrics are considered, simulations can be carried out without problems. However, some simulation problems arise with the Method of Moments (MoM) code of FEKO software when the nite size of very thin dielectric layers is taken into account. The meshing process of a thin layer is dicult because the number of triangular mesh elements is large and with a high aspect ratio. This increases the computational requirements (time and memory) and can introduce numerical errors that generate incorrect simulation results. Finite dielectrics are necessary since the chip size is an important parameter that aects the antenna performance. To support this armation, the same antenna design was fabricated on three chips with dierent widths (2 mm, 4 mm and 6 mm). Frequency shifts higher than 500 MHz were observed due to the variation of the eective relative permittivity. One option to avoid these problems is the use of a simulation model without the

70

3.2. Prototype antennas fabrication and measurement

With SiO2 layer 0 -5 200 175 150
Rin (W)

Without SiO2 layer

S11 (dB)

-10 -15 -20 -25 -30 1 50 0 2 4 6 f (GHz) 8 10

125 100 75 50 3 1 2 4 6 f (GHz) 8 10

Radiation Efficiency (%)

2 4 6 f (GHz) 8 10

2.5 2 1.5 1 0.5 0 1 2 4 6 f (GHz) 8 10

Xin (W)

-50

-100 -150 1

Figure 3.7: Antenna input parameters and radiation eciency with and without the SiO2 layer. The silicon conductivity is = 5 S/m. The antenna is a meandered dipole and the chip dimensions are 5.25 mm x 1.9 mm.

SiO2 layer. Due to the thinness of this layer, the results would be quite similar to the real ones (gure 3.7). Nevertheless, if we want to include the SiO2 eects in the simulations, there are other numerical methods (FDTD or FEM) that are more suitable for the study of complex dielectric structures. In our case we decided to use the hybrid Method of Moments Finite Elements Method (MOM-FEM) FEKO code [70]. A precise simulation model was necessary in order to achieve good agreement between simulations and measurements, for this reason, the parameters of the wafers used in the antennas fabrication process were measured. Wafers thickness measurements gave the same value as the theoretical one (500 m). The SiO2 layer thickness ranged from 5.8 m to 6.7 m, and hence simulations were made using the theoretical value of 6 m. Important dierences were observed in the conductivity ranging from 11.5 cm to 13.4

Chapter 3. Small silicon integrated antennas for the remote sensors

s = 5 S/m s = 8 S/m 0 -5
S11 (dB)

71

s = 6 S/m s = 9 S/m 200 175 150

Rin (W)

s = 7 S/m s = 10 S/m

-10 -15 -20 -25 -30 1 50 0 2 4 6 f (GHz) 8 10

125 100 75 50 3 1 2 4 6 f (GHz) 8 10

Radiation Efficiency (%)

2 4 6 f (GHz) 8 10

2.5 2 1.5 1 0.5 0 1 2 4 6 f (GHz) 8 10

Xin (W)

-50

-100 -150 1

Figure 3.8: Eects of the silicon conductivity in antenna parameters. Input reection coecient, input impedance and eciency are shown. The antenna is a meandered dipole and chip dimensions are 5.25 mm x 1.9 mm.

cm. Consequently, a conductivity value of 12.5 cm was used instead of the 20 cm of the wafer specications. As can be seen in gure 3.8, the eects of conductivity are important: when conductivity is increased, radiation eciency and gain are degraded and the impedance curves become smoother. Dipole congurations with a dierential feeding schema were used, taking 100 as the dierential reference impedance. A dierential antenna topology was selected due to the fact that it can be easily connected to dierential ended active circuits without the need for a balun. It is important to point out that good broadband matching levels were achieved due to the high silicon losses. Therefore, the main problem of integrated antennas is not the matching levels but the low gain and eciency values.

72

3.2. Prototype antennas fabrication and measurement

Antenna under test

2 Coaxial cables to the VNA

GSGSG Differential probe

C su Pr pp m ob et e or all st t ic ati ch on uc k

PV

Antenna under test

PVC support
(a) (b)

Figure 3.9: Silicon antennas measurement setup. (a) A 1 cm PVC support was used to separate the antenna from the metallic chuck. (b) The dierential dipole antennas were fed with a Picoprobe GSGSG probe. To validate the simulated results, measurements of the fabricated prototypes were carried out in a M150 Cascade Microtech probe station [96] using a Picoprobe GSGSG probe [97] to feed the dierential dipole antennas. The antenna was separated from the probe station with a 1 cm PVC support. Otherwise the metallic chucks, directly in contact with the antenna, would invalidate the measured results. This PVC support was mechanized in order to transmit the vacuum to the antenna and avoid problems related to unwanted antenna movements during measurement. Figure 3.9 shows the measurement setup. Dipole antennas with a dierential feeding schema were measured. Typically, a balun is used to convert the single ended signal of the VNA into a dierential signal. However, in the measurement of the frequency dependence of input parameters, the use of a balun can be a problem since broadband baluns are necessary. These baluns are expensive and can introduce errors. For this reason in the input parameters measurement we used a two port VNA and the mixed mode S parameters [98]. The dierential input reection coecient can be calculated as Sdd = (S11 S21 S12 + S22 ) 2 (3.1)

where S11 , S21 , S12 and S22 are the scattering parameters of the two single ended ports measurement. With this method, broadband measurements of dierential antennas input parameters can be carried out avoiding the problems related to the baluns limited bandwidth

Chapter 3. Small silicon integrated antennas for the remote sensors

Spectrum analyzer

73

Microstrip patch receiving antenna

Differential probe

1.25 m PVC Support

Signal generator

Balun

Antenna under test Probe station metallic chuck

Figure 3.10: Propagation measurement setup used for the gain measurement of the fabricated prototypes. Gain measurements were carried out in the 5.8 GHz ISM band.

As previously said, one of the most limiting factors of the systems based on silicon integrated antennas are the low antenna gain values due to the high silicon losses. To obtain the fabricated prototypes gain, propagation measurements were carried out. In propagation measurements we could not use the mixed mode S parameters because a VNA with more than 2 ports was not available; therefore in these measurements we used a balun. Since these measurements were made at a single frequency we could use a simple microstrip narrowband balun. Antenna gain was measured using the measurement setup shown in gure 3.10. A signal generator and a 5.8 GHz balun were used to feed the transmitting antenna (gure 3.11(a)). The receiving antenna was a 5.8 GHz microstrip patch built on Rogers RO4003C substrate (r = 3.38, tan = 0.0027) (gure 3.11(b)). Measurements were performed inside the probe station, trying to maximize the clearance area near the integrated antenna and with a clear line of sight between the transmitting and the receiving antennas. Although the probe station was adapted to minimize its eects on the antenna input parameters, it still aected the measurements. For this reason, two scenarios were considered in simulations. The rst one was a simulation scenario that emulates the measurement setup: under the antenna there was a PVC block that represents the support and a PEC surface that models the eects of the probe station metallic chuck. In this scenario, good agreement between simulations and measurements was achieved (gure 3.12). Three identical antennas from dierent wafers were measured, proving the good repeatability of the fabrication process and measurements. In the second scenario the

74 3.3. Miniaturization techniques applied to low resistivity silicon integrated antennas

Signal Ge nerator

BALUN

(a)

(b)

Figure 3.11: Dierent components used for the propagation measurements. (a) Signal generator with a balun. (b) Receiving patch antenna fabricated on Rogers RO4003C substrate (r = 3.38, tan = 0.0027). antenna was analyzed in free space in order to evaluate its performance when it is not aected by the probe station.

3.3

Miniaturization techniques applied to low resistivity silicon integrated antennas

The sensors will use the 5.8 GHz ISM frequency band to transmit the collected information to the central unit. At 5.8 GHz the wavelength is much larger than the chip dimensions, for this reason miniaturization techniques are necessary. In this section, the performance oered by various miniaturization techniques when they are applied to silicon integrated antennas will be analyzed. It is important to keep in mind that we will work with a high-loss substrate and this will aect the performance oered by these techniques. The achieved miniaturization factors will be presented, the eects on the radiation eciency will be analyzed, and these results will be compared with the results obtained with a low-loss substrate.

3.3.1

Antenna designs

Dierent miniaturization techniques (gure 3.13) were used to obtain resonant dipoles around 5 - 6 GHz with a chip size of 12.5 mm x 2 mm. Two top-loading techniques were used, one based on metallic circles and the other based on capacitive loads. We also built several meandered dipoles with dierent number of meanders and strip widths (100 m

Chapter 3. Small silicon integrated antennas for the remote sensors

Simulated Measured 2 0 -5
S11 (dB)

75

Measured 1 Measured 3

-10 -15 -20 -25 -30 1 2 4 6 f (GHz) 200 175 150

Rin (W)

10

50 0
Xin (W)

-50

125 100 75

-100 -150 1

6 f (GHz)

10

50

6 f (GHz)

10

Figure 3.12: Input parameters of a meandered dipole antenna fabricated on silicon. Solid line shows the simulated results. Dashed lines show the measured results of the same antenna fabricated on dierent wafers. A 100 impedance was used as dierential reference impedance.

and 200 m). Figure 3.14 shows the simulated input parameters and the radiation eciency of one of the meandered dipoles included in this study. The optimal matching point is near 5 GHz and the eciency maximum is around 15 % at 5.3 GHz. This eciency is higher than the eciencies shown in gure 3.7 because the antenna of gure 3.7 was designed using the small chip size, and the antenna of gure 3.14 was designed using the large chip size. In the larger chips, less miniaturization is required, and as will be demonstrated later in this section, this is translated into an improvement in eciency levels. A photograph of some of the fabricated prototypes can be seen in gure 3.15.

76 3.3. Miniaturization techniques applied to low resistivity silicon integrated antennas

d L2

w1

L1

L1

w2

L1

L1

Figure 3.13: Topologies included in the miniaturization study.

Simulated 0 -5
S11 (dB)

Measured 200 175 150

Rin (W)

-10 -15 -20 -25 -30 1 50 0 2 4 6 f (GHz) 8 10

125 100 75 50 18 1 2 4 6 f (GHz) 8 10

Radiation Efficiency (%)

2 4 6 f (GHz) 8 10

15 12 9 6 3 0 1 2 4 6 f (GHz) 8 10

Xin (W)

-50

-100 -150 1

Figure 3.14: Measured and simulated results of a meandered dipole (gure 3.13) with N = 8 meanders, conductor width = 100 m and L1 = 9.2 mm.

Chapter 3. Small silicon integrated antennas for the remote sensors

77

Figure 3.15: Antennas with dierent miniaturization techniques. The chip dimensions are 12.5 mm x 2 mm.

3.3.2

Miniaturization study: Measurements and simulation results

To compare the miniaturization oered by the dierent structures it is advisable to use the resonance frequency instead of the S11 minimum position since the former is not dependent on the reference impedance. Dierent dipole lengths (L1 ) for the dierent topologies were used to guarantee resonance frequencies (fres ) of around 5 - 6 GHz, therefore they can not be compared directly. For this reason, the following gure of merit (F ), valid for dipole congurations, is dened as F = fres L1 (3.2)

where it is assumed that the resonance frequency is inversely proportional to the dipole length. Low F values are desirable to achieve high miniaturization. Using F, the miniaturization factor M F can be dened as M F (%) = 100 1 F Ff lat
dipole

(3.3)

where Ff latdipole is the measured gure of merit of a simple at dipole built on silicon (reference antenna). The M F gives information about the length and area reductions (considering a constant chip width) oered by the dierent designs. High M F values are desirable. The measured M F s obtained with the dierent topologies are plotted in gure 3.16(a). The reference antenna is a simple at dipole with Ff latdipole = 68 GHzmm. These miniaturization factors have been measured in the probe station scenario. Figure 3.16(b) shows the total simulated eciency at the resonance frequency of the same topologies.

78 3.3. Miniaturization techniques applied to low resistivity silicon integrated antennas

45
L2=3.5 W2=0.2 L2=3 W2=0.35 L2=2 W2=0.2 N=4 N=6 N=2 N=2 d=0.6 N=4 d=1.4 N=8 N=6

N=13

45 40 35
d=0.6 d=1.4

Miniaturization Factor MF (%)

40 35 30 25 20 15 10 5 0

Total Efficiency (%)

N=8

N=13

30 25 20 15 10 5

N=2 N=2 L2=2 W2=0.2

N=4 N=4

N=6 N=6

N=8 N=8

N=13

N=13 L2=3 W2=0.35 L2=3.5 W2=0.2

100mm meanders (L1 = 9.2 mm) Top capacitive load (L1 = 8 mm)
(a)

200mm meanders (L1 = 9.2 mm) Top circle load (L1 = 12 mm)
(b)

Figure 3.16: Results of the miniaturization study in the probe station scenario. (a) Measured miniaturization factor (M F ). (b) Simulated total eciency of the dierent topologies. N is the number of meanders. The dimensions of the other variables (dened in gure 3.13) are in mm. The reference antenna for the M F calculation is a simple at dipole with Ff latdipole = 68 GHzmm.

In meandered dipoles resonance frequency decreases as the number of meanders increases. This is due to the increase in physical conductor length. However, when the number of meanders is high the eciency of this miniaturization is reduced because close opposite currents tend to cancel each other out. Measurements show that thin meandered dipoles (100 m) provide more size reduction (> 40 % of the total area), whereas miniaturization factors for thick meandered dipoles (200 m) are smaller (27 %). As expected, when conductors are thin and the number of meanders (N ) is increased, losses are higher and the eciency is degraded. Therefore, there is a trade-o between antenna miniaturization and radiation eciency which can be veried comparing gure 3.16(a) (positive slope as N increases) and gure 3.16(b) (negative slope as N increases). Measured miniaturization factors for the dipoles with capacitive loads range from 30 % to 40 %, which are similar to the values obtained with the 100 m meandered dipoles. Again the drawback is that radiation eciency is reduced as the miniaturization factor is increased. Circle loaded dipoles show the best eciency levels, but the miniaturization obtained with these structures is less than 15 %. With this topology, there is not a clear trade-o between miniaturization and eciency. When the loading circles diameter is increased,

Chapter 3. Small silicon integrated antennas for the remote sensors

L2=3.5 W2=0.2 L2=3 W2=0.35 L2=2 W2=0.2 N=2 N=4 N=2 d=1.4 d=0.6 N=4 N=8 N=13 N=8 N=6

79

45

N=13

45 40 35
d=0.6 N=2 N=4 N=2 L2=2 W2=0.2 N=4 N=6 N=6 N=8 N=13 N=8 N=13 L2=3.5 W2=0.2 d=1.4

Miniaturization Factor MF (%)

40 35 30 25 20 15 10 5 0

Total Efficiency (%)

30 25 20 15 10 5

N=6

L2=3 W2=0.35

100mm meanders (L1 = 9.2 mm) Top capacitive load (L1 = 8 mm)
(a)

200mm meanders (L1 = 9.2 mm) Top circle load (L1 = 12 mm)
(b)

Figure 3.17: Results of the miniaturization study in a free space scenario. (a) Simulated miniaturization factor (M F ) (b) Simulated total eciency of the dierent topologies. N is the number of meanders. The dimensions of the other variables (dened in gure 3.13) are in mm. The reference antenna for the M F calculation is a simple at dipole with Ff latdipole = 89 GHzmm. both miniaturization and eciency are increased, however it must be taken into account that the maximum diameter is limited by the chip width. It is important to point out that the simulated results in a free space scenario (gure 3.17) are similar to those obtained in the probe station (checked by measurements). Hence, it is expected that similar results will be obtained in the real operating scenario of the antenna. Radiation eciency degradation when miniaturization techniques are applied is a well known eect. However, in silicon integrated antennas this eciency degradation is much more signicant than in low-loss substrate due to the high silicon conductivity. To prove this, the same antenna designs were simulated using a low-loss substrate with a similar dielectric constant (RO3010, r = 10.2, tan = 0.0023)[68]. The miniaturization factors obtained with this new substrate (gure 3.18(a)) are similar to the result obtained with the low resistivity silicon substrate (gure 3.16(a)). By comparing the radiation eciency of the antennas built on the silicon substrate (gure 3.16(b)) and the eciency of the antennas built on the low-loss substrate (gure 3.18(b)), it can be concluded that the eciency degradation introduced by miniaturization techniques is emphasized when low resistivity substrates are used. Indeed, in the case of the thin meandered dipoles, in low-loss substrate the radiation eciency is

80 3.3. Miniaturization techniques applied to low resistivity silicon integrated antennas degraded less than 15 % when the number of meanders is increased from 2 to 13, whereas in the case of standard CMOS silicon substrate the eciency reduction is near 50 %. Table 3.1 shows the additional losses introduced by miniaturization. With the low-loss substrate a maximum degradation equal to 0.5 dB is observed, meanwhile when the silicon substrate is used, the degradation is higher than 2.5 dB.

Table 3.1: Radiation eciency and additional losses due to miniaturization of a 100 m meandered dipole when the number of meanders is increased. N =2 23.0 % 0 dB 93.3 % 0 dB N =4 18.6 % 0.8 dB 90.8 % 0.1 dB N =6 17.4 % 1.2 dB 88.7 % 0.2 dB N =8 16.0 % 1.5 dB 86.2 % 0.3 dB N = 13 12.7 % 2.6 dB 82.7 % 0.5 dB

Standard Silicon Low-loss Substrate

Eciency Additional losses with respect to N = 2 Eciency Additional losses with respect to N = 2

50

L2=3.5 W2=0.2 L2=3 W2=0.35 L2=2 W2=0.2 N=6 N=4 N=8 N=2 N=6 N=4 d=1.4 N=8

Miniaturization Factor MF (%)

45 40 35 30 25 20 15 10 5 0

N=13

100
d=0.6 d=1.4 N=4 N=4

Radiation Efficiency (%)

95 90 85 80

N=2 N=2 L2=2 W2=0.2

N=6 N=6

N=13

N=8 N=13 N=8

N=2

L2=3 W2=0.35

N=13

d=0.6

L2=3.5 W2=0.2

100mm meanders (L1 = 9.2 mm) Top capacitive load (L1 = 8 mm)
(a)

75

200mm meanders (L1 = 9.2 mm) Top circle load (L1 = 12 mm)
(b)

Figure 3.18: Results of the miniaturization study in the probe station scenario when antennas are fabricated on a low-loss substrate (RO3010, r = 10.2, tan = 0.0023). (a) Simulated miniaturization factor (M F ). (b) Simulated radiation eciency of the dierent topologies. N is the number of meanders. The dimensions of the other variables (dened in gure 3.13) are in mm. The reference antenna for the M F calculation is a simple at dipole with Ff latdipole = 75.5 GHzmm.

Chapter 3. Small silicon integrated antennas for the remote sensors

81

3.4

Improving silicon integrated antennas by substrate micromachining

As it has been shown in the previous section, one of the main drawbacks of antenna integration are the low radiation eciency levels due to the silicon losses and the eciency degradation introduced by miniaturization techniques. In literature several solutions to improve antenna eciency can be found, such as the use of a higher substrate resistivity, addition of low-loss insulating layers, substrate thinning or selective removal of the substrate by means of micromachining. For instance, in [99] proton implantation of selected substrate areas is used to increase the silicon resistivity. The main problem of this technique is that rare special equipments are necessary. Another solution is to increase the distance between the radiating element and the lossy silicon. Thicker SiO2 layers [81] or an additional layer of a low-loss material such as benzocyclobutene (BCB) can be used to insulate the antenna. These materials have low dielectric constants and very low losses, but are dicult to apply with a thickness greater than 20 - 30 m, therefore eciency improvements are limited. In [80, 87, 100] dierent examples of application of these materials as insulation layers can be found. In [84] the silicon substrate thickness is reduced to improve the on-chip antenna eciency. Another solution is to apply micromachining techniques to selectively etch silicon of the back side of the antenna. For example, [94] uses a micromachining process to improve the eciency of a dipole antenna in the 5.8 GHz band. Other examples of micromachined antennas can be found in [87, 88, 101] In previously reported designs [87, 94] a single rectangular shaped mask has been used for the selective silicon etching process. In this section, we will carry out a comparative study of several etching patterns in order to maximize the gain improvement without a signicant increase in the antenna fragility. The antenna performance will be analyzed in various operating scenarios to prove that similar gain improvements can be expected in real applications.

3.4.1

Antenna designs, denition of the etching areas and fabrication process

The radiating element is a meandered dipole designed to work at frequencies near 5 GHz (gure 3.19(a)). The small chip dimensions (1.9 mm x 5.25 mm) were used for the prototypes fabrication. The best option to improve the radiation eciency would be to remove all the silicon

82

3.4. Improving silicon integrated antennas by substrate micromachining

M1

M2

M3

M4

M5

M6

M7

M8

(a)

(b)

Figure 3.19: Design of the fabricated prototype antennas. (a) Image of the fabricated meandered dipole antenna. (b) Schematic representation of the etching patterns used for the micromachining process. Black rectangles indicate the areas where silicon is etched.

under the antenna, but the etching of large areas is problematic because the resulting membrane structure is fragile and can be easily broken. Etching an array of smaller areas can provide similar improvements without compromising the mechanical robustness. The etching patterns under study (gure 3.19(b)) try to avoid the elimination of a large, continuous silicon area, keeping silicon walls to guarantee the structure robustness. Fabrication started with p-type silicon wafers having a resistivity of 20 Ohmcm and a thickness of 500 m. A 2 m-thick layer of SiO2 was grown with a wet oxidation process and a 4 m-thick SiO2 was deposited by Plasma Enhanced Chemical Vapor Deposition (PECVD). The resulting thickness of SiO2 (6 m) is similar to the eective thickness of the dielectric stack that some CMOS technologies have under the top metal level, so the results of the present study are valid for antennas made with standard technologies. Next, a 2 m-thick aluminum (Al) was deposited by sputtering. The Al layer was patterned by standard photolithography and wet etch. On the back side of the wafers, a

Chapter 3. Small silicon integrated antennas for the remote sensors

83

(a)

(b)

Figure 3.20: Prototype antennas with micromachining. (a) Top view of one of the antennas. (b) Rear view of some of the antennas.

1 m-thick Al layer was deposited and patterned in the same way as above. Subsequently, the SiO2 from the back side was removed with a dry etching process and a Deep-Reactive Ion Etching (DRIE) process was used to partially remove the silicon beneath the antenna structures. The fabrication was nished by removing the back side Al with a wet etching process. More information about these fabrication processes can be found in [102]. Figure 3.20 shows some of the fabricated prototypes.

3.4.2

Performance improvement due to the micromachining process: Measurements and simulation results

In gure 3.21 the input reection coecient of the antenna with dierent etching patterns is plotted. A slight frequency shift to higher frequencies can be seen due to the fact that the elimination of silicon reduces the eective relative permittivity. When micromachining is applied impedance curves also become slightly sharper due to the lower losses, reducing the matching bandwidth of the antenna. The eects of the dierent micromachining congurations on the input parameters are similar. From gure 3.21(b) it is seen that the above mentioned eects become more important when the etching depth is increased. Micromachining eects on the input reection coecient are not so important to

84
0 -5

3.4. Improving silicon integrated antennas by substrate micromachining

0 -5

S11 (dB)

-10 -15 -20 -25 -30 1 2

M1 M2 M3 M4 M5 M6 M7 M8

S11 (dB)

-10 -15 -20 -25

100 200 300 400 450

WM

6 f (GHz)
(a)

10

-30 1

WM

6 f (GHz)
(b)

10

Figure 3.21: Simulated input reection coecient. (a) Dierent etching congurations are applied. The etching depth is 450 m. (b) Dierent etching depths (m) are applied. The M2 pattern is used for the micromachining process. The wafers thickness is 500 m. WM = Without Micromachining. motivate the use of these techniques; it is in the antenna radiation parameters where a signicant performance improvement is obtained. In gure 3.22(a) the eects of the etching process in the radiation eciency can be clearly seen because an eciency improvement is observed when micromachining is applied. Depending on the etching pattern, this improvement can be higher than a factor of 2. It is important to point out that our band of interest (5.8 GHz) is near the eciency maximum. Figure 3.22(b) shows the eciency at 5.8 GHz for the dierent etching patterns as a function of the etching depth. The wafers thickness is 500 m in all cases. The eciency improves when the depth is increased, and again, this improvement also depends on the geometry of the etched areas. Similar conclusions can be obtained from gure 3.22(c). A gain improvement of up to 4 dB can be achieved in the normal direction with the best etching pattern (M1). For the same etching depth (i.e. 450 m), the gain improvement can range from 1.5 dB to 4 dB depending on the etching pattern used in the micromachining process. To verify these results some gain measurements of the fabricated prototypes were carried out. The measured and simulated gain improvements are summarized in table 3.2. Although measured gains are slightly higher than simulator predictions there is a good agreement in the trends. With an etching depth of 450 m measured gain improvements of near 4 dB are obtained. Higher etching depths would improve the gain even more, but the resulting structures would be more fragile. With 50 m of silicon in the cavities, the resulting structures are strong enough to withstand the measurements in the probe station

Chapter 3. Small silicon integrated antennas for the remote sensors

85

Radiation Efficiency (%)

5 4 3 2 1 0 1 2

M1 M2 M3 M4 M5 M6 M7 M8
WM

6 f (GHz)
(a)

10

Radiation Efficiency (%)

Gain Improvement (dB)

5 4 3 2 1

M1 M2 M3 M4 M5 M6 M7 M8
WM

5 4 3 2 1

M1 M2 M3 M4 M5 M6 M7 M8

0 50 100

200

300 Depth (mm)

(b)

400

500

0 50 100

200 300 Depth (mm)

(c)

400

500

Figure 3.22: Simulated radiation parameters when micromachining techniques are applied. (a) Radiation eciency as a function of frequency. The etching depth is 450 m. (b) and (c) radiation eciency and gain improvement as a function of the etching depth. The operating frequency is 5.8 GHz and the wafers thickness is 500 m. The reference gain for the gain improvement plot is the gain of the antenna without micromachining (18.95 dB). WM = Without Micromachining.

86

3.4. Improving silicon integrated antennas by substrate micromachining

Table 3.2: Measured and simulated gain improvement for the dierent etching patterns. The last column shows the etched area. The etching depth is 450 m and the wafers thickness is 500 m. Etching pattern M1 M2 M3 M4 M5 M6 M7 M8 Simulated Gain Improvement 3.5 3.1 3.1 2.6 2.1 1.9 1.9 1.5 Measured Gain Improvement 3.9 3.6 3.3 2.8 2.6 2.4 2.0 1.8 Etched Area (mm2 ) 2.28 3.39 3.30 1.98 2.70 2.43 2.15 2.00

without the need to ll the cavities with another dielectric material. We tried to completely remove the silicon from the patterns but in some of the prototypes the remaining dielectric membrane supporting the antenna broke, probably due to internal mechanical stress in the oxide layers. Intuitively, it could be deduced that the best etching pattern is the one that eliminates more silicon under the antenna. However, this is only partially true because the geometry used for the etching process is also important. According to table 3.2, the geometry with the highest gain improvement is not the one that eliminates most silicon. The amount of silicon etched in M2 is 50 % higher than in M1 but the measured gain improvement in M1 is 0.3 dB better than in M2. In a similar way, the amount of silicon etched in M4 and M8 is nearly the same, but M4 gain improvement is 1 dB better than M8. Based on these results, some general guidelines for the use of micromachining techniques can be given in order to maximize the gain improvement of silicon integrated dipole antennas. First of all, structures with horizontal (normal to the dipole orientation) trenches show better performance than structures with vertical ones. This is a reasonable result because horizontal trenches limit the current circulation in the substrate. Secondly, it is better to eliminate silicon near the feeding area, where currents and elds are stronger than in the areas far away from the feeding point. This can be veried in gure 3.23, where only a small part of the silicon chip around the position y is etched. When the etched area is under the feeding, gain is up to 2 dB better than when it is under the antenna ends.

Chapter 3. Small silicon integrated antennas for the remote sensors

3

87

Gain Improvement (dB)

2.5 2 1.5 1 0.5 0 -1.5 -1 -0.5 0 0.5 Dy (mm)

(b)

Dy

1.5

(a)

Figure 3.23: Simulated gain improvement as a function of the etched area position. (a) Pattern used in the micromachining process. (b) Gain improvement plot. The reference gain is the gain of the antenna without micromachining (18.95 dB).

3.4.3

Antenna performance improvement due to micromachining in a free space scenario

In the previous section the micromachining eects have been analyzed through simulations and measurements carried out inside the probe station. One could think that the obtained results only apply in that particular case. In this section, simulations are used to assess the antenna performance in free space. Comparing the S11 parameters in a free space scenario (gure 3.24(a)) with the results of the previous section (gure 3.21), the main dierence is a slight frequency shift to higher frequencies. This is related to the fact that, in the probe station scenario, the antenna is on a supporting material with a relative permittivity higher than one, which tends to reduce the resonance frequency. The gain improvement at 5.8 GHz with respect to an antenna without micromachining is shown in gure 3.24(b). The improvement is similar to the values obtained in the probe station scenario. Comparing gure 3.22 and gure 3.24, it can be seen that if the etching patterns are sorted according to their gain improvement, the same result is obtained in both scenarios. Consequently, it is expected that this improvement will also be valid in the real operating scenario of the antenna which will depend on the nal application. In this section it has been shown that with micromachining techniques a gain improvement near 4 dB is obtained. This gain improvement is relevant. If these antennas

88
0 -5

3.5. Antenna performance degradation due to the presence of coils

5

Gain Improvement (dB)

4 3 2 1

S11 (dB)

-10 -15 -20 -25 -30 2

M1 M2 M3 M4 M5 M6 M7 M8

M1 M2 M3 M4 M5 M6 M7 M8

WM

6 f (GHz)
(a)

10

0 50 100

200 300 Depth (mm)

(b)

400

500

Figure 3.24: Performance improvement due to micromachining in a free space scenario. (a) Simulated input reection coecient. The etching depth is 450 m. (b) Gain improvement as a function of the etching depth. The operating frequency is 5.8 GHz and the wafers thickness is 500 m. WM = Without Micromachining. are used in our wireless sensor network, the power required for the communication between nodes (keeping the same system performance) can be reduced by more than fty percent thus increasing battery life. These techniques can be also useful to mitigate the additional losses introduced when miniaturization techniques are applied to reduce antenna dimensions.

3.5

Antenna performance degradation due to the presence of coils for an inductive link energy coupling

According to [103], previously presented remote and implantable sensors are relatively large due to the use of a sizeable battery for powering the sensor for a long time. To solve this problem it would be of interest to have a way to transmit energy to the sensor and sporadically reload its battery without having physical access to it, thus allowing for smaller batteries. In the normal system operation mode the remote sensor continuously sends information to an external receiver using the RF link (a range of over a meter [104]), but when the battery is at, it can be reloaded with a short range inductive coupling (a range of a few centimeters [103]). The main drawback of this conguration is that the presence of the coil for the low frequency inductive coupling will signicantly degrade the RF integrated antenna performance. The coil is usually one of the bigger chip elements, and hence one of the elements that can introduce higher degradation levels to the RF link. This section will be

Chapter 3. Small silicon integrated antennas for the remote sensors

89

RF antenna Coil

Distance d

Figure 3.25: Fabricated prototype antenna for the evaluation of the coil eects on the RF link. The chip dimensions are 3.8 mm x 5.2 mm.

Figure 3.26: Prototype antennas. The chip dimensions are 3.8 mm x 5.2 mm. The distances d between the RF antenna and the coil are 320 m, 570 m, 820 m, 1.07 mm, 1.32 mm and 1.57 mm. focused on the study of the eects of this coil. One of the main parameters that denes the performance degradation in the RF link is the distance d between the coil and the RF antenna (gure 3.25). For this reason several prototypes changing this distance d were fabricated and measured (gure 3.26). As can be seen in these gures, the radiating element is a meandered dipole. The chip dimensions are 3.8 mm x 5.2 mm.

3.5.1

Performance degradation due to the coil eects: measurements and simulation results

In spite of the presence of the coil, antennas are well matched to a 100 dierential reference impedance (gure 3.27), showing that the coil does not signicantly change the antenna input parameters. Where the degradation is more signicant, is in the radiation

90

3.5. Antenna performance degradation due to the presence of coils

0 -5
S11 (dB)

-10 -15 -20 -25 -30 1 2 4 6 f (GHz) 8 10

Figure 3.27: Measured input reection coecient of an integrated antenna fabricated together with a coil for an inductive link (gure 3.25). The distance between the antenna and the coil is d = 320 m.

parameters. Figure 3.28(a) shows the simulated and measured gain of the fabricated prototypes. The x-axis represents the separation d between the antenna and the coil, ranging from 320 m to 1.57 mm. Good agreement between simulations and measurements is achieved. Measurements were repeated twice, showing that the measurement setup oers good repeatability in the results. The measured gain when the coil is at the maximum distance
-10 -15 -20 -25 -30
6

Gain Degradation (dB)

Antenna Gain (dB)

Simulation Avg Measurements Measurement 1 Measurement 2

5 4 3 2 1 0 0.5

Simulation Avg Measurements Measurement 1 Measurement 2

0.5

0.75 1 1.25 Distance d (mm)

(a)

1.5

0.75 1 1.25 Distance d (mm)

(b)

1.5

Figure 3.28: Radiation parameters as a function of the distance d between the RF antenna and the coil. (a) Antenna gain. The simulated gain without the coil is 17.7 dB. (b) Gain degradation. The reference gain is the gain when the coil is at d = 1.57 mm. Operating frequency f = 5.8 GHz.

Chapter 3. Small silicon integrated antennas for the remote sensors

91

Table 3.3: Measured gain degradation and system range reduction as a function of the distance d between the RF antenna and the coil.

Distance d (mm) 0.32 0.57 0.82 1.07 1.32 1.57

Gain degradation (dB) 5.5 3.9 2.2 1.2 0.8

System range reduction (%) 47 36 23 12 9

(d = 1.57 mm) is near -19 dB and the gain when the coil is closest to the antenna is near -24.5 dB. The simulation of the antenna without the coil gives a gain value near -17.7 dB. To reduce the total chip dimensions, it is desirable to have the coil close to the antenna, however when the distance between both elements is reduced the gain degradation is more signicant (gure 3.28(b)). Table 3.3 summarizes the measured performance degradation when the coil approaches the antenna. Taking a distance of 1.57 mm as reference, a degradation of 2.2 dB can be seen as this distance is reduced to 820 m. If the distance is further reduced to obtain a smaller chip, the gain degradation can be higher, obtaining 5.5 dB when the distance is equal to 320 m. In the last column of table 3.3 the system range reduction has been calculated assuming a free space propagation model. All the other system parameters are kept constant when the separation d is changed. Reductions of nearly 50 % of the system range can be observed. These results prove that when the inductive coil and the RF antenna are integrated in the same chip, there will be a clear trade-o between chip dimensions (limited by the separation between the coil and the RF antenna) and the system performance.

3.5.2

Performance degradation due to the coil eects in a free space scenario

In this section, the coil eects when the antenna is in a free space scenario will be analyzed by means of simulations. In this scenario, the simulated antenna gain without the coil is equal to 21.6 dB. In the probe station this gain is higher because the support and the metallic chuck modify the radiation pattern shape increasing the gain of the dipole in the broadside direction. Figure 3.29 shows the simulated gain degradation due to the coil eects in a free space scenario as a function of the distance d between the coil and the RF antenna. Despite

92

3.5. Antenna performance degradation due to the presence of coils

6

Gain Degradation (dB)

5 4 3 2 1 0 0.5 0.75 1 1.25 Distance d (mm) 1.5

Figure 3.29: Simulated gain degradation in a free space scenario as a function of the distance d between the RF antenna and the coil. The reference gain (-25.5 dB) is the gain when the coil is at d = 1.57 mm. Operating frequency f = 5.8 GHz. the fact that the absolute gain values are not the same in both scenarios, the degradation levels due to the presence of the coil are similar. Therefore it is predictable that similar degradation levels will be also obtained in the operating scenario of the real application.

3.5.3

Eects of the performance degradation in a real system link budget

Now, we will show the coil eects on the performance of a real communication system: a remote sensor network similar to the system that has inspired this thesis. The system parameters will be based on the results presented by Lin in [75]. However, we are working with low power sensors and for this reason the transmitted power will be lower than the value used in [75] in order to reduce the power consumption. The receiver antenna will be a patch antenna with a gain of 6 dB. System parameters are summarized in table 3.4. With the information of table 3.4, and assuming a free space propagation model, the Table 3.4: System parameters used for the link budget. Link Margin Tx Power Rx Antenna Gain Rx Sensitivity Thermal Noise Data rate (100 kbits/sec) Eb/No Rx Noise Figure 12 dB -10 dBm 6 dB -97.8 dBm -173.8 dBm/Hz 50 dBHz 18 dB 8 dB

Chapter 3. Small silicon integrated antennas for the remote sensors Table 3.5: System range for dierent chip congurations. Antenna Without coil With the coil at 1.57 mm With the coil at 320 m Antenna measured gain (dB) -17.7 -19.0 -24.5 System Range (m) 6.6 5.7 3

93

values of table 3.5 have been obtained. The original RF link range (without the parasitic coil) is greater than 6.5 m. However when the coil is integrated with the antenna in the same chip, this range is reduced to 5.7 m, and if the distance between the antenna and the coil is reduced to 320 m, the system range is further degraded to 3 m. In this section we have studied the eects of a coil, but the same conclusions are also valid for any other element integrated with the antenna, such as the sensor or the other electronic circuits, because in the real application the antenna will not be alone in the chip. Furthermore it is important to point out that, as was shown in the previous section, this system performance degradation could be mitigated by the use of micromachining techniques to improve the antenna radiation eciency.

3.6

Chapter conclusions

In this chapter silicon integrated antennas have been presented as a possible solution for the remote sensor antenna. Silicon integrated antennas allow the integration of all the elements into a single chip, obtaining a real system-on-chip solution and reducing the packaging costs and system dimensions. To minimize costs and make the integration feasible, the antenna fabrication process must be compatible with the standard CMOS technologies. This means that low resistivity silicon must be used as substrate, which signicantly degrades the antenna radiation eciency. The communication between the sensors and the central unit uses the 5.8 GHz ISM band. At this frequency antennas are big and dicult to integrate, and hence miniaturization techniques must be applied. With some simple techniques, like meandered lines or loaded structures, size reductions near 40 % can be obtained. However, it must be taken into account that with the low resistivity silicon substrates used for the antennas fabrication, the trade-o between miniaturization and eciency degradation becomes much more signicant. Low eciency levels could be dened as the main problem of silicon integrated antennas. The use of micromachining techniques to eliminate part of the silicon under the

94

3.6. Chapter conclusions

antenna is a good solution to improve these eciencies. The micromachining eects are more signicant in the radiation parameters than in the input parameters, and antenna gain improvements near 4 dB have been measured. The higher the etching depth, the better the eciency improvement. However, when the depth is increased the structure is more fragile and the fabrication process is more dicult (we had fabrication problems with etching depths greater than 450 m). In the micromachining process, not only the total amount of etched silicon is important, but the way this silicon is eliminated is also signicant. The most sensitive area is the area near the feeding point, where generally elds and currents are stronger, for this reason it is important to increase the amount of etched silicon in this area. Furthermore, measured and simulated results show that the congurations with trenches perpendicular to the dipole orientation oer better gain improvement. Measurements show that with the same etching depth (450 m), gain dierences greater than 2 dB can be observed depending on the etching pattern. The eects of this gain improvement can be relevant to system performance. For instance if these antennas are used in the sensors of the sensor network, keeping the same system performance, the power delivered by the sensor to the antenna can be reduced by more than fty percent when micromachining techniques are applied, signicantly increasing battery life. In real applications, the antenna will be integrated together with other elements built into the same chip. Normally these elements are small, but in some cases, their dimensions can be signicant, and therefore, the presence of these parasitic elements will degrade the RF antenna performance. In this chapter, the eects of having a coil for an inductive energy coupling near the antenna have been analyzed. The presence of the coil degrades the antenna gain taking into account that when the separation between both elements is reduced the chip can be smaller, but the performance degradation is higher. Reductions of 50 % in the system range have been measured as this separation is reduced from 1.57 mm to 320 m. To sum up, the results of this chapter demonstrate that silicon integrated antennas in the 5.8 GHz ISM frequency band are a feasible solution and that they can be fabricated using standard CMOS technologies together with other elements such as a coil or a sensor. Therefore, silicon integrated antennas are suitable for the sensors of our sensor network. The main problem is the low eciency levels that degrade the system performance, however measurements show that these eciency levels can be improved with micromachining techniques.

Chapter

Human body eects on antennas performance

In a remote sensor network, some of the wireless sensors could be implanted inside the body. The human body is a complex dielectric structure of materials with high conductivity and permittivity values that can adversely aect antenna performance. So the study of the body tissues eects is extremely important to ensure the correct operation of wireless communication systems operating in the vicinity of or within the body. This chapter will present a review of previous works and existing regulations. Then, the problems related to simulation and measurement of the body eects on antennas performance will be discussed and a detailed study of the propagation losses introduced by the body tissues in dierent ISM bands will be presented. The dierences between several body models will be also analyzed. Finally, measurements with body tissue simulating liquids will be carried out to demonstrate the simulations validity.

4.1

Introduction to the human body eects

There exists a great diversity of applications where antennas interact with human tissues, but basically two scenarios can be dened: The case where both antennas are outside the body, and the case where one of the antennas is implanted inside the body. This section will present a review of previous works and existing regulations for both scenarios.

4.1.1

Antennas in the vicinity of the human body

During the last decade there has been an important increase in public awareness of 95

96

4.1. Introduction to the human body eects

radiation hazards associated with mobile phones and in particular, with base stations. Consequently, multiple references analyzing the possible eects of cellular base stations on humans have appeared. As an example, recently the Spanish Ministry of Industry, Tourism and Commerce, and the College of Telecommunication Engineers (COIT) have jointly published a leaet [105] providing some safety guidelines related to cellular base stations, in an attempt to answer the questions that people typically ask about the possible eects of base station radiation on the human body. [106] investigates the minimum necessary distance between a human body and a cellular base station in order to comply with the threshold values of the Institute of Electrical and Electronic Engineers (IEEE) and the International Commission on Non-Ionizing Radiation Protection (ICNIRP). In [107] human exposure to electromagnetic elds is analyzed in a realistic urban environment, incorporating possible signal reections of walls and buildings. The analysis of the mobile terminal eects on the head tissues is also an important issue because the separation between the mobile phone and the users head during a call is minimal. [108] shows an extensive study of the electromagnetic absorption into the human head and neck due to mobile phones radiating at 835 MHz and 1900 MHz. The elds for various head models (adult and child head) are analyzed and compared with the regulation limits. Other references where the mobile phones eects are studied can be found in [109113], where dierent antenna congurations are considered. Another eld of interest where antennas are close to the human body are wearable devices [114] and Body Area Networks (BAN). In these cases the propagation channel is quite dierent from a free space propagation model. For this reason, there are multiple references where the problem of BAN channel modeling is analyzed [115117]. Another sensitive issue is the design of the antenna because small, exible and low prole antennas are usually required to improve user mobility and comfort. Textile antennas made of electro-textile materials are a good solution for these applications [118120]. In the previous examples the antenna is used as a radiating element for information transmission. However, in some applications, like breast tumor detection, antennas near the body are used with imaging purposes [121124]. Guidelines and regulations for antennas in the vicinity of the human body In certain European countries a social concern has been raised about the possible health hazards resulting from RF elds exposure. As a consequence, several international radiation safety standards, such as the European Commission Recommendation related to electromagnetic exposure (ERC, July 1999 [125]), have been published. Most of them are

Chapter 4. Human body eects on antennas performance

97

based on the ICNIRP guidelines [126] and the IEEE standard of 1999 [127] (updated in [128] and [129]). In the United States, the Federal Communication Commission (FCC) has provided some guidelines to assist local government ocials in devising ecient procedures to ensure that antennas comply with FCC limits for human exposure to radiofrequency [10]. In the frequency range from 100 KHz to 10 GHz, restrictions are generally based on the Specic energy Absorption Rate (SAR) values in order to prevent excessive and harmful heating of body tissues [125]. The European Commission gives the following denition: The Specic energy Absorption Rate (SAR) averaged over the whole body or over parts of the body is dened as the rate at which energy is absorbed per mass unit of body tissue and it is expressed in watts per kilogram (W/kg). Whole body SAR is a widely accepted measure for relating adverse thermal eects to RF exposure. Besides the whole body average SAR, local SAR values are necessary to evaluate and limit excessive energy deposition in small parts of the body resulting from special exposure conditions. All SAR values are to be averaged over any six minute period [125]. This denition can be expressed as [130] SAR = d dt dW dm = d dt dW dV (4.1)

where dW is the incremental energy absorbed by (dissipated in) an incremental mass (dm) contained in a volume element (dV ) of a given density () . The SAR values are related to the electric eld at a given point by [130] SAR = | E |2 (4.2)

where is the tissue conductivity and E is the RMS electric eld strength. Finally, under ideal non-thermodynamic circumstances, SAR can be related to the increase in temperature by [130] SAR = where T is the change in temperature, c T t (4.3)

t is the duration of exposure and c is the specic

heat capacity. The SAR limitations according to the ICNIRP and IEEE guidelines are summarized in table 4.1. These limitations are mainly based on the body temperature rise due to exposure, taking into account that a maximum increase lower than 1o C is desirable. The IEEE standard of 1999 [127] presented some local SAR limits dierent from the ICNIRP guidelines, but in the 2005 revision of the standard, the peak spatial average SAR values

98

4.1. Introduction to the human body eects Table 4.1: SAR: Whole body and maximum local limits [125128]. Exposure Situation Occupational General Public Whole Body Average SAR (W/Kg) 0.4 0.08 Local SAR (Head and trunk) (W/kg) 10 2

were changed with the aim of harmonizing the basic restrictions with the ICNIRP values. These limits also agree with those dened by the European Commission in [125]. In table 4.1 two exposure situations are dened. The occupationally exposed population consists of adults who are generally exposed under known conditions and are trained to be aware of potential risks and to take appropriate precautions. By contrast, the general public comprises individuals of all ages and of varying health status, including particularly susceptible groups or individuals, taking into account that in many cases members are unaware of their exposure to electromagnetic elds. Moreover, members of the general public can not reasonably be expected to take precautions to minimize or avoid exposure, thus more stringent restrictions are adopted for the general public than for the occupationally exposed population [126]. Finally, it is important to point out that the aforementioned documents are the most important recommendations and guidelines, however other documents which give specic details can be found. For instance [130] shows several measurement methods to evaluate the SAR inside the human body, [131] presents an extensive review of experimental studies about the biological eects of RF exposure and [132] gives information about the measures employed by dierent countries. In this last document it is stated that the Spanish law [133] includes the contents of the European Commission recommendation of July, 1999 [125].

4.1.2

Antennas implanted inside the human body

This section will analyze the scenario where one of the antennas is implanted inside the body and a wireless link with an external antenna is desired. The implanted antenna (with a thin insulation layer of biocompatible material) will be in contact with the human tissues, critically aecting the antenna performance. For this reason the antenna must be designed taking into account the conditions that it will nd inside the body. Another sensitive issue is signal propagation through the body. Body tissues are highly lossy, especially at high frequencies, therefore, they will introduce additional losses degrading the implanted antenna gain and the wireless link performance. Most

Chapter 4. Human body eects on antennas performance

99

Figure 4.1: Recent implantable cardiac debrillators provide in-home monitoring via wireless base stations that relay data to doctors [136].

of the eorts of this chapter will be focused on the analysis and quantication of these propagation losses. In spite of all these drawbacks, implantable antennas are an important eld of interest that is growing rapidly. During the second half of the past century an important eort was put in the design of implantable biomedical devices for the treatment of various illnesses [134]. In 1952, Dr. P. Zoll reported the rst external pacemaker [135], and some years later, the rst implantable pacemaker was presented. As time went by, these primitive devices evolved to the current ones, with better performance and longer lifetime. Other examples of applications with implantable devices are monitoring systems, drug delivery systems and debrillators [134]. In these implantable devices a wireless link with an external receiver can be used to monitor the device battery level and the clinical parameters of the patient. As an example of application, an implantable sensor with a wireless communication link could be used in a glucose monitoring system [137], automatically alerting the emergency systems when a problem is detected (gure 4.1). These implantable sensors with wireless links will be one of the key points for future e-health systems, especially for the treatment of chronic diseases. One of the most important companies in this area is Medtronic [20], a company founded more than 40 years ago that oers a wide catalog of solutions for the treatment of dierent illnesses (gure 4.2). The most commonly used frequency band for implantable antennas is the Medical Implant Communications Service (MICS) band allocated in 1999 by the FCC in the 402 405 MHz frequency range. It is a frequency band for ultra-low power, unlicensed, mobile radio services for transmitting data in diagnostic or therapeutic applications

100

4.1. Introduction to the human body eects

(a)

(b)

(c)

Figure 4.2: Medtronic oers a wide catalog of solutions for the treatment of chronic diseases [20]. (a) Real time glucose monitoring system. (b) Implantable cardioverter debrillator with wireless telemetry system. (c) Implantable cardiac monitor.

associated with implantable medical devices. Another option is the use of low frequency (13.56 MHz) inductive links [103, 138], but low ranges are obtained (4 cm) and the external coil must touch the patient body near the internal coil [103]. In [139] and [140] two antenna designs for the MICS band with a superstrate layer to improve the antenna performance were presented (gure 4.3). Due to their shape and dimensions (4 cm x 3.6 cm x 0.4 cm for the microstrip antenna and 2.4 cm x 2 cm x 0.25 cm for the PIFA), these antennas are suitable for big medical devices implanted in the human chest, like pacemakers or debrillators. In [141] a similar design is studied, analyzing the eects of the pacemaker and battery pack on the antenna performance. Other examples of antennas in the MICS band can be found in [142147]. The aforementioned examples, and the existing commercial solutions, typically use the MICS bands, however there exists a growing interest in higher frequencies, such as the 2.45 GHz ISM band, because smaller antennas can be obtained. For instance in [104] an implantable antenna for an intracranial pressure monitoring system (gure 4.4(a)) was presented. [148] and [149] analyze the cavity slot of gure 4.4(b) and [150] shows an implantable wireless telemetry board with an ISM band antenna for invivo transocular transmissions.

Chapter 4. Human body eects on antennas performance

101

(a)

(b)

Figure 4.3: Antennas for implantable devices. (a) Microstrip antenna. [139]. (b) PIFA antenna. [140]. In [144, 151] a dual band PIFA antenna for the MICS and the 2.45 GHz ISM bands is shown (gure 4.4(c)). This dual band design can be used with the ultra low power implantable transceiver Zarlink ZL70101 [152] that transmits information in the MICS band and uses a wake-up signal at 2.45 GHz. Finally, of interest is the study presented in [153], where the radiation characteristics of an ingestible wireless device in dierent frequency bands are analyzed.

Guidelines and regulations for antennas implanted inside the human body As in the rst scenario, or even more because here the antennas are inside the body, the safety issues related to the eects of electromagnetic elds on the body tissues are a point of interest and awareness. As was previously said, the most commonly used frequency bands, especially in commercial applications, is the MICS band. It is a free band for implantable devices that can be used without a specic license. Nevertheless, according to [19], the maximum

102

4.1. Introduction to the human body eects

(a)

(b)

(c)

Figure 4.4: Examples of antennas operating in the 2.45 GHz ISM band. (a) PIFA antenna for an intracranial pressure monitoring system [104]. (b) Cavity slot antenna to be implanted in the patient arm [148]. (c) Dual band antenna for the MICS and ISM bands [144].

power in any 300 KHz bandwidth radiated by medical radio transmitters operating at 402 - 405 MHz shall not exceed 25 w EIRP. The FCC also species how the EIRP measurements must be carried out. The radiated emissions and EIRP levels may be determined by measuring the radiated eld from the equipment under test at a distance of 3 meters [19]. Furthermore radiated emissions and EIRP measurements must be made in accordance with a commission-approved human body simulator and test techniques. A suitable tissue substitute material is dened in the OET Bulletin 65 Supplement C [154]. Soon afterwards the denition of the MICS band, some regulatory arrangements

Chapter 4. Human body eects on antennas performance

103

supporting MICS devices were done in order to enable the use of implantable devices worldwide. For instance the International Telecommunication Union Radiocommunications sector (ITU-R) developed a recommendation for the MICS band [155] taking into account that this band has to be shared with the Meteorological Aids Service (allocated in the same frequency range some time ago). However, it is not a problem because with the EIRP limit of MICS systems and the interference mitigation techniques used by the MICS equipment, no harmful interference should occur between both systems [155]. The European Radiocommunication Commission also allocated the same band for the same services in the recommendation ERC/REC70-03 [156]. [156] covers all kind of Short Range Devices (SDR), but annex 12 is focused on medical implants, under the name of Ultra Low Power Active Medical Implants (ULP-AMI). In this recommendation the maximum EIRP power of 25 w dened by the FCC [19] is also used. The European Telecommunications Standards Institute (ETSI) developed harmonized standards for these devices [157] where some additional information about the measurements methods to test the ULP-AMI limits is given. The aforementioned standards and recommendations are for the MICS band (402 405 MHz), but in recent years an important eort has been put into the design of antennas for implantable devices working at higher frequencies. Some examples that use the 2.45 GHz ISM band have been shown in gure 4.4. To the authors knowledge, there is not a well dened specic regulation for implantable devices at 2.45 GHz. For this reason, generally the SAR limits given for the rst scenario where the antennas were in the vicinity of the human body are also used when one of the antennas is inside the body. However, in implantable sensors the transmitted power levels are usually very low in order to increase the battery life, and hence the SAR levels are below these limits. Some other authors like K. Ito in [148], also use the power limit of 25 w for the 2.45 GHz ISM band.

4.2

Human body models

Implanted antennas performance evaluation is dicult. In simulations accurate models of the human body are necessary. In measurements, physical models called phantoms lled with special liquids are normally used. These liquids have electrical properties similar to body tissues and allow the evaluation of the antenna and the whole system performance under test conditions similar to the real body. In this section, dierent solutions for modeling the body eects will be analyzed.

104

4.2. Human body models

Figure 4.5: One layer model used by K. Ito for the antenna simulation inside the human body [148].

4.2.1

Simulations with human body models

In simulations, the simplest and fastest, but least accurate option is the use of a single layer structure of an homogenous material with properties similar to body tissues. As can be seen in gure 4.5, the antenna is simulated inside a block of this dielectric material [140, 141, 148]. However, in the one layer model there is an additional question that must be answered: What are the properties of the equivalent material used to model the overall behavior of the complex human body dielectric structure? In literature multiple solutions can be found. For instance, [149] and [158] used a material called 2/3 muscle, that is a material with the electrical properties of the muscle tissue multiplied by 2/3. In spite of the fact that 2/3 muscle material has been used by several authors, the Oce of Engineering and Technology (OET) of the FCC published some guidelines for the SAR evaluation process (OET Bulletin 65 supplement C [154]). This document gives the properties of the equivalent material that must be used (table 4.2). The conductivity of this equivalent material is slightly higher than the conductivity of muscle and other body tissues, therefore the propagation losses simulated with this equivalent material will be slightly overestimated. Another possible option is the use of a multilayer structure of nite or innite dimensions. Normally, a three layer (muscle - fat - skin) model is used [139, 159, 160].

Chapter 4. Human body eects on antennas performance

105

Table 4.2: Equivalent tissue dielectric parameters for head and body models [154]. Head r (S/m) 52.3 0.76 45.3 0.87 43.5 0.87 41.5 0.98 40.5 1.20 40.3 1.29 40.0 1.40 39.2 1.80 38.5 2.40 35.3 5.27 Body r (S/m) 61.9 0.80 58.2 0.92 56.7 0.94 55.0 1.06 54.0 1.30 53.8 1.40 53.3 1.52 52.7 1.95 52.0 2.73 48.2 6.00

Frequency (MHz) 150 300 450 915 1450 1610 1800 - 2000 2450 3000 5800

Table 4.3: Electrical properties of body tissues in dierent ISM frequency bands [161]. Relative Permittivity r 433 915 2450 5800 46.08 41.33 38.01 35.11 5.57 5.46 5.28 4.95 56.87 54.99 52.73 48.48 Conductivity (S/m) 433 915 2450 5800 0.70 0.87 1.46 3.72 0.04 0.05 0.10 0.29 0.80 0.95 1.74 4.96

Frequency (MHz) Skin Fat Muscle

The electrical properties of body tissues can be found in [161], and are summarized in table 4.3, where the skin, fat and muscle electrical parameters in some ISM bands are shown. The human head also can be modelled wit a multilayer model, like in [139] where a spherically six layer (brain - cerebrospinal uid - dura - bone - fat - skin) head model is used. If the antenna is outside the body, similar results are obtained with the one layer and the three layer models. However, as it will be shown later in this chapter, a three layer model is necessary when the antenna is inside the body because the antenna behavior can dier depending on the tissue layer where it is implanted. Finally, there is the option of using 3D anatomical models [153, 162165]. These realistic models oer the most accurate results because almost all the details of the body shape are taken into account. There are two kinds of anatomical models: the homogeneous models and the inhomogeneous models. In the rst ones, the model consist of a uniform dielectric material block with the shape of a human being, meanwhile in the second ones the dierent body tissues are taken into account. Inhomogeneous models can be created with the information of Magnetic-Resonance Imaging (MRI) [164] images where the shape of the internal organs can be obtained.

106

4.2. Human body models

(a)

(b)

Figure 4.6: The human body is an inhomogeneous structure. (a) MRI image of the thesis author. (b) Cut of the FEKO inhomogeneous body model. The simulation model takes into account most real body details.

The body tissues properties at dierent frequencies appear in [161]. Figure 4.6(a) shows a MRI image of the thesis author and gure 4.6(b) shows a cross section of a FEKO inhomogeneous model. Comparing both gures, it can be seen that the inhomogeneous model takes into account most real body details. Furthermore, models with dierent positions and kinds of person (male, female, adult, child, ....) can be used to evaluate the eects of dierent body parameters in the system performance. Many examples of human body models can be found in literature (gure 4.7) and some commercial softwares, like FEKO [70], Semcad X [166] or Empire XCcel [167], also oer their own models. In particular, the FEKO software used in this thesis provides several models of specic parts of the body, like a model of the head and the shoulders for mobile phones SAR evaluation, and three whole body models: one with a supercial mesh (SEP model), and two with a volumetric mesh to be solved with Finite Elements Methods (FEM models). The SEP model is a homogenous anatomical model. One of the FEM models is homogenous (334,733 tetrahedra) whereas the other is inhomogeneous (2.2 millions tetrahedra). Both are suitable up to 1 GHz [70]. Table 4.4 shows the peak memory usage and the simulation time necessary for the simulation of a 900 MHz half wavelength dipole antenna in front of a human body. When the inhomogeneous model is used computational requirements are high, the simulation

Chapter 4. Human body eects on antennas performance

107

(a)

(b)

(c)

Figure 4.7: Examples of human body models found in literature. (a) 3D human body in dierent positions [164]. (b) Cut of an inhomogeneous body model [168]. (c) 3D inhomogeneous head model [139]. Table 4.4: Computational requirements of FEKO body models. Model Homogeneous SEP model Homogeneous FEM model Inhomogeneous FEM model Peak Memory (GB) 9.45 2.61 15.62 Simulation time (s) 7669 1678 33618

needs 15.6 GB of memory and the analysis of each frequency takes 33600 seconds. With the homogeneous models the requirements are smaller, for instance the FEM homogeneous model only needs 2.6 GB of memory and the simulation time is below 1700 seconds. These results show that generally the simulations made with homogeneous models are faster and need less memory than the simulations made with inhomogeneous models. Therefore there

108

4.2. Human body models

exists a clear trade-o between the simulations accuracy (inhomogeneous models) and the simulations requirements (homogeneous models). Furthermore, it is important to take into account that in some scenarios homogeneous models can be perfectly valid, for instance, when a dipole antenna is placed in front of a human body (gure 4.8). In this kind of simulation where the antenna is external to the body, homogenous and inhomogeneous models give the same results. Nevertheless, when the antenna is implanted inside the body normally an inhomogeneous model has to be used to obtain accurate results. The main problem of 3D anatomical models is that computational requirements are high due to the models complexity and the need for a ne mesh to take into account all the body details. For this reason, in spite of the better accuracy of the anatomical models, the aforementioned one layer and three layers models are still very popular.

Figure 4.8: Radiation pattern of a dipole antenna in front of a human body at a distance of 10 cm. The inhomogeneous FEM model was used, but the same results were obtained with the homogeneous one. A 30 dB dynamic range has been used for the 3D far eld plot.

Chapter 4. Human body eects on antennas performance

109

4.2.2

Measurements using human body phantoms

Implanted antennas performance evaluation is not a trivial issue because it is not possible to carry out measurements in the real operating scenario, so a scenario that emulates the environment of the real human body is necessary. The most widespread solution is the use of a container with a regular or an anatomical shape lled with an special liquid [169]. The antenna is sunk into the liquid and measurements are carried out [139, 140, 148, 149, 158]. These liquids, also known as body tissues simulating liquids, must have dielectric properties similar to the human body. There are some commercial liquids [166] with the dielectric properties of table 4.2 ready to carry out this kind measurements, and SPEAG also oers complete solutions to automatically carry out these measurements, like the Dasy 5 measurement setup shown in gure 4.9 [166]. If a container with a regular shape is used (for instance a cuboid), measurements are equivalent to the one layer model simulations, meanwhile if the container has an anatomical shape, measurements are equivalent to simulations with an homogeneous anatomical body

Figure 4.9: Dasy 5 measurement setup. It is designed for SAR measurements, but it could be also used for the measurement of implanted antennas [166].

110

4.3. Study of the propagation losses introduced by body tissues

model. In section 4.4 some measurements of a patch antenna using body tissue simulating liquids with a regular shape container will be presented to demonstrate the validity of the simulator predictions. Recently Karacolak [144, 151] has been working on the use of tissue simulating gels with dielectric properties similar to the body tissues. This is an option of interest because it is a rst attempt to measure using multilayer structures. There are also some authors that have carried out measurements with real tissues, but only those of animals [150, 170].

4.3

Study of the propagation losses introduced by body tissues

One of the main problems related to implanted antennas are the additional propagation losses introduced by the body tissues that critically degrades system performance. This section will focus on the analysis of these losses. The propagation losses of body implanted antennas in the ISM bands of 433 MHz, 915 MHz, 2450 MHz and 5800 MHz will be studied. The results of the 433 MHz band can be also considered valid for the 402 - 405 MHz MICS band. It is important to point out that the aim of this study is the analysis of the body eects on propagation losses, not the design of the implanted antenna, for this reason simple straight dipoles will be used as transmitting and receiving antennas. Straight dipoles may not be the optimal solution, but they are typically used as reference antennas. Two body models will be considered, one based on a single equivalent layer and the other based on a three layer structure, showing the advantages and limitations of each one. The study with anatomical models will not be carried out due to the high system requirements and simulation times. Firstly, the one layer and three layer models will be used to evaluate the eects of introducing the antenna inside the body. The system performance as a function of several parameters like the external antenna distance, implanted antenna depth and operating frequency will be evaluated. Then, the eects of the dierent tissues thickness will be studied using the three layer model, and nally the consequences of adding a coating layer to the antenna, necessary for biocompatibility issues, will be discussed.

4.3.1

Free space wireless link

One of the main parameters that denes the performance of a wireless system is the

Chapter 4. Human body eects on antennas performance antenna pair gain (Ga) [75]. It can be calculated using the scattering parameters as

111

|S21 |2 (4.4) (1 |S11 |2 )(1 |S22 |2 ) where the term in the denominator is used to subtract the mismatch losses of the Ga = transmitting and receiving antennas. The antenna pair gain is mainly related to the propagation losses and the antennas gain. In a free space scenario, where both antennas are outside the body and far eld conditions are assumed, the antenna pair gain can be computed as Ga(f s) = GT GAir GR the air propagation gain or free space propagation gain dened as GAir = 4r
2

(4.5)

where GT is the transmitting antenna gain, GR is the receiving antenna gain and GAir is

(4.6)

where is the free space wavelength and r is the distance between both antennas.

4.3.2

One layer model

Now, the eects of having one of the antennas implanted inside the body will be studied using the one layer model. Firstly, the simulation scenario will be presented and then the simulated results obtained with the FEKO MOM code [70] will be analyzed. Simulation scenario The system under study is based on two dipole antennas, one implanted inside the human body, and one outside the body. Half wavelength resonant dipoles are used as transmitting and receiving antennas to avoid precision errors when subtracting the mismatch losses in equation 4.4. The simulation scenario and the denition of the dierent parameters are shown in gure 4.10. In order to reduce the simulation computational requirements an innite dielectric structure was used to model the body, however the results were also checked with nite dielectrics. The equivalent tissue properties can be found in table 4.2. Simulation results The insertion of the antenna inside the human body has two main eects: A decrease in the resonant antenna dimensions and a degradation of the gain due to the high permittivity and conductivity of body tissues.

112

4.3. Study of the propagation losses introduced by body tissues

L1 External dipole
H1

Implanted dipole

H2

L2

T = 70 mm

Body equivalent tissue

Figure 4.10: One layer scenario. The transmitting and receiving antennas are half wavelength resonant dipoles. The equivalent tissue properties are shown in table 4.2.

Figure 4.11(a) plots the antenna pair gain as a function of the internal antenna depth when the external antenna is 400 mm away from the body surface. The antenna pair gain decreases when the frequency is increased. For instance, if the implanted antenna depth is equal to 10 mm, the antenna pair gain decreases almost 20 dB when the frequency is changed from 2.45 GHz to 5.8 GHz. This degradation is due to the increase in the free space propagation losses (the frequency is multiplied by 2, therefore losses are increased in 6 dB) and the increase in the body conductivity (table 4.3) that introduces approximately 14 dB of additional losses.

433 MHz
-30 H1 = 400 mm

915 MHz

2450 MHz
140 120 100

5800 MHz
H1 = 400mm

Antenna Pair Gain (dB)

-40 -50 -60 -70 -80 0 5 10 H2 (mm)

(a)

L2 (mm)

80 60 40 20

15

20

10 H2 (mm)
(b)

15

20

Figure 4.11: Simulation results for the one layer model (gure 4.10). (a) Antenna pair gain. (b) Implanted dipole length to keep the antenna resonant at the desired frequency.

Chapter 4. Human body eects on antennas performance

113

Figure 4.11(b) shows the required implanted dipole length to keep the structure resonant at dierent depths. It was obtained empirically through an iterative simulation process. This length is inversely proportional to the frequency and it is signicantly smaller than one half of the free space wavelength because of the high permittivity of body tissues. With the one layer model the dipole length does not signicantly change with depth (gure 4.11(b)). According to these results, a clear trade-o between antenna dimensions and antenna performance exists. When frequency is increased, antenna dimensions are reduced, but the antenna pair gain decreases degrading the system performance.

4.3.3

Three layer model

In this section we will study the propagation losses using a three layer model to simulate more accurately the real scenario and we will highlight the one layer model limitations.

Simulation scenario In this scenario the body eects were modeled using a three layer model made of skin, fat and muscle (gure 4.12). Depending on the body area where the antenna is implanted the thickness of each layer can be dierent. For this reason thickness values used in literature may vary from author to author [137, 139, 144, 171, 172]. We used the thickness shown in gure 4.12. The body tissues properties can be found in table 4.3.
L1 H1 Skin Fat Muscle TSKIN = 3 mm H2 Implanted T = 7 mm FAT dipole L2 TMUSCLE = 60 mm

External dipole

Figure 4.12: Three layer (skin - fat - muscle) model simulation scenario. As in the previous scenario, the transmitting and receiving antennas are half wavelength resonant dipoles. The body tissues properties are shown in table 4.3.

114 Simulation results

4.3. Study of the propagation losses introduced by body tissues

Figure 4.13 shows the simulation results with the three layer model. The antenna pair gain (gure 4.13(a)) signicantly depends on the layer where the antenna is positioned. In the implanted antenna gain plot (gure 4.13(b)), the eects of the three layers can be clearly distinguished. According to table 4.3, skin and muscle tissues have higher conductivity values than fat, therefore the implanted antenna gain is lower in these layers. It must be noted that for the higher frequency bands, where the fat layer is electrically thicker, more than 10 dB of dierence can be observed. The air propagation gain (Gair ) is shown in gures 4.13(c) and 4.13(d). Simulated air propagation gain has been obtained subtracting the internal and external antennas gain from the antenna pair gain (Ga). The theoretical results using a free space propagation model (equation 4.6) are plotted with a yellow line. When the implanted antenna depth (H2 ) is changed, at plots are obtained, proving that with typical antenna depths of some millimeters, the air propagation gain does not signicantly depend on this parameter. When the distance to the external antenna (H1 ) is changed, an inversely proportional quadratic dependence is observed, showing a good agreement with the theoretical model. These results prove that when far eld conditions are fullled, body losses can be included in the implanted antenna gain and use a simple free space propagation model to evaluate the system link budget. When the implanted dipole is in the fat layer, the eective permittivity is signicantly lower than in the other layers and the dipole length (L2 ) has to be increased by approximately a factor of 2 to keep the same resonance frequency (gure 4.13(e)) Again, we can see the aforementioned trade-o between antenna dimensions and performance. When the antenna is in the fat layer, its gain is higher (gure 4.13(b)), but it has to be longer (gure 4.13(e)). In the coverage maps of gure 4.14, the antenna pair gain is plotted as a function of the implanted antenna depth (H2 ) and the distance to the external receiver (H1 ). These maps can be useful in the system design process. From the system specications, the required antenna pair gain is calculated, and then the system range for a given frequency band and an implanted antenna depth can be easily predicted with the information of gure 4.14. As an example, if the antenna pair gain has to be higher than 35 dB to assure the proper system operation, we can predict that it will be impossible to have the system operating at 5.8 GHz.

Chapter 4. Human body eects on antennas performance

433 MHz
-30

115
5800 MHz
H1 = 400 mm

915 MHz

2450 MHz
-10 -15

H1 = 400 mm

Antenna Pair Gain (dB)

-40

Gain (dB)

-50 -60 -70 -80 0 5 10 15 20

-20 -25 -30 -35 -40 0 5 10 15 20

H2 (mm)
(a)

H2 (mm)
(b)

Air Propagation Gain (dB)

Air Propagation Gain (dB)

-15 -20 -25 -30 -35 -40 0 5

H1 = 400 mm

0 -5 -10 -15 -20 -25 -30 -35 -40 100 150

H2 = 2.5 mm

H2 (mm)
(c)

10

15

20

200

250 300 H1 (mm)

350

400

(d)

140 120 100

H1 = 400mm

L2 (mm)

80 60 40 20 0 0 5 10 15 20

H2 (mm)
(e)

Figure 4.13: Simulation results with the three layer model (gure 4.12). Yellow lines show the theoretical results. (a) Antenna pair gain. (b) Implanted antenna gain. (c) Air propagation gain as a function of the implanted antenna depth. (d) Air propagation gain as a function of the distance to the external antenna. (e) Implanted dipole length to keep the antenna resonant at the desired frequency.

116

4.3. Study of the propagation losses introduced by body tissues

433 MHz Antenna Pair Gain (dB) 15 10 5 100 150 200 250 300 H1 (mm) 350 400 15 10 5 100 150 200 250 300 H1 (mm) 350 400 915 MHz Antenna Pair Gain (dB)

H2 (mm)

H2 (mm)

-25 -30 -35 -40 -45 -50 -55

2450 MHz Antenna Pair Gain (dB) 15 15

5800 MHz Antenna Pair Gain (dB)

H2 (mm)

10 5 100 150 200 250 300 H1 (mm) 350

H2 (mm)

-60 -65 -70

10 5

400

100

150

200 250 300 H1 (mm)

350

400

Figure 4.14: Coverage maps showing the antenna pair gain (Ga) at dierent frequencies as a function of the implanted antenna depth and the distance to the external antenna. Coverage maps are useful to predict the system performance as a function of the frequency band and the antennas position.

One layer model vs. three layer model This section shows a comparison between the results obtained with the one layer model and the three layer model. The objective is to point out the limitations of the one layer model, which are similar to the limitations existing in the measurements carried out using phantoms lled with an homogeneous body tissue simulating liquid. With reference to the antenna pair gain, in the one layer model (gure 4.11(a)) it is lower than in the three layer model (gure 4.13(a)). This was expected since the conductivity values used in the one layer model are slightly higher than the values used in the three layer model. Furthermore, it must be noted that the one layer model does not take into account the gain improvement when the antenna is in the fat layer. It tends to overestimate losses, so it can be accepted as a conservative approach from the point of view of the antenna pair gain. In the three layer model dierences in implanted dipole length depending on the tissue

Chapter 4. Human body eects on antennas performance

117

where the antenna is implanted are signicant and can be greater than a factor of 2 (gure 4.13(e)). This eect is not considered by the one layer model where the dipole length is kept almost constant (gure 4.11(b)). This is an important limitation of the one layer model and the measurements made using body tissue simulating liquids. If the antenna design is carried out taking into account the one layer model, the phantom measurements can agree with simulations, however, if the antenna is implanted in the fat layer, it will be detuned and the system might not work properly. The conclusion is that a three layer model must be used in the implantable antenna design process, especially if the antenna has to be implanted in the fat layer.

4.3.4

Eects of the layers thickness

We will show the eects of modifying the thickness of the skin and fat layers. The skin layer can change from 1 mm to 5 mm depending on the body part, and the fat layer from 3 mm to 15 mm. The thicker the fat layer, the wider the depth range where the gain is maximum and the higher the maximum value of this gain (gure 4.15(a)). This is due to the fact that the antenna is further separated from muscle and skin, so the losses introduced by these tissues are reduced. Furthermore, as the thickness of fat layer increases, the depth range
Tfat= 3 mm Tfat = 7 mm Tfat = 15 mm
-10 -15 H1 = 400 mm

70 60

Tfat = 10 mm 1 Layer Model H1 = 400 mm

Gain (dB)

L2 (mm)
5 10 15 H2 (mm)
(a)

50 40 30 20

-20 -25 -30 0

20

25

10 0

10 15 H2 (mm)
(b)

20

25

Figure 4.15: Eects of the fat layer thickness. Yellow lines show the results obtained with the one layer model. Operating frequency f = 915 MHz. (a)Implanted antenna gain. (b) Implanted dipole length to keep the antenna resonant at the desired frequency.

118

4.3. Study of the propagation losses introduced by body tissues

Tskin = 1 mm Tskin = 4 mm
-10 -15 H1 = 400 mm

Tskin = 2 mm Tskin = 5 mm
70 60

Tskin = 3 mm 1 Layer Model

H1 = 400 mm

Gain (dB)

-20 -25 -30

L2 (mm)
0 5 10 15 H2 (mm)
(a)

50 40 30 20

20

25

10 0

10 15 H2 (mm)
(b)

20

25

Figure 4.16: Eects of the skin layer thickness. Yellow lines show the results obtained with the one layer model. Operating frequency f = 915 MHz. (a) Implanted antenna gain. (b) Implanted dipole length to keep the antenna resonant at the desired frequency.

where the dipole has to be longer is wider (gure 4.15(b)). Comparing the one layer model and the three layer model, the dierences between both models become more signicant when the fat layer is thicker. When the skin layer thickness is increased the depth range where the antenna gain is higher changes accordingly to the fat layer position (gure 4.16(a)). The same conclusion can be drawn for the depth range where the antenna has to be longer (gure 4.16(b)).

4.3.5

Eects of a coating layer

In previous sections the antenna was directly in contact with the body. In real applications an insulation or coating layer of biocompatible material will be necessary to protect the antenna from the body tissues. Now, the eects of this coating layer will be analyzed. In [173] Merli shows a list of materials that can be used as insulation layers. In our study, a low-loss coating layer with a relative permittivity equal to r = 4 was used. Figure 4.17 shows the resonant dipole gain and length in the 2450 MHz ISM band when the coating layer thickness varies from 0.1 mm to 0.8 mm. Results with no coating are also plotted. With regard to the antenna gain, the presence of a coating layer is advantageous,

Chapter 4. Human body eects on antennas performance

0.1 mm 0.6 mm
-10 -15 H1 = 400 mm

119
0.4 mm No coating
H1 = 400 mm

0.2 mm 0.8 mm
30 25

Gain (dB)

-20 -25 -30

L2 (mm)

20 15 10

10 H2 (mm)
(a)

15

20

10 H2 (mm)
(b)

15

20

Figure 4.17: Eects of a coating layer on the antenna performance. Operating frequency f = 2450 MHz. (a) Implanted antenna gain. (b) Implanted dipole length to keep the antenna resonant at the desired frequency. especially if the antenna is in the muscle or skin layers where gain improvement of around 5 dB can be obtained with coating layers thinner than 1 mm. When the antenna is in the fat layer, the gain improvement is not so signicant because losses are not so great in this layer. From the point of view of the implanted dipole length, the use of a coating layer has a positive and a negative aspect. The positive one is that the antenna becomes less sensitive to the surrounding material as the coating layer thickness increases, and therefore, the antenna detuning in the dierent tissues will be less signicant. The negative aspect is that when the coating layer is applied the eective permittivity decreases and consequently the miniaturization obtained by the high relative permittivity of body tissues is less important. Variations in the dipole length higher than a factor of 2 can be observed depending on the coating layer thickness (gure 4.17(b)).

4.4

In-body antenna measurements with body tissue simulating liquids

All the previous results are based on FEKO [70] simulations using innite dielectrics to model body eects, but FEKO results have not been validated with measurements. In this section some antenna measurements with an antenna immersed in various liquids will

120

4.4. In-body antenna measurements with body tissue simulating liquids

be shown in order to demonstrate the validity of FEKO simulations when the antenna is inside a dielectric material. Measurements were carried out around the 2.45 GHz ISM band using a homogeneous liquid inside a rectangular receptacle, and hence they are equivalent to the one layer model used in the previous section. The antenna was measured immersed in two dierent liquids to show the simulations validity regardless of the conductivity and permittivity values of the liquid. First of all, measurements were made with distilled water ( liquid MSL 2450 [166] (
r

= 80,

tan = 0.1188 at 2.45 GHz [174176]), and then with the SPEAG body tissue simulating
r

= 52.7, = 1.95 S/m at 2.45 GHz). The MSL 2450 liquid

conductivity and permittivity at 2.45 GHz are similar to human body dielectric properties. The objective of this section is to show the validity of FEKO simulations when the antenna is inside a liquid (phantom), and not the design of the radiating element. Therefore a simple patch antenna with a superstrate was used in order to minimize the problems related to the measurement setup. In a patch antenna a dierential feeding system is not necessary, avoiding the need for a balun and the possible problems related to unbalanced currents. Furthermore, the patch ground plane minimizes the undesired eects of the coaxial cable. It is true that the study presented in section 4.3 was based on a dipole conguration and in this section a patch antenna is used to prove the FEKO simulations validity. However, if FEKO correctly models the body eects on a patch antenna, it is expected that it will also correctly model the same eects on a dipole. Moreover, it is important to point out that both kinds of antenna could be used in implantable devices. Patch congurations are useful for pacemakers, meanwhile a dipole antenna could be more suitable for a glucose sensor. The (
r

microstrip

patch

was

made

on

Taconic

TFR-45

[177]

substrate

= 4.5, tan = 0.0035), and a superstrate of the same material was used to insulate

the antenna from the liquid and improve its performance (gure 4.18). The patch and the ground plane dimensions were 28 mm x 35 mm and 80 mm x 80 mm respectively. The feeding point was at the edge of the patch to improve the matching levels when the antenna is immersed in liquid. Figure 4.19 shows the measurement setup. The antenna under test was inside a rectangular plastic container lled with liquid. A circular plastic box with holes was used to hold the antenna at a xed height. The antenna was fed using a SMA cable with a silicone gasket to avoid liquid leakages through the hole created to insert the cable into the container. One of the main advantages of this setup is that the antenna depth can be

Chapter 4. Human body eects on antennas performance

121

(a)

(b)

Figure 4.18: Microstrip patch antenna with the feeding point at the edge of the patch to improve the matching levels when it is immersed in liquid. (a) Patch antenna without superstrate. (b) Patch antenna with superstrate.

Figure 4.19: Measurement setup. The antenna under test was immersed in liquid. The depth can be easily changed by adjusting the liquid level. A ruler was used to measure the antenna depth.

122
0 -2
S11 (dB)

4.4. In-body antenna measurements with body tissue simulating liquids

0 -2
S11 (dB)

-4 -6 -8

-4 -6 -8

-10

Simulated Measured

1.5

2 f (GHz)
(a)

2.5

-10

Simulated Measured

1.5

2 f (GHz)
(b)

2.5

Figure 4.20: Input reection coecient of a microstrip patch with superstrate (gure 4.18(b)) when the antenna depth is 8 mm. The measurement setup is shown in gure 4.19. (a) Antenna in distilled water. (b) Antenna in the SPEAG body tissue simulating liquid MSL 2450 [166]. easily changed by adjusting the liquid level. The input reection coecient of the microstrip patch was measured with the antenna immersed in distilled water (gure 4.20(a)) and in the SPEAG MSL 2450 liquid [166] (gure 4.20(b)). In both scenarios the input reection coecient is better than -6 dB, and good agreement between measurements and simulations is achieved demonstrating that the antenna input parameters predicted by FEKO when the antenna is immersed in a liquid are correct. Only a systematic frequency shift of less than 5 % is observed. This is mainly due to simulation inaccuracies and tolerances in the fabrication process and in the liquids permittivity and conductivity values. Furthermore, simulations do not take into account the frequency dependence of the liquids parameters. Figure 4.20 demonstrates the validity of FEKO predictions of the input parameters. However in implantable antennas the most limiting parameter is usually the antenna gain. For this reason, measurements to prove the validity of FEKO predictions of this parameter were carried out. The antenna gain was obtained using the measurement setup shown in gure 4.21. The measured and simulated gain plots can be seen in gure 4.22. The antenna depth was changed from 1 mm to 20 mm. Good agreement between simulations and measurements is achieved with both liquids. The general trend is that when the depth is increased, the gain is reduced because the electromagnetic waves have to propagate a higher distance through the lossy media. The gain in the MSL liquid is lower than in distilled water. This was expected because the MSL liquid conductivity is higher

Chapter 4. Human body eects on antennas performance

123

External antenna

VNA

Implanted antenna (immersed in liquid)

Figure 4.21: Measurement setup used to carry out propagation measurements. The receiving antenna was a microstrip patch with superstrate designed to operate in the desired frequency band (2.45 GHz ISM band) in free space. The distance between both antennas was 1.25 m. The implanted antenna depth was modied by changing the liquid level.

124
4 2 0 -2 -4 -6 -8 -10 -12 -14 -16

4.5. Chapter conclusions

4 2 0 -2 -4 -6 -8 -10 -12 -14 -16

Antenna Gain (dB)

Antenna Gain (dB)

Simulated Measured

Simulated Measured

8 10 12 14 16 18 20 Depth (mm)
(a)

8 10 12 14 16 18 20 Depth (mm)
(b)

Figure 4.22: Antenna gain in the normal direction of the microstrip patch with superstrate (gure 4.18(b)) as a function of the depth. The measurement frequency is near the 2.45 GHz ISM band. (a) Antenna in distilled water. (b) Antenna in the SPEAG body tissue simulating liquid MSL 2450 [166].

than the water conductivity. Even with the MSL liquid, acceptable measured gain values near -10 dB are obtained. This is mainly due to the superstrate eects that signicantly improve the antenna gain. In this section we have compared the simulated and measured S11 parameter and gain when the antenna is immersed in two dierent liquids. Generally good agreement between simulations and measurements is achieved, demonstrating the validity of the FEKO simulation models used in section 4.3, and therefore, the validity of the propagation losses study.

4.5

Chapter conclusions

In some applications sensors are near or inside the body. The human body is a complex dielectric structure with high permittivity and conductivity values that signicantly aects the antenna performance. In this chapter, these eects have been studied. There are several regulations and guidelines that dene the maximum allowed power levels for this kind of applications. In the few gigahertz frequency range, these maximum power levels are usually given as SAR limitations, 2 W/kg being the maximum peak SAR value permitted for the general public. There is a specic frequency band for implantable devices, the 402 405 MHz MICS band with a maximum EIRP = 25 w, however recent works use higher frequencies in order to reduce antenna dimensions.

Chapter 4. Human body eects on antennas performance

125

The simulation of implantable antennas is complex because the body eects must be taken into account. In simulations the body eects are modeled using numerical models with properties similar to the human body called phantoms. Generally, the more detailed the simulation model, the more accurate the results but the higher the computational requirements. Furthermore, it is true that an inhomogeneous model can be necessary for implantable devices, but when the antenna is outside the body, a homogenous model is usually enough to obtain good results. In measurements, special liquids with electrical properties similar to the body tissues known as body tissue simulating liquids are used. These measurements are equivalent to the homogeneous simulation models. A study to quantify the losses when an antenna is implanted inside the body has been presented in section 4.3. Simulations show that losses increase with frequency and antenna depth. Moreover, depending on the layer where the antenna is implanted its performance can be quite dierent. When the antenna is in the fat layer, its gain is better than when it is in the muscle or skin layers (with dierences near 10 dB) but its length has to be longer to keep the same resonance frequency. The one layer model does not take into account these variations in antenna performance. This is the main limitation of the one layer model and of the measurements made using body tissue simulating liquids. The eects of an insulation layer, necessary for biocompatibility issues, have been also analyzed. The presence of the coating layer is advantageous because it improves the antenna gain and the antenna becomes less sensitive to the surrounding materials properties. When the antenna is in the muscle or skin layers, gain improvements around 5 dB have been obtained with coating layers thinner than 1 mm. The negative aspect is that the antenna has to be larger because the miniaturization obtained by the high permittivity of body tissues is less important. Finally some measurements of an antenna immersed in dierent liquids have been shown. FEKO simulations and measurements agree well, demonstrating the validity of FEKO predictions, and hence, the validity of the propagation losses study.

Chapter

Conclusions and future work

5.1 Conclusions

This thesis has been focused on the design of antennas for a distributed sensor network made up of remote sensors connected to a central unit with a wireless link. The central unit collects the sensors information and sends the data to the Internet with a WLAN connection that uses the 2.45 GHz and 5 GHz frequency bands dened by the 802.11 standards. The wireless link between the sensors and the central unit uses the 5.8 GHz ISM band, although it is also possible to use the 2.45 GHz ISM band to reduce propagation losses. The sensor antenna and the central unit antenna are quite dierent. In the central unit a multiband antenna is necessary to give service to the 802.11 and the ISM bands with a single radiating element. Multiband functionality is not necessary in the sensors, but antenna dimensions are critical. The design objectives of both antennas are completely dierent, so for this reason they have been analyzed separately in two chapters. Chapter 2 shows several multiband monopole congurations suitable for the central unit antenna. The designs presented in this chapter are based on two dierent concepts: Fractal geometries. Structures loaded with resonators. Some fractal shapes, like the Sierpinski gasket, have been used in multiband antenna designs due to their self similarity property at dierent scale levels. Fractal shapes have been typically applied to the monopole element, but the results of this thesis show that 127

128

5.1. Conclusions

similar geometries can also be used in the ground plane to improve antenna performance. A fractal-based ground plane can reinforce the monopole resonances improving the existing bands, or even creating new operating bands. Two antenna designs have been presented, one where a triangular monopole is mounted on a fractal ground plane inspired by the Sierpinski gasket and one where a similar ground is used with a Sierpinski monopole. Both antennas fulll the needs of the sensor network central unit, with good matching levels and similar radiation patterns in the dierent bands, therefore both solutions could be applied. Nevertheless, the design that uses the fractal shape in both elements is more exible and it oers higher bandwidth in the lower band. The second option is the use of resonator loaded structures. The multiband behavior of a monopole antenna can be improved loading the antenna with resonators. Furthermore, with the o-resonance loading principle the high losses of the resonators at their resonances can be avoided, introducing new bands without signicantly degrading the in-band radiation eciency. Based on these ideas several antenna designs have been presented. The rst is a dual band antenna based on capacitively loaded open loop resonators that covers the 2.45 GHz and 5 GHz bands of the 802.11 standard and the 5.8 GHz ISM band. The antenna operating principles have been analyzed, showing that the position of the bands can be easily controlled by adjusting the monopole and resonators dimensions. Next, two new designs that increase the number of operating bands have been shown. The rst makes use of the SRR dual band behavior to achieve triple band performance. The second is an asymmetric structure that combines the resonances of two dierent resonators to obtain triple band performance. The third operating band available in these antennas could be used to improve the sensor network performance adding a WiMAX connection useful when the central unit is far from a WLAN access point. All the solutions presented in chapter 2 are suitable for the central unit. If we have access to the ground plane design, the fractal-based solutions can be used, while if we do not have access to the ground plane design, then the solutions based on resonator loaded structures should be used. Moreover, both techniques could be combined to further improve the multiband performance using loaded monopoles on fractal-based ground planes. Chapter 3 is focused on the remote sensors antenna design. A good option to obtain small and cheap sensors is the integration of the antenna and the electronic circuits into the same chip, obtaining a system-on-chip solution. To minimize costs, the antenna fabrication process must be compatible with the standard CMOS technologies. The main drawback is that the low resistivity silicon substrates typically used in CMOS technologies degrade the antenna performance.

Chapter 5. Conclusions and future work

129

Small sensors are desirable; therefore, miniaturization must be applied to reduce antenna dimensions. This chapter shows the performance oered by several miniaturization techniques when they are applied to antennas fabricated on silicon. The main problem is that the losses associated with miniaturization are emphasized by the use of a low resistivity substrate. For instance, the antenna gain decreases in 2.5 dB when the number of meanders is increased from 2 to 13. Micromachining techniques are a good solution to reduce the silicon losses and improve antenna eciency. The best option would be to eliminate all the silicon under the antenna, but the resulting membrane structure is fragile and can be easily broken. In this thesis it has been shown that the etching of an array of small cavities results in a good gain improvement without compromising the structure robustness. Measurements show that not only the total amount of etched silicon is important, but the way how this silicon is eliminated is also relevant. The best etching patterns are the ones that eliminate more silicon near the feeding point and create horizontal trenches. With an etching depth of 450 m, gain improvements up to 4 dB have been measured. The antenna will be fabricated in a chip together with other electronic circuits that can degrade the RF link performance. This degradation has been analyzed with a particular example: the degradation introduced by a coil necessary for an inductive energy coupling. This is one of the worse cases because coil dimensions are similar to the antenna. Measurements show that the gain degradation depends on the separation between both elements. When this separation is reduced to obtain smaller chips, the RF link degradation is greater, with measured degradations of near 7 dB when the separation between both elements is 320 m. Some sensors can be implanted inside the human body. The analysis of this scenario is complex due to the presence of the body tissues. In simulations, complex numerical models have to be used to take into account the body eects. In measurements phantoms lled with body tissue simulating liquids are needed to emulate the real operating scenario of the antenna. Chapter 4 studies the body eects when the radiating element is implanted. The main eects of having the antenna inside the body are a reduction of the antenna dimensions and a degradation of the antenna gain due to the human tissues high permittivity and conductivity values. The body eects in dierent ISM bands have been analyzed, showing that there exists a trade-o between antenna dimensions and system performance. When frequency is increased, antennas are smaller but the gain is lower, with dierences near 20 dB when the antenna depth is 10 mm. Simulations also show that the antenna behavior

130

5.2. Future work

can change depending on the tissue layer where it is implanted. If it is in the fat layer, antenna gain is better, but it has to be larger to keep the same resonance frequency. The main limitations of the one layer model have been analyzed. From the point of view of propagation losses the one layer model can be accepted as a conservative approach because it tends to overestimate the losses. However this model does not take into account the antenna detuning when it is moved from one tissue layer to another. Therefore, if the antenna design is carried out with the one layer model and nally it is implanted near or inside the fat layer, detuning problems can appear. For this reason multilayer models are recommended for the design of implantable antennas. The eects of an insulation layer, necessary for biocompatibility issues, have also been analyzed. The insulation layer improves the antenna gain and antenna dimensions are less sensitive to the layer where it is implanted. The main drawback is that the antenna is larger. The propagation losses study of this chapter is based on simulated results, however the validity of FEKO simulations has been demonstrated with measurements of a patch antenna immersed in a body tissue simulating liquid. Finally, it is important to point out that the results of this thesis have permitted the author to publish 5 journal papers and 7 conference papers (1 best paper award).

5.2

Future work

In this thesis the design of multiband antennas and silicon integrated antennas for a sensor network have been analyzed. The thesis is nished, but research never ends. For this reason, a list of topics that could be studied in future is given here: Use of other fractal shapes with self-similarity properties for the design of multiband antennas. Use of the resonator loading techniques with other kinds of antenna, like monopoles with small ground planes or PIFA antennas. Combine the use of fractal shapes with resonator loaded structures to improve multiband behavior. Study of multiband antennas performance degradation when they are in the vicinity of human tissues. Use of other fabrication processes (not necessarily compatible with standard CMOS

Chapter 5. Conclusions and future work

131

technologies) to improve antenna performance. It is important to evaluate the tradeo between performance and costs. Implementation of a measurement setup to carry out measurements of silicon integrated antennas outside the probe station. Use of an external antenna coupled to the chip (e.g. with an inductive coupling) to improve the radiation parameters without the need for physical connections between the chip and the external antenna. Study of silicon integrated antennas performance degradation when they are implanted inside the human body. Design of implantable antennas optimized for the 2.45 GHz and 5.8 GHz ISM bands.

References

[1] Apple, http://www.apple.com. [2] Nokia, http://www.nokia.com. [3] Sony - Ericsson, http://www.sonyericsson.com. [4] Hewlett Packard, http://www.hp.com. [5] V. W. S. Tang, Y. Zheng, and J. Cao, An Intelligent Car Park Management System based on Wireless Sensor Networks, 1st International Symposium on Pervasive Computing and Applications, pp. 6570, 2006. [6] S. Lee, D. Yoon, and A. Ghosh, Intelligent Parking Lot Application Using Wireless Sensor Networks, International Symposium on Collaborative Technologies and Systems (CTS), pp. 4857, 2008. [7] S. V. Srikanth, P. J. Pramod, K. P. Dileep, S. Tapas, M. U. Patil, and C. B. N. Sarat, Design and Implementation of a Prototype Smart PARKing (SPARK) System Using Wireless Sensor Networks, International Conference on Advanced Information Networking and Applications Workshops (WAINA), pp. 401406, 2009. [8] E+E Electronik: Wireless Sensor for Humidity and Temperature (EE240), http: //www.epluse.com. [9] ADF 7242 : Low Power IEEE 802.15.4/Proprietary GFSK/FSK Zero-IF 2.4 GHz Transceiver IC, Analog Devices datasheets. [10] Federal Communications Commission (FCC), http://www.fcc.gov. [11] International Telecommunication Union (ITU), http://www.itu.int. 133

134

References

[12] Part 18: Industrial, Scientic and Medical Equipment, FCC Rules and Regulations, vol. 1, pp. 877884. [13] Part 15: Radio Frequency Devices, FCC Rules and Regulations, vol. 1, pp. 751 870. [14] IEEE Standards Association, http://standards.ieee.org. [15] Bluetooth SIG, http://www.bluetooth.com. [16] Zigbee Alliance, http://www.zigbee.org. [17] S. Po, G. Dagang, M. Hapipi, and F. Hock, Incorporating Sensors into Smart Shirt for Wireless Sentinel Medical Detection and Alarm, Journal of Physics: Conference Series, vol. 34, pp. 10681072, 2006. [18] Banner Engineering: Wireless Sensors Networks, http://www.

bannerengineering.com. [19] Part 95: Personal Radio Devices, Subpart E, I, FCC Rules and Regulations, vol. 5, pp. 527586. [20] Medtronic, http://www.medtronic.com. [21] R. Tummala and M. Swaminathan, Introduction to system-on-package (SOP):

Miniaturization of the Entire System, McGraw-Hill, 2008. [22] C. A. Balanis, Antenna Theory: Analysis and Design, John Wiley & Sons, 1997. [23] A. Cardama, Ll. Jofre, J. M. Rius, J. Romeu, S. Blanch, and M. Ferrando, Antenas, Edicions UPC, 2002. [24] J. D. Kraus and R. J. Marhefka, Antennas for All Applications, John Wiley & Sons, 2002. [25] D. A. Sanchez-Hernandez, Multiband integrated antennas for 4G terminals, Artech House, 2008. [26] R. H. DuHamel and D. E. Isbell, Broadband Logarithmically Periodic Antenna Structures, IRE National Convention Record, pp. 119128, 1957. [27] V. H. Rumsey, Frequency Independent Antennas, IRE National Convention Record, pp. 114118, 1957.

References

135

[28] H. Y. Wang, Dual-Resonance Monopole Antenna with Tuning Stubs, IEEE Proceedings - Microwave, Antennas and Propagation, vol. 153, no. 4, pp. 395399, 2006. [29] H. Y. Wang and M. Zheng, Tripe-Band Wireless Local Area Network Monopole Antenna, IET Microwaves, Antennas Propagation, vol. 2, no. 4, pp. 367372, 2008. [30] S. M. Kim, K. S. Yoon, and W. G. Yang, Dual Wideband Sleeve Dipole Type Antenna for S-DMB and 2.4/5 GHz WLAN, Microwave and Optical Technology Letters, vol. 49, no. 10, pp. 24402443, 2007. [31] M. John and M. J. Ammann, Integrated Antenna for Multiband Multi-National Wireless Combined with GSM1800/PCS1900/IMT2000 + Extension, Microwave and Optical Technology Letters, vol. 48, no. 3, pp. 613615, 2006. [32] E. H. Lim and K. W. Leung, Dual-Wideband Rectangular Dielectric Resonator Antenna for WLAN Communications, Microwave and Optical Technology Letters, vol. 48, no. 2, pp. 378380, 2006. [33] Y. F. Lin, H. M. Chen, C. Y. Lin, and S. C. Pan, Planar Inverted-L Antenna With a Dielectric Resonator Feed in a Mobile Device, IEEE Transactions on Antennas and Propagation, vol. 57, no. 10, pp. 33423346, 2009. [34] J. Anguera, I. Sanz, J. Mumbru, and C. Puente, Multiband Handset Antenna with a Parallel Excitation of PIFA and Slot Radiators, IEEE Transactions on Antennas and Propagation, vol. 58, no. 2, pp. 348356, 2010. [35] A. Cabedo, J. Anguera, C. Picher, M. Ribo, and C. Puente, Multiband Handset Antenna Combining a PIFA, Slots, and Ground Plane Modes, IEEE Transactions on Antennas and Propagation, vol. 57, no. 9, pp. 25262533, 2009. [36] R. A. Bhatti, Y. Im, and S. Park, Compact PIFA for Mobile Terminals Supporting Multiple Cellular and Non-Cellular Standards, IEEE Transactions on Antennas and Propagation, vol. 57, no. 9, pp. 25342540, 2009. [37] J. Soler, F. J. Gonzalez, C. Puente, and J. Anguera, Technology Letters, vol. 41, no. 6, pp. 434437, 2004. [38] H. Peitgen, H. Jurgens, and D. Saupe, Chaos and Fractals: New Frontiers of Science, Springer Science, 2004. Advances in Loading

Techniques to Design Multifrequency Monopole Antennas, Microwave and Optical

136

References

[39] P. S. Addison, Fractals and Chaos: An Illustrated Course, IOP Publishing Ltd, 1997. [40] B. B. Mandelbrot, Fractals: Form, Chance and Dimension, W. H. Freeman, 1977. [41] B. B. Mandelbrot, The Fractal Geometry of Nature, W. H. Freeman, 1982. [42] C. Puente, J. Romeu, R. Pous, J. Ramis, and A. Hijazo, Small but Long Koch Fractal Monopole, Electronics Letters, vol. 34, no. 1, pp. 910, 1998. [43] C. Puente, J. Romeu, and A. Cardama, The Koch Monopole: a Small Fractal Antenna, IEEE Transactions on Antennas and Propagation, vol. 48, no. 11, pp. 17731781, 2000. [44] C. Puente, J. Romeu, R. Pous, X. Garcia, and F. Benitez, Fractal Multiband Antenna Based on the Sierpinski Gasket, Electronics Letters, vol. 32, no. 1, pp. 12, 1996. [45] C. Puente, J. Romeu, R. Pous, and A. Cardama, On the Behavior of the Sierpinski Multiband Fractal Antenna, IEEE Transactions on Antennas and Propagation, vol. 46, no. 4, pp. 517524, 1998. [46] C. Puente, J. Romeu, R. Bartoleme, and R. Pous, Perturbation of the Sierpinski Antenna to Allocate Operating Bands, Electronics Letters, vol. 32, no. 24, pp. 21862188, 1996. [47] C. Puente, M. Navarro, J. Romeu, and R. Pous, Variations on the Fractal Sierpinski Antenna Flare Angle, IEEE Antennas and Propagation Society International Symposium, vol. 4, pp. 23402343, 1998. [48] C. Puente, C. Borja, M. Navarro, and J. Romeu, An Iterative Model for Fractal Antennas: Application to the Sierpinski Gasket Antenna, IEEE Transactions on Antennas and Propagation, vol. 48, no. 5, pp. 713719, 2000. [49] J. Romeu and J. Soler, Generalized Sierpinski Fractal Multiband Antenna, IEEE Transactions on Antennas and Propagation, vol. 49, no. 8, pp. 12371239, 2001. [50] J. Soler, D. Garcia, C. Puente, and J. Anguera, Novel Combined Mod-p Multiband Antenna Structures Inspired on Fractal Geometries, Technology Letters, vol. 41, no. 5, pp. 423426, 2004. [51] G. F. Tsachtsiris, C. F. Soras, M. P. Karaboikis, and V. T. Makios, Analysis of a Modied Sierpinski Gasket Monopole Antenna Printed on a Dual Band Wireless Microwave and Optical

References

137

Devices, IEEE Transactions on Antennas and Propagation, vol. 52, no. 10, pp. 25712579, 2004. [52] S. R. Best, On the Radiation Pattern Characteristics of the Sierpinski and Modied Parany Gasket Antennas, IEEE Antennas and Wireless Propagation Letters, vol. 1, pp. 3942, 2002. [53] K. C. Hwang, A Modied Sierpinski Fractal Antenna for Multiband Application, IEEE Antennas and Wireless Propagation Letters, vol. 6, pp. 357360, 2007. [54] J. Anguera, C. Puente, C. Borja, and R. Montero, International Symposium, vol. 3, pp. 162165, 2001. [55] J. Anguera, E. Martinez, C. Puente, C. Borja, and J. Soler, Broad-Band DualFrequency Microstrip Patch Antenna with Modied Sierpinski Fractal Geometry, IEEE Transactions on Antennas and Propagation, vol. 52, no. 1, pp. 6673, 2004. [56] J. Anguera, E. Martinez, C. Puente, C. Borja, and J. Soler, Broadband Bowtie Microstrip Patch

Antenna Based on the Sierpinski Fractal, IEEE Antennas and Propagation Society

Triple-Frequency Microstrip Patch Radiator Combining a Dual-Band Modied Sierpinski Fractal and a Monoband Antenna, IEEE Transactions on Antennas and Propagation, vol. 54, no. 11, pp. 33673373, 2006. [57] J. Soler, J. Parron, and J. Gemio, New Multiband Asymmetric Top-Loaded Mod-p Antenna for Terrestrial and Satellite Communication Services, IEEE Antennas and Propagation Society International Symposium (AP-S), pp. 14, 2008. [58] M.C. Huynh and W. Stutzman, Antenna (PIFA) Performance, Ground Plane Eects on Planar Inverted-F IEE Proceedings - Microwaves, Antennas and

Propagation,, vol. 150, no. 4, pp. 209213, 2003. [59] F. Yang and Y. Rahmat-Samii, Reection Phase Characterizations of the EBG Ground Plane for Low Prole Wire Antenna Applications, IEEE Transactions on Antennas and Propagation, vol. 51, no. 10, pp. 26912703, 2003. [60] M. F. Abedin and M. Ali, Modifying the Ground Plane and Its Eect on Planar Inverted-F Antennas (PIFAs) for Mobile Phone Handsets, IEEE Antennas and Wireless Propagation Letters, vol. 2, no. 1, pp. 226229, 2003. [61] N. Fortino, J. Y. Dauvignac, G. Kossiavas, and R. Staraj, Design Optimization of UWB Printed Antenna for Omnidirectional Pulse Radiation, IEEE Transactions on Antennas and Propagation, vol. 56, no. 7, pp. 18751881, 2008.

138

References

[62] R. Zaker, C. Ghobadi, and J. Nourinia, Bandwidth Enhancement of Novel Compact Single and Dual Band-Notched Printed Monopole Antenna With a Pair of L-Shaped Slots, IEEE Transactions on Antennas and Propagation, vol. 57, no. 12, pp. 3978 3983, 2009. [63] M. Gopikrishna, D. D. Krishna, C. K. Anandan, P. Mohanan, and K. Vasudevan, Design of a Compact Semi-Elliptic Monopole Slot Antenna for UWB Systems, IEEE Transactions on Antennas and Propagation, vol. 57, no. 6, pp. 18341837, 2009. [64] F. J. Herraiz-Martinez, D. Segovia-Vargas, L. E. Garcia-Munoz, and V. GonzalezPosadas, Dual-Frequency Printed Dipole Loaded with Metamaterial Particles, IEEE Antennas and Propagation Society International Symposium (AP-S), pp. 14, 2008. [65] F. J. Herraiz-Martinez, L. E. Garcia-Munoz, D. Gonzalez-Ovejero, V. GonzalezPosadas, and D. Segovia-Vargas, Dual-Frequency Printed Dipole Loaded with Split Ring Resonators, IEEE Antennas and Wireless Propagation Letters, vol. 8, pp. 137140, 2009. [66] R. D. Straw, The ARRL Antenna Book, ARRL: The National Association for Amateur Radio, 2007. [67] R. Villarino, G. Junkin, J. Parron, and J. M. Gonzalez-Arbesu, Smaller Resonators for Articial Magnetic Surfaces, International Conference on Electromagnetics in Advanced Applications (ICEAA), pp. 169172, 2007. [68] Rogers Corporation, http://www.rogerscorp.com. [69] P. Miskovsky, J.M. Gonzalez-Arbesu, and J. Romeu, Antenna Radiation Eciency Measurement in an Ultrawide Frequency Range, IEEE Antennas and Wireless Propagation Letters, vol. 8, pp. 7275, 2009. [70] FEKO: Comprehensive Electromagnetic Solutions, http://www.feko.info. [71] A. Garcia-Lamperez and M. Salazar-Palma, Dual Band Filter with Split-Ring Resonators, IEEE MTT-S International Microwave Symposium Digest, pp. 519 522, 2006. [72] J. W. Fan, C. H. Liang, and X. W. Dai, Design of Cross-Coupled Dual-Band Filter with Equal-Length Split-Ring Resonators, Progress In Electromagnetics Research, vol. 75, pp. 285293, 2007.

References

139

[73] P. Gay-Balmaz and O. J. F. Martin, Electromagnetic Resonances in Individual and Coupled Split-Ring Resonators, Journal of Applied Physics, vol. 92, no. 5, pp. 29292936, 2002. [74] D. Liu, B. Gaucher, U. Pfeier, and J. Grzyb, Advanced Millimeter-Wave

Technologies: Antennas, Packaging and Circuits, John Willey and Sons, 2009. [75] J. Lin, H. Wu, Y. Su, L. Gao, A. Sugavanam, J. E. Brewer, and K. O. Kenneth, Communication Using Antennas Fabricated in Silicon Integrated Circuits, IEEE Journal of Solid-State Circuits, vol. 42, no. 8, pp. 16781687, 2007. [76] J. Gemio, J. Parron, J. Sacristan, F. Vila, F. Serra-Graells, A. Baldi, and P. de Paco, Miniaturization Techniques Applied to Standard CMOS Technology Integrated Antennas, Fourth European Conference on Antennas and Propagation(EUCAP), pp. 14, 2010. [77] K. O. Kenneth, K. Kim, B. A. Floyd, J. L. Mehta, H. Yoon, C. Hung, D. Bravo, T. O. Dickson, X. Guo, R. Li, N. Trichy, J. Caserta, W. R. Bomstad, J. Branch, D. Yang, J. Bohorquez, E. Seok, L. Gao, A. Sugavanam, J. J. Lin, J. Chen, and J. E. Brewer, On-Chip Antennas in Silicon ICs and Their Application, IEEE Transactions on Electron Devices, vol. 52, no. 7, pp. 13121323, 2005. [78] B. A. Floyd, C. Hung, and K.O. Kenneth, Intra-Chip Wireless Interconnect

for Clock Distribution Implemented with Integrated Antennas, Receivers, and Transmitters, IEEE Journal of Solid-State Circuits, vol. 37, no. 5, pp. 543552, 2002. [79] J. Buechler, E. Kasper, P. Russer, and K. M. Strohm, Silicon High-ResistivitySubstrate Millimeter-Wave Technology, IEEE Transactions on Microwave Theory and Techniques, vol. 34, no. 12, pp. 15171521, 1986. [80] Y. Pinto, C. Person, D. Gloria, A. Cathelin, D. Belot, and R. Plana, Hybrid Integrated Antenna for Automotive Short Range Radar Applications at 79 GHz, 38th European Microwave Conference (EuMC), pp. 773776, 2008. [81] Y. P. Zhang, M. Sun, and L. H. Guo, On-Chip Antennas for 60 GHz Radios in Silicon Technology, IEEE Transactions on Electron Devices, vol. 52, no. 7, pp. 16641668, 2005. [82] S. Hsu, K. Wei, C. Hsu, and H. Ru-Chuang, A 60 Ghz Millimeter Wave CPW Fed Yagi Antenna Fabricated by Using 0.18 m CMOS technology, IEEE Electron Device Letters, vol. 29, no. 6, pp. 625627, 2008.

140

References

[83] J. J. Lin, L. Gao, A. Sugavanam, X. Guo, R. Li, J. E. Brewer, and K. O. Kenneth, Integrated Antennas on Silicon Substrates for Communication over Free Space, IEEE Electron Device Letters, vol. 25, no. 4, pp. 196198, 2004. [84] J. J. Lin, X. Guo, R. Li, J. Branch, J. E. Brewer, and K. O. Kenneth, 10x Improvement of Power Transmission Over Free Space Using Integrated Antennas on Silicon Substrates, IEEE Custom Integrated Circuits Conference, pp. 697700, 2004. [85] B. Jiang, J. Mao, and W. Yin, An Ecient Ladder Reector Antenna for Interchip Communications, IEEE Antennas and Wireless Propagation Letters, vol. 7, pp. 777780, 2008. [86] A. Shamim, L. Roy, N. Fong, and N. G. Tarr, 24 GHz On-Chip Antennas and Balun on Bulk Si for Air Transmission, IEEE Transactions on Antennas and Propagation, vol. 56, no. 2, pp. 303311, 2008. [87] E. Ojefors, K. Grenier, L. Mazenq, F. Bouchriha, A. Rydberg, and R. Plana, Micromachined Inverted F Antenna for Integration on Low Resistivity Silicon Substrates, IEEE Microwave and Wireless Components Letters, vol. 15, no. 10, pp. 627629, 2005. [88] I. Papapolymerou, R. Franklin Drayton, and L. P. B. Katehi, Micromachined Patch Antennas, IEEE Transactions on Antennas and Propagation, vol. 46, no. 2, pp. 275283, 1998. [89] P. V. Bijumon, Y. Antar, A. P. Freundorfer, and M. Sayer, Dielectric Resonator Antenna on Silicon Substrate for System On-Chip Applications, IEEE Transactions on Antennas and Propagation, vol. 56, no. 11, pp. 34043410, 2008. [90] C. Cao, Y. Ding, X. Yang, J. J. Lin, A. K. Verma, J. Lin, F. Martin, and K.O. Kenneth, A 24 GHz Transmitter with an On-Chip Antenna in 130nm CMOS, Symposium on VLSI Circuits, pp. 148149, 2006. [91] C. Cao, Y. Ding, X. Yang, J. J. Lin, H. Wu, A. K. Verma, J. Lin, F. Martin, and K.O. Kenneth, A 24 GHz Transmitter with On-Chip Dipole Antenna in 0.13 m CMOS, IEEE Journal of Solid-State Circuits, vol. 43, no. 6, pp. 13941402, 2008. [92] J. Gemio, J. Parron, and J. Soler, Human Body Eects on Implantable Antennas for ISM Bands Applications: Models Comparison and Propagation Losses Study, Progress In Electromagnetics Research, vol. 110, pp. 437452, 2010.

References

141

[93] M. Klemm and G. Troester, EM Energy Absorption in the Human Body Tissues due to UWB Antennas, Progress in Electromagnetics Reasearch, vol. 62, pp. 261 280, 2006. [94] J. Yan and R.D. Murch, Fabrication of a Wideband Antenna on a Low-Resistivity Silicon Substrate Using a Novel Micromachining Technique, IEEE Antennas and Wireless Propagation Letters, vol. 6, pp. 476479, 2007. [95] Instituto de Microelectronica de Barcelona - Centro Nacional de Microelectronica (IMB-CNM), http://www.cnm.es. [96] Cascade Microtech, http://www.cmicro.com. [97] Picoprobe, by GGG Industries, http://www.ggb.com. [98] W. Fan, A. Lu, L. L. Wai, and B. K. Lok, Characterization of Dierential Structures, Packaging Technology (EPTC), pp. 533537, 2003. [99] K. T. Chan, A. Chin, Y. B. Chen, Y. D. Lin, T. S. Duh, and W. J. Lin, Integrated Antennas on Si, Proton-Implanted Si and Si-on-Quartz, International Electron Devices Meeting (IEDM), pp. 40.6.140.6.4, 2001. [100] P. Abele, E. Ojefors, K. B. Schad, E. Sonmez, A. Trasser, J. Konle, and H. Schumacher, Wafer Level Integration of a 24 GHz Dierential SiGe-MMIC Oscillator with a Patch Antenna Using BCB as a Dielectric Layer, 33rd European Microwave Conference, pp. 293296, 2003. [101] E. Ojefors, H. Kratz, K. Grenier, R. Plana, and A. Rydberg, Micromachined Loop Antennas on Low Resistivity Silicon Substrates, IEEE Transactions on Antennas and Propagation, vol. 54, no. 12, pp. 35933601, 2006. [102] R. C. Jaeger, Introduction to Microelectronic Fabrication: Modular Series on Solid State Devices, Volume 5, Prentice Hall, 2001. [103] M. M. Ahmadi and G. A. Jullien, A Wireless-Implantable Microsystem for Mixed-Mode S-parameter

5th Conference on Electronics

Continuous Blood Glucose Monitoring, IEEE Transactions on Biomedical Circuits and Systems, vol. 3, no. 3, pp. 169180, 2009. [104] R. Warty, M. R. Toghi, U. Kawoos, and A. Rosen, Characterization of Implantable Antennas for Intracranial Pressure Monitoring: Reection by and Transmission Through a Scalp Phantom, IEEE Transactions on Microwave Theory and Techniques, vol. 56, no. 10, pp. 23662376, 2008.

142

References

[105] Antenas de TELEFONIA MOVIL: Preguntas frecuentes, Ministerio de Industria, Turismo y Comercio; Colegio Ocial de Ingenieros de Telecomunicacion (COIT). [106] J. Cooper, B. Marx, J. Buhl, and V. Hombach, Determination of safety distance limits for a human near a cellular base station antenna, adopting the IEEE standard or ICNIRP guidelines, Bioelectromagnetics, vol. 23, pp. 429443, 2002. [107] P. Bernardi, M. Cavagnaro, S. Pisa, and E. Piuzzi, Human Exposure to Radio Base-Station Antennas in Urban Environment, IEEE Transactions on Microwave Theory and Techniques, vol. 48, no. 11, pp. 19962002, 2000. [108] 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 Transactions on Microwave Theory and Techniques, vol. 44, no. 10, pp. 18841897, 1996. [109] Q. Yu, O. P. Gandhi, M. Aronsson, and D. Wu, An Automated SAR Measurement System for Compliance Testing of Personal Wireless Devices, IEEE Transactions on Electromagnetic Compatibility, vol. 41, no. 3, pp. 234245, 1999. [110] K. R. Mahmoud, M. El-Adawy, S. M. M. Ibrahem, R. Bansal, and S. H. ZainudDeen, Investigating the Interaction Between a Human Head and a Smart Handset for 4G Mobile Communication Systems, Progress In Electromagnetics Research C, vol. 2, pp. 169188, 2008. [111] G. Lazzi, S. S. Pattnaik, C. M. Furse, and O. P. Gandhi, Comparison of FDTD Computed and Measured Radiation Patterns of Commercial Mobile Telephones in Presence of the Human Head, IEEE Transactions on Antennas and Propagation, vol. 46, no. 6, pp. 943944, 1998. [112] L.C. Kuo, Y.C. Kan, and H.R. Chuang, Analysis of a 900/1800-MHz Dual-band Gap Loop Antenna on a Handset with Proximate Head and Hand Model, Journal of Electromagnetic Waves and Applications, vol. 21, no. 1, pp. 107122, 2007. [113] K. Chim, K. C. L Chan, and R. D. Murch, Investigating the Impact of Smart Antennas on SAR, IEEE Transactions on Antennas and Propagation, vol. 52, no. 5, pp. 13701374, 2004. [114] P. Bonato, Wearable Sensors and Systems: From Enabling Technology to Clinical Applications, IEEE Engineering in Medicine and Biology Magazine, vol. 29, no. 3, pp. 2536, 2010.

References [115] I. Khan, Y. I. Nechayev, and P. S. Hall, pp. 573580, 2010.

143 On-Body Diversity Channel

Characterization, IEEE Transactions on Antennas and Propagation, vol. 58, no. 2,

[116] Q. Wang, T. Tayamachi, I. Kimura, and J. Wang, An On-Body Channel Model for UWB Body Area Communications for Various Postures, IEEE Transactions on Antennas and Propagation, vol. 57, no. 4, pp. 991998, 2009. [117] S. L. Cotton and W. G. Scanlon, Characterization of the on-Body Channel in an Outdoor Environment at 2.45 GHz, 3rd European Conference on Antennas and Propagation (EuCAP 2009), pp. 722725, 2009. [118] Y. Ouyang and W. J. Chappell, High Frequency Properties of Electro-Textiles for Wearable Antenna Applications, IEEE Transactions on Antennas and Propagation, vol. 56, no. 2, pp. 381389, 2008. [119] S. Zhu and R. Langley, 926935, 2009. [120] T. Maleszka, M. Preisner, and P. Kabacik, Meshed Ground Plane Structures for Textile Antennas, 3rd European Conference on Antennas and Propagation (EuCAP), pp. 713717, 2009. [121] M. Klemm, I. Craddock, J. Leendertz, A. Preece, and R. Benjamin, Experimental and Clinical Results of Breast Cancer Detection Using UWB Microwave Radar, IEEE Antennas and Propagation Society International Symposium (AP-S), pp. 14, 2008. [122] M. Klemm, I. Craddock, J. Leendertz, A. Preece, and R. Benjamin, Radar-Based Breast Cancer Detection Using a Hemispherical Antenna Array - Experimental Results, IEEE Transactions on Antennas and Propagation, vol. 57, no. 6, pp. 16921704, 2009. [123] G. Ruvio and M. J. Ammann, A Miniaturized Antenna for UWB-Based Breast Imaging, 3rd European Conference on Antennas and Propagation (EuCAP), pp. 18641867, 2009. [124] M. Guardiola, S. Capdevila, and L. Jofre, (EuCAP), pp. 18551859, 2009. UWB BiFocusing Tomography for Dual-Band Wearable Textile Antenna on an EBG

Substrate, IEEE Transactions on Antennas and Propagation, vol. 57, no. 4, pp.

Breast Tumor Detection, 3rd European Conference on Antennas and Propagation

144

References

[125] COUNCIL RECOMMENDATION of 12 July 1999 on the Limitation of Exposure of the General Public to Electromagnetic Fields (0 Hz to 300 GHz), Ocial Journal of the European Communities, 1999. [126] International Commission on Non-Ionizing Radiation Protection (ICNIRP) Guidelines for Limiting Exposure to Time-Varying Electric, Magnetic and Electromagnetic Fields (up to 300 GHz), Health Physics, vol. 74, no. 4, pp. 494522, 1998. [127] IEEE Standard for Safety Levels with Respecto to Human Exposure to Radio Frequency Electromagnetic Fields, 3 KHz to 300 GHz, IEEE Standard C95-1-1999. [128] IEEE Standard for Safety Levels with Respecto to Human Exposure to Radio Frequency Electromagnetic Fields, 3 KHz to 300 GHz, IEEE Standard C95-1-2005. [129] IEEE Draft Standard for Safety Levels with Respect to Human Exposure to Radio Frequency Electromagnetic Fields, 3 kHz to 300 GHz-Amendment 1: Species Ceiling Limits for Induced and Contact Current, Claries Distinctions between Localized Exposure and Spatial Peak Power Density, IEEE Standard C95-1a-2010. [130] IEEE Recommended Practice for Measurements and Computations of Radio Frequency Electromagnetic Fields With Respect to Human Exposure to Such Fields,100 kHz - 300 GHz, IEEE Standard C95.3-2002 (R2008). [131] Exposure to High Frequency Electromagnetic Fields, Biological Eects and Health Consequences (100 kHz-300 GHz), International Commission on Non-Ionizing Radiation Protection, ICNIRP 16/2009. [132] Implementation Report on the Council Recommendation Limiting the Public Exposure to Electromagnetic Fields (0 Hz to 300 GHz), European Comission. [133] REAL DECRETO 1066/2001, de 28 de septiembre, por el que se aprueba el Reglamento que establece condiciones de protecci on del dominio p ublico radioel ectrico, restricciones a las emisiones radioel ectricas y medidas de protecci on sanitaria frente a emisiones radioel ectricas., Spain, Bolet n Ocial del Estado (BOE num. 234), 29 / 09 / 2001, pp. 3621736227. [134] W. Greatbatch and C. F. Holmes, History of Implantable Devices, IEEE

Engineering in Medicine and Biology Magazine, vol. 10, no. 3, pp. 3841, 1991. [135] P. Zoll, 2376, 1952. Resuscitation of the Heart in Ventricular Standstill by External

Stimulation, The New England Journal of Medicine, vol. 56, no. 10, pp. 2366

References

145

[136] D. Halperin, T. Kohno, T. S. Heydt-Benjamin, K. Fu, and W. H. Maisel, Security and Privacy for Implantable Medical Decives, IEEE Pervasive Computing, vol. 7, no. 1, pp. 3039, 2008. [137] D. Panescu, Emerging Technologies: Wireless Communication Systems for IEEE Engineering in Medicine and Biology

Implantable Medical Devices,

Magazine, vol. 27, no. 2, pp. 96101, 2008. [138] U. Jow and M. Ghovanloo, Modeling and Optimization of Printed Spiral Coils in Air, Saline, and Muscle Tissue Environments, IEEE Transactions on Biomedical Circuits and Systems, vol. 3, no. 5, pp. 339347, 2009. [139] J. Kim and Y. Rahmat-Samii, Implanted Antennas Inside a Human Body:

Simulations, Designs, and Characterizations, IEEE Transactions on Microwave Theory and Techniques, vol. 52, no. 8, pp. 19341943, 2004. [140] J. Kim and Y. Rahmat-Samii, Planar Inverted-F Antennas on Implanted Medical Devices: Meandered Type Versus Spiral Type, Microwave and Optical Technology Letters, vol. 48, no. 3, pp. 567572, 2006. [141] P. Soontornpipit, C. M. Furse, and Y. C. Chung, Design of Implantable Microstrip Antenna for Communication with Medical Implants, IEEE Transactions on Microwave Theory and Techniques, vol. 52, no. 8, pp. 19441951, 2004. [142] J. Abadia, F. Merli, J. F. Zurcher, J. R. Mosig, and A. K. Skrivervik, 3D-Spiral Small Antenna for Biomedical Transmission Operating within the MICS Band, 3rd European Conference on Antennas and Propagation (EuCAP), pp. 18451849, 2009. [143] C. M. Lee, T. C. Yo, C. H. Luo, C. H. Tu, and Y. Z. Juang, Compact Broadband Stacked Implantable Antenna for Biotelemetry with Medical Devices, Electronics Letters, vol. 43, no. 12, pp. 660662, 2007. [144] T. Karacolak, A. Z. Hood, and E. Topsakal, Design of a Dual-Band Implantable Antenna and Development of Skin Mimicking Gels for Continuous Glucose Monitoring, IEEE Transactions on Microwave Theory and Techniques, vol. 56, no. 4, pp. 10011008, 2008. [145] F. Merli, L. Bolomey, E. Meurville, and A. K. Skrivervik, Implanted Antenna for Biomedical Applications, IEEE Antennas and Propagation Society International Symposium (AP-S), pp. 14, 2008.

146

References

[146] W. S. Chen and B. Y. Lee, Novel Printed Monopole Antenna for PDA Phone and WLAN Applications, Journal of Electromagnetic Waves and Applications, vol. 23, pp. 20732088, 2009. [147] T. Chien, C. Cheng, H. Yang, J. Jiang, and C. Luo, Development of Nonsuperstrate Implantable Low-Prole CPW-Fed Ceramic Antennas, Wireless Propagation Letters, vol. 9, pp. 599602, 2010. [148] K. Ito, Wei Xia, M. Takahashi, and K. Saito, An Implanted Cavity Slot Antenna for Medical Communication Systems, 3rd European Conference on Antennas and Propagation (EuCAP), pp. 718721, 2009. [149] W. Xia, K. Saito, M. Takahashi, and K. Ito, Performances of an Implanted Cavity Slot Antenna Embedded in the Human Arm, IEEE Transactions on Antennas and Propagation, vol. 57, no. 4, pp. 894899, 2009. [150] E. Y. Chow, C. Yang, A. Chlebowski, S. Moon, W. J. Chappell, and P. P. Irazoqui, IImplantable Wireless Telemetry Boards for In Vivo Transocular Transmission, IEEE Transactions on Microwave Theory and Techniques, vol. 56, no. 12, pp. 3200 3208, 2008. [151] T. Yilmaz, T. Karacolak, and E. Topsakal, Characterization and Testing of a Skin Mimicking Material for Implantable Antennas Operating at ISM Band (2.4 GHz2.48 GHz), IEEE Antennas and Wireless Propagation Letters, vol. 7, pp. 418420, 2008. [152] Zarlink Semiconductor, http://www.zarlink.com. [153] L. Xu, M. Q. H. Meng, H. Ren, and Y. Chan, Radiation Characteristics of Ingestible Wireless Devices in Human Intestine Following Radio Frequency Exposure at 430, 800, 1200, and 2400 MHz, IEEE Transactions on Antennas and Propagation, vol. 57, no. 8, pp. 24182428, 2009. [154] Evaluating Compliance with FCC Guidelines for Human Exposure to IEEE Antennas and

Radiofrequency Electromagnetic Fields. Supplement C: Additional Information for Evaluating Compliance of Mobile and Portable Devices with FCC Limits for Human Exposure to Radiofrequency Emissions, Oce of Engineering and Technology, Federal Communications Commission, 2001. [155] Recommendation ITU-R RS.1346 :Sharing Between the Meteorological Aids Service and Medical Implant Communication Systems (MICS) Operating in the Mobile

References

147

Service in the Frequency Band 401-406 MHz, International Telecommunication Union : Radiocommunication Sector (ITU-R). [156] ERC Recommendation 70-03 Relating to the Use of Short Range Devices (SDR), European RadioCommunications Committee (ERC). [157] ETSI EN 301 839. Electromagnetic Compatibility and Radio Spectrum Matters (ERM); Short Range Devices (SRD); Ultra Low Power Active Medical Implants (ULP-AMI) and Peripherals (ULP-AMI-P) Operating in the Frequency Range 402 MHz to 405 MHz; Part 1: Technical Characteristics and Test Methods, European Telecommunications Standards Institute (ETSI). [158] N. Haga, K. Saito, M. Takahashi, and K. Ito, Characteristics of Cavity Slot Antenna for Body-Area Networks, IEEE Transactions on Antennas and Propagation, vol. 57, no. 4, pp. 837843, 2009. [159] H. Usui, M. Takahashi, and K. Ito, Radiation Characteristics of an Implanted Cavity Slot Antenna into the Human Body, IEEE Antennas and Propagation Society International Symposium (AP-S), pp. 10951098, 2006. [160] L. Roelens, W. Joseph, and L. Martens, Characterization of the Path Loss Near Flat and Layered Biological Tissue for Narrowband Wireless Body Area Networks, International Workshop on Wearable and Implantable Body Sensor Networks (BSN), 2006. [161] Federal Communications Commission, Compilation of the Dielectric Properties of Body Tissues at RF and Microwave Frequencies, http://www.fcc.gov/oet/ rfsafety/dielectric.html. [162] M. Christopoulou, S. Koulouridis, and K. S. Nikita, Parametric Study of Power Absorption Patterns Induced in Adult and Child Head Models by Small Helical Antennas, Progress In Electromagnetics Research, vol. 94, pp. 4967, 2009. [163] M. B. Manapati and R. S. Kshetrimayum, SAR Reduction in Human Head

from Mobile Phone Radiation using Single Negative Metamaterials, Journal of Electromagnetic Waves and Applications, vol. 23, pp. 13851395, 2009. [164] Y. Liu, Z. Liang, and Z. Yang, Computation of Electromagnetic Dosimetry

for Human Body Using Parallel FDTD Algorithm Combined with Interpolation Technique, Progress in Electromagnetics Research, vol. 82, pp. 95107, 2008.

148 [165] A. Hirata, H. Sugiyama, and O. Fujiwara,

References Estimation of Core Temperature

Elevation in Humans and Animals for Whole-Body Averaged SAR, Progress In Electromagnetics Research, vol. 99, pp. 5370, 2009. [166] Schmid and Partner Engineering AG (SPEAG), http://www.speag.com. [167] Empire XCcel, http://www.empire.de. [168] A. Khaleghi and I. Balasingham, Improving In-Body Ultra Wideband

Communication Using Near-Field Coupling of the Implanted Antenna, Microwave and Optical Technology Letters, vol. 51, no. 3, pp. 585589, 2009. [169] H. H. Chou, H. T. Hsu, H. T. Chou, K. H. Liu, and F. Y. Kuo, Reduction of Peak SAR in Human Head for Handset Applications with Resistive Sheets (R-Cards), Progress In Electromagnetics Research, vol. 94, pp. 281296, 2009. [170] T. Karacolak, R. Cooper, and E. Topsakal, Electrical Properties of Rat Skin and Design of Implantable Antennas for Medical Wireless Telemetry, IEEE Transactions on Antennas and Propagation, vol. 57, no. 9, pp. 28062812, 2009. [171] A. Christ, T. Samaras, A. Klingenbock, and N. Kuster, Characterization of

the Electromagnetic Near-Field Absorption in Layered Biological Tissue in the Frequency Range from 30 MHz to 6,000 MHz, Physics in Medicine and Biology, vol. 51, pp. 49514965, 2006. [172] G. P. Chen and Z. Q. Zhao, Ultrasound Tomography-Guide TRM Technique for Breast Tumor Detecting in MITAT System, Journal of Electromagnetic Waves and Applications, vol. 24, pp. 14591471, 2010. [173] F. Merli, B. Fuchs, and A. K. Skrivervik, Inuence of Insulation for Implanted Antennas, 3rd European Conference on Antennas and Propagation (EuCAP), pp. 196199, 2009. [174] Complex Dielectric Constant of Water, http://www.random-science-tools. com/electronics/water_dielectric.htm. [175] T. Meissner and F. J. Wentz, The Complex Dielectric Constant of Pure and Sea Water From Microwave Satellite Observations, IEEE Transactions on Geoscience and Remote Sensing, vol. 42, no. 9, pp. 18361849, 2004. [176] C. A. Balanis, Advanced Engineering Electromagnetics, John Wiley & Sons, 1989. [177] Taconic, Advanced Dielectric Division, http://www.taconic-add.com.