Body Matched Antennas For Microwave Medical Applications: Xuyang Li

Karlsruher Forschungsberichte aus dem Band
Institut für Hochfrequenztechnik und Elektronik

72
Xuyang Li
Body Matched Antennas for

Microwave Medical Applications
Xuyang Li

Karlsruher Forschungsberichte
aus dem Institut für Hochfrequenztechnik und Elektronik
Herausgeber: Prof. Dr.-Ing. Thomas Zwick
Band 72
by
Xuyang Li
Dissertation, Karlsruher Institut für Technologie (KIT)
Fakultät für Elektrotechnik und Informationstechnik, 2013
Impressum
Karlsruher Institut für Technologie (KIT)

KIT Scientific Publishing
Straße am Forum 2
D-76131 Karlsruhe
KIT Scientific Publishing is a registered trademark of Karlsruhe
Institute of Technology. Reprint using the book cover is not allowed.
www.ksp.kit.edu
This document – excluding the cover – is licensed under the

Creative Commons Attribution-Share Alike 3.0 DE License
(CC BY-SA 3.0 DE): http://creativecommons.org/licenses/by-sa/3.0/de/
The cover page is licensed under the Creative Commons
Attribution-No Derivatives 3.0 DE License (CC BY-ND 3.0 DE):
http://creativecommons.org/licenses/by-nd/3.0/de/
Print on Demand 2014

ISSN 1868-4696
ISBN 978-3-7315-0147-3
Editor‘s Preface
Originally motivated by the ever increasing demand on the high data

rate in the field of mobile communication, intensive research has been
underway since many years on a new technology: the ultra wideband
(UWB) technology. The transmit signal spreads in a huge bandwidth,
which results in an extremely low power spectrum density and there-
fore low interference to other radio communication services. The
frequency spectrum released in many parts of the world since some
years ago paves the way for the future commercial UWB systems and
their research works. Meanwhile, the research focus on applications
relevant to UWB has shifted definitively to radar. Of these research
works, the medical diagnosis based on UWB technology plays an es-
sential role. Though in general the UWB technology cannot achieve
the performance of the existing technologies such as magnetic reso-
nance imaging (MRI), computed tomography (CT scan) and sonog-
raphy using ultrasound, the imaging systems based on microwaves
could well be a complementary method to these existing medical di-
agnosis systems. On the one hand, the new UWB systems can be
considerably much cheaper and more portable than MRT and CT. On
the other hand, microwaves have an advantage compared to ultra-
sound, in that they can penetrate through the bone and “see” the
tissues behind it. Besides that, the UWB technology enables wireless
data transmission from implantable sensors inside the human body
to outside equipment with relatively high data rates. However, all of
the medical applications based on UWB technology have a common
problem: the relatively high signal attenuation caused by the hu-
man tissues. A basic approach to solving this problem is to place the
antenna directly on the body. In this way, the strong reflection from
the boundary of air and human body can be significantly reduced.
Moreover, the approach allows a reduction of the antenna size due
to the reduced wavelength, which is not only essential for implanted
devices but also for a medical imaging system, since multichannel
systems with an antenna array are required.
In his dissertation Mr. Li developed the concepts and basic techniques
for the realization of body-matched antennas for medical applications.
The major challenges of this work were caused by the fact that these
antennas will directly radiate into a medium different from vacuum/air.
This does not only lead to different antenna concepts and designs but
also does not allow an antenna characterization by the standard an-
tenna measurement setups. Therefore Xuyang Li also had to develop a
novel antenna characterization setup for body-matched antennas, and
I am positive that his work will attract much interest from the research
community worldwide. With his new miniaturization and optimization
strategies for ultra-wideband body-matched antennas Mr. Li achieved
significant size reductions with concurrent efficiency maximization so
I strongly believe that his work will draw attention in the research
community worldwide. My personal wish for Mr. Li is that his creativ-
ity and great engineering talent together with his organizational skills
and team spirit will continue to earn him both scientific and economic
success.
Prof. Dr.-Ing. Thomas Zwick

– Director of the IHE –
Forschungsberichte aus dem
Institut für Höchstfrequenztechnik und Elektronik (IHE)
der Universität Karlsruhe (TH) (ISSN 0942-2935)
Herausgeber: Prof. Dr.-Ing. Dr. h.c. Dr.-Ing. E.h. mult. Werner Wiesbeck
Band 1 Daniel Kähny

Modellierung und meßtechnische Verifikation polarimetrischer,
mono- und bistatischer Radarsignaturen und deren Klassifizierung
(1992)
Band 2 Eberhardt Heidrich
Theoretische und experimentelle Charakterisierung der
polarimetrischen Strahlungs- und Streueigenschaften von Antennen
(1992)
Band 3 Thomas Kürner
Charakterisierung digitaler Funksysteme mit einem breitbandigen
Wellenausbreitungsmodell (1993)
Band 4 Jürgen Kehrbeck
Mikrowellen-Doppler-Sensor zur Geschwindigkeits- und
Wegmessung - System-Modellierung und Verifikation (1993)
Band 5 Christian Bornkessel
Analyse und Optimierung der elektrodynamischen Eigenschaften
von EMV-Absorberkammern durch numerische Feldberechnung (1994)
Band 6 Rainer Speck
Hochempfindliche Impedanzmessungen an
Supraleiter / Festelektrolyt-Kontakten (1994)
Band 7 Edward Pillai
Derivation of Equivalent Circuits for Multilayer PCB and
Chip Package Discontinuities Using Full Wave Models (1995)
Band 8 Dieter J. Cichon
Strahlenoptische Modellierung der Wellenausbreitung in
urbanen Mikro- und Pikofunkzellen (1994)
Band 9 Gerd Gottwald
Numerische Analyse konformer Streifenleitungsantennen in
mehrlagigen Zylindern mittels der Spektralbereichsmethode (1995)
Band 10 Norbert Geng
Modellierung der Ausbreitung elektromagnetischer Wellen in
Funksystemen durch Lösung der parabolischen Approximation
der Helmholtz-Gleichung (1996)
Band 11 Torsten C. Becker
Verfahren und Kriterien zur Planung von Gleichwellennetzen für
den Digitalen Hörrundfunk DAB (Digital Audio Broadcasting) (1996)
Band 12 Friedhelm Rostan

Dual polarisierte Microstrip-Patch-Arrays für zukünftige
satellitengestützte SAR-Systeme (1996)
Band 13 Markus Demmler
Vektorkorrigiertes Großsignal-Meßsystem zur nichtlinearen
Charakterisierung von Mikrowellentransistoren (1996)
Band 14 Andreas Froese
Elektrochemisches Phasengrenzverhalten von Supraleitern (1996)
Band 15 Jürgen v. Hagen
Wide Band Electromagnetic Aperture Coupling to a Cavity:
An Integral Representation Based Model (1997)
Band 16 Ralf Pötzschke
Nanostrukturierung von Festkörperflächen durch elektrochemische
Metallphasenbildung (1998)
Band 17 Jean Parlebas
Numerische Berechnung mehrlagiger dualer planarer Antennen mit
koplanarer Speisung (1998)
Band 18 Frank Demmerle
Bikonische Antenne mit mehrmodiger Anregung für den räumlichen
Mehrfachzugriff (SDMA) (1998)
Band 19 Eckard Steiger
Modellierung der Ausbreitung in extrakorporalen Therapien
eingesetzter Ultraschallimpulse hoher Intensität (1998)
Band 20 Frederik Küchen
Auf Wellenausbreitungsmodellen basierende Planung terrestrischer
COFDM-Gleichwellennetze für den mobilen Empfang (1998)
Band 21 Klaus Schmitt
Dreidimensionale, interferometrische Radarverfahren im Nahbereich
und ihre meßtechnische Verifikation (1998)
Band 22 Frederik Küchen, Torsten C. Becker, Werner Wiesbeck
Grundlagen und Anwendungen von Planungswerkzeugen für den
digitalen terrestrischen Rundfunk (1999)
Band 23 Thomas Zwick
Die Modellierung von richtungsaufgelösten Mehrwegegebäude-
funkkanälen durch markierte Poisson-Prozesse (2000)
Band 24 Dirk Didascalou

Ray-Optical Wave Propagation Modelling in Arbitrarily
Shaped Tunnels (2000)
Band 25 Hans Rudolf
Increase of Information by Polarimetric Radar Systems (2000)
Band 26 Martin Döttling
Strahlenoptisches Wellenausbreitungsmodell und Systemstudien
für den Satellitenmobilfunk (2000)
Band 27 Jens Haala
Analyse von Mikrowellenheizprozessen mittels selbstkonsistenter
finiter Integrationsverfahren (2000)
Band 28 Eberhard Gschwendtner
Breitbandige Multifunktionsantennen für den konformen Einbau
in Kraftfahrzeuge (2001)
Band 29 Dietmar Löffler
Breitbandige, zylinderkonforme Streifenleitungsantennen für den
Einsatz in Kommunikation und Sensorik (2001)
Band 30 Xuemin Huang
Automatic Cell Planning for Mobile Network Design: Optimization
Models and Algorithms (2001)
Band 31 Martin Fritzsche
Anwendung von Verfahren der Mustererkennung zur Detektion
von Landminen mit Georadaren (2001)
Band 32 Siegfried Ginter
Selbstkonsistente Modellierung der Erhitzung von biologischem
Gewebe durch hochintensiven Ultraschall (2002)
Band 33 Young Jin Park
Applications of Photonic Bandgap Structures with Arbitrary
Surface Impedance to Luneburg Lenses for Automotive Radar (2002)
Band 34 Alexander Herschlein
Entwicklung numerischer Verfahren zur Feldberechnung
konformer Antennen auf Oberflächen höherer Ordnung (2002)
Band 35 Ralph Schertlen
Mikrowellenprozessierung nanotechnologischer Strukturen am
Beispiel von Zeolithen (2002)
Band 36 Jürgen von Hagen

Numerical Algorithms for the Solution of Linear Systems of
Equations Arising in Computational Electromagnetics (2002)
Band 37 Ying Zhang
Artificial Perfect Magnetic Conductor and its Application
to Antennas (2003)
Band 38 Thomas M. Schäfer
Experimentelle und simulative Analyse der Funkwellenausbreitung
in Kliniken (2003)
Band 39 Christian Fischer
Multistatisches Radar zur Lokalisierung von Objekten im Boden (2003)
Band 40 Yan C. Venot
Entwicklung und Integration eines Nahbereichsradarsensorsystems
bei 76,5 GHz (2004)
Band 41 Christian Waldschmidt
Systemtheoretische und experimentelle Charakterisierung
integrierbarer Antennenarrays (2004)
Band 42 Marwan Younis
Digital Beam-Forming for high Resolution Wide Swath Real
and Synthetic Aperture Radar (2004)
Band 43 Jürgen Maurer
Strahlenoptisches Kanalmodell für die Fahrzeug-Fahrzeug-
Funkkommunikation (2005)
Band 44 Florian Pivit
Multiband-Aperturantennen für Basisstationsanwendungen
in rekonfigurierbaren Mobilfunksystemen (2005)
Band 45 Sergey Sevskiy
Multidirektionale logarithmisch-periodische
Indoor-Basisstationsantennen (2006)
Band 46 Martin Fritz
Entwurf einer breitbandigen Leistungsendstufe für den
Mobilfunk in Low Temperature Cofired Ceramic (2006)
Band 47 Christiane Kuhnert
Systemanalyse von Mehrantennen-Frontends (MIMO) (2006)
Band 48 Marco Liebler
Modellierung der dynamischen Wechselwirkungen von
hoch-intensiven Ultraschallfeldern mit Kavitationsblasen (2006)
Band 49 Thomas Dreyer

Systemmodellierung piezoelektrischer Sender zur Erzeugung
hochintensiver Ultraschallimpulse für die medizinische Therapie (2006)
Band 50 Stephan Schulteis
Integration von Mehrantennensystemen in kleine mobile Geräte
für multimediale Anwendungen (2007)
Band 51 Werner Sörgel
Charakterisierung von Antennen für die Ultra-Wideband-Technik
(2007)
Band 52 Reiner Lenz
Hochpräzise, kalibrierte Transponder und Bodenempfänger
für satellitengestützte SAR-Missionen (2007)
Band 53 Christoph Schwörer
Monolithisch integrierte HEMT-basierende Frequenzvervielfacher
und Mischer oberhalb 100 GHz (2008)
Band 54 Karin Schuler
Intelligente Antennensysteme für Kraftfahrzeug-Nahbereichs-
Radar-Sensorik (2007)
Band 55 Christian Römer
Slotted waveguide structures in phased array antennas (2008)
Fortführung als
"Karlsruher Forschungsberichte aus dem Institut für Hochfrequenztechnik
und Elektronik" bei KIT Scientific Publishing
(ISSN 1868-4696)
Karlsruher Forschungsberichte aus dem
(ISSN 1868-4696)
Die Bände sind unter www.ksp.kit.edu als PDF frei verfügbar

oder als Druckausgabe bestellbar.
Band 55 Sandra Knörzer

Funkkanalmodellierung für OFDM-Kommunikationssysteme
bei Hochgeschwindigkeitszügen (2009)
ISBN 978-3-86644-361-7
Band 56 Thomas Fügen
Richtungsaufgelöste Kanalmodellierung und Systemstudien
für Mehrantennensysteme in urbanen Gebieten (2009)
ISBN 978-3-86644-420-1
Band 57 Elena Pancera
Strategies for Time Domain Characterization of UWB
Components and Systems (2009)
ISBN 978-3-86644-417-1
Band 58 Jens Timmermann
Systemanalyse und Optimierung der Ultrabreitband-
Übertragung (2010)
ISBN 978-3-86644-460-7
Band 59 Juan Pontes
Analysis and Design of Multiple Element Antennas
for Urban Communication (2010)
ISBN 978-3-86644-513-0
Band 60 Andreas Lambrecht
True-Time-Delay Beamforming für ultrabreitbandige
Systeme hoher Leistung (2010)
ISBN 978-3-86644-522-2
Band 61 Grzegorz Adamiuk
Methoden zur Realisierung von dual-orthogonal, linear
polarisierten Antennen für die UWB-Technik (2010)
ISBN 978-3-86644-573-4
Band 62 Jutta Kühn

AIGaN/GaN-HEMT Power Amplifiers with Optimized
Power-Added Efficiency for X-Band Applications (2011)
ISBN 978-3-86644-615-1
(ISSN 1868-4696)
Band 63 Małgorzata Janson

Hybride Funkkanalmodellierung für ultrabreitbandige
MIMO-Systeme (2011)
ISBN 978-3-86644-639-7
Band 64 Mario Pauli
Dekontaminierung verseuchter Böden durch
Mikrowellenheizung (2011)
ISBN 978-3-86644-696-0
Band 65 Thorsten Kayser
Feldtheoretische Modellierung der Materialprozessierung
mit Mikrowellen im Durchlaufbetrieb (2011)
ISBN 978-3-86644-719-6
Band 66 Christian Andreas Sturm
Gemeinsame Realisierung von Radar-Sensorik und
Funkkommunikation mit OFDM-Signalen (2012)
ISBN 978-3-86644-879-7
Band 67 Huaming Wu
Motion Compensation for Near-Range Synthetic Aperture
Radar Applications (2012)
ISBN 978-3-86644-906-0
Band 68 Friederike Brendel
Millimeter-Wave Radio-over-Fiber Links based on
Mode-Locked Laser Diodes (2013)
ISBN 978-3-86644-986-2
Band 69 Lars Reichardt
Methodik für den Entwurf von kapazitätsoptimierten
Mehrantennensystemen am Fahrzeug (2013)
ISBN 978-3-7315-0047-6
Band 70 Stefan Beer
Methoden und Techniken zur Integration von 122 GHz
Antennen in miniaturisierte Radarsensoren (2013)
ISBN 978-3-7315-0051-3
Band 71 Łukasz Zwirełło
Realization Limits of Impulse-Radio UWB Indoor
Localization Systems (2013)
ISBN 978-3-7315-0114-5
Band 72 Xuyang Li
Body Matched Antennas for Microwave Medical Applications (2014)
ISBN 978-3-7315-0147-3
Zur Erlangung des akademischen Grades eines
DOKTOR-INGENIEURS
von der Fakultät für

Elektrotechnik und Informationstechnik,
am Karlsruher Institut für Technologie (KIT)
genehmigte
DISSERTATION
von
Dipl.-Ing. Xuyang Li
geb. in Zhejiang, China
Tag der mündlichen Prüfung: 28. 05. 2013

Hauptreferent: Prof. Dr.-Ing. Thomas Zwick
Korreferent: Prof. Dr. rer. nat. habil. Matthias Hein
Abstract
The microwave technique is a currently very attractive technique for medi-

cal applications. It uses non-ionizing electromagnetic waves and has good
penetration capability into human tissues (in the GHz range). Since the ul-
tra wideband (UWB) technology features high resolution imaging and low
system complexity, the making of a flexible yet portable device for medical
diagnosis can be achieved. Moreover wireless data transmission of physio-
logical and vital signs of patients for monitoring purposes are also feasible
due to the high data transmission capability of wideband microwave signals.
With these, a new future telemedicine home system can be envisioned, in
which the vital signs of the patients can be detected and monitored in real
time and hence improving their mobility as well as their quality of life. In
these microwave medical systems, it is ultimately the performance of the an-
tennas for radiating and receiving the signals that is the governing factor of
the overall system performance.
In this thesis, new concepts of body-matched antennas for microwave medi-
cal applications were developed, where the antennas are placed directly on the
human body or implanted in the human body. The antenna designs are mainly
for microwave medical diagnosis and data telemetry systems. Two types of
antennas were designed - on-body matched antennas for microwave medical
diagnosis and implantable antennas for data telemetry. These antennas are
designed to be matched to the human body to strongly reduce the reflection
at the boundary of air and skin, hence the dielectric properties of different
human tissues and complex multiple-layered human body must be taken into
account. Challenges of the antenna design with regard to the efficient design
process, miniaturization, application specific performance, characterization
and verification of the antennas arise and are the main focus as well as con-
tributions of the thesis.
The basis of the antenna design and matching comes from the study of the
properties of the different human tissues and the microwave propagation in
the human body. Following a systematic antenna design and optimization
procedure, the resulting antennas were then further miniaturized for compact-
ness. Since the antennas operate in the near-field, novel measurement veri-
fication systems were also designed to measure the radiation performance of
the antennas. The resulting set of antennas developed were compact in size
yet maintained a high radiation performance. They are found to be suited
as implantable antennas for IMDs and as on-body matched antennas for mi-
crowave medical imaging, operating at different frequency bands. The ap-
plicability of the on-body matched antennas for microwave imaging is also
demonstrated by means of a measurement demonstrator system with an an-
tenna array for the detection of hemorrhagic stroke. The results show the
high detection capability for stroke due to the low operational frequency, high
front-to-back ratio, and the small antenna size. All these contributions enable
a portable system to complement the current medical applications with the
objective to provide more advanced healthcare systems.
Vorwort
Die vorliegende Arbeit entstand während meiner Zeit als wissenschaftlicher

Mitarbeiter am Institut für Hochfrequenztechnik und Elektronik (IHE) des
Karlsruher Instituts für Technologie (KIT). Hiermit möchte ich mich herzlich
bei allen Personen bedanken, die zum Gelingen dieser Arbeit beigetragen
haben.
Mein besonderer Dank gilt meinem Doktorvater, Herrn Prof. Thomas Zwick,
für die konstruktiven inhaltlichen Anregungen zu dieser Arbeit und die vielfäl-
tige fachliche sowie außerfachliche Unterstützung meiner wissenschaftlichen
Tätigkeit. Bedanken möchte ich mich auch bei Herrn Prof. Matthias Hein
für die freundliche Übernahme des Korreferats und die zügige Erstellung des
Zweitgutachtens. Im Weiteren danke ich Herrn Prof. Werner Wiesbeck her-
zlich, der meine fachlichen und persönlichen Fähigkeiten stets gefördert hat.
Ein Dankeschön an M.Sc. Yoke Leen Sit, Dr.-Ing. Grzegorz Adamiuk,
Dr.-Ing. Juan Pontes und Dipl.-Ing. Kai-Philipp Pahl, für ihre sorgfältige
Durchsicht und die wertvollen Ratschläge. Ferner möchte ich meinen Zim-
merkollegen, allen wissenschaftlichen Mitarbeitern und allen Kollegen aus
Verwaltung und Technik meinen Dank aussprechen für die außergewöhn-
liche Atmosphäre und die hervorragende Zusammenarbeit. Schließlich gilt
mein ausdrücklicher Dank meinen Studierenden, die im Rahmen ihrer Ab-
schlussarbeiten mit hohem Engagement und Kreativität einen wichtigen Teil
zu meiner Arbeit geleistet haben.
Nicht zuletzt geht mein herzlicher Dank an meine Familie für ihre uneingesch-
ränkte Förderung meines Auslandsstudiums und ihre liebevolle Unterstützung
auf dem Weg zu meiner Promotion.
Karlsruhe, im November 2013
Xuyang Li
Contents
List of Symbols and Acronyms v
1. Introduction 1
1.1. State-of-the-art microwave medical applications . . . . . . . 1
1.2. Telemedicine using microwaves and its challenges . . . . . . 5
1.3. Goal and organization of this work . . . . . . . . . . . . . . 9
2. Analysis of microwave propagation in a multilayer human body 11

2.1. Microwave propagation in a lossy medium . . . . . . . . . . 12
2.2. Dielectric properties of human tissues . . . . . . . . . . . . 16
2.2.1. Debye equation and Cole-Cole equation . . . . . . . 16
2.2.2. Permittivity of human tissues . . . . . . . . . . . . . 17
2.2.3. Attenuation, penetration depth and wave impedance
in human tissues . . . . . . . . . . . . . . . . . . . 18
2.3. Analysis of microwave propagation in the human bladder . . 22
2.3.1. Multilayer model of the human abdomen . . . . . . 23
2.3.2. Concept of the radar system . . . . . . . . . . . . . 24
2.3.3. Characteristics of propagating microwave signals
in the multilayer abdomen . . . . . . . . . . . . . . 26
2.3.4. Distortion and fidelity of UWB pulses in human tissues 28
2.3.5. Analysis of the system dynamic range for urine
detection in the human bladder . . . . . . . . . . . . 31
2.4. Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
3. Characterization and measurement systems of body-matched

antennas 39
3.1. Characterization methods of body-matched antennas . . . . 40
3.1.1. Antenna terms in a lossy medium . . . . . . . . . . 40
i
Contents
3.2. Measurement techniques for the characterization of body-matched

antennas . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
3.2.1. E-field probe based antenna measurement system . . 45
3.2.2. Planar-rectangular near-field measurement system . 55
3.3. Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
4. Implantable antennas for wireless communication of IMDs 65

4.1. Overview of wireless communication of IMDs . . . . . . . . 65
4.1.1. Frequency bands and power limitations . . . . . . . 66
4.1.2. Challenges of the implantable antenna design . . . . 67
4.2. Design of stripline-fed double-layer slot antennas . . . . . . 69
4.2.1. Stripline-fed slot antenna with equally-sized
meandered strips (M3-1) . . . . . . . . . . . . . . . 72
4.2.2. Stripline-fed slot antenna with modified slotted ground
(M3-2) . . . . . . . . . . . . . . . . . . . . . . . . 77
4.2.3. Stripline-fed slot antenna with modified meandered
strips (M3-3) . . . . . . . . . . . . . . . . . . . . . 80
4.3. Verification of the stripline-fed double-layer slot antennas . . 81
4.4. Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . 88
5. On-body matched antennas for medical diagnosis 91

5.1. Motivation and design challenges of on-body matched antennas 92
5.2. Design procedure . . . . . . . . . . . . . . . . . . . . . . . 93
5.3. Differentially-fed slot antennas for the operational band from
1 to 7 GHz . . . . . . . . . . . . . . . . . . . . . . . . . . . 95
5.3.1. Double-elliptical slot antenna . . . . . . . . . . . . 96
5.3.2. Sector-like slot antenna . . . . . . . . . . . . . . . . 107
5.3.3. Stepped-slot antenna . . . . . . . . . . . . . . . . . 111
5.3.4. Stepped-slot antenna with slotline feed network . . . 114
5.3.5. Summary of the differentially-fed slot antennas . . . 120
5.4. On-body matched antennas with a lowest operational frequency
of 0.5 GHz . . . . . . . . . . . . . . . . . . . . . . . . . . . 121
5.4.1. Dual-band aperture-coupled Bowtie antenna . . . . . 122
5.4.2. Compact double-layer folded Bowtie antenna . . . . 127
5.5. Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . 141
ii
Contents
6. Body-matched antennas based microwave medical imaging 145

6.1. Measurement setup for microwave medical imaging . . . . . 146
6.1.1. Brain phantom for stroke detection . . . . . . . . . . 146
6.1.2. Array with on-body matched antennas . . . . . . . . 150
6.1.3. Measurement setup with brain phantom . . . . . . . 151
6.2. Image reconstruction with beamforming algorithm . . . . . 153
6.2.1. Calibration methods . . . . . . . . . . . . . . . . . 154
6.2.2. Delay and Sum beamformer . . . . . . . . . . . . . 157
6.3. Analysis of calibrated signals and 2D imaging results . . . . 159
6.3.1. Summary . . . . . . . . . . . . . . . . . . . . . . . 165
7. Summary and conclusions 169
A. Parameters of tissues based on Cole-Cole equation 175
Bibliography 177
Own publications 189
iii
Acronyms and Symbols
Acronyms
AUT Antenna Under Test
BW Bandwidth
CIR Channel Impulse Response
Co-pol Co-polarization
CPW Coplanar Waveguide
CST Computer Simulation Technology
CT Computed Tomography
DAS Delay and Sum
EM Electro-Magnetic
E-field Electric Field
FCC Federal Communications Commission
FIR Finite Impulse Response
ICNIRP International Commission on Non-Ionizing Radiation
Protection
IF Intermediate Frequency
IFFT Inverse Fast Fourier Transformation
IMDs Implanted Medical Devices
IR Impulse-Radio
IR-UWB Impulse-Radio Ultra Wideband
ISM Industrial, Scientific and Medical Radio
LNA Low Noise Amplifier
MAMI Multistatic Adaptive Microwave Imaging
Matlab MATrix LABoratory
MedRadio Medical Device Radio
v
List of Symbols and Acronyms
MICS Medical Implant Communication Service

MIST Microwave Imaging via Space-Time
MRI Magnetic Resonance Imaging
NF-FF Near-Field to Far-Field
PEG Polyethylene Glycol
PN Pseudo-Noise
Radar Radio Detection and Ranging
RF Radio Frequency
RFID Radio-Frequency Identification
SA Synthetic Aperture
SAR Specific Absorption Rate
SCR Signal-to-Clutter Ratio
SNR Signal-to-Noise Ratio
TEM Transverse Electromagnetic
UWB Ultra Wideband
VNA Vector Network Analyzer
X-pol Cross-polarization
Symbols
Lower case letters
c0 speed of light in vacuum ( 2; 997925 108 m/s)

cm speed of light in medium
d distance
dp distortion of pulse
e Euler’s constant (2:718 : : :)
f frequency
fH highest operational frequency
fL lowest operational frequency
hc .t/ channel impulse response
hs thickness of substrate
k wave number
vi
ksp fraction of the resonant wavelength

r distance in spherical coordinate
rA distance to antenna
t time
uT .t/ transmitted signal in the time domain
uR .t/ received signal in the time domain
x .t/ input signal of the beamformer
y .t/ output signal of the beamformer
x; y; z Cartesian coordinate
Capital letters
B Magnetic flux density

C Antenna pattern
CF Coherence factor of the beamformer
Ci Heat capacity of tissues
Cs Circumference of the slot
D Dielectric flux density
Da Attenuation
DA Dimension of antenna
E Electric field intensity
F Plane wave spectrum
Fp Fidelity of pulse
F=B Front-to-back ratio
G Antenna gain
Gp Antenna peak gain
H Magnetic field intensity
H .f / Transfer function in the frequency domain
Hc .f / Transfer function of the channel
HRx .f / Transfer function of the receiving antenna
Hs .f / Transfer function of the system
HTx .f / Transfer function of the transmitting antenna
I Current
vii
J Conduction current density

L Scan length of planar-rectangular near-field measurement
Lsp Effective electrical length of the meandered strips
Prad Radiated power
Rff Fraunhofer distance
RL Loss resistance of an antenna
Rnf Fresnel distance
Rrad Radiation resistance of an antenna
S Reflection characteristic of the human abdomen model
SL Power density in lossy medium
S11 Input reflection coefficient at port 1
S21 Forward transmission coefficient from port 1 to port 2
XA Reactance of an antenna
X .f / Transfer function of a transmitted signal
Y .f / Transfer function of a received signal
ZA Antenna impedance
Zm Characteristic wave impedance in medium
Z0 Characteristic impedance of feed lines (microstrip, slotline, etc.)
(50 in this work)
ZF0 Characteristic wave impedance in free space ( 377 )
Greek symbols
˛ Attenuation constant
˛n Distribution parameter in the Cole-Cole equation
ˇ Phase constant
ı Loss tangent
ıp Penetration depth
" Magnitude of the dispersion in the Cole-Cole equation
T Change in temperature
t Time period
" Permittivity
"r Relative permittivity
viii
"r;eff Effective relative permittivity

"1 Permittivity at ! ! 1 (highest frequency) in the Debye
equation
"s Permittivity at ! ! 0 (lowest frequency) in the Debye equation
"0 Permittivity of free space (8:854 10 12 F/m)
' Azimuth angle of spherical coordinate
Complex propagation constant
Reflection coefficient
Radiation efficiency of an antenna
p Penetration efficiency
Wave length
m Wave length in medium
Permeability
0 Permeability of free space (4 10 7 V.s/(A.m))
Pi (3:14159 : : :)
Polar angle of spherical coordinate
Mass density
Conductivity
i Static ionic conductivity in the Cole-Cole equation
Time delay of signal
0 Relaxation time in the Cole-Cole equation
ij Time delay in the beamformer
! Angular frequency
!ij Beamformer apodization weights
Mathematical notation and symbols

p
j Imaginary unit j D 1
s Variable
sE Vector
s Complex variable
r Nabla
r sE Divergence of vector field
ix
r sE Rotation of vector field

jj Absolute value
log Logarithm
tan Tangent
x
1. Introduction
It was in 1840 that the pioneers Recamier and Pravaz experimentally demon-
strated the use of the electric current at sub-microwave frequency to generate
heat for the destruction of uterine cancer [Guy84]. However, the interest in
using microwaves with a short wavelength for medical applications only be-
gan in World War II. Hollman proposed the use of microwaves at 1.2 GHz to
heat the deep tissues for therapeutic applications in Germany in 1938 [Hol38].
In 1939, Hemingway and Stenstrom reviewed the short-wave diathermy in
the United States, which is a method of applying heat using microwaves for
therapeutic purposes in medicine [Ste23]. However, the development and
clinical trials of these concepts were strictly limited to low frequencies (be-
low 100 MHz) at that time, since the technologies for radio frequency (RF)
hardware had not been developed. Today, thanks to the rapid development in
semiconductor technology and various signal processing techniques, there is
a growing interest in the research and development of medical applications
based on microwave techniques.
1.1. State-of-the-art microwave medical

applications
So far conventional technologies such as magnetic resonance imaging (MRI)
using magnetic fields, mammography and computed tomography (CT scan)
using X-ray, sonography using ultrasound are widely used for medical di-
agnosis (such as cancer, bone imaging, etc.). These technologies provide
good sensitivity regarding the image contrast between the different tissues
and good spatial resolution in the sub-millimeter range. However, the MRI
system is not readily available at all hospitals due to the bulky machinery re-
quired, the high manufacturing cost and the relatively long examination time.
1
1. Introduction
Mammography and CT scan, using ionizing radiation, are techniques with

intrinsic hazards and increase the risk of cancer incidence. Moreover, the
painful examinations associated with mammography have to be tolerated by
the patients. The ultrasound technique performs without ionization and is
preferred for medical diagnosis. However, it cannot penetrate bones and air
due to their strong reflections that cause acoustic shadows [MV98].
Compared to problems and risks associated with these technologies, the mi-
crowave technique for medical applications uses non-ionizing electromag-
netic waves, which makes it less harmful for the patients compared to mam-
mography and CT scan. Microwaves at low frequencies (below 2 GHz) fea-
ture good penetration ability into all human tissues including the bones, in
which ultrasound has difficulty to penetrate. A high resolution for microwave
medical imaging can be achieved by using wideband microwave signals. Fur-
thermore, microwaves within the same frequency range allow a combination
of diagnosis and wireless data transmission, which is not possible using any
other technologies for medical applications.
Nowadays, the major microwave applications in the medical field are in data
telemetry, medical diagnosis and treatment (see in Figure 1.1). These three
applications will be briefly introduced in the following sections.
Figure 1.1.: Three major microwave medical applications.
Data telemetry
Data telemetry in this work refers to wireless data transmission using mi-
crowaves between implanted medical devices (IMDs) and external devices.
Due to the rising quality of life and a growing market for health care prod-
ucts, IMDs have gained much interest for healthcare providers in the recent
2
1.1. State-of-the-art microwave medical applications
years. Examples of such devices are implantable cardioverter defibrillators

[WDJC 96], bladder stimulators [BSAGC 00] and pacemakers [VC09], which
are widely used. In the case of the traditional IMDs, the wires used to con-
nect to the devices for the acquisition of diagnosis signals increase the pain
and risk of infection in the patients. Therefore, a wireless communication link
between implants and external devices is strongly desirable. With the wire-
less link, the continuous monitoring of the state of implanted devices (e.g.
battery state) can also be achieved. The external devices can then act as a
controller for the implanted devices.
Potential applications of the data telemetry are such as in glucose monitor-
ing for diabetics, cochlear implants, deep brain stimulation [KHT08, QLD03,
TDGL11], etc. For glucose monitoring for diabetics, the glucose level is
recorded by an implanted biosensor and transmitted by an implantable an-
tenna to the external devices. In the application of cochlear implants, the au-
dio signal can be transmitted directly by an external antenna attached behind
the ear to the receiver in the cochlea (cochlear implant). Other applications
using data telemetry such as wireless monitoring of body temperature, blood
pressure and heart beat rate are addressed in the literature.
To date, there are no other techniques that are more suitable than the mi-
crowave technique for the data telemetry between IMDs and external devices.
This is because microwaves enable wireless communication links through the
human tissues, which significantly improves the quality of life of the patients
compared to the use of traditional IMDs with wires.
Medical diagnosis using microwaves
The applications for medical diagnosis are in the detection of breast cancer,
stroke, water accumulation in human body, etc. Among these, one of the most
important applications of medical diagnosis is the detection of breast cancer,
which is the most prevalent form of cancer among women [FABC 07]. Ap-
proximately one million women around the world suffer from breast cancer
[MJ13]. Therefore, technologies with high accuracy and sensitivity to detect
the presence of tumors are required. An almost pain-free examination with
a short examination time and a portable apparatus is especially desirable for
the detection of early-stage breast cancer.
3
1. Introduction
Stroke detection is the third leading cause of death in the world and is re-
garded as a challenging issue in the medical world [RFFea07]. In general,
stroke causes the loss of brain functionality due to a disturbance of the blood
flow to the brain. It is classified into two major categories of ischemic and
hemorrhagic stroke, where both require very different medical treatment (ther-
apies) [Lan09, TP08]. Since the diagnosis and determination of the stroke
categories must be done within a few hours, flexible and portable devices that
are accessible to the public are strongly in demand.
The spectrum of applications of medical diagnosis is extended to the detec-
tion and localization of water accumulations in the human body [1]. This
application is crucial for people with diseases such as pulmonary edema due
to heart failure [FG95] and urinary incontinence. For example, by monitoring
the urine in the human bladder, a permanent catheterization can be avoided
and the quality of life of these patients can be significantly improved. Portable
and low-cost device is required to enable each patient to be equipped with this
apparatus.
Microwave signals provide high potential for the imaging of the breast, the
bladder or the brain [KCLC 09, ARTN11, GJR10]. This is due to the fact that
microwaves feature non-ionizing radiation, high range resolution and high
penetration ability, which have been discussed and compared with conven-
tional medical imaging techniques at the beginning of this chapter. Moreover,
a low system complexity of the microwave system, which enables the real-
ization of a portable apparatus, is emphasized after reviewing the different
medical applications.
Microwave medical diagnosis is based on the concept of observing the scat-
tered signals caused by the dielectric contrast, which is the difference of
the dielectric properties between the malignant tissues and the surrounding
healthy tissues [FLHS02, LBVVH05]. In microwave medical diagnosis, most
researches are focused on two main approaches: radar (Radio Detection and
Ranging) imaging and microwave tomography. In radar imaging, wideband
antennas for high range resolution are applied for synthetic aperture (SA)
based operation or multistatic operation based on fixed array [20] [Nik11,
KCLC 09]. Microwave tomography is based on the reconstruction of the dis-
tribution of the complex permittivity by solving the inverse electromagnetic
4
1.2. Telemedicine using microwaves and its challenges
scattering problem. Recently, the microwave holography in medical near-

field imaging has been proposed in [ARNT11], which uses a reference wave
and sophisticated holographic reconstruction algorithms.
Medical treatment using microwaves
Medical treatment using microwaves is based on using the heat generated by

microwave radiation to increase the local temperature to destroy the abnor-
mal tissues (e.g. malignant tissues). This technique is more sensitive and
effective compared to ionizing radiation (i.e. X-ray) and chemical toxins (i.e.
Chemotherapy) [Roe99]. The related applications range from hyperthermia
in the treatment of breast cancer, transurethral microwave thermotherapy to
microwave ablation [Kap96, CBHVV04, SDMS05].
In the framework of this thesis, medical treatment using microwaves will not
be considered. The focus of this thesis is on the telemedicine system using
microwaves for data telemetry and diagnosis.
1.2. Telemedicine using microwaves and its

challenges
Telemedicine for the improvement of the quality of health care for patients
was introduced two decades ago. Telemedicine refers to the use of telecom-
munication for the transmission of health information to deliver clinical health-
care from a distance. In this way, the access to the medical services of patients
can significantly be improved.
A vision of the future telemedicine for a healthcare system for nursing home
residents using microwaves is illustrated in Figure 1.2. The medical diagnosis
system serves to monitor health problems such as stroke for prompt diagnosis
and treatment. On the other hand, the data transmission between IMDs and
external devices are performed simultaneously. In this way, the combination
between medical diagnosis and data telemetry using microwaves contributes
very positively to the existing healthcare services. The details of the vision
are briefly introduced in the following paragraphs.
5
1. Introduction
Figure 1.2.: Vision of the telemedicine for nursing home residents using mi-
crowaves and the goal as well as organization of the work to re-
alize these applications.
In the data telemetry of this healthcare system, the physiological data (temper-
ature, blood pressure, glucose concentration) or vital signs (such as respira-
tion, heart beating, etc.) are monitored by sensors integrated on the implants.
The implants are wirelessly powered by an antenna at a certain frequency
(usually very low frequency in the MHz range). The wireless data transmis-
sion between implants and external medical devices is performed at a high
frequency (in the GHz range). However, the wireless transmission system
is normally in low-power stand-by modus and can be awaken by an external
signal for operating data transfer at full power at the frequency band licensed
for medical devices. At the medical center, the received data is forwarded to
the healthcare practitioner to evaluate the patient’s status. In the case where
6
1.2. Telemedicine using microwaves and its challenges
abnormalities are detected, the doctor is immediately informed so that neces-

sary actions can be initiated in time.
Simultaneously, as shown in Figure 1.2, medical diagnosis of stroke using mi-
crowaves can be performed flexibly and monitored frequently. This diagnosis
functions as a complementary method in home healthcare systems compared
to traditional technologies (i.e. MRI and CT scan), which can only be per-
formed in hospital due to the bulky devices and high cost. The low system
complexity of microwave imaging system enables a flexible and even portable
diagnostic device. This system also features a low interference to other mi-
crowave systems due to the low regulated radiation power in the respective
frequency bands [Com01]. With the help of further image processing, the
diagnosis can be evaluated and wirelessly transmitted to the medical center.
However, several technical difficulties in the proposed telemedicine system
using microwaves must be overcome, which will be described in the following
paragraphs.
From the system point of view, the challenges of wireless data transmission
between the implants and external devices are for example low-power con-
sumption, which is critical for the lifespan of the implanted devices. More-
over, wireless data transmission can be strongly limited by radiation regula-
tions in terms of radiation power and operational frequency range. Further
challenges are such as robustness against the movement and position of the
patients as well as secure and reliable communication [NMdSC 07, SCK07].
For microwave medical diagnosis, the signal-to-noise ratio (SNR) of the re-
ceived signal is a major issue. Since the reflections from the targets (e.g.
tumors) are very weak due to the high signal attenuation in human tissues in
the GHz range, the SNR of the whole system is thus very low. Therefore,
more energy is required to penetrate into human tissues.
With these requirements in mind, one of the main obstacles towards the real-
ization of the vision shown in Figure 1.2 is the development of suitable an-
tennas for IMDs and medical diagnosis using radar imaging. For such appli-
cations, the antennas for radiating and receiving the signals is the governing
factor of the SNR of the overall system.
In this work, two types of antennas for the purpose of medical applications
are considered. They are implantable antennas and on-body matched an-
7
1. Introduction
tennas. The implantable antennas are used for the data transmission between
IMDs and external devices, while the on-body matched antennas are used in
diagnosis applications (such as cancer, brain diseases, etc.).
For wireless data transmission of IMDs with external devices, the implantable
antennas are located together with IMDs in the human body and surrounded
by different human tissues depending on the applications (refer to skin, fat,
muscle, etc.).
Different from the on-body matched antenna, antennas can be built for mi-
crowave medical diagnosis, which are placed at a distance from the body
(termed as off-body antenna). In the case of off-body antenna, a strong re-
flection of the incident waves on the air-skin interface occurs and more than
50% of the energy is back scattered, since the relative permittivity of skin is
very large (around 40 at 1 GHz) compared to free space. Furthermore, the
distance between antenna and human body introduces additional free space
attenuation. Therefore, the energy penetrating into the human body is lower
than 10% of the total energy of the incident waves. Moreover, the reflections
from the air-skin interface are dominant, which overlap the weak useful re-
flection. Thus, for medical diagnosis, the antenna is required to be matched to
the human body. Matching the antennas on the human body (refer to human
skin for the on-body matched antenna) has the purpose of reducing the strong
reflection on the air-skin interface and to allow more energy to be radiated
into the human body for obtaining a stronger signal for processing.
Though the implantable antennas and on-body matched antennas are used
for different medical applications, their common ground is that they are both
matched to the human body and are termed as body-matched antennas in
this thesis. Therefore, the body-matched antennas refer to both the implantable
and on-body matched antennas. Many challenges arise for the body-matched
antennas in terms of the design, characterization, fabrication and verification,
which are briefly discussed in the following paragraphs.
The design of body-matched antennas differs from that for the regular an-
tennas for free-space operation. This is because it suffers from the problems
of lossy human tissues and their different permittivities resulting in totally
different wave propagation behavior. The geometry and dielectric properties
of the human body must be considered for the optimization and character-
8
1.3. Goal and organization of this work
ization of these antennas. Therefore, the performance of the antennas (e.g.

impedance matching, bandwidth, radiation pattern, etc.) is strongly depen-
dent on the dielectric properties of tissues and structure of the human body.
As a result, the antennas will suffer from reduced efficiency, radiation pattern
fragmentation and variations in feed-point impedance.
Since the antennas are placed directly on or in the human body, the radiation
into a multilayer lossy medium must be considered. The computational load
of the simulation of the antenna model together with the human body becomes
very large in terms of computer memory. The design procedure then becomes
more complicated than in the case of simple free-space operation scenarios.
Moreover, the size of the antennas strongly affects the performance of diag-
nosis imaging systems or the size of IMDs. Thus miniaturization techniques
are required to significantly reduce the size of the radiating structures. Fur-
thermore, many features such as having low profile (fabrication complexity),
low weight and low cost are required for medical applications.
The characterization of the body-matched antennas must be done with the
consideration of the dielectric properties of human tissues. Thus, the antenna
cannot be verified in an anechoic chamber. The verification in the measure-
ment taking into account the presence of the human body (e.g. using tissue-
simulating liquid) becomes also sophisticated and the results change with the
scenarios and setups. Regarding the radiation regulation for the human body,
many requirements (i.e. Specific Absorption Rate (SAR) value) must also be
fulfilled.
1.3. Goal and organization of this work

The goal of this work is to design body-matched antennas for the microwave
healthcare system for nursing home residents (see in Figure 1.2). The focus
is to miniaturize the antenna size yet maintaining high efficiency, high gain as
well as stable radiation pattern over the frequency of the implantable and on-
body matched antennas. The operational frequencies of the antennas range
from 0.5 GHz to 10 GHz and the different frequency bands corresponding to
the different applications will be given in the following chapters. With regard
to the design of antennas, planar antennas are chosen due to an easy fabrica-
9
1. Introduction
tion and to achieve low profile as well as low-cost devices. The miniaturiza-
tion techniques of the antenna design in this thesis relies on the optimization
of the feeding- and antenna structures as well as regulation of the current dis-
tribution on the radiating elements. With these achievements, it is expected
that these antennas will significantly enhance the performance of the IMDs
as well as imaging systems for diagnosis and thus contribute very positively
to the future microwave telemedicine systems.
With this goal in mind, the organization of this work is outlined (referring to
Figure 1.2) as follows:
Before the design of the body-matched antennas is introduced, the dielec-
tric properties of the different human tissues and human model are studied
in chapter 2. Based on the dielectric properties of human tissues, the mi-
crowave propagation in the human body is then investigated with respect to
their reflection, attenuation and transmission. A model of the human bladder
with different tissue layers including frequency dependent dielectric prop-
erties is applied for the estimation of the required SNR and dynamic range
of the imaging system. In the following chapter, different characterization
methods of the body-matched antennas are described. For the verification of
the body-matched antennas, two antenna measurement systems with tissue-
simulating liquid are given, which allow the measurements of the near-field
and far-field radiation pattern of the body-matched antennas. Chapter 4 pro-
vides the design and the miniaturization concepts of implantable antennas for
IMDs. That is followed by on-body matched antennas for microwave imag-
ing at different operational frequency bands in chapter 5. The main focus
is the miniaturization of the antennas without the significant degradation of
the antenna performance. The applicability of the on-body matched antennas
for microwave imaging is shown in Chapter 6, by means of a measurement
demonstrator for the detection of the hemorrhagic stroke. Finally, Chapter 7
gives conclusions of this thesis to microwave medical applications.
10
2. Analysis of microwave propagation
in a multilayer human body
In the modeling of the human body for microwave medical applications, the
human body must be considered as a multilayer model consisting of different
lossy human tissues. The human tissues are in general frequency dispersive,
meaning that they distort the signal when it propagates through them. There-
fore, it is of great importance to analyze the microwave propagation in the
human body taking into account the frequency dependent dielectric proper-
ties of different tissues.
This chapter provides the investigation of the propagation characteristics of

the electromagnetic (EM) signals in tissues. For that purpose, the dielectric
properties of different tissues are analyzed. A full consideration of the propa-
gation characteristics of the signals (from 1 to 10 GHz) in human tissues with
respect to the frequency dispersion, attenuation, reflection and transmission is
provided. Then, with the above knowledge, the performance as well as chal-
lenges for the microwave medical system are analyzed with a realistic human
abdomen model for the detection of urine accumulation. Regarding the fre-
quency band used, the ultra wideband (UWB) pulse from 3.1 to 10.6 GHz
according to [FCC02] is investigated in terms of attenuation and distortion.
Based on these results, the optimal operational frequency bands are discussed
for different medical applications. This chapter ends with a short conclu-
sion about microwave propagation in the human body and the requirements
of antenna characteristics for the optimal system performance for microwave
medical applications.
11
2. Analysis of microwave propagation in a multilayer human body
2.1. Microwave propagation in a lossy medium

In general, a homogeneous, isotropic and lossy medium can be described with
a complex permittivity " and permeability , which can be written as
" D "0 j"00 ; (2.1)

D 0 j00 ; (2.2)
where "00 and 00 describe the dielectric loss and magnetization loss, respec-
tively. 00 is not considered in this thesis (assumed to be 0), since it does not
exist in biological tissues. Furthermore, assuming that the human tissues are
non-magnetic [VRK06], the permeability can be written as D 0 .
With regard to the permittivity, there are the loss due to dielectric damping of
vibrating dipole moments and the conductive loss (or Joule loss) [Poz, ZL07].
However, it is impossible to distinguish between these losses. Since the con-
duction current is dominant in most biological tissues, the dielectric damping
is assumed to be 0 and hence the loss is specified only by the effective con-
ductivity loss.
Assuming that the medium has a conductivity , a conduction current density
JE exists and together with dielectric flux density D,
E can be written as
JE D E;
E (2.3)
E D "0 "r E;
D E (2.4)
where E is the electric field intensity, "0 and "r is the permittivity of free
space and the relative permittivity of a lossy medium, respectively. Taking
into account the time varying nature of the electromagnetic waves, the electric
field (E-field) intensity is described as
EE .x; y; z; t/ D EE .x; y; z/ e j!t ; (2.5)
where ! D 2f is the angular frequency and f is the frequency.

Based on (2.3) and (2.4), the Maxwell’s equation in differential form for mag-
12
netic field intensity HE can be modified to
@DE
r HE D C JE
@t
D j!"0 "r EE C EE

E
D j!"0 "r j E: (2.6)
!"0
Thus, the complex permittivity can be modified as

" D "0 "r j : (2.7)
!"0
The loss tangent can be calculated by

tan ı D : (2.8)
!"r "0
Consider now the Maxwell’s differential form equation for EE
@BE
r EE D D j!0 HE ; (2.9)
@t
assuming that the medium is source-free, the wave equation for EE can be
obtained by modifying (2.9) as follows:

r r EE D r j!0 HE D j!0 r HE

r r EE r 2 EE 2
D ! 0 "0 "r j EE
!"0

2 E
2
r E C ! 0 "0 "r j EE D 0: (2.10)
!"0
To simplify (2.10), the wave number in lossy medium can be written as

s

kD! 0 "0 "r j : (2.11)
!"0
13
Having the wave number k, the complex propagation constant for the medium
can be then determined by
s

D ˛ C jˇ D j k D j! 0 "0 "r j ; (2.12)
!"0
where ˛ and ˇ are respectively the attenuation constant and phase constant.
Separating the real and the imaginary part of the complex propagation con-
stant yields two terms:
v " #
u r 2
! t "r
u
˛ D c0 2 1 C !"0 "r 1 ; (2.13)
v " #
u r 2
! t "r
u

ˇ D c0 2 1 C !"0 "r C 1 : (2.14)
It can be seen that the conductive loss results in a non-zero attenuation con-
stant and also modifies the phase constant.
After introducing the attenuation constant ˛ and phase constant ˇ, the fol-
lowing terms will be discussed with regard to a lossy medium.
Wavelength
The wavelength of a propagating signal in a lossy medium can be written as

0 vu 2s 31
2
2 B u "r
u
m D D 2c0 = @! t 4 1 C C 15A : (2.15)
C
ˇ 2 !"0 "r
Penetration depth
The penetration depth ıp describes the penetration ability of EM waves into a

lossy medium. This is defined as the depth, at the point where the amplitude
of the field decreased to 1=e of the original value and can be described by
14
1 c0
ıp D D v " #: (2.16)
˛ u r 2
! t "r
u
1C2 !"0 "r 1
Attenuation
Assuming that the waves propagate in the +z direction, the attenuation of the
E-field (in the z direction) can be written as
˛z
jE .z/j jE .0/j e ˛z
Da D D De ; (2.17)
jE .0/j jE .0/j
where the attenuation in dB is:
˛z
Da jdB D 20 log10 .e / D z ˛ 20 log10 .e/ D 8:686 ˛ z: (2.18)
Therefore the attenuation constant in dB is expressed as 8:686 ˛.
Wave impedance
The wave impedance in a lossy medium is also modified to reflect a complex

value:
s v
u 0
Zm D Du h i: (2.19)
" j !" 0
t
"0 "r
Since wideband signals are in the scope of this thesis, the frequency disper-
sion cannot be neglected. The complex permittivity of a lossy medium from
(2.7) can thus be modified to

.!/
" .!/ D "0 "r .!/ j ; (2.20)
!"0
where the "r .!/ and .!/ are the frequency dependent relative permittivity
and conductivity, respectively.
In conclusion, the attenuation constant ˛, the phase constant ˇ and the wave
impedance can be determined if the complex permittivity (dielectric property)
15
is known. Since the complex permittivities of different human tissues are the
basis for the analysis of the microwave propagation in the human body, they
will be introduced in the following sections.
2.2. Dielectric properties of human tissues

The dielectric properties of human tissues describe the level of interaction of
the EM waves with the molecules of the tissues. For the numerical analysis
of the propagation of EM waves in tissues or for determining the specific
absorption rates (SAR) i.e. the absorption of EM energy by the tissues, the
dielectric properties of human tissues are the key parameters. Therefore, it
is of significance to determine the dielectric properties of human tissues. For
that purpose, the well-known Debye equation and the Cole-Cole equation
[GGC96, GLG96b], which predict the permittivity of human tissues based
on experimental data of human tissues, will be introduced in the following
section.
2.2.1. Debye equation and Cole-Cole equation

The dielectric properties of human tissues are frequency dependent due to the
frequency-dependent (dispersion) polarization of atoms, electrons and ions
caused by the E-fields in tissues [Rei98, MD99].
Assuming that an ideal, noninteracting dipoles are considered in a dielectric
relaxation respons, the frequency dispersion of the dielectric properties of
human tissues can be described by the well-known Debye equation [Deb60]:

"s "1
" D "0 "r j D "1 C ; (2.21)
!"0 1 C j! 0
where "1 and "s are the permittivity at ! ! 1 (highest frequency) and
! ! 0 (lowest frequency), respectively. 0 is the relaxation time, which is
the required time for a stimulated dipole to return to its original state. It must
be emphasized that only a single relaxation time (first order approximation of
different relaxation regions, which will be mentioned in the next paragraph)
is considered in this equation. Thus, the Debye equation is not sufficient to
16
predict the dispersion in a wide frequency range. In reality, several types of

relaxation processes with regard to different polarizations exist. The relax-
ation process can hence be extended to a wider frequency range.
The dielectric behavior of human tissue in a broadband frequency range can
be separated into four dispersion regions (three main relaxation regions: dis-
persion for low, medium and high frequencies respectively; one minor region)
classified according to the different mechanisms of polarization [Hur85]. Bas-
ed on many reported experimental data of various tissues, the Cole-Cole equa-
tion [GGC96] [GLG96a] [GLG96b], which is the improved version of the
Debye equation, provides multiple dispersion terms. These terms predict the
frequency dependency within each dispersion region.
The Cole-Cole equation is given in (2.22), where ˛n is the distribution pa-
rameter (a measure of the broadening of the dispersion) and n indicates four
different dispersion regions, " D "s "1 denotes the magnitude of the dis-
persion, i is the static ionic conductivity.

X "n i
" .!/ D "0 "r j D "1 C .1 ˛ /
C : (2.22)
!"0 n 1 C j!
0
n j!"0
n
2.2.2. Permittivity of human tissues

Based on the Cole-Cole equation, the permittivity of different tissues can be
predicted in a wideband frequency range (relevant frequency range in this
thesis is from 1 to 10 GHz) (Parameters referring to appendix A). Since the
detection of urine accumulation in the human bladder and the detection of
stroke are within the scope of this thesis, the dielectric properties of eight
types of tissues (muscle, skin, bone, fat, skull, white matter, grey matter and
blood) are generated from the Cole-Cole equation. The dielectric properties
of human urine and distilled water on the other hand, were measured with the
Agilent 85070E Dielectric Probe Kit.
Figure 2.1 (a) shows the relative permittivity ("r ) of these ten tissues over fre-
quency. Starting with the tissues which have the lowest relative permittivity,
fat has the lowest relative permittivity of around 5, since its water content is
almost negligible. Furthermore, the dispersion of the permittivity of fat is not
remarkable in the frequency range from 1 to 10 GHz. This is followed by the
17
skull (cortical bone) and bone. At 5 GHz, the skull has a relative permittivity
of 10, while that of the bone is around 18.
The "r of white matter, skin, grey matter and muscle are between 30 and 50.
They have a large relative permittivity due to the high water content present
in these tissues. Grey matter exhibits a slightly higher relative permittivity
than white matter. Those with the relative permittivity of above 50 are blood,
urine and distilled water. Water and urine show similar relative permittivity
between 70 and 80. It can also be seen that the frequency dependency of
the relative permittivity of tissues with high water content varies very signifi-
cantly over the 1 to 10 GHz range.
From Figure 2.1 (a), it can be concluded that the blood in the human brain
and the urine in human bladder have high relative permittivity compared to
their surrounding tissues. This high contrast of relative permittivity to the
surrounding tissues provides the potential for detecting the position or volume
of the blood and urine accumulation in the human body.
The losses of the respective materials are related to the conductivity as shown
in Figure 2.1 (b). Fat, skull, bone and white matter exhibit a low conductiv-
ity, while the respective quantities for water, urine, blood and muscle show
larger values. Urine and blood have high conductivities of 7.5 and 5.5 S/m
at 5 GHz, respectively. Therefore, a large attenuation of the EM signals in
water, urine and muscle is expected. The conductivity of urine is higher than
that of distilled water. This is because urine contains more ions and electrons
than distilled water and this results in a higher conductivity. Furthermore, it
must be noted that the frequency dependency on the conductivity of the tis-
sues with high water content is very strong from 1 to 10 GHz. This is because
the vibrating distance of ions becomes shorter with increasing frequency and
hence the conductivity is improved.
2.2.3. Attenuation, penetration depth and wave

impedance in human tissues
Based on the complex relative permittivity of different human tissues, the at-
tenuation constant and penetration depth of EM waves in these tissues can be
determined using (2.13) and (2.16) to investigate the influence of the conduc-
18
80
dis.water
70 urine
muscle
skin
60 bone
Relative permittivity εr
fat
50 skull
white matter
grey matter
40
blood
30
20
10
0
1 2 3 4 5 6 7 8 9 10
Frequency f in GHz
(a) Relative Permittivity
15
dis.water
urine
muscle
12 skin
bone
fat
Conductivity σ in S/m
skull
9
white matter
grey matter
blood
6
0
1 2 3 4 5 6 7 8 9 10
Frequency f in GHz
(b) Conductivity
Figure 2.1.: Relative permittivity and conductivity of different human tissues

over frequency (distilled water and urine: measured by the au-
thor; other tissues: predicted by the Cole-Cole equation).
19
tivity on the microwave propagation. In Figure 2.2, the attenuation constant

and penetration depth related to different tissues are presented.
The results show that EM waves typically undergo much higher attenuation
in human tissues than in free space, e.g. attenuation constant caused by
oxygen absorption and water vapor absorption at 1 GHz being 8 10 8 and
1 10 8 dB/cm [LHC93], respectively. However, the attenuation constant
caused by fat, bone and white matter at 1 GHz is around 1 dB/cm. The tissues
with high water content (muscle, blood and urine) show a higher attenuation
constant. Furthermore, this value increases strongly with frequency, which
introduces distortion for wideband signals due to the different signal attenu-
ations at different frequencies. This is investigated quantitatively in section
2.3 with respect to the fidelity of an ultra wideband (UWB) pulse.
The high signal attenuation caused by human tissues reduces the performance
of any kind of microwave system for medical applications, where signal pen-
etration into and reflection from the tissues are required. Figure 2.2 (b) shows
that the signal from 1 to 3 GHz has a large penetration depth in different tis-
sues (e.g. 4 cm in muscle at 1 GHz), while at above 5 GHz ıp decreases (1 cm
in muscle at 5 GHz). Therefore, a very high system dynamic range is required
to capture the weak reflections from the tissues at high frequencies. More-
over, the different attenuations of the signal at different frequencies result in
the distortion of the frequency spectrum of the received signal. The received
signal level can be very different over a wider frequency range (e.g. the whole
UWB frequency range from 3.1 to 10.6 GHz according to [FCC02]), which
must be considered in the design of the system.
After the discussion about the attenuation and penetration depth of microwave
signal in different human tissues, the wave impedance in tissues is analyzed.
This is of significance for the impedance matching of the body-matched an-
tennas in the later chapters. Taking a plane wave for example, the wave
impedance of a plane wave in different human tissues becomes complex due
to the complex permittivity of the tissues. For the purpose of comparison,
only the absolute value of the wave impedance is provided in Figure 2.3. The
values in all tissues are much lower than in free space (ZF0 =377 ). In this
case, the impedance matching from the 50 feed line of the antennas to a
low wave impedance in tissues can be easily achieved. However, the different
20
25
dis.water
urine
muscle
20 skin
Attenuation constant α in dB/cm
bone
fat
skull
15
white matter
grey matter
blood
10
0
1 2 3 4 5 6 7 8 9 10
Frequency f in GHz
(a) Attenuation constant
20
15 dis.water
10 urine
muscle
skin
5 bone
Penetration depth δp in cm
4 fat
3 skull
2 white matter
grey matter
blood
1
1 2 3 4 5 6 7 8 9 10
Frequency f in GHz
(b) Penetration depth
Figure 2.2.: Attenuation constant and penetration depth of microwaves in di-

fferent human tissues over frequency.
21
180
dis.water
urine
160 muscle
skin
bone
Wave impedance |Z | in Ω
140
fat
skull
m
120 white matter

grey matter
blood
100
80
60
40
1 2 3 4 5 6 7 8 9 10
Frequency f in GHz
Figure 2.3.: Wave impedance in different human tissues over frequency.
wave impedances in different tissues result in different impedance matchings

for a given body-matched antenna. Therefore, the impedance matching of
the antennas placed in different tissues will be changed strongly. The re-
sulted impedance mismatch reduced considerably the radiation efficiency of
the body-matched antennas.
With the knowledge of the dielectric properties of different human tissues as
well as the corresponding quantities (such as attenuation, penetration depth
and wave impedance), the microwave propagation in the human body con-
sisting of different tissues can now be analyzed.
2.3. Analysis of microwave propagation in the

human bladder
In this section, urine detection in the human bladder motivates the analysis
of the propagation of microwave signals (GHz range) in the human body.
The detection and the monitoring of urine in the human bladder is of signifi-
22
2.3. Analysis of microwave propagation in the human bladder
cance in medical fields and has been discussed in the introduction. It can be
achieved by detecting the reflected microwave signals from the boundaries of
human tissues such as fat and bladder muscle. Thus, the microwave signals
have to penetrate different tissues of the abdomen. Due to the high attenua-
tion and frequency dispersion of the microwave signals in human tissues, the
detection possibility of the reflected signals is strongly degraded. Therefore, a
quantitative investigation of the microwave propagation in the human body is
significant for the prediction of performance of microwave medical systems
also to help design more effective antenna systems to counter the propagation
loss.
Based on the dielectric properties of the different human tissues introduced
in the last section, the propagation characteristics of microwave signals are
investigated with respect to their reflection, attenuation and transmission. A
model of the human bladder with different tissue layers including the fre-
quency dependent dielectric properties is used. Furthermore, a system con-
cept of the UWB radar for the detection of water accumulation in the human
bladder is introduced. The UWB pulse attenuation and distortion in different
human tissues are then investigated and evaluated.
2.3.1. Multilayer model of the human abdomen
Since the permittivity of different human tissues has been investigated, a

multilayer model of the human abdomen with realistic dielectric properties
can be implemented. According to the Visible Human Project male dataset
[USNLoM], a multilayer structure of the human abdomen can be modeled as
shown in Figure 2.4. This dielectric model consists of 6 different layers of
human tissues. The connective tissue between skin and muscle is modeled as
fat, since it is composed mostly of fatty tissue. To simplify this model for the
analysis excluding a sophisticated geometry, each layer is assumed to be flat.
The typical thickness of each tissue is based on the investigations on adults
(20-60 years old) [CKK04] and are summarized in Table 2.1.
23
Figure 2.4.: Multilayer model of human abdomen.
Table 2.1.: Typical thicknesses of tissues in the human abdomen (full bladder)
based on adults between 20-60 years old.
tissue thickness (mm)
skin 0.3
connective tissue (fat) 10
muscle 10
fat 2
bladder muscle 2
urine 30
bladder muscle 2
2.3.2. Concept of the radar system

For the analysis of microave propagation in the human bladder, an appropriate
concept of radar system is required. In [1], [22, 23], the system performance
of three different radar concepts with different signal modulations were in-
vestigated by the author. They are impulse-radio (IR) approach, frequency
sweep approach with Vector Network Analyzer (VNA) (details will be given
in chapter 6) and the Pseudo-noise (PN) approach. These concepts are widely
used under the research for medical applications in the recent years. The per-
formance of these three radar concepts have been verified in measurements
24
by the author. It has been found that the impulse-radio (IR) approach has the
lowest system dynamic range of the three radar systems, but it also has the
lowest system complexity [Vat10]. Using the frequency sweep approach with
VNA and the Pseudo-noise (PN) approach a high system dynamic range can
be achieved [Sch10].
To obtain the characteristics of microwave propagation in the human ab-
domen, the analysis both in the time and the frequency domain (for the IR
approach and the frequency sweep approach, respectively) is considered in
the following section.
With regard to the IR approach, the UWB pulse with a frequency range from
3.1 to 10.6 GHz is used. The block diagram of the proposed Impulse-Radio
Ultra Wideband (IR-UWB) radar system is shown in Figure 2.5. The trans-
mitter consists of a trigger, a pulse generator and a transmit antenna. The
signal is received by a separate antenna, which is connected to a sampling
unit. The reflected signal is sampled by a device with a high sampling rate
and a data acquisition unit. In this IR-UWB concept, a correlation analysis
in the time domain is performed to enhance the SNR of the received signal.
For that purpose, the received UWB signal is correlated with the reference
signal from the transmitter. Since the human tissues are dispersive materi-
als, the received signal is distorted as it propagates through the tissues, which
results in a degraded correlation function between transmitted and received
pulses. The degradation of the received UWB pulse can be evaluated in terms
of distortion and fidelity. The distortion dp is mathematically defined as the
variation of the received signal uR .t/ with respect to the transmitted signal
uT .t/ [LS94]:
ˇ ˇ2
ˇ ˇ
C1 ˇ
uR .t C / uT .t/
Z ˇ
dp D min i1=2 ˇˇ dt: (2.23)
ˇ ˇ
ˇ hR
ˇ C1 i1=2 hR
1 2 C1 2
ˇ 1 ju R .t/j 1 juT .t/j
ˇ
Based on the distortion, Fp quantifies the fidelity of the received pulse com-
pared to the transmitted signal and is determined by the peak of the cross-
correlation function as
25
Figure 2.5.: Block diagram of the IR-UWB radar system for urine detection
in the human bladder (dA : distance between transmitting and
receiving antennas; r: distance between antennas and bladder
model).
C1
uR .t C / uT .t/
Z
Fp D max hR i1=2 hR i1=2 dt: (2.24)
1 C1 2 C1 2
1 juR .t/j 1 juT .t/j
Since the sign of the correlation can be usually neglected, the absolute value
of Fp is used. It can be concluded that the minimum distortion is obtained
when the fidelity between the two signals is maximum.
With the introduction of the model of the human abdomen and the related
radar concepts, the characteristics of microwave propagation can be analyzed
as shown in the following sections.
2.3.3. Characteristics of propagating microwave signals

in the multilayer abdomen
During the propagation of microwave signals in the human abdomen, reflec-
tions occur at the boundaries of different tissues. The magnitude of the UWB
signals is dependent on the dielectric contrast of tissues. The reflection coef-
ficient of the plane wave can be determined using the Fresnel equation. In the
case of normal incidence to the surface of the two dielectrics, whose permit-
tivities are denoted as "1 and "2 , the normal incident reflection coefficient can
26
be written as
p p
"1 "2
Dp p : (2.25)
"1 C "2
Microwave signals propagating through the multilayer model of the human

abdomen (refer to Figure 2.4) will experience reflections at the boundaries
between the various tissues. The reflection (calculated by 2.25) and transmis-
sion properties of microwave signals at the boundaries of different tissues are
shown in Figure 2.6. An analysis is performed in the frequency domain (from
1 to 10 GHz) for a better comparison. The first strong reflection occurs at the
boundary between air and skin, where half of the energy is reflected back.
Since the contrast between skin and fat as well as between fat and muscle
are very high, strong reflections are also caused at these two boundaries. The
reflection from the boundary between the muscle and urine is weak, since
the two tissues show little difference in the dielectric property. The bladder
muscle has a thickness of approximately 0.2 mm and has a low contrast with
respect to urine. Hence for the detection of urine in the human bladder, the
reflection from the boundary between fat and bladder muscle can be used
directly.
100 100
90 90
Transmission coefficient in %
Reflection coefficient in %
80 80
70 70
60 60
50 50
40 40
30 air / skin 30 air / skin
skin/ fat skin / fat
20 20
fat / muscle fat / muscle
10 muscle / urine 10 muscle/ urine
0 0
1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10
Frequency f in GHz Frequency f in GHz
(a) Reflection coefficient (b) Transmission coefficient
Figure 2.6.: Simulated reflection and transmission coefficient of different tis-

sue boundaries over frequency (The frequency dependance of di-
electric property is included).
27
The signal attenuation, the penetration depth and the attenuation constant of
4 human tissues (skin, fat, muscle and urine) in the frequency range from 1 to
10 GHz are shown in Figure 2.2. In the fat tissue the signal has a penetration
depth of 9.3 cm at 3 GHz, while a penetration depth of around 1.9 cm in
skin tissue is identified. In urine, an extremely low penetration depth (0.95
cm at 3 GHz) is observed. The value, as expected, decreases with increasing
frequency. At 8 GHz a penetration depth of 2.6 cm in the fat tissue is noted.
In Figure 2.2 (a) the attenuation constant of different tissues is shown. It pro-
vides an estimate of the amount of the signal attenuation during propagation
through the tissues. The results show that e.g. urine has an attenuation of 9.1
dB/cm at 3 GHz, which increases up to 25.6 dB/cm at 8 GHz. The signal
attenuation in the skin and muscle tissue are also very high (at 4.6 dB/cm and
4.8 dB/cm at 3 GHz, respectively). The fat tissue causes relatively low signal
attenuation while its attenuation constant is 0.91 dB/cm at 3 GHz. The reason
for this is the low water content in fat.
2.3.4. Distortion and fidelity of UWB pulses in human

tissues
In the IR-UWB system, the correlation between the received signal and tem-
plate pulse is calculated to enhance the SNR of the received signal. As has
been shown, the complex permittivity of human tissues changes relatively
strongly over the considered frequency range. Hence the transmitted pulse is
distorted due to its large bandwidth. As a consequence the pulse undergoes
changes in its form, amplitude and width. This makes the pulse detection
complicated and decreases the resolution of the UWB system. Therefore, the
signal distortion of UWB signals is quantitatively analyzed as follows.
In order to investigate the dispersive effect only regarding the dielectric prop-
erties of human tissues, the radiation properties of the antennas are not con-
sidered. Therefore, a plane wave is used as the excitation signal (TEM mode)
in the simulation shown in Figure 2.7. The analysis is performed with a
tissue-phantom of a single tissue with frequency-dependent dielectric prop-
erty in the commercial software Computer Simulation Technology (CST) Mi-
crowave Studio. In the simulation, a open boundary condition is applied and
28
the reflection at the boundary is not considered. The transmitted signal (pulse)
has the frequency bandwidth from 3.1 to 10.6 GHz. Different human tissues
are investigated separately. The signal is fed directly to the surface of the hu-
man tissue, which has a size of 5050100 mm3 . Several probes, which are
placed in the tissue, are used to obtain the pulses at different distances to the
feed port in the tissue.
Figure 2.7.: Diagram of the simulation configuration with plane wave and
probes.
Figure 2.8 (a) shows the attenuation of the transmitted pulse in human tissues.
To calculate the attenuation of the pulse, the amplitude of the pulse (frequency
range from 3.1 to 10.6 GHz) is evaluated. The amplitude of the pulse in the
fat tissue decreases slightly as the thickness increases. A thickness of 10 mm
of fat tissues causes an attenuation of 2.75 dB, while in the case of muscle
and urine the resulting attenuation is 13.7 dB and 19.7 dB, respectively. It
agrees with the analysis for a single frequency that the tissue with a higher
water content causes higher signal attenuation. Moreover, the curves are not
completely linear, since the surface waves cannot be completely eliminated in
simulation. Since the presence of the surface waves must be also considered
during the microwave propagation in the human body in the realistic scenario,
this effect can be tolerated in this analysis.
The fidelity and distortion of the UWB pulse are shown in Figure 2.8 (b)
and (c). The pulses propagating through urine, skin and muscle exhibit a
strong decrease of the fidelity and an increase of the distortion with increasing
thickness of the tissue. However, at a thickness of 10 mm, the fidelity of the
29
80
70 fat
muscle
60
Attenuation in dB
skin
50 urine
40
30
20
10
0
0 5 10 15 20 25 30 35 40 45 50
Distance d in mm
(a) Attenuation
1
0.9
0.8
0.7
0.6
Fidelity
0.5
0.4 fat
0.3 muscle
0.2 skin
0.1 urine
0
0 5 10 15 20 25 30 35 40 45 50
Distance d in mm
(b) Fidelity
1.8
1.6 fat
muscle
1.4 skin
1.2 urine
Distortion
1
0.8
0.6
0.4
0.2
0
0 5 10 15 20 25 30 35 40 45 50
Distance d in mm
(c) Distortion
Figure 2.8.: Simulated attenuation, fidelity and distortion of UWB pulse in

various human tissues (muscle, skin, bone, fat and distilled wa-
ter).
30
signals in these three tissues is still better than 0.8.

In reality the tissues that cause the high distortion (i.e. skin, muscle) are
relatively thin (in the range of 10 mm). The resulted fidelity is still very
high (refer to Figure 2.8 (b)). On the other hand, it can be noted that the
tissues with low water content (e.g. fat tissue) show very low distortion of the
UWB pulse. It can therefore be concluded that although some tissues cause
distortion of the signals, the effect is acceptable and the correlation receiver
can be still applied for the reception of the signal.
2.3.5. Analysis of the system dynamic range for urine

detection in the human bladder
After the investigation of the microwave propagation in single human tissue,
the system dynamic range of the system together with the bladder model is
estimated. The goal is to predict the required system dynamic range for the
detection of the urine in the human bladder.
To demonstrate the frequency dependency of the radar link budget, the re-
ceived signal is evaluated in the frequency domain. If the transmitted signal
is denoted by X.f / and the received signal by Y .f /, then the whole prop-
agation process can be described with the following mathematical formula
[SW05]:
Y .f / X .f /
p D HRx .f / Hc .f / HTx .f / j! p ; (2.26)
Z0 Z0
where HTx .f / and HRx .f / are the transfer functions of the transmitting and
receiving antennas, respectively, and Hc .f / is the channel transfer function
of the scenario, which includes the propagation, reflections, dispersion and
attenuation of EM waves. The reflection from each tissue boundary is taken
into account. The received signal is hence a superposition of temporarily
shifted echoes from all boundaries.
In the simulation model, a plane wave is used as the excitation signal. The an-
tenna transfer functions (Tx and Rx) are then included in the model, however
the near-field effect of the antennas (affecting very slightly the signal attenu-
ation of the reflections from the boundaries of the tissues) is not considered
to simplify the simulation model [Vat10], since the focus is on the estimation
31
of the signal attenuation caused by human tissues. With this assumption the
obtained transfer function of the whole system Hs .f / can be written as
Y .f / e j!r=c0
Hs .f / D D j! HTx .f / S HRx .f / ; (2.27)
X .f / 2 rc0
where S is the reflection characteristic of the human abdomen model with
multiple tissue layers.
It must be noted that in medical scenarios, the antennas are the critical ele-
ments of the overall system and they directly affect the recovered signals in
terms of SNR and distortion. To demonstrate the best performance of the
simulation model, typical Vivaldi-antennas are applied, which are matched in
free space (off-body antenna). The antenna features a size of 78 75 mm2 ,
fed by aperture coupling and optimized for the frequency range from 2.5 to
12.5 GHz [SW05]. The transfer function of the antenna in main beam direc-
tion in the H-plane is shown in Figure 2.9. The drawbacks of the off-body
antennas will be given together with the results, which will motivate the de-
sign of the on-body matched antennas in chapter 5.
−25
−30
H(f) in dB
−35
−40
−45
2 4 6 8 10 12
frequency f in GHz
Figure 2.9.: The transfer function H.f / of the used Vivaldi antenna in main
beam direction, H-plane, co-polarization.
In the system concept, two identical Vivaldi antennas are used at the collo-
cated transmitter and receiver side. The main beam of the antennas is oriented
perpendicularly to the surface of the bladder model and vertical polarization
is applied. The antennas are placed at a distance of 50 mm from the human
abdomen model [20].
32
The transfer function of the whole system Hs in the considered scenario in-
cluding the antenna performance, free space propagation, attenuation, reflec-
tion and transmission in human tissues at different frequencies is depicted in
Figure 2.10. The initial attenuation value of lower than 0 dB results from
j!r=c0
the implementation of the term j! HTx .f / e 2c0 HRx .f / at the starting
point (d = -50 mm; antenna transfer function in the main beam direction is
used). The different attenuations at the starting point at different frequen-
cies are caused by the frequency dependency of the antenna transfer func-
tion. Furthermore, three important reflections from the boundaries between
fat, bladder muscle and urine are taken into account.
At 2 GHz, the attenuation of the reflection from the boundary between fat
and bladder muscle is at the level of 58 dB. The reflection from the boundary
between bladder muscle and urine is estimated to 76 dB. Due to the high
attenuation caused by the urine, the attenuation of the reflection from the
boundary between bladder and bladder muscle (at the rear side of the bladder)
is 120 dB. At 3 GHz, the reflection from the boundary between fat and bladder
muscle and boundary between bladder muscle and urine are at the level of
50 dB and 70 dB, respectively. The reason for this is the higher amplitude
of the antenna transfer function at 3 GHz (the antenna is optimized for the
frequency range from 2.5 to 11 GHz). It must be noted that the attenuation
increases strongly with increasing frequency above 3 GHz (note the scale in
the case of 7 GHz). The attenuation of the reflection from the boundaries
between bladder muscle and urine at 7 GHz is estimated to be 95 dB. At this
level of attenuation, the reflection is difficult to be detected due to the limited
sensitivity at the receiver.
The reflection from the boundaries between urine and bladder muscle (at the
rear side of the bladder) in the whole frequency range is not detectable, since
the attenuation is larger than 100 dB from 2 to 7 GHz. However, the reflection
from the boundary between urine and bladder muscle (at the front side of
the bladder) before the transmission through urine can be detected from 2
to 7 GHz. The variation of the front bladder muscle can be also utilized to
estimate the volume of urine in the human bladder.
It can be also seen that a strong reflection exists at the boundary between air
and skin, which introduces a reduction of 6.5 dB (incident and back-scattered
33
(a) f = 2 GHz
(b) f = 3 GHz
(c) f = 7 GHz
Figure 2.10.: Signal attenuation (system transfer function) of the different re-
flections from the boundaries in the human abdomen model (re-
fer to Figure 2.4) at different frequencies.
34
2.4. Summary
wave) in the system transfer function. This indicates that 75% of the transmit-
ted energy is scattered from the skin. The free space attenuation at a distance
of 50 mm introduces a decrease of 6 dB in system transfer function. Further-
more, the problem of the mismatch of the antenna placed close to the human
body has not yet been considered. Therefore, the off-body antenna causes a
high decrease (> 12.5 dB) of the system transfer function.
From these results it can be concluded that a very high system dynamic range
(larger than 80 dB) is needed to capture the weak reflections for the detection
of urine of the human bladder, since the reflected signals after transmission
through the human tissues are very weak and hardly detectable. For that
purpose, the frequency sweep approach can be adopted, since a high system
dynamic range can be obtained using large input power and low the noise
level (using small intermediate frequency (IF) bandwidth). However, the IR
approach is not suitable for this detection due to the system dynamic range
being strictly limited by quantization resolution of sampling unit [Vat10].
2.4. Summary
The dielectric properties of human tissues predicted from Cole-Cole equation
showed that the high water-content tissues such as muscle and urine exhibit
high relative permittivity, very high signal attenuation and low penetration
depth. On the other hand, the EM waves in tissues with low water-content
(e.g. fat and connective tissue) undergo a low attenuation. Moreover, be-
cause of the high dielectric contrast between tissues with high and low water-
content, strong reflections occur at their boundary, which can be utilized for
the detection purpose.
After the quantitative study of the dielectric properties of human tissues, the
performance of a microwave medical system for the detection of the urine in
the human bladder, using an IR-UWB concept, has been investigated. The in-
troduced model predicts that for a typical human body the distortion is small
enough for the reception of the signal. Furthermore, the predictions of the
presented model can be applied for the calibration in the radar signal pro-
cessing, which further increases the performance of the system and detection
capability.
35
In the concept for the analysis of the system dynamic range for urine detec-
tion in the human bladder, two Vivaldi-antennas matched in free space are
used. The results show that there will be a decrease of at least 12.5 dB of
the received signal level, using the traditional antennas matched in free space.
Considering this drawback, a concept of the on-body matched antennas men-
tioned in the introduction arises, which can improve the microwave system
performance for the medical diagnosis. More advantages of on-body matched
antennas will be given in chapter 5. By using an antenna array, which will
be discussed in chapter 6, the detection ability of weak reflections in such a
system can be significantly improved.
With the knowledge of this feasibility study based on the model of the hu-
man abdomen for the detection of urine in the human bladder using a realistic
model and parameters, the required system dynamic range and suitable fre-
quency range can be quantitatively predicted. It has been shown in signal
attenuation of reflections from the abdomen model that the UWB band from
3.1 to 10.6 GHz licensed for communication according to [FCC02] is not an
optimal frequency range for microwave medical diagnosis due to the high
signal attenuation in human tissues at the higher frequencies. Nevertheless,
the higher frequencies are needed to enlarge the bandwidth to achieve a fine
range resolution. A system dynamic range of 80 dB below 5 GHz is needed
for the detection of the weak reflections at the boundary between bladder
muscle and urine and is still achievable. However, the reflected signals above
7 GHz are not detectable and hence not usable. Therefore, the highest opera-
tional frequency of about 7 GHz can be considered. On the other hand, lower
frequencies are preferred to be applied due to their good penetration ability.
A big challenge at the lower frequencies is the large wavelength and hence
large antenna size. This leads to the challenge of miniaturizing the on-body
matched antennas, which will be given in chapter 5.
Regarding the suitable frequency band for on-body matched antennas in imag-
ing systems for diagnosis, frequency bands such as 1-7 GHz or even 0.5-
2 GHz for different applications are proposed (see chapter 5). For medical
data transmission using microwaves, low frequencies (low signal attenuation)
for implantable antennas are desirable (see chapter 4), since a long opera-
tional range is required for the wireless communication with the external de-
36
2.4. Summary
vices. However, the band is constrained to be very narrow (within 100 MHz)
according to the spectrum regulation of the government.
Since the body-matched antennas (on-body matched and implanted antennas)
are directly in contact or embedded in the human tissues, special character-
ization methods and measurement systems with tissue-simulating liquid are
required and will be, before the developments of different antennas, intro-
duced in the following chapter.
37
3. Characterization methods and
measurement systems of
body-matched antennas
This chapter deals with the methods for the characterization and measurement
verification of body-matched antennas. In the first section, different charac-
terization methods of body-matched antennas are introduced to evaluate the
performance of the to-be-designed antennas. Taking the lossy medium into
account, the important terms of body-matched antennas are discussed.
The second part of this chapter provides the verification systems of the body-
matched antennas. Since the antennas are matched to the lossy human body,
the antennas have to be surrounded by a tissue-simulating liquid for the mea-
surement verification. Due to a high signal attenuation in the tissue-simulating
liquid and the increasing signal attenuation with frequency, the measurement
distance and highest frequency are both limited. Therefore, two special an-
tenna measurement systems are introduced to experimentally verify the ra-
diation pattern of the body-matched antennas. The E-field probe based an-
tenna measurement system allows the direct measurement of the pattern of the
body-matched antennas at a short distance (< 60 mm). The far-field pattern
of the antennas can be obtained by near-field to far-field (NF-FF) transforma-
tion of the electric near-fields, which can be measured by a planar-rectangular
near-field measurement system.
39
3. Characterization and measurement systems of body-matched antennas
3.1. Characterization methods of

body-matched antennas
The analysis of antennas, which are surrounded by lossy medium is very dif-
ferent from the antennas in free space, since the lossy medium results in a dif-
ferent wavelength and introduces additional conduction current. Many quan-
tities (e.g. wavelength, wave number, attenuation constant, phase constant,
etc.) that are real in free space become complex in a lossy medium [Moo63,
Kar04]. Therefore, different characterization methods of body-matched an-
tenna such as reflection coefficient, bandwidth, efficiency and radiation pat-
tern are provided in the following section, some of those definitions are mod-
ified for antennas in lossy medium.
3.1.1. Antenna terms in a lossy medium

Impedance and reflection coefficient
An equivalent circuit of a transmitting antenna is shown in Figure 3.1, where

the antenna is fed by a feed line with characteristic impedance of Z0 . The
antenna impedance can be expressed as
Z A D RL C Rrad C jXA ; (3.1)
where RL and Rrad are the loss resistance and the radiation resistance of the
antenna, respectively. XA is the reactance of antenna.
Figure 3.1.: Equivalent circuit of the transmitting antenna.
40
3.1. Characterization methods of body-matched antennas
The antenna is assumed to be placed in a lossy medium with an infinitely

large dimension according to the spherical coordinates shown in Figure 3.2.
According to the definition in the case of the antenna in free space, the radi-
ation resistance can be modified in terms of radiated power Prad in far-field
and the input current I [Bal05]:
SL si n r 2 dd'
RR
Prad
Rrad D 2 D ; (3.2)
I I2
Figure 3.2.: Illustration of the spherical coordinate system.
where the power density in lossy medium SL can be written as

2 ˇ ˇ2 3
ˇ Eˇ
E
6ˇ ˇ 7 2˛r
SL D Re 4 5 e ; (3.3)
Zm
which is modified by introducing a decay factor e 2˛r (˛ in lossy medium

is a negative value ) to compensate for the attenuation caused by the lossy
medium (Zm is the wave impedance in a lossy medium). Although the atten-
uation of the waves in a lossy medium is already compensated, the radiation
resistance and E-field are still different compared to the one of the antenna in
free space. It is because the current distribution of the antenna is altered due
to the presence of the lossy medium with high relative permittivity.
41
With the radiation resistance and reactance of the body-matched antenna, the
antenna efficiency and the reflection coefficient can be determined as
Rrad
D ; (3.4)
Rrad C RL
Z A Z0
D : (3.5)
Z A C Z0
To achieve a low reflection coefficient, it is required to match the impedance

of the power source with that of the antenna. Ideally, the line impedance Z0
must be complex conjugate to the antenna’s impedance Z A . To achieve that,
in practice, various techniques e.g. matching networks are used.
The value of the magnitude of is equal to the input reflection coefficient
S11 . A good impedance matching is evaluated and compared to -10 dB re-
garding reflection coefficient.
Radiation pattern in near-field and far-field
The antenna radiation pattern is the specification of the angular dependence of

the radiated EM waves from the antenna. Based on the spherical coordinate
system (see Figure 3.2), the antenna pattern is related to the solid angles as
well as ' and is defined with regard to the E-fields as
ˇ ˇ ˇ
ˇE ˇ
ˇE .r; ; '/ˇ
ˇ ˇ
C .r; ; '/ D ˇ (3.6)
ˇ ˇ
ˇE ˇ
ˇE .r; ; '/ˇ
ˇ ˇ
max r D const:
ˇ
Depending on the distance to the antenna, a distinction must be made between

near-field and far-field radiation pattern. For that purpose, the electromag-
netic field radiated by an antenna can be simplified into three characteristic
regions: reactive near-field, radiating near-field and far-field.
The reactive near-field (Rayleigh zone) is caused by charges and currents on
the antenna and the reactive power is radiated. Increasing the distance to
the antenna (Fresnel distance Rnf ), the radiation fields (Fresnel zone) begin
but still depend on the distance to the antenna. At a certain distance away
42
3.1. Characterization methods of body-matched antennas
from the antenna, the radiated E-fields and H-fields are not affected by the
charges and currents. This range is called far-field (Fraunhofer zone) and
the corresponding distance is termed as Fraunhofer distance Rff . In the far-
field, the radiated wave of the antenna can be considered as a plane wave.
E-fields and H-fields are in phase and the antenna pattern at this region does
not depend on the distance. These three different ranges can be determined
by the dimension of the antenna DA and wavelength m as follows:
m
Reactive near-field: rA ; (3.7)
2
m 2DA2
Radiating near-field: rA ; (3.8)
2 m
2DA2
Far-field: rA : (3.9)
m
Since the human-body matched antennas are mostly operated in near-field

range, the near-field antenna patterns must be characterized. In this thesis,
the near-field pattern of the human-body matched antenna refers to the radi-
ating near-field of the antenna. Since the radiation pattern is governed by the
distance to the antenna, the near-field pattern must be specified for a certain
distance to the antenna.
Polarization
As discussed in the previous section, the radiated EM waves from the antenna
in the far-field consist of two orthogonal components: the electric and mag-
netic field vectors, which oscillate harmonically. The polarization of the EM
waves describes the time varying direction and relative magnitude of the EM
waves (normally referred to as the E-field).
Polarization can be classified as linear, circular or elliptical. In this thesis
only linear polarization is considered. The field is linearly polarized, when
the electric field vector always along a certain axis.
43
SAR
The Specific Absorption Rate (SAR) is a measure of the heating effect of EM

waves on human tissues. There are two major methods to obtain the SAR
value of a medical device. The technique which is based on equation (3.10)
determines the change in temperature (T ) within a time period (t), while
the measurement of the E-field intensity is based on (3.11).
T
SAR D Ci : (3.10)
t
jEj2
SAR D : (3.11)

The former technique has the drawback that most temperature sensors do not
have the required sensitivity to promptly evaluate the minimal temperature
gradient in a short exposure time, which is desired to minimize thermal dif-
fusion effects. Therefore, the electric field measurement is preferred. The
parameters required for the calculation are shown in equation (3.11), where
jEj is the magnitude of the RMS E-field and is the mass density.
Typically, the 1 g or 10 g SAR is used, which is the averaged SAR value over
a volume of a mass of 1 or 10 g tissue [Com01].
In this section, it has been discussed that the body-matched antennas can be
characterized in terms of reflection coefficient, efficiency and radiation pat-
tern. Additionally, the near-field pattern of the body-matched antennas in
lossy medium and the SAR value in human tissues are of interest. More char-
acterization methods of body-matched antennas (i.e. penetration efficiency
and front-to-back ratio) will be discussed together with antenna configura-
tions in chapter 5. Regarding these antenna terms, two measurement systems
for the validation of the radiation pattern are introduced in the following sec-
tions.
44
3.2. Measurement techniques for the characterization of body-matched antennas
3.2. Measurement techniques for the

characterization of body-matched
antennas
Due to the presence of human tissues, the antennas cannot be characterized
in free space and must instead be surrounded by the corresponding tissues.
However, In vivo measurements of each antenna with humans are not realis-
tic and dead tissues (such as pork and beef used to emulate the human muscle)
are not representative, because the "r and of tissues changes strongly after
the death. Therefore, most measurements in the literature take place in vitro,
since the dielectric properties of different human tissues were already inves-
tigated and modeled in [GLG96a, GLG96b] (see chapter 2).
In this thesis, different tissue-simulating liquids are used for the characteriza-
tion of different body-matched antennas. For the on-body matched antenna, a
solution of Polyethylene Glycol (PEG) 400 and water is applied as matching
liquid and introduced in the following section, while a sugar-water solution
is used for implantable antennas at 2.45 GHz (see chapter 4). The PEG-water
solution can approximate the dielectric properties of tissues in a wide fre-
quency range and the sugar-water solution is suitable for narrowband opera-
tion. The impedance matching and bandwidth can be measured directly by
immersing the antenna into the tissue-simulating liquid.
For the characterization of the antenna radiation pattern at a short distance,
an E-field probe based antenna measurement system has been developed. A
2D planar near-field antenna measurement is introduced to obtain the far-
field pattern of the body-matched antennas by applying the NF-FF trans-
formation. Since there is no standard antenna available to be immersed in
a tissue-simulating liquid for the measurement, on-body matched antennas,
which will be introduced in chapter 5, will be used for the measurement ver-
ification of the radiation pattern.
3.2.1. E-field probe based antenna measurement system

Since the body-matched antennas are directly in contact with human body
hence has a short distance (in cm range) to the target inside the tissues, the ra-
45
diation pattern of the antennas at a short distance must be investigated. How-

ever, the measurement distance is limited in the range from 40 mm (due to
the geometric limitation of the antenna) to 60 mm (due to the high signal at-
tenuation in liquid). At most frequencies, this distance corresponds to the
near-field range of the antennas. For some extremely small antennas with
wideband characteristic, it must be emphasized that this distance is in the
far-field range at low frequencies and near-field range for high frequencies.
The measurement system is characterized from 0.5 to 8 GHz with radiation
pattern measurement both in the E-plane and H-plane of the antennas.
Tissue-simulating liquid
To verify the antennas in the measurement with regard to the impedance

matching and radiation pattern, a tissue-simulating liquid is used to replace
the human body. For the verification of the on-body matched antennas, a
solution is composed of distilled water and Polyethylene Glycol (PEG) 400.
The characteristics of the sugar-water solution for implantable antennas will
be given in chapter 4.
PEG is a transparent liquid and is soluble in water. The different dielectric
properties of the solution based on different weight ratios of distilled water
and PEG are measured by a dielectric probe (85070E, Agilent) and shown in
Figure 3.3. The solution can emulate tissues such as fat, skin and muscle of
different permittivities. Since the on-body matched antennas are directly in
contact with the human body, an averaged dielectric property of skin and fat
(shown in Figure 3.3) is used to approximate the dielectric property of the
human body. It can be observed in Figure 3.3 that the solution of PEG and
distilled water (weight ratio of 6:4) has very similar relative permittivity to
the arithmetic mean of relative permittivity of human skin and fat (averaged
tissue) [GGC96]. Therefore, the 60% PEG-water solution is used to charac-
terize the on-body matched antennas for medical diagnosis (see in chapter 5)
[Sam12].
Furthermore, the signal attenuation constant versus frequency of the solution
in Figure 3.4 shows that the pure PEG liquid has a very low attenuation con-
stant. By mixing it with distilled water, the attenuation constant of the PEG-
water solution increases significantly due to the increased intensity of water
46
70
PEG 40%
60 PEG 50%
r
PEG 60%
Relative permittivity ε
50 PEG 100%
averaged tissue
40
30
20
10
0
0 1 2 3 4 5 6 7 8 9 10
Frequency f in GHz
(a) Relative permittivity
6
PEG 40%
5 PEG 50%
PEG 60%
4 PEG 100%
averaged tissue
3
0
0 1 2 3 4 5 6 7 8 9 10
Frequency f in GHz
(b) Conductivity
Figure 3.3.: Measured relative permittivity and conductivity of the PEG-water

solutions with different water content over frequency (averaged
tissue: averaged dielectric property of skin and fat predicted by
Cole-Cole equation).
dipoles and the related polarization process, which can also be observed in
the increasing conductivity resulting in attenuation. Moreover, the solutions
indicate a slight difference of their attenuation constant in the frequency range
from 1 to 3 GHz, while a large difference is present at high frequencies. In ad-
dition, the 60% solution indicates 5.8 dB/cm attenuation constant at 5 GHz,
which results in a 23.2 dB attenuation at this frequency at the distance of
47
40 mm between the two antennas. Compared to the attenuation constant of

averaged tissue, 60% PEG-water solution has similar value in the frequency
range from 1 to 6 GHz, however the difference increase over the frequency up
6 GHz. In conclusion, the large relative permittivity of the liquid is obtained
by mixing the PEG and water at the expense of a high signal attenuation con-
stant, which must be tolerated in the measurement.
20
PEG 40%
Attenuation constant α in dB/cm
PEG 50%
15 PEG 60%
PEG 100%
averaged tissue
10
0
0 1 2 3 4 5 6 7 8 9 10
Frequency f in GHz
Figure 3.4.: Signal attenuation constant of the PEG-water solutions with dif-
ferent water content over frequency (averaged tissue: averaged
dielectric property of skin and fat predicted by Cole-Cole equa-
tion).
Measurement setup of the E-field probe based antenna

measurement system
The measurement setup is shown in Figure 3.5. All equipment such as the
VNA and the motors are controlled by a laptop via USB interface. Two linear
stages with step motors with an accuracy in the m range are applied to adjust
the position of the E-field probe in x and y directions. A rotary table with
a step motor (M-062) serves to rotate the Antenna Under Test (AUT) for
measurement of the radiation pattern, hence a 360ı rotation in xz-plane can
be performed.
To increase the dynamic range of the measurement system, (to counter the
high signal attenuation in the liquid phantom), a power amplifier (Hittite)
48
with an amplification of 19 dB from DC to 15 GHz is used. An additional

low noise amplifier (LNA) can be used optionally at port 2. The dotted and
solid lines in Figure 3.5 indicate the control signals for the devices and the
RF signals, respectively.
The AUT is connected to port 1 of the VNA and the E-field probe to port 2.
Since only a measurement distance of 40-60 mm can be achieved due to the
high signal attenuation, the size of the probe is very critical for the accuracy
of the measured radiation pattern. The effective area of the probe should be as
small as possible compared to the aperture of the AUT [33]. Furthermore, the
measured pattern of the antenna is specified for a certain distance, since the
measured pattern is dependent on the measurement distance due to the short
measurement distance (in most cases in near-field range).
Figure 3.5.: Schematic illustration of the antenna measurement setup using

tissue-simulating liquid to emulate human tissues.
An open-ended E-field probe, fabricated from a 50 semi-rigid cable (5.6

mm diameter coaxial cable), is used as a coupling element in lossy medium
[Smi75]. A certain length of the outer conductor and dielectric (Teflon) have
been stripped off, exposing the inner conductor, as depicted in Figure 3.6
49
(a). The sensitivity of the probe can be improved by increasing the length
or the diameter of the inner conductor [BLM06]. However the geometry of
the E-field probe influences the spatial resolution as well as the accuracy of
the measurement system therefore the following measurement was taken to
obtain an optimal geometry. The S11 of the probes with different lengths of
exposed inner conductor in Figure 3.6 (b) shows that the E-field probe is not
matched in air but an acceptable impedance matching is achieved in PEG-
water solution from 1 to 8 GHz. The S11 at low frequencies is improved by
increasing the length of exposed inner conductor. With a longer exposed inner
conductor, the probe is more sensitive to capture the radiated E-fields from
AUT. However, the measurement accuracy in terms of spatial resolution can
be degraded due to the large size of the probe antenna, which is comparable
to the measurement distance. A length of 4 mm is used to achieve a high
sensitivity of the probe, while a high accuracy of the measurement due to
spatial resolution can be also provided. Its impedance matching is sufficient
for capturing the E-field of the radiated wave of the AUT.
The E-field probe is inherently sensitive to the polarization of the radiated
waves according to the orientation of the inner conductor. The Ey component
is measured based on the configuration as shown in Figure 3.5. Therefore, the
AUT can be characterized with the E-plane and H-plane patterns by rotating
the AUT in the xy-plane. Since the probe is fixed during the measurement,
the pattern of the probe does not effect the measurement results and an ad-
ditional calibration with pattern compensation of the probe is not required.
A comparison of the results between the different frequencies is extremely
difficult, since the signal attenuation in the tissue-simulating liquid cannot be
calculated exactly [33]. Therefore, for different frequencies, the measured
S21 is normalized to its maximum, respectively.
The parameters of the antenna measurement system are summarized in Ta-
ble 3.1. To achieve a maximum dynamic range of the measurement system,
the transmitting power of the VNA is set to its maximum 5 dBm and the IF
bandwidth to 100 Hz. This configuration results in a dynamic range of 100
dB. However, the frequency, at which the complete pattern in 360ı of E-plane
or H-plane can be measured, is still limited up to 4 GHz (at the measurement
distance of 40 mm) due to the signal attenuation in PEG-water solution.
50
−2
−4
S11 in dB
−6
−8 4 mm, in air
1.5 mm, in PEG
4 mm, in PEG
−10
10 mm, in PEG
15 mm, in PEG
−12
0 1 2 3 4 5 6 7 8
Frequency f in GHz
(a) Photo (b) Measured S11
Figure 3.6.: Photo and measured S11 of the E-field probes with different
lengths of exposed inner conductor (a: photo; b: S11 in air
and the PEG-water solution of the E-field probes with different
lengths of exposed inner conductor).
(a) Measurement setup (b) AUT and probe
Figure 3.7.: Photos of the antenna measurement setup with tissue-simulating

liquid.
51
Table 3.1.: Parameters of the E-field probe based antenna measurements sys-
tem.
Transmitting power 5 dBm IF bandwidth 100 Hz
Frequency range 0.5 to 4 GHz Number of frequency points 751
Motors in use 3 Angular step 4ı
A planar stepped-slot antenna optimized in the frequency range from 1 to 7

GHz [3] (the design will be explained in detail in chapter 5) is used as AUT
for verification of the measurement system. For total immersion into liquid,
the antenna has to be watertight so that the cavities on the top side will not
be affected by the liquid as shown in Figure 3.8. Styrofoam is used to fill the
cavities. The prototype is then wrapped in an absorber material with glue and
copper foil to minimize the radiation to the surrounding environment. There-
fore, the radiation occurs only in the half plane of the main beam direction.
(a) (b) (c) (d)
Figure 3.8.: Photos of the prototype of the stepped-slot antenna (see in chapter
5). (a: antenna with Styrofoam to maintain the free space region;
b: antenna with absorber material; c: antenna with pine tar for
isolation to the liquid and d: antenna with copper foil).
52
Verification of the E-field probe based antenna measurement

system
To evaluate the performance of the E-field probe based antenna measurement

system, the measured radiation pattern using AUT (the planar stepped-slot
antenna) will be compared with the simulated results.
The measured S21 in main beam direction between the AUT and the probe at
different distances are shown in Figure 3.9. In the measured S21 , the influence
of the cable and the power amplifier are eliminated, however, the attenuation
caused in the PEG-water solution is still included. The signal attenuation
increases strongly versus frequency. At a distance of 20 mm, the AUT cannot
be rotated completely in 360ı due to the dimension of the AUT. Therefore, a
distance of 40 mm was chosen, so that the measured E-field is in the radiating
near-field for this antenna and the radiation pattern in 360ı up to 4 GHz can be
obtained completely. The S21 of the co-polarization (co-pol) and the cross-
polarization (x-pol) at the distance of 40 mm show that the probe guarantees
a polarization suppression of more than 20 dB.
0
co−pol, 20 mm
co−pol, 40 mm
−20
co−pol, 60 mm
x−pol, 40 mm
−40
S21 in dB
−60
−80
−100
0 1 2 3 4 5 6 7 8
Frequency f in GHz
Figure 3.9.: Measured S21 at different distances between AUT and probe
(main beam direction).
Figure 3.10 (a) and (b) show the simulated normalized pattern of the planar
stepped-slot antenna at a distance of 40 mm from the antenna immersed in
53
PEG-water solution. This distance corresponds to the thicknesses of fat and

muscle tissues of the human abdomen. Since the simulated radiation pattern
up to higher frequencies (noise is not considered in CST simulation) can be
obtained, the results range from 1 to 9 GHz. In the E-plane, the main beam
direction is constant over the whole frequency range due to the very stable
phase center. The beamwidth of the radiated E-field however decreases with
frequency. At 6 and 9 GHz, two sidelobes are observed. In the H-plane,
variation of main beam direction over the frequency is observed. The beam
is significantly wider than the beam in the E-plane.
The measured normalized patterns of the planar stepped-slot antenna at dif-

ferent frequencies are shown in Figure 3.10 (c) and (d). The results are lim-
ited to 4 GHz due to the high signal attenuation in PEG solution at the higher
frequencies. The measured patterns at 1 and 3 GHz both in the E-plane and H-
plane are similar compared to the simulated results. The measured pattern in
the H-plane with a wide beamwidth is relatively constant over the frequency.
Side lobes in the E-plane are observed. The slight asymmetry of the pattern
in the E-field is caused by the slightly asymmetric connection of antenna el-
ement and feed network. The difference between simulated and measured
results are mostly caused by the different boundary conditions. In the simula-
tion, the boundary is set to be an ideal open boundary without reflections. The
volume of the tissue-simulating liquid in the measurement however is limited
and thus the tank has only a dimension of 406020 cm3 . Weak reflections
at the boundary of the tank contributes to the difference. The reflection from
the surface of the tank at the low frequencies is of significance due to rela-
tively low signal attenuation. Another reason could be the water leakage into
the antenna (details see chapter 5).
In conclusion, the agreement between simulated and measured results con-

firms that the E-field probe based near-field measurement system can be used
for the verification of designed body-matched antennas in terms of impedance
matching and radiation pattern.
54
0 0
−5 −5
Normalized pattern in dB
−10 −10
−15 −15
−20 1 GHz −20 1 GHz

3 GHz 3 GHz
−25 6 GHz −25 6 GHz
9 GHz 9 GHz
−30 −30
−90 −60 −30 0 30 60 90 −90 −60 −30 0 30 60 90
Angle in degree Angle in degree
(a) Simulated, E-plane (b) Simulated, H-plane
0 0
−5 −5
−10 −10
−15 −15
−20 1.0 GHz −20 1.0 GHz

2.0 GHz 2.0 GHz
−25 3.0 GHz −25 3.0 GHz
4.0 GHz 4.0 GHz
−30 −30
−90 −60 −30 0 30 60 90 −90 −60 −30 0 30 60 90
Angle in degree Angle in degree
(c) Measured, E-plane (d) Measured, H-plane
Figure 3.10.: Simulated and measured normalized radiation pattern of the pla-
nar stepped-slot antenna in the E-plane and H-plane at a distance
of 40 mm.
3.2.2. Planar-rectangular near-field measurement

system
Though the body-matched antennas are used for near-field range operation,
the far-field pattern and gain of the antennas can be characterized as a mea-
sure of the antenna performance. In simulation, the far-field pattern and gain
of the antennas can easily be calculated by excluding the conductivity of the
medium. In this way, the signal attenuation caused by the lossy medium is not
55
included. However, in most cases, the pattern in the far-field cannot be mea-
sured directly, since the tissue-simulating liquid (PEG-water solution) causes
a high signal attenuation and thus, the measurement distance is strongly lim-
ited.
To counter this limitation, a different measurement setup is required. A
planar-rectangular near-field antenna measurement system is designed to ob-
tain the far-field pattern of the body-matched antennas. The electric near-field
of the AUT immersed in the PEG-water solution is measured and the far-field
pattern is processed through NF-FF transformation. The principle of the NF-
FF Transformation and the planar-rectangular near-field measurement system
setup are provided as follows.
Principle of NF-FF Transformation
The planar NF-FF transformation is based on the plane wave spectrum ap-
proach using Fourier transform. The E-field in the far-field region can be
determined by
1 1
E y; z/ D 1
Z Z
j kE rE
E.x; FE .kx ; ky /e d kx d ky ; (3.12)
4 2 1 1
where FE is the plane wave spectrum, kx and ky are the wave numbers of
the plane wave spectrum in x and y directions, respectively [Bal05]. rE D
rE.x; y; z/ is the distance to the origin of the coordinate system.
In the framework of the proposed near-field antenna measurement, a planar-
rectangular scanning for the acquisition of the E-field in the near-field region
of the AUT is applied. According to the coordinate system for the antenna
measurement setup (refer to Figure 3.12), two components of the E-fields
in the x and y directions can be obtained from the measurements. Assuming
that Ex0 and Ey0 are the measured tangential electric near-fields over the plane
scanned by the beams .z D 0/, the aforementioned plane wave spectrum (Fx
and Fy in x and y directions) can be written as [Bal05] [GMP07]
56
Z 1 Z 1
Fx .kx ; ky / D Ex0 .x; y; z D 0/e j.kx xCky y/ dxdy; (3.13)
1 1
Z 1 Z 1
Fy .kx ; ky / D Ey0 .x; y; z D 0/e j.kx xCky y/ dxdy: (3.14)
1 1
With the known plane wave spectrum, the E-field in the far-field region can be
obtained. To present far-field pattern in terms of spherical angles, the E-fields
for two planes (E-plane and H-plane) can be modified to
ke jkr
E .r; ; '/ ' j ŒFx cos' C Fy si n'; (3.15)
2 r
ke jkr
E .r; ; '/ ' j cosŒ Fx si n' C Fy si n': (3.16)
2 r
Taking the medium with high permittivity into account, the wavenumber can
be written as
nq o

Re "r j !" 0
k D 2f ; (3.17)
c0
The imaginary part of the wave number is ignored and the signal attenua-
tion is not considered in the transformation algorithm, since the results are
normalized in spherical coordinate system at each frequency.
Moreover, in the planar-rectangular scanning, the dimension of the scan area
is limited by the real size of the measurement system and the strong signal
attenuation in the tissue-simulating liquid, in which the antenna is immersed.
Therefore, the missing information outside the scan area limits the maximum
achievable angle max after NF-FF Transformation [GMP07]. Furthermore,
the truncation error for the given size of the scanning area restricts the max
[DFGC 08]. The maximum achievable angle max is illustrated in Figure 3.11
(a), where the DA1 and DA2 are the apertures of the AUT and the probe an-
tenna in one direction (x or y), L is the length of the scan area. max can be
estimated by (3.18). In Table 3.2, different max are provided with respect to
57
the L and rA (assuming the aperture of the measured antenna DA1 = 35 mm,
DA1 D DA2 ) [Sam12].

L DA1 DA2
max D arctan : (3.18)
2rA
(a) Scan scenario (b) Scan area (xy-plane with dot-

ted line in (a))
Figure 3.11.: Schematic representation of the scenario for the determination

of the max and the scan area.
To achieve a large max , a very large scan area and huge tank with PEG-water
solution are required, which is difficult to be realized. Therefore, planar-
rectangular scanning is more suitable for antennas with high gain and small
beamwidth.
Table 3.2.: Maximum achievable angle max in the far-field pattern depending
on L and rA .
L in mm 104 104 112 112 120 120
rA in mm 15 20 15 20 15 20
max in degree 48.57 40.36 54.46 46.40 59.04 51.34
58
Figure 3.12.: Block diagram of the planar-rectangular near-field antenna mea-

surement system.
Measurement setup of the planar-rectangular near-field

measurement
The planar-rectangular near-field measurement system uses a similar setup as

the E-field probe based measurement system and is illustrated in Figure 3.12.
The design of the overall measurement system with its important parameters
is given in Table 3.3. The AUT is connected to port 1 of the VNA and the
probe with port 2. Two identical antennas (the planar stepped-slot antenna)
are used as AUT and probe, since the antenna is required to be immersed in
the liquid and traditional horn antennas or open-ended waveguide probes are
not suitable. The PEG-water solution is also used as tissue-simulating liquid.
The relative permittivity of the PEG-water solution is taken into account in
the transformation algorithm.
Since the electric near-field is measured and the far-field pattern is obtained
through a NF-FF Transformation, the near-field and far-field range must be
analyzed for a certain antenna. The far-field range (Fraunhofer distance) can
be calculated using (3.9), where DA is the largest aperture of the antenna
(diagonal: 43.6 mm). "r is the relative permittivity of the tissue-simulating
liquid. As an approximation, the "r is assumed to be 20 for the whole fre-
quency range from 1 to 8 GHz, which is used in the antenna design in the
59
Table 3.3.: Design parameters of the planar-rectangular near-field measure-

ment system.
frequency range 0.5 - 8 GHz
frequency points 751
scan length in x direction 104 mm
scan length in y direction 104 mm
scan step in x and y direction 4 mm
measurement distance in z direction 15 mm
simulation (see chapter 5). The reactive near-field and far-field range of the
antenna are given in Table 3.4. The measurement should be performed in
the radiating near-field range of the antenna [32]. Regarding the Rnf (Fresnel
distance) and the Rff (Fraunhofer distance) in the frequency range from 1 to
8 GHz, a measurement distance of 15 mm between the AUT and the probe
antenna is chosen.
Table 3.4.: Reactive near-field and far-field range of the measurement antenna
at different frequencies based on a frequency independent "r D 20.
f (GHz) 1 3 5 7 8
Rnf (mm) 10.68 3.56 2.13 1.53 1.34
Rff (mm) 56.67 170.03 283.52 396.03 453.15
Moreover, the scan steps in the x (x) and the y (y) direction both are
chosen to be 4 mm so that a maximum uniform sample spacing m =2 is main-
tained. The scanning area is set to be 104104 mm2 . A planar stepped-slot
antenna with a size of 2635 mm2 is used as AUT for the antenna measure-
ment. This configuration results in maximum achievable angles of 60ı and
48.6ı in the E-plane and H-plane, respectively.
Verification of the planar-rectangular near-field measurement
For the NF-FF Transformation, the E-fields in both the co- and cross-polariza-
tion of the AUT are required. The measured raw data at different frequen-
60
cies are shown in Figure 3.13. Since the maximum magnitude of S21 in Co-
polarization decreases strongly with the frequency due to the increase of the
signal attenuation over frequency, the near-field pattern is normalized at each
frequency, respectively. Moreover, for a better comparison between co- and
cross-polarization, the magnitude of S21 is normalized to the maximum of
co-polarization at each frequency.
It can be observed that at 1 GHz the maximum magnitude of S21 in co-
polarization is normalized to be 0 dB, where the receiving antenna is placed
in the center of the scan zone. On the edge of the scan area, the S21 drops
to -49 dB. However, the measurement in cross-polarization indicates -30 dB
of S21 in the center of the scan area and the value increases to a maximum
(-15 dB) at (x=12 mm, y=12 mm). At 1 and 4 GHz, 4 maximums can be seen
in cross-polarization. Moreover, these results show that the AUT exhibits a
very good cross-polarization suppression [32].
Furthermore, the raw data at the lower frequencies show a wider beamwidth
than at the higher frequencies, which must result in a wide pattern in far-
field at the low frequencies after NF-FF transformation. This agrees with the
theoretical analysis that at the low frequencies the antenna provides a wide
pattern due to the small electrical dimensions of the antenna.
In Figure 3.14, the normalized patterns processed by the NF-FF transforma-
tion are provided in the E-plane and H-plane. The results are compared with
the simulated pattern at different frequencies. In simulation, the dielectric
property of the PEG-water solution is also implemented. The scan area, scan
step and antenna distance remain as in the measurement setup for the verifi-
cation.
It can be seen that the far-field radiation pattern at the higher frequencies
(i.e. 6 and 8 GHz) could also be obtained using this planar-rectangular near-
field antenna measurement system and NF-FF transformation, which cannot
be measured directly in the E-field probe based measurement system due to
the extreme high signal attenuation (in the far-field range). The processed far-
field patterns based on the simulated and measured data agree with each other
at 1 and 4 GHz. At 6 and 8 GHz, differences between simulated and measured
results are observed, which are caused by the different boundaries of the lossy
medium in the simulation and measurement. However, a remarkable similar-
61
(a) Co-polarization, 1 GHz (b) Cross-polarization, 1 GHz
(c) Co-polarization, 4 GHz (d) Cross-polarization, 4 GHz
(e) Co-polarization, 8 GHz (f) Cross-polarization, 8 GHz
Figure 3.13.: Measured raw data S21 at different frequencies over the whole
scan area: (a) in co-ploarization and (b) in cross-polarization.
62
3.3. Summary
ity of the pattern can be confirmed. In conclusion, the planar-rectangular

near-field antenna measurement system enables to verify the far-field radia-
tion pattern of the body-matched antennas.
0° 0°
30° −30° 30° −30°
60° −60° 60° −60°
simulated 1 GHz simulated 1 GHz

measured 1 GHz measured 1 GHz
measured 4 GHz measured 4GHz
90° −90° 90° −90°
−30 −20 −10 0 −30 −20 −10 0
(a) E-plane, 1 and 4 GHz (b) H-plane, 1 and 4 GHz

0° 0°
30° −30° 30° −30°
60° −60° 60° −60°

measured 6 GHz measured 6GHz
90° −90° 90° −90°
−30 −20 −10 0 −30 −20 −10 0
(c) E-plane, 6 and 8 GHz (d) H-plane, 6 and 8 GHz
Figure 3.14.: Far-field patterns obtained using the NF-FF transformation of

simulated and measured data at different frequencies in the E-
plane and H-plane, co-polarization (normalized in dB).
3.3. Summary
Different characterization methods for the evaluation of the body-matched
antennas have been discussed in this chapter. The body-matched antennas
can be characterized like the antennas for free-space operation in terms of
impedance matching, antenna gain and radiation efficiency. Regarding the
radiation pattern, the near-field and far-field patterns can be used. For the
body-matched antenna, in particular the near-field pattern is of significance,
since the body-matched antennas are used at a short distance to the target. In
63
addition, the SAR value is a special characteristic of antennas for medical ap-
plications and required to be evaluated to guarantee the safety for the human
body with regard to the RF radiation.
To verify the characteristics such as impedance matching and radiation pat-
tern of the body-matched antennas, two special antenna measurement sys-
tems are developed: the E-field probe based measurement system and the
planar-rectangular near-field measurement system. To approximate the di-
electric properties of human tissues (surrounding medium of body-matched
antennas) in the measurement, tissue-simulating liquid (PEG-water solution)
is used. The body-matched antennas must be immersed into the liquid in the
measurement. The results show that the E-field probe based antenna mea-
surement system enables a measurement of the radiation pattern at a short
distance to the body-matched antennas. However, the measurement distance
is limited (40-60 mm) in terms of SNR of the received signal due to the high
signal attenuation in tissue-simulating liquid. To obtain the far-field pattern of
the antennas, the planar-rectangular near-field measurement system is applied
to measure the electric near-fields. Then the far-field pattern is obtained by
using a NF-FF transformation. The verified results (compared to the simula-
tion results) confirm that these measurement systems are feasible to measure
the radiation far-field pattern of the body-matched antennas in the E- and
H-plane. In the following chapters, the emphasis will be on the near-field
pattern of these antennas, since the antennas will be operated at a short dis-
tance. Therefore, the E-field probe based antenna measurement system will
be used for the verification of the to-be-designed antennas.
After considering the characterization methods of body-matched antennas
and the development of the measurement systems, the implantable antennas
for data telemetry and on-body matched antennas for medical diagnosis will
be introduced in the following two chapters.
64
4. Implantable antennas for wireless
communication of IMDs
This chapter deals with one of the major microwave applications in the med-
ical field: data transmission between implanted and external devices. Based
on this application, this chapter focuses on the design of implantable antennas
in terms of their small size, high efficiency and robust performance. First, an
overview of the state-of-art research and challenges for the wireless commu-
nication of the IMDs are given. Based on these challenges and requirements
for the implantable antennas, the design of the stripline-fed double-layer slot
antennas is then provided together with miniaturization techniques to opti-
mize the antenna structures. The proposed antennas are intended to be im-
planted in the muscle tissues in the human body. The simulation results are
validated in the measurement using sugar-water solution, which emulates the
muscle tissues. This chapter is concluded with a performance comparison of
the three antennas with regard to their bandwidth, size, efficiency and sensi-
tivity against permittivity variation of the muscle tissues.
4.1. Overview of wireless communication of

IMDs
For most modern IMDs, using a wireless communication link between im-
plants and external devices is the optimum choice compared to using wired
connection. Contrary to traditional methods of using wires for the connection
with the implanted devices for the acquisition of diagnosis signals, implants
based on wireless communication require minimum incision into the body.
Inspections of IMDs can be then carried out without frequent chirurgical op-
eration. Wireless communication of IMDs enables the reduction of costs (e.g.
65
4. Implantable antennas for wireless communication of IMDs
maintenance cost of devices), lowers the risk of infection due to the wires,
which leads to the increase of comfort for the patients. Furthermore, such
telemetry systems of IMDs have the possibility of monitoring the patient’s
physical state as well as controlling the functionalities of the devices at a cer-
tain distance between the external devices and the patient. The data can then
be transfered immediately to the medical center. In emergency situations,
it allows the immediate detection and treatment of physical abnormalities,
which results in time saving and eliminating hospitalization costs by home
care monitoring [SSC94].
To limit the radiation of RF devices and to guarantee a safe level of human
exposure to RF emissions, regulations are set up by the Federal Communica-
tions Commission (FCC) with regard to the operational frequency bands and
power limitations, which will be discussed in the following sections. Then,
the challenges of wireless performance for IMDs will be described.
4.1.1. Frequency bands and power limitations

Wireless IMDs are allowed to operate in several frequency bands according
to the different standards in different countries. Depending on the data rate,
operational distance and power transfer capability, different radiation regula-
tions are applied. The FCC has distributed the spectrum allocations shown in
Table 4.1 for medical applications [Com01, Ser09, ITU01].
Table 4.1.: Frequency bands and power limitations for medical devices ac-
cording to the FCC.
Band Frequency Maximum EIRP
MedRadio 401-406 MHz -16 dBm
ISM 2.4-2.5 GHz 20 dBm
As established by the FCC, the operational frequency bands for medical appli-
cations are mainly the Medical Device Radiocommunications Service (MedRa-
dio) from 401 MHz to 406 MHz and the Industrial, Scientific and Medical Ra-
dio (ISM) band from 2.4 to 2.5 GHz. The MedRadio band is used by a variety
of devices in and near the human body to extend the previously established
66
4.1. Overview of wireless communication of IMDs
Medical Implant Communication Service (MICS), which mostly comprises

communication of IMDs. This band is preferred where low interference is
required. However, the ISM Band has the advantage of smaller achievable
antenna size due to shorter wavelength and authorization of unlicensed use.
Furthermore, these medical devices have to adhere to the power limitation
set by the FCC in Table 4.1 to minimize the interference to other existing
RF systems. This is also due to the impact of EM fields on human tissues,
which causes hazardous dielectric heating. This then leads to tissue burns
and disturbance to the human body functionality caused by the rise of the
temperature as well as blood pressure [oNiRPI98]. The heating effect can
be quantitatively determined in terms of SAR. The International Commission
on Non-Ionizing Radiation Protection (ICNIRP) has established the European
limit for SAR of 2 W/kg averaged over 10 g of tissues. In the United States the
requirements are stricter, as the FCC allows a maximum power of 1.6 W/kg
averaged over 1 g of tissues.
With the limitations of frequency bands and radiation power of the IMDs in
mind, the main challenges of telemetry systems with IMDs come down to the
design of implantable antennas, which is the focus of this chapter. A well
designed implantable antenna should be light, small yet fulfills the techni-
cal requirements of efficient RF radiation and transmission. The design of
small-sized implantable antennas in the ISM band is considered, which al-
lows a miniaturization of the implants. The details about the challenges of
the antenna design can be found in the following section.
4.1.2. Challenges of the implantable antenna design

The implantable antennas should fulfill various requirements such as hav-
ing a high efficiency, stable pattern and low SAR value. The high radiation
efficiency in turn contributes to the low energy consumption of the overall im-
plants and therefore a long operating time of the battery, which then reduces
the times of changing the battery by means of chirurgical operation. SAR is
related to the exposure of the human tissues to RF. A low SAR value means a
low exposure to RF. Therefore the SAR value is capped at a maximum allow-
able safe value for RF medical devices. The SAR distribution is also depen-
dent on the geometrical model of the human body and the radiation pattern
67
of the implantable antenna. Therefore, highly directional radiation pattern of

the implantable antenna must be prevented and an omni-directional pattern is
desirable.
Another challenge is that the radiation performance of the implantable anten-
nas is dependent on the location in the body, where it is embedded. Since
the implantable antennas are totally surrounded by human tissues, the char-
acteristics of the implantable antennas such as resonance frequency, band-
width, efficiency and input impedance are strongly dependent on the dielec-
tric properties of the surrounding tissues. Most of the implantable antennas
are designed using a resonant structure with narrow bandwidth. The problem
with using a resonant structure is that the shift of the resonance frequency of
the antenna due to the change of the relative permittivity of the surrounding
tissues (change of water content of tissues or difference between different in-
dividuals), causes significant degradation of the performance of the devices
(e.g. efficiency). The reduced efficiency results in a short battery lifetime of
IMDs. Furthermore, the radiating element usually must be electrically iso-
lated to prevent a short circuit by ions in the tissue. Requirements for the
coating materials used include bio-compatibility, mechanical robustness and
long-term durability. In conclusion, the robustness of the antenna in terms of
impedance matching and radiation efficiency is extremely important, which
should be validated in a realistic scenario.
For the verification of the antenna performance, in vivo measurements on hu-
mans are difficult to be permitted, whereas measurements carried out on live
test subjects (e.g. rats) are expensive. Therefore, in vitro measurements are
often applied, where only slight differences are observed in comparison with
the in vivo measurements [GLG96a]. As the "r and of tissues change sig-
nificantly after death, fresh meat such as beef and pork cannot be used as the
phantom. Currently, liquid phantoms are preferred to emulate the dielectric
property of human tissues for implantable antennas due to their high stability
of the dielectric property and low cost.
For implantable antennas, it is desirable that the size is as small as possible
to minimize the site of implantation. Since the size of an antenna is directly
proportional to the operating wavelength, further miniaturization techniques
must be employed. Therefore different methods for miniaturization have been
68
4.2. Design of stripline-fed double-layer slot antennas
addressed in the literature. For devices to be placed between skin and muscle,
for instance, a shorted pin on a spiral structure to the ground is introduced
[KRS04]. A smaller size of the proposed antenna was achieved compared to
the microstrip antenna. A four layer stacked triple band antenna was intro-
duced in [HLCC 11]. The electric path of the surface current has been length-
ened by the design of a meandered structure in [KHT08] and the antenna size
was significantly reduced. However, the antenna suffers from a limited band-
width of several tens of MHz. Moreover, a lower radiation efficiency of the
many implantable antennas have to be tolerated.
The goal of this chapter is to design a miniaturized implantable antenna with a
high efficiency, omni-directional radiation properties and robust performance
with the slight changing of the dielectric properties of tissues. To be able to
design a small-sized antenna, the ISM band (2.4 to 2.5 GHz) is chosen instead
of the MedRadio band. For the operation of the antenna in the ISM band,
the antenna must have a bandwidth larger than 100 MHz. In the following
section, the details of the design principles will be given.
4.2. Design of stripline-fed double-layer slot

antennas
First of all, the location of the implantable antenna becomes critical for the
antenna design, since the implantable antennas are completely surrounded by
human tissues. Table 4.2 shows the dielectric properties of different tissues at
the ISM band (2.45 GHz). The relative permittivity of blood, liver and muscle
are very similar, while a large difference between fat and muscle is observed.
Thus, the location of the implantable antennas must be determined at first.
It can be found that a large number of common implanted devices such as
pacemakers, defibrillators, and radio-frequency identification (RFID) implants
are located in the muscle tissues. Therefore, the main objective of this chapter
is to design implantable antennas, which can be embedded in or between the
muscle tissues. It is desirable that the antenna has an omnidirectional pattern
so that signals can be sent to the base station (external devices) regardless of
the patient’s orientation towards the base station.
69
Table 4.2.: Dielectric properties of human tissues at 2.45 GHz [GGC96,

GLG96a].
Tissue "r in S/m
Blood 58.26 2.545
Heart 54.81 2.256
Muscle 52.72 1.73
Liver 43.03 1.686
Fat 5.28 0.104
In the literature, several designs of implantable antennas with a three dimen-

sional structure are addressed in [MBZC 11]. In [WSTI09], an implanted
three-dimensional H-shaped cavity slot antenna is proposed for short range
wireless communications. However, a high fabrication complexity is asso-
ciated with it. To reduce the fabrication complexity and cost of implantable
antennas, antenna designs with planar structures are considered in this thesis.
Regarding the feeding technique of the implantable antenna, a stripline-fed
technique is used instead of a microstrip feed to minimize the spurious radia-
tion (e.g. leaky waves) along the feed line, which is surrounded by the muscle
tissues with a high relative permittivity. This is because the microstrip feed-
ing line leads to an additional loss thus causing undesired reduction of the
total radiation efficiency. Figure 4.1 illustrates the E-field distributions of
microstrip and stripline in the muscle tissues. Figure 4.1 (a) shows the mi-
crostrip located between the substrate ("r;sub = 10.2) and muscle ("r = 52.72),
which results in strong surface and leaky waves. The surface waves are partly
absorbed in the muscle tissues, while the leaky waves cause the distortion of
the radiation pattern. Therefore, the realized gain and efficiency of the im-
plantable antenna will be severely reduced. Moreover, the wave impedance
of the microstrip line changes, hence the impedance matching at the antenna
port is degraded, with the change of the dielectric properties of the muscle tis-
sues (change of the water content of different individuals). The antenna fed
by slotline and coplanar waveguide (CPW) suffers from the same problem in
the design of implantable antennas [SFC04].
70
In the case of a stripline configuration as shown in Figure 4.1 (b), the feed is
enclosed by two ground planes. No surface or leaky waves occur and hence
the crosstalk to the muscle tissues is prevented, i.e. no distortion of the radia-
tion pattern is introduced by the feed line. Furthermore, the wave impedance
of the stripline is independent of the dielectric property of the surrounding
tissues, since the electric fields are totally enclosed between the two ground
planes. This results in the elimination of dispersion and the propagation of
pure transverse electromagnetic (TEM) waves.
(a) Microstrip (b) Stripline
Figure 4.1.: Schematic illustration of the E-field distributions of microstrip (a)

and stripline (b) between the substrate ("r;sub = 10.2) and muscle
tissues ("r = 52.72).
For a size reduction of the antenna, a simple method is to increase the relative
permittivity "r and the thickness hs of the substrate. However, the increase
of "r;sub decreases the achievable bandwidth [KKYP01]. With increasing hs ,
fringing fields are enforced, leading to extension of the electric field lines over
the actual antenna geometry. The antenna appears electrically larger, resulting
in a shift of resonance frequency into the lower frequency range. However, for
large hs , the surface waves dominate. A part of the input power is converted in
surface waves instead of being radiated. In addition, reflection and scattering
at the substrate’s physical boundaries increase the cross-polarization level,
71
thus reducing the gain and the radiation efficiency. Also, the radiation pattern
is distorted. Therefore, "r and hs must be chosen carefully.
Another technique for size reduction is to increase the electrical length of the
radiator by introducing slots or fractal structures. Two planar implantable an-
tennas with meandered strips have been presented in [KHT08] and [KRS06].
In [KKYP01] the rectangular stripline-fed meandered slot antenna is investi-
gated. A significant reduction of the size of the antenna is obtained. However,
the antenna features a very narrow band. A slight frequency shift caused by
the change of the dielectric properties of tissues lead to the impedance mis-
match of the antenna and thus a reduced antenna efficiency. In this chapter, a
slot antenna combined with a meandered structure is proposed to maintain a
certain bandwidth of the antenna and to reduce the antenna size at the same
time.
In the following sections, three different stripline-fed slot antennas are intro-
duced with the consideration of the challenges and problems mentioned in
section 4.1.2. The antennas are designed and characterized at the ISM band.
Among these antennas, the basic model is the antenna with equal-sized mean-
dered strips. The size of this antenna is then reduced by optimizing the shape
of the meandered strips and the ground, respectively, which results in two
miniaturized antennas. All three antennas are validated in the measurements
and discussed in the final section of this chapter.
4.2.1. Stripline-fed slot antenna with equally-sized

meandered strips (M3-1)
The implantable antenna (M3-1) comprises two planar layers with the same
substrate. The three structures of the antenna (top side, bottom side and radia-
tor between the two substrates) are shown in Figure 4.2. The rectangular slots
(Structure 2) at the top and the bottom are ground planes with the same struc-
ture. Structure 1 consists of meandered strips connected with the stripline of
50 line impedance.
Benefiting from the meandered radiator, the size of the antenna is reduced,
since the electrical length of the structure is increased. The slotted grounds at
the top and bottom side enable a large bandwidth [39]. After the investigation
72
(a) Top view (Structure 1) (b) Top view (Structure 2)
(c) Side view of the whole structure (d) Perspective view of the whole
structure
Figure 4.2.: Layout of the antenna M3-1 (Grey color denotes the metal).
of several designs of meandered strips with regard to the inductive behav-

ior and mutual coupling between meandered strips, three-armed meandered
strips show the best performance with respect to the impedance matching and
size reduction of the whole antenna. The radiation efficiency of the antenna
is increased by using a stripline feed to prevent the radiation along the feed
line (as discussed in the last section). The three dimensional (3D) view of the
double-layer slot antenna is shown in Figure 4.2 (d). The substrate Rogers
RT 6010 with the dielectric constant of 10.2, tan ı of 0.0023 and thickness of
1.27 mm is chosen for the antenna design and fabrication.
With regard to the meandered strips, the excitation mode is equivalent to that
of a stub antenna with the same length as a straight elongated meandered
strip. As can be deduced from the alignment of the surface current vectors in
the simulation, the antenna in Figure 4.2 excites a resonance of 43 wavelength
73
at 2.45 GHz. The effective electrical length of the meandered strips can be
calculated by
c0
Lsp D ksp D ksp p ; (4.1)
"r;eff f
where ksp is the fraction of the resonant wavelength (in this case is 34 ) and
f is the resonance frequency. The real electrical length of the meandered
strips is slightly different from the predicted one, since the parasitic effects
between the strips must be taken into account. The lumped circuit model
shown in Figure 4.3 aids the comprehension of the meandered antenna’s non-
ideal electric behavior [ESSK00]. In this model, each strip is represented by
the inductance of its magnetic field. The magnetic field lines rotate around
the metallic strips. At the edges of meandered strips charges accumulate to
induce a capacitive effect. Cascading each meandered segment results in the
final equivalent circuit. The open ended meandered strip is represented by
capacitances in order to model the fringing fields with the substrate. As the
structure is symmetrical to the xy-plane, Figure 4.3 depicts one half of the
equivalent circuit.
Figure 4.3.: Equivalent lumped circuit model of the antenna M3-1 with ca-
pacitances and inductances.
Since the antenna is implanted in the human muscle, in simulation, a muscle

phantom ("r =52.72, =1.73 S/m) emulating the dielectric property of real
muscle tissues is used (dielectric property as muscle at 2.45 GHz).
At first, the antenna parameters are optimized without considering the con-
ductivity of the phantom, since then the radiation pattern and efficiency can be
observed without signal attenuation caused by the lossy phantom. Therefore,
only the relative permittivity ("r =52.72) is applied in the simulation. After the
first optimization step without the loss of the phantom, this loss is then added,
to observe its impact on the designed antenna.
74
The resonance frequency of the slot antenna is related to the size of the slot (ls
and ws ), on which the size of whole antenna is dependent. Furthermore, by
optimizing the parameters of the meandered strips, the resonance frequency
can be also decreased. Therefore, the size of the slot can be reduced by in-
troducing the meandered strips while the resonance frequency is maintained.
Furthermore, increasing the length of the meandered strips le and decreasing
ld leads to a low resonance frequency, since the electrical length of meandered
strips is elongated. Moreover, the simulation results show that a large w2
leads to a low resonance frequency. Since the coupling within the meandered
strips is reduced by their increased distance (w2 ), the electrical length of the
strips is hence increased. The impedance matching becomes also better and
a strong resonance is obtained. However, the bandwidth is slightly reduced.
Therefore, a compromise has to be made between the resonance frequency
and bandwidth. The resonance frequency of the antenna shifts according to
the change of the dielectric properties of the muscle tissues (different individ-
uals). Therefore, a large bandwidth of the antenna enables the robustness of
the impedance matching (to be under -10 dB) of the implantable antenna at
the ISM band despite of a slight shift of the resonance frequency.
Observing the current distribution of the antenna in the simulation, the ground
size (wg , lg and the thickness of the slot edge) can be reduced without sig-
nificant modification to the antenna characteristics, since the strong surface
current is concentrated around the slot edge.
A semi-rigid coaxial cable (Farnell, RG402U) is used, which has an inner
diameter of 0.9 mm and an outer diameter of 3.58 mm, as the antenna feed.
The simulation result in Figure 4.4 (a) shows that the antenna fed by the
coaxial cable with a length of 5 cm has a similar impedance matching with
the one fed by a waveguide port in CST.
Now that the antenna characteristics have been determined, the conductivity
of muscle tissue (=1.73 S/m) is considered in the simulation to include the
losses caused by the phantom. From the result in Figure 4.4 (b), the change of
S11 and the resonance frequency is indistinguishable. Therefore, only a slight
optimization of some parameters is required to obtain a resonance frequency
at 2.45 GHz.
75
0 0
waveguide port σ= 0 S/m
−5 coaxial cable −5 σ= 1.73 S/m
−10 −10
S11 in dB
S11 in dB
−15 −15
−20 −20
−25 −25
−30 −30
−35 −35
1.4 1.6 1.8 2 2.2 2.4 2.6 2.8 3 3.2 1.4 1.6 1.8 2 2.2 2.4 2.6 2.8 3 3.2
(a) (b)
Figure 4.4.: Simulated influences of excitation port (a) and conductivity of

surrounding tissue (b) in reflection coefficient of the stripline-fed
slot antenna (M3-1).
After the optimization procedure of all parameters, the final optimized pa-
rameters are given in Table 4.3. The M3-1 antenna has an overall size of
14:5 12:5 2:54 mm3 . The final version of this antenna has a relative band-
width of 19.5%, ranging from 2.16 to 2.68 GHz in simulation.
Table 4.3.: Parameters of the antenna M3-1.

Parameters wg lg lgs ws ls wf
Value in mm 12.5 14.5 4 12 10.2 0.45
Parameters lf w1 w2 w3 ld le
Value in mm 4.325 0.5 1.17 0.65 1 9.5
Though a compact size of the M3-1 antenna has been achieved, an even
smaller antenna size is of significance for the IMDs. Based on this model,
two further miniaturization techniques are introduced in the following sec-
tions. It has already been discussed that the coupling between the meandered
strips and the ground is essential for the radiation of the E-fields. To optimize
this coupling, the slot in section 4.2.2 and the meandered strips in section
4.2.3 are modified. The goal is to lower the resonance frequency by optimiz-
76
ing the electrical coupling between meandered strips and slot. In this way,
a small-sized antenna can be achieved, while maintaining the resonance fre-
quency.
4.2.2. Stripline-fed slot antenna with modified slotted

ground (M3-2)
To miniaturize the antenna M3-1, the circumference of the slotted ground is
increased by introducing two strips on the upper edge of both ground layers,
which are symmetrical to the center of the antenna structure. In this way, the
electrical coupling between the meandered strips and ground is significantly
increased, which improves the radiation of the E-fields. In order to maintain a
symmetrical radiation characteristic, the dimensions of the strips are selected
to be identical and the position must be symmetrical to the center point of the
slot.
The new antenna is referred to as the M3-2 antenna and its structure is shown
in Figure 4.5. The configuration of Structure 1 and Structure 2 is the same as
M3-1 antenna (refer to 4.2 (c)).
(a) Structure 1 (b) Structure 2
Figure 4.5.: Layout of the antenna M3-2.
To optimize the additionally introduced strips, the influences of the parame-

ters (es and xs ) in terms of S11 is investigated. The distance xs is varied in
77
a way that the strips are approximately above the middle of one meandered
segment, i.e. at xs D3.8 mm, the strip is located above the first short upper
meandered strip, continuing with xs D2.7 mm, where the strip is above the
second long meandered strip. The results in Figure 4.6 show a decrease of
the resonance frequency with a smaller xs . Furthermore, regarding the im-
pact of the additional metal structures on the radiation pattern, the radiation
maximum is split in two maximums in the xz-plane in the case of a large
xs D3.8 mm in simulation (see in Figure 4.7). It is because the intensity of
the surface current in the center of the meandered structure decreases. With
decreasing xs , however, the surface current on the inner two meandered arms
and on the ground layer (see in Figure 4.8) increases, thus forcing the peaks
to converge.
0 0
−5 −5
−10 −10
−15 −15
S11 in dB
S11 in dB
−20 −20
xs=3.8 mm es=2 mm
−25 −25
xs=2.7 mm es=4 mm
−30 −30
xs=2.3 mm es=6 mm
−35 xs=1.3 mm −35 es=8 mm
−40 −40
2 2.2 2.4 2.6 2.8 3 2 2.2 2.4 2.6 2.8 3
(a) With varied xs (b) With varied es
Figure 4.6.: Simulated S11 of antenna M3-2 with varied xs and es .
Increasing another important parameter, es , the resonance frequency decreases

(see in Figure 4.6 (b)). However, the larger the es , the more skewed the
main beam direction from the normal radiation direction towards x direction
[You12].
The fully optimized parameters are given in Table 4.4. The antenna has a size
of 12.94 11.66 mm2 . In comparison to the formerly proposed M3 antenna,
a size reduction of 17:7 % has been achieved.
78
90° 90° 90°

120° 60° 120° 60° 120° 60°
150° 30° 150° 30° 150° 30°
0 −2 −4 0 −2 −4 0 −2 −4
±180° −6 0° ±180° −6 0° ±180° −6 0°
−150° −30° −150° −30° −150° −30°
−120° −60° −120° −60° −120° −60°

−90° −90° −90°
(a) xs =1.3 mm (b) xs =2.3 mm (c) xs =3.8 mm
Figure 4.7.: Simulated radiation pattern of antenna M3-2 with varied xs in

yz-plane (normalized in dB).
Figure 4.8.: Surface current distribution of the antenna M3-2 (meandered

strip with one ground on the bottom side visible): xs D 1:3 mm
and es D 6 mm.

Parameters wg lg ws ls w1 w2 w3 lgs
Values in mm 11.66 12.94 10.68 8.62 0.43 0.98 0.82 3.82
Parameters wf lf ld le xs es xd -
Values in mm 0.45 4.30 0.91 8.13 1.33 6.38 0.73 -
79
4.2.3. Stripline-fed slot antenna with modified

meandered strips (M3-3)
By applying the second miniaturization technique, the third antenna model

(M3-3) is developed. Here, the meandered strips were folded to extend the
electrical length of the meandered strips without enlarging its size and to im-
prove the coupling between meandered strips and slotted ground.
The influence of the additional meandered strips on the S11 -parameters are
shown in Figure 4.10. It is obvious that the enlarged electrical length of the
meandered strips is responsible for the decrease of the resonance frequency.
The parameter w4 contributes to the impedance matching and the resonance
frequency. A strong decrease of the resonance frequency is observed with
the increase of w4 , while the impedance matching is degraded. However,
modifying the width w5 , its impact on impedance matching and bandwidth is
not significant.
Figure 4.9.: Geometry of the M3-3 antenna.
As can be seen from the parameters in Table 4.5, the optimized M3-3 antenna
has an overall size of 12.9 10.15 mm2 , i.e. a size reduction of 27:8 % has
been achieved compared to the M3-1 antenna.
80
4.3. Verification of the stripline-fed double-layer slot antennas
0 0
−5 −5
−10 −10
S11 in dB
S11 in dB
−15 −15
w4=1 mm w5=0.1 mm
−20 w4=2 mm −20 w5=0.2 mm
w4=3 mm w5=0.3 mm
−25 −25
w4=4 mm w5=0.4 mm
−30 −30
2 2.2 2.4 2.6 2.8 3 2 2.2 2.4 2.6 2.8 3
(a) With varied w4 (b) With varied w5
Figure 4.10.: Simulated S11 of antenna M3-3 with varied w4 and w5 .

Parameters wg lg ws ls w1 w2 w3 lgs
Value in mm 10.15 12.9 9.45 8.4 0.32 1.68 0.77 3.82
Parameters wf lf ld le1 le2 w4 w5 -
Value in mm 0.45 4.06 0.79 7.35 8.27 1.37 0.21 -
4.3. Verification of the stripline-fed

double-layer slot antennas
In this section, the characteristics of the three stripline-fed double-layer slot
antennas will be verified by measurements. For the verification purpose, a so-
lution of water and sugar is chosen to emulate the muscle tissue. The sugar-
water solution is very suitable to emulate the muscle tissues (high relative
permittivity) within a narrow bandwidth. This is because water and sugar
have a dielectric constant of "r of 77.2 and 3 at 2.45 GHz [TRVdVG95], re-
spectively. The relative permittivity and conductivity of the muscle tissues
can be exactly obtained by mixing sugar and water with a certain weigh ratio.
However, the sugar-water mixture cannot emulate each tissue with arbitrary
ratio. To emulate tissues with low relative permittivity (e.g. bone), a high
81
weight ratio (> 1:1) of sugar is required, which is not completely soluble in
water at room temperature.
To approximate the dielectric property of muscle tissues at 2.45 GHz, the
sugar content is finally experimentally determined (40.9%). Table 4.6 depicts
the weight ratio of the sugar-water solution for the tissue-equivalent phan-
tom. As can be seen in Figure 4.11, the approximation of the sugar-water
solution to muscle is sufficiently precise [GGC96, GLG96a]. Furthermore,
an attenuation constant of 4.89 dB/cm at 2.45 GHz can be determined.
Table 4.6.: Recipe for the sugar-water solution to emulate muscle tissues.
Ingredient weight ratio in %
Water 59.1
Sugar 40.9
70 10
60
8
r
Relative permittivity ε
50
40 6
30 4
20
2
10 muscle muscle
sugar−water solution sugar−water solution
0 0
1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10
(a) Relative permittivity (b) Conductivity
Figure 4.11.: Dielectric properties of the sugar-water solution vs. muscle tis-
sue.
The antennas have been fabricated by etching the two substrates separately
and then bonding them with glue. To minimize the influence caused by the
additional dielectric (glue) between the substrates, the effect of the glue is
also examined in the simulation. Figure 4.12 shows the simulation results
82
of the M3-1 antenna in the case of a 0.2 mm thick adhesive layer placed be-
tween the two substrates. The conductivity of the muscle tissues is taken into
account in the simulation. The blue curve illustrates the S11 of the antenna
with a homogeneous distribution of the glue between the two substrates. Dis-
tributing the glue only along the borderline of the substrate with a width of
2 mm, the S11 (red color) is strongly degraded due to the presence of the air
gap between two substrates. The green curve shows S11 of the antenna, while
the glue is distributed only near the boundary of the rectangular substrate. In
this case, its influence on S11 is very slight. Thus, the last method is chosen
for gluing the substrates together.
The antennas are fed by a semi-rigid coaxial cable, which has an inner diam-
eter of 0.9 mm and an outer diameter of 3.58 mm, respectively. Figure 4.13
shows the prototypes of stripline-fed slot antennas. The size reduction of the
M3-2 and M3-3 can be clearly seen from the photos. In Figure 4.13 (a), the
substrates in the upper row have the Structure 1 (refer to Figure 4.2) on one
side and are free of copper on the other side, while the substrates in the lower
row are soldered with both Structure 1 and Structure 2. The fabricated proto-
types with semi-rigid cables are shown in Figure 4.13 (b). The antennas are
immersed in the sugar-water solution to measure the impedance matching of
the antennas.
Figure 4.14 provides the measured S11 compared with the simulated results.
The measured S11 of the three antennas agrees very well with the simulated
results. The resonance frequency of the measured S11 of the M3-1 antenna
is the same as the simulated one. However, a strong shift of the resonance
frequency of the measured S11 of the M3-2 antenna is identified. The S11 in
the whole ISM band from 2.4 to 2.5 GHz is not sufficiently good (under -10
dB), though the simulated antenna is matched in this frequency range. After
fabricating four prototypes, M3-2 indicates very high sensitivity of the perfor-
mance to the manufacturing tolerance. This is due to the displacement of the
two strips placed on the upper edge of both ground layers, which influences
the S11 and the radiation pattern. Moreover, the shift of the resonance fre-
quency of the antenna M3-3 is acceptable, since the antenna is still matched
from 2.4 to 2.5 GHz.
83
−5
−10
S11 in dB
−15
−20
−25
−30
−35
1.6 1.8 2 2.2 2.4 2.6 2.8 3 3.2
Frequency f in GHz
Figure 4.12.: Simulated S11 of the antenna M3-1 with different glue distri-
butions between substrates: homogeneous distribution between
the substrates (blue), along the borderline of the substrates with
a width of 2 mm (red), near the boundary of the substrate (green)
and without glue (black).
(a) Etched substrates (b) Fabricated antenna prototypes
Figure 4.13.: Photos of the prototypes of the stripline-fed slot antennas.
In the next step, the radiation patterns of the antennas are investigated. It is
assumed that the distance between the implantable antenna and the external
receiver is more than 50 mm. The radiation pattern of the antennas are charac-
terized at a distance of 70 mm (due to the geometrical limitation for the 360ı
rotation), which corresponds to the far-field of the antennas (Rff = 18 mm).
The experimental setup for antenna pattern measurements is illustrated in Fig-
ure 4.15. For the measurements, the coaxial cable is bent (see Figure 4.13) to
84
0 0
−5 −5
−10 −10
S11 in dB
S11 in dB
−15 −15
−20 −20
−25 −25
−30 simulated −30 simulated

measured measured
−35 −35
0 0.5 1 1.5 2 2.5 3 3.5 4 0 0.5 1 1.5 2 2.5 3 3.5 4
(a) M3-1 antenna (b) M3-2 antenna
−5
−10
S11 in dB
−15
−20
−25
−30 simulated
measured
−35
0 0.5 1 1.5 2 2.5 3 3.5 4
Frequency f in GHz
(c) M3-3 antenna
Figure 4.14.: Simulated and measured S11 of the three stripline-fed slot an-
tennas.
position the antenna on the platform with motors. The antennas are suspended
vertically into sugar-water solution. The used cables are wrapped with tape
and polythene foil to waterproof them. The dimensions of the sugar-water
container are 406010 cm3 . Based on the respective antenna configura-
tions (see Figure 4.2, 4.5 and 4.9), the E-plane corresponds to the yz-plane,
whereas the H-plane is located in the xz-plane.
It can be observed in the measurement that the measured results are strongly
influenced by the accuracy of the location of the probe and AUT. The mis-
85
(a) Antenna alignment for E-plane
Figure 4.15.: Measurment setup of radiation pattern for implantable antennas
alignment of the probe and AUT can result in a shift of phase and amplitude,
thus further distorting the radiation pattern.
The measured radiation patterns in the E-plane and H-plane of three antennas
are shown in Figure 4.16, for co- and cross-polarization. It can be seen that
the main beam of the radiation pattern is normal to the surface of the antenna
structure (in ˙z directions).
A directive pattern of three antennas in the E-plane is observed. The radiation

at 0ı and 180ı (˙x axis) is very low. No sidelobes exist in the E-plane. The
radiation pattern of the antenna M3-2 is not exactly symmetric. A difference
of 8 dB between the two main beam directions at 90ı and -90ı can be seen.
Moreover, in the E-plane, a cross-polarization suppression of more than 20 dB
in main beam direction of antenna M3-1 and M3-3 is achieved.
In the H-plane, omni-directional patterns of all the three antennas can be iden-
tified. A cross-polarization suppression of 10 dB can be estimated. Different
main lobes at multiples of 45ı can be observed in cross-polarization. This is
because of the surface current existing at the corners of the slotted ground,
which causes radiation in cross-polarization in H-plane (xz-plane).
86
0° 0°
30° −30° 30° −30°
60° −60° 60° −60°
0 −10
0 −10 −20 −30 −40
90° −20 −30 −90° 90° −90°
−40
120° −120° 120° −120°

Co−pol Co−pol
X−pol X−pol
150° −150° 150° −150°
±180° ±180°
(a) M3-1, E-plane (b) M3-1, H-plane

0° 0°
30° −30° 30° −30°
60° −60° 60° −60°
0 −10 0 −10
−20 −30 −40 −20 −30 −40
90° −90° 90° −90°
120° −120° 120° −120°

Co−pol Co−pol
X−pol X−pol
150° −150° 150° −150°
±180° ±180°
(c) M3-2, E-plane (d) M3-2, H-plane

0° 0°
30° −30° 30° −30°
60° −60° 60° −60°
0 −10
0 −10 −20 −30 −40
90° −20 −30 −90° 90° −90°
−40
120° −120° 120° −120°

Co−pol Co−pol
X−pol X−pol
150° −150° 150° −150°
±180° ±180°
(e) M3-3, E-plane (f) M3-3, H-plane
Figure 4.16.: Measured radiation pattern of the implantable antennas at a dis-

tance of 70 mm in the E-plane and H-plane at 2.45 GHz (nor-
malized in dB).
87
4.4. Summary
The design concept of the M3-1 antenna relies on a stripline-feed, meandered

strips and a slotted ground to achieve high radiation efficiency, large band-
width (more than 10%). The radiation efficiency of the antenna is increased
by using a stripline-feed, since no surface- and fringing waves are present
along the stripline surrounded by a lossy medium and hence additional loss
along the feed line is prevented. The meandered strips contribute to the re-
duction of the antenna size, while the slotted ground improves the antenna
bandwidth.
Based on this antenna, the miniaturization of the antenna size to achieve the
same resonance frequency was demonstrated by the optimization of the an-
tenna structures. It has been shown that by introducing strips on the ground
planes (M3-2 antenna) or extending the meandered strips (M3-3 antenna), the
stripline-fed slot antenna can be further miniaturized compared to the size of
the M3-1 antenna. The size of the M3-2 antenna was reduced at the expense
of a reduced bandwidth and radiation efficiency. The M3-3 antenna with the
modified meandered strips is successfully optimized. By the optimization
of the antenna structures with regard to the coupling effect and the careful
regulation of the current distribution, the antenna size of M3-3 is strongly de-
creased, while the bandwidth and high radiation efficiency of the antenna are
still maintained.
A short overview of the three implantable antennas is given in Table 4.7 with
respect to bandwidth (BW), relative bandwidth (rel. BW), peak gain (Gp )
( D 0 S/m), radiation efficiency () in simulations and measurements. Ad-
ditionally, for comparison, characteristics of two planar implantable antennas
optimized for ISM band in [AKS13] and [SKRC 11] are given. It can be seen
that the antenna M3-3 features the smallest dimensions and the largest band-
width in both simulation and measurement. The antenna M3-2 has the highest
gain in the simulation, however, has the narrowest bandwidth and lowest ra-
diation efficiency with respect to the simulation results. In conclusion, the
M3-3 antenna shows the best radiation properties.
The M3-1 and M3-3 antennas fulfilled the requirements mentioned at the be-
ginning of this chapter for IMDs operated at the ISM band. The performance
88
4.4. Summary
Table 4.7.: Comparison of the M3-1, M3-2, M3-3, reference antenna 1 in

[AKS13] and reference antenna 2 in [SKRC 11].
Antenna M3-1 M3-2 M3-3 ref. 1 ref. 2
Size (mm2 ) 14.512.5 12.911.6 12.910.1 1111 24.625.9
Simulated 2.16-2.68 2.36-2.74 2.27-2.89 2.34-2.52 2.30-2.57
BW (GHz) = 0.52 = 0.37 = 0.625 = 0.18 = 0.27
Rel. BW 19.5% 17.7% 32.15% 7.4% 11%
Simulated 98.9% 84.6% 94.2% - -
Measured 2.25-2.6 2.53-2.78 2.36-2.87 - 2.20-2.55
BW (GHz) = 0.35 = 0.25 = 0.51 - = 0.35
Rel. BW 14.4% 9.4% 19.5% - 14.2%
of these implantable antennas is successfully verified by measurements. The

characterized radiation patterns are omni-directional in the H-plane and direc-
tional in the E-plane with high directivity, which enables the establishment
of a robust wireless link between the IMDs and external medical devices.
Moreover, the developed implantable antennas with compact size will lead
to small-sized IMDs, which will significantly improve the applicability of
IMDs.
This closes the chapter on the design of the implantable antennas for wire-
less communication of IMDs. Regarding the second medical application (i.e.
medical diagnosis), the design details of the on-body matched antennas can
be found in the next chapter.
89
5. On-body matched antennas for
medical diagnosis
In this chapter, we move on from the implantable antennas to on-body matched

antennas. These on-body matched antennas address the drawbacks of off-
body antennas (as discussed in chapter 2) for microwave medical diagnosis
purposes. In contrast with implantable antennas discussed in the previous
chapter, the antennas for medical diagnosis are required to feature a very
wide bandwidth to allow a high-resolution imaging based on radar imaging.
In microwave tomography, traditionally mono-frequency procedure is applied
and narrow band antennas can be used. However, a multifrequency (or wide-
band) processing improves the inversion results hence the imaging quality
[GMZC 10]. Therefore, the goal of this chapter is to develop wideband and
multi-band antennas.
Different on-body matched antennas operating in two different frequency

ranges for medical diagnosis are presented. First, the challenges and design
principles of the on-body matched antennas are described. In the follow-
ing sections, different differentially-fed wideband slot antennas with differ-
ent slot-shapes for the miniaturization of the antenna size are discussed and
compared. These antennas are characterized from 1 to 7 GHz. To enhance
the sensitivity of the system with respect to weak reflections from targets in
human tissues, a lower operational frequency is required. Therefore, a dual-
band slotted Bowie antenna (characterized from 0.5 to 0.7 GHz and 1.3 to
4 GHz) and a folded Bowtie antenna (characterized from 0.5 to 2 GHz) are
designed with the focus on miniaturization and stable radiation properties.
91
5. On-body matched antennas for medical diagnosis
5.1. Motivation and design challenges of

on-body matched antennas
Microwave medical diagnosis via imaging for the identification of e.g. tu-
mors is based on the concept of observing the reflected signal from the tar-
get (in some methods, the transmission is also utilized). The reflections are
caused by the contrast of the dielectric properties between the tumor and the
surrounding healthy tissues. For the detection of these reflections, high ra-
diation penetration of the signals from the antenna into the human body is
desirable to achieve a high SNR, since the reflections are very weak due to
the high signal attenuation in tissues (see chapter 2). For such applications,
the performance of the antennas is the governing factor of the overall sys-
tem performance. Matching the antennas on the human body has the purpose
of reducing the strong reflection from the air-skin interface (drawback of the
off-body antennas discussed in chapter 1) and to allow more energy to pen-
etrate into the human body to obtain a stronger signal for image processing.
Another advantage of matching the antenna to the human body is that the size
can be rescaled according to the relative permittivity of the human skin, since
the high relative permittivity leads to a wavelength shorter than in free space.
However, the antenna design for medical diagnosis differs from that for reg-
ular antennas for free-space operation due to the presence of the complex
human body, which can be modeled as a lossy medium. Therefore several
challenges arise in the design stage. The general design challenges of body-
matched antennas have been mentioned in the introduction. In detail, the
following challenges and requirements are associated with the design of on-
body matched wideband antennas for medical diagnosis:
Operational frequency The operational frequency of the antenna should

preferably include low frequencies (around 1 GHz) considering the high-
contrast characteristic and low signal attenuation in human tissues.
Operational bandwidth The operational bandwidth must be as large

as possible to achieve a high range resolution, since the range resolution
is inversely proportional to the operational bandwidth. That means the
larger the bandwidth, the finer the range resolution.
92
5.2. Design procedure
Radiation property The radiation property must not change signifi-

cantly in the whole frequency band. The near-field pattern must be
investigated. A high radiation efficiency of the antennas is desirable.
Front-to-back ratio A high front-to-back ratio of the antenna radiation

and low side- and back-radiation are desired to minimize the antenna
coupling and the interference to other microwave sub-systems. This
also allows a high portion of energy to be coupled into the human body.
Reflection between air and human skin Since the antenna should be
optimized for operation directly on the skin, a good contact between the
antenna and skin is needed. The sensitivity of the antenna performance
regarding imperfect contact must be investigated. In the case of a non-
flat surface such as the skin on head with hairs, matching materials
should be used.
Antenna size For medical diagnosis, an antenna array with a num-

ber of antenna elements both for transmitting and receiving is typically
applied. The small-sized antenna allows the construction of an array
with a large number of elements within limited area/space for gather-
ing more useful reflections. In this way, a microwave image with high
quality (e.g. signal-to-clutter ratio (SCR)) can be obtained. More de-
tails will be discussed in chapter 6.
Verification by measurement The variation of the permittivity of tis-

sues results in a significant change of performance of the antenna.
Hence a similar scenario and boundary conditions in the simulation
and measurement are required. The antenna must be water-tight for the
measurement of the radiation pattern. The measurements of antennas
(i.e. S11 ) are performed in tissue-simulating liquid as well as on the
human skin.
5.2. Design procedure

Since the on-body matched antennas are in direct contact with the human
body, the behavior of the antennas depends on the dielectric properties and
93
geometry of the human body. Therefore, the antennas should be optimized

together with the human body model in the design procedure. However, a
simulation of antennas with the human body model results in high number
of mesh cells and a simulation time of a few days is expected. Due to the
limitation of the memory capacity and computational capability of the com-
puter, it is not recommended to optimize the antenna together with a whole
complex human model with a high space resolution. The duration of antenna
optimization will be extremely long. Therefore, an efficient optimization pro-
cedure for the on-body matched antennas is required.
The optimization procedure of the antenna in the configuration shown in Fig-
ure 5.1 is performed in CST Microwave studio with the following steps:
Step 1: The antenna is constructed together with a free space region on

one side. On the other side of the antenna is a tissue-simulating phan-
tom as shown in Figure 5.1 (a). The phantom has the averaged relative
permittivity of skin and fat [GLG96a]. The frequency dispersion and
conductivity of the phantom material ("r = 20 or 35 depending on the
frequency bands, =0 S/m) is not considered. The open boundary con-
dition is used in the simulation so that the reflection on the boundary
is ignored. Therefore, the impedance matching, efficiency and gain of
the antenna can be investigated. Since many iterations are required, the
considerations above can significantly reduce the number of mesh cells
thus reducing the computing time per iteration.
Step 2: The antenna is simulated with the free space region on one side
and a realistic tissue-simulating phantom including its frequency dis-
persion and conductivity. A parameter tuning is performed with the ex-
perience from the parameter study in step 1. The impedance matching
and radiation pattern of the antenna are investigated. The parameters
of the antenna are then updated.
Step 3: A further study (i.e. SAR) of the antenna together with a com-
plicated multilayer human phantom shown in Figure 5.1 (b) is per-
formed. In this configuration, the realistic permittivities of different
tissues and the multi-reflections of different layers are taken into ac-
count.
94
5.3. Differentially-fed slot antennas for the operational band from 1 to 7 GHz
(a) With tissue-simulating phantom (b) With multilayer human phantom
Figure 5.1.: Simulation scenarios of the on-body matched antenna with

tissue-simulating phantom and multilayer human phantom.
The final optimized parameters can be imported into the models in step 1 and
2 to obtain different characteristics of the antenna (gain, radiation pattern,
etc.). After the optimization steps, the antennas are fabricated for the verifi-
cation by measurement with respect to the impedance matching and radiation
pattern in the tissue-simulating liquid. The tissue-simulating liquid emulates
the dielectric properties of the human tissues. A slight modification of the
antenna dimensions can be made after the analysis of the measured results.
After introducing of the challenges and the design principle of the on-body
matched wideband antennas, different concepts of developing these antennas
are described in the following sections. The antennas are characterized in
different operational frequency bands with the focus on the miniaturization
of the antennas.
5.3. Differentially-fed slot antennas for the

operational band from 1 to 7 GHz
Taking the challenges and requirements in section 5.1 into account, a planar
antenna structure is applied for the easy and low-cost fabrication of the anten-
nas. Moreover, the planar antenna can be also easily contacted to the nearly
planar surface of human body. Regarding the radiation pattern, most on-body
95
antennas suffer from the problem of the beam width and main beam direction
varying over frequency as well as over the changing permittivity of the hu-
man tissues. To achieve a stable main beam direction in the whole frequency
range, symmetrical radiating element with differential feed is adopted. The
goal is to obtain a sufficiently large bandwidth and high radiation efficiency
as well as a good penetration into human tissues.
In this section, the design of UWB slot antennas based on the concept in
[ABWZ09, Ada10] using a new design procedure and miniaturization tech-
nique, is investigated. The antenna is characterized from 1 to 7 GHz (lower
frequencies refer from 1 to 3 GHz, while higher frequencies are from 3 GHz
in the later discussion). Due to the large wavelength at 1 GHz, three different
slots for the miniaturization of the antenna size are investigated and verified
both in simulation and measurement: double elliptical slot, sector-like slot
and stepped slot, which are discussed in succession in the following sections.
Then, an optimized feed network is introduced to reduce the overall size of
the antennas.
5.3.1. Double-elliptical slot antenna

To maintain a symmetrical structure, two elliptically shaped monopoles on
the top and a double-elliptical slotted ground at the bottom of the antenna are
designed. The double-elliptical shape is used to miniaturize the antenna.
In Figure 5.2, the layout of the double-elliptical slot antenna fed by microstrip
lines is shown. The antenna consists of two elliptically shaped monopoles on
the top side of the substrate surrounded by a double-elliptically slotted ground
plane (the inverse area of the double elliptical slot) at the bottom side. In the
preliminary design, the elliptically shaped monopoles are differentially fed by
two tapered microstrip lines with 50 line impedance at the ports. The input
signals at the two ports have a 180ı phase difference. Therefore, the radiated
co-polarized E-fields related to both monopoles are in phase [31]. Instead of
the differential feed with microstrip lines, a differentially-fed network (refer
to Figure 5.4) used to feed the two monopoles will be introduced later.
The symmetrical arrangement of the two monopoles each within one slot
helps to keep the current distribution in the radiation zone symmetrical around
the center of the antenna. Such a current distribution results in a symmetri-
96
(a) Elliptically shaped monopoles (top (b) Double elliptical slot (bottom view)
view)
Figure 5.2.: The layout of the double-elleptical slot antenna.
cal radiation pattern, which can be optimized to be very stable over the op-
erational frequency range. In this way the phase center (whose position is
frequency independent) of the radiation is located exactly in between the two
monopoles at the center of the structure due to the symmetry of the current
distribution [ABWZ09]. This indicates that the radiating impulse of this an-
tenna features a low distortion and a high fidelity.
The lowest operational frequency of the slotted wideband antenna can gener-
ally be written as
c0
fL D p (5.1)
Cs "r;eff
where Cs is the circumference of the slot and "r;eff is the effective relative
permittivity of the substrate. Therefore, to achieve a low operation frequency,
either a large circumference of the slot or a high-permittivity substrate should
be used.
Compared to a single elliptical slot with the same dimensions, the double-
elliptical slot on the ground plane with the same length and width increases
the circumference, thus lowering the operation frequency without changing
the overall size of the antenna. As a result, the lower operational frequency is
further decreased and the overall bandwidth is increased.
97
On the other hand, a high-permittivity substrate is desirable for the miniatur-

ization of the antenna. However, the relative permittivity of the commercially
available low-profile substrates is limited up to around 10 due to the high
fabrication cost. To increase the relative permittivity of the substrate (larger
than 15), ceramic body combining using sintering technology is proposed in
[dCHSC 11]. The size of the antenna can be rescaled due to the high relative
permittivity and therefore be miniaturized. In this work, the substrate Rogers
RT 6010 (PTFE/Ceramic Laminates) with the dielectric constant of 10.2 and
thickness of 1.27 mm is chosen for the antenna design and the fabrication.
This substrate does not feature an extremely high relative permittivity com-
pared to ceramic substrates. However, the antenna structure can easily be
etched. The miniaturization of the antenna is based on the optimization of
the antenna structure including the feed network with regard to the current
distribution of the antenna.
Since the antenna is matched to the human body, the high relative permittivity
of the tissues contribute to the high effective permittivity "r;eff of the antenna.
In simulation, a phantom emulating the human body is placed on the ground
side as in Figure 5.3 (a), leaving free space on the top side of the substrate.
The phantom ("r D 20, D 0 S/m) has the average relative permittivity of
skin and fat at 4 GHz (center frequency of the operational band from 1 to
7 GHz) [GLG96a] and the dispersion of the phantom material is not consid-
ered for the characterization of the antenna. This configuration results in a
shorter wavelength within the phantom (as compared to in free space) there-
fore allowing for a smaller-sized antenna. The direction of the radiation is
perpendicular to the ground plane (+z direction) and goes into the phantom.
By matching the antenna to the phantom with a high relative permittivity, a
directive antenna pattern with high front-to-back ratio can be achieved.
In the optimization procedure, it can be found that the impedance matching at
the higher frequencies can be easily achieved. However, an unstable radiation
pattern occurs because the wavelength at the higher frequencies is 3 to 4 times
smaller than that at the lower frequencies. The current distribution at the
higher frequencies changes and a high antenna mode of operation, with side-
/grating lobes can be observed. The main pattern in yz-plane is split into
different maxima (sidelobes and grating lobes). This can be minimized by
98
reducing the size of the slot and the distance of the two elliptically shaped
monopoles in the y direction. Overall, the slot width is increased only in
the x direction so that a small distance between two monopoles and the slot
edges can be maintained. In this way, the grating lobes are suppressed at the
higher frequencies and a desirable radiation pattern can be realized for a wide
frequency range.
(a) Without feed network (b) With feed network
Figure 5.3.: The arrangement of the double-elleptical slot antenna and phan-
tom.
In practice, two differential signals must be fed to the two ports of the an-
tenna. Two RF cables for two ports must also be exactly the same to maintain
their 180ı phase difference. To eliminate this feeding issue, an additional
differential feed network is desirable to be connected directly to the antenna
element. The differential feed network [ABWZ09] is a wideband divider with
differential outputs applied to excite the differential signals. It is made from
the same substrate as the antenna to minimize the attenuation caused by the
interconnection between the feed network and the antenna. The feed network
with a size of 52261.27 mm3 is shown in Figure 5.4 (a) and is optimized
for the operation from 1 to 10 GHz.
To connect the feed network with the antenna element, the ground plane of
the antenna element is moved to the top side as shown in Figure 5.4 (b). The
connection of the feed network and the antenna is in such a way that the ports
2 and 3 in Figure 5.4 (a) are connected to the respective port of the same label
in Figure 5.4 (b). For the differential feed setup, the microstrip line at port 2
is connected to the monopole while the ground of port 3 is connected to the
99
(a) Feed network (top and bottom sides superimposed) (b) Radiator element on top side
Figure 5.4.: The layout of the double elliptical slot antenna and differential
feed network (condcutors are grey; unit: mm).
monopole. The feed signal is channeled into the feed network from port 1.
Figure 5.5 illustrates the perspective view of the slot antenna together with
the feed network.
The arrangement of the antenna element and feed network is illustrated in
Figure 5.3 (b). With this configuration, the signals are radiated perpendicu-
larly to the surface of the underside of the antenna (without the ground plane)
into human body. The overall size of the antenna is 263552 mm3 .
(a) Left side view (b) Right side view
Figure 5.5.: The perspective view of double-elliptical slot antenna with feed
network (circular points indicate the electrical connections be-
tween feed network and radiating element).
100
Characteristics of the double-elliptical slot antenna
In the following section, the characteristics of the optimized antenna model

in simulation will be given to demonstrate the achievements of the discussed
design principles. The simulation results will be then verified by measure-
ment.
The simulated input reflection coefficient (S11 and S22 ) at the differential
feed ports of the antenna without the feed network (see Figure 5.2) are the
same due to the symmetry of the antenna structure. The result (see Figure 5.6
(a)) shows that a very good impedance matching (-10 dB) is achieved in the
frequency range from 1.2 to 9 GHz. The curve of S11 without strong reso-
nance is observed due to the smooth impedance matching between microstrip
feed line and elliptical monopoles.
In Figure 5.6 (b), the impedance matching of the antenna with the feed net-
work (as shown in Figure 5.4) from 1.1 to 9 GHz can be observed. However,
the feed network introduces oscillations/resonances due to the feed structure
and interconnection with the antenna element. The measurement is performed
in the PEG-water solution as mentioned in Chapter 3. A slight difference
between the simulated and measured curves, especially around 2 GHz, is ob-
served. It is because the required interpolation of material property in CST
leads to slightly different dielectric property of the PEG-water solution com-
pared to the measured one. Moreover, the dispersion of the PEG-water solu-
tion at the open boundary in simulation causes also different results. All the
body-matched antennas are associated with these two reasons causing slightly
different results between simulation and measurement, which will not be re-
peated in the following chapters. However, the simulated and measured result
in PEG-water solution agree well in terms of nulls. At the higher frequencies,
the S11 is improved to be better than -10 dB. The measured results on the
skin and in PEG-water solution are consistent with each other, which shows
the applicability of the antenna in direct contact with the skin for medical
diagnosis. In the later paragraphs, the differentially-fed slot antennas are re-
ferred to the antennas with feed network, if the antennas without feed network
are not mentioned.
Since the dielectric properties of human tissues vary across individuals, the
sensitivity of the on-body matched antenna to a dielectric property variation is
101
0 0
−5 −5
−10 −10
S11 in dB
S11 in dB
−15 −15
−20 −20
simulated in PEG
−25 −25 measured in PEG
measured on skin
−30 −30
1 2 3 4 5 6 7 8 9 1 2 3 4 5 6 7 8 9
(a) Without feed network (simulated) (b) With feed network
Figure 5.6.: Simulated and measured S11 of double-elliptical slot antenna

without (fed by microstrip lines) and with feed network.
investigated in the simulation. The variation of relative permittivity of phan-

tom from 20 to 30 are within the normal range of averaged tissue (skin and
fat). The results shown in Figure 5.7 confirm that the impedance matching
of the proposed antenna is very stable over a large range (from 15 to 30) of
relative permittivity of human tissues. This is because the radiating element
is located on the top side of the antenna between free space and substrate,
hence the change of the relative permittivity of the phantom does not cause
a significant variation of the current distribution of the radiating elements es-
pecially at the higher frequencies. However, it also can be observed that the
S11 of the antenna input port at around 2 GHz goes slightly up to -8 dB with
high relative permittivity of the phantom.
To investigate the power that penetrates into the phantom (human body), the
penetration efficiency p is defined as
Pbody
p D ; (5.2)
Pbody C Pair
where Pbody and Pair are the total radiated power into the human body and in
free space region (refer to 5.3 (b)), respectively.
The ratio of the radiation energy into the human body can be also described by
the front-to-back ratio. The front-to-back ratio indicates the ratio of radiated
102
−5
−10
in dB
−15
ε =15
11
r
S
−20 ε =20
r
ε =25
−25 r
ε =30
r
−30
1 2 3 4 5 6 7 8 9
Frequency f in GHz
Figure 5.7.: Simulated S11 of the double-elliptical slot antenna with varied
relative permittivity "r of phantom ( =0 S/m).
power between the front and rear of a directional antenna (the front side of
the antenna points to the human body), which is defined as
Pbody
F=B D : (5.3)
Pair
It can be seen from (5.3) that a high penetration efficiency yields a high front-
to-back ratio.
The penetration efficiency of the double-elliptical slot antenna with varying
relative permittivity of the phantom is investigated. The results (see Figure
5.8) show that, in case of "r D 1, a penetration efficiency of only about 50%
is obtained. Increasing the relative permittivity of the phantom to 10, p is
mostly above 80%. However, p increases slightly with the increasing relative
permittivity (from 10 to 40) of the phantom. The reason is, as discussed in the
last paragraph, that the radiator is placed between free space region and the
substrate with "r;sub D 10:2 and thus the dielectric contrast of materials placed
on the two sides of radiator cannot be changed significantly by the relative
permittivity of phantom. At 7 GHz, the radiation into free space increases at
the higher "r (e.g. 30). On the other hand, p for a relative permittivity of
the phantom from 10 to 40 is relatively high (> 0.7) in the whole operating
frequency range. It can be concluded that the high p is achieved due to
103
the configuration of the antenna in direct contact with the human body. The
relative permittivity of tissues (larger than 20) will not affect strongly p of
the proposed slot antenna.
With "r D 20, 80% of the radiated energy (refer to Figure 5.9) is coupled into
the phantom, which indicates a front-to-back ratio of larger than 6 dB. From 3
to 6 GHz, the antenna has a p of about 0.9 (F/B >9.5 dB). It can be therefore
noted that the double-elliptical slot antenna in direct contact with human body
features a high penetration efficiency and front-to-back ratio.
40 1
Relative permittivity of phantom εr
35
0.8
30
25 0.6
20
0.4
15
10
0.2
5
1 ηp 0
1 2 3 4 5 6 7 8 9
Frequency f in GHz
Figure 5.8.: Simulated penetration efficiency p (in linear scale) of the

double-elliptical slot antenna over frequency with varied relative
permittivity "r of phantom ( =0 S/m).
Although the radiation gain (far-field) cannot be achieved in the real scenario
due to the presence of lossy human tissues, this is a measure of the designed
antenna regarding its radiation efficiency and directivity. The simulated max-
imum gain of the antenna in the main beam direction with the phantom is
shown in Figure 5.9. The results show that the antenna has a relatively high
and very constant gain (> 8 dBi) from 2 to 8 GHz. A smaller gain is observed
at the lower frequencies. It is due to the antenna being electrically small
in terms of wavelength at the lower frequencies, which results in a broad
beamwidth of the radiation pattern without a high directivity. On the other
104
hand, the radiation above 7 GHz is not applicable, since the reflection from
the targets is too weak to be detected due to the very high signal attenuation.
Yet for the operational band from 1 to 7 GHz, the antenna features a very high
radiation gain compared to conventional wideband antennas in free space.
15 1
0.9
0.8
Penetration efficiency ηp
0.7
10
0.6
Gain in dBi
0.5
0.4
5
0.3
0.2
Gain 0.1
Penetration efficiency
0 0
1 2 3 4 5 6 7 8 9
Frequency f in GHz
Figure 5.9.: Simulated maximum gain and penetration efficiency of the

double-elliptical slot antenna with the phantom ("r D 20, D 0
S/m).
Since the antenna is in direct contact with the human body, it is of significance
to characterize the near-field pattern. However, the standard free space an-
tenna measurement cannot be used and hence the measurement is performed
in the PEG-water solution with a similar dielectric property as the human
body. For the measurement, the antenna has to be watertight (shown in Fig-
ure 3.8) so that the free space region on the top side will not be affected by
the tissue-simulating liquid.
The simulated and measured near-field (Rff =63 mm at 1 GHz, r= 40 mm) pat-
tern in the E-plane and H-plane in PEG-water solution at different frequencies
are shown in Figure 5.10 and 5.11, respectively. This distance corresponds to
the average thickness of fat and muscle tissues of the human abdomen. The
results are limited to 4 GHz due to a high signal attenuation in PEG-water
solution at the higher frequencies in the measurement and the VNA used for
105
measurement has a limited dynamic range. From the results, it can be ob-
served that the differential feed and the symmetry of the antenna geometry
provide a good symmetrical near-field pattern of the antenna. The radiation
direction of the main beam is found to be very stable over a broad frequency
range due to the phase center being in the middle of the antenna.
Slight differences (e.g. directivity and side lobes) between simulated and
measured results are caused by different boundary condition. To reduce the
simulation time, the calculated volume is strongly reduced using open bound-
ary (reflections at the boundary are not considered) in CST compared to the
measurement scenario. Moreover, an automatic interpolation of the frequency-
dependent permittivity of the medium must be tolerated in the simulation tool
to enable the convergence of the solver. However, the similar run of the curves
between simulated and measured results is used for the verification purpose.
In the E-plane (yz-plane), a narrow beam is identified with increasing fre-
quency. At 3 GHz and 4 GHz, the pattern features sidelobes as expected in
the analysis mentioned in the design principle of this antenna. A wider pat-
tern is observed in the simulation than in the measurement. This is because
the measured scenario is not exactly the same as the open boundary in the
simulation. It is also due to the high sensitivity of the measured results to the
geometrical arrangement of the AUT as mentioned in 3.2.1. In the H-plane
(xz-plane), a very similar shape of the near-field patterns can be seen at dif-
ferent frequencies. The beamwidth of the near-field pattern becomes wider
with increasing frequency.
It can be concluded that the double-elliptical slot antenna in direct contact
with human body exhibits a very good impedance matching from 1.35 to
9 GHz. This antenna features good impedance matching, very high gain,
front-to-back ratio and hence very good penetration ability into human body
based on the simulation results. Observing the near-field pattern of the an-
tenna, a stable main beam direction is expected due to the symmetric radiat-
ing structure. Regarding the simulated gain and pattern, the operational band
of the antenna is limited to 7 GHz due to strong sidelobes and grating lobes at
the higher frequencies. Based on the analysis in chapter 2, the reflections at
the higher frequencies above 7 GHz are too weak to be detected. Hence, the
double-elliptical slot antenna with a size of 263552 mm3 can be operated
106
0° 0°
30° −30° 30° −30°
60° −60° 60° −60°
simulated 1GHz simulated 3GHz

measured 1GHz measured 3GHz
90° −90° 90° −90°
−30 −20 −10 0 −30 −20 −10 0
(a) 1 and 2 GHz (b) 3 and 4 GHz
Figure 5.10.: Simulated and measured near-field pattern of the double-

elliptical slot antenna (r= 40 mm), E-plane (yz-plane) for co-
polarization (normalized in dB).
in frequency range of 1.35 to 7 GHz for medical diagnostic purposes.
0° 0°
30° −30° 30° −30°
60° −60° 60° −60°

90° −90° 90° −90°
−30 −20 −10 0 −30 −20 −10 0
Figure 5.11.: Simulated and measured near-field pattern of the double-

elliptical slot antenna (r= 40 mm), H-plane (xz-plane) for co-
5.3.2. Sector-like slot antenna

To further increase the circumference of the slot and hence decrease the low-
est operational frequency of the antenna from the previous section, the an-
tenna is modified by using a sector-like slot. The sector-like slot antenna
is also fabricated on the substrate Rogers RT 6010 ("r =10.2 , d =1.27 mm,
tan ı =0.0023) and has two monopoles and a sector-like slot on the top of the
107
Table 5.1.: Design parameters of the sector-like slot antenna.

parameter w l s1 s2 e1 e2 r ˛
value (mm) 35 35 1:8 2:2 4:5 3 16 15ı
substrate as shown in Figure 5.12 [34].
(a) Top view (b) fabricated an- (c) fabricated

tenna (microstrip antenna (ground side)
side)
Figure 5.12.: Layout and photos of the sector-like slot antenna (conductors
are grey).
The parameter s2 in Figure 5.12 is important for the radiation at high frequen-
cies since a strong surface current at high frequencies is observed at the two
steps being closed to the monopoles. The impedance matching of the input
port in the whole frequency band is achieved by optimizing e1 , e2 , ˛, s1 and
s2 . The optimized values of the parameters are given in Table 5.1. The steps
between the two sectors are introduced to suppress the high mode current
distribution at high frequencies that result in significant sidelobes and grating
lobes in the radiation pattern as mentioned in the previous section. By using
these steps, the distance between two monopoles s1 can be reduced (maintain
the radius of the slot), which leads to suppress the sidelobes and grating lobes
at the higher frequencies [Yan11] [34].
The two monopoles are fed by the same differential feed network from the
previous section. The arrangement of the antenna and the feed network is
shown in Figure 5.12 (b) and (c). The arrangement of antenna, feed network
108
−5
−10
in dB
11 −15
S
−20
simulated in PEG
−25 measured in PEG
measured on skin
−30
1 2 3 4 5 6 7 8 9
Frequency f in GHz
Figure 5.13.: Simulated and measured S11 in PEG-water solution of the

sector-like slot antenna.
and phantom material is the same as shown in Figure 5.3 (b).

The simulated and measured results have very similar characteristics from 1
to 9 GHz as shown in Figure 5.13. The simulated S11 is mostly under -10 dB
from 1.07 to 9 GHz except for the jumps at 4.3, 7 and 8.2 GHz. Compared
to this, significant improvement in the impedance matching at these three
frequencies is observed in the measured S11 . However, the S11 at 1.8 GHz,
due to the high "r of the PEG solution at low frequencies, reaches -6.5 dB.
The simulated maximum gain and penetration efficiency of the antenna are
shown in Figure 5.14. The sector-like slot antenna exhibits a slightly higher
maximum gain compared to the double-elliptical slot antenna. From 2.5 to
9 GHz, the antenna has a maximum gain almost above 10 dBi. This is due to
the fact that the sector-like slot antenna features a larger size in one dimension
and thus its directivity increases slightly.
Moreover, the penetration efficiency of the sector-like slot antenna is almost
larger than 0.9 (front-to-back ratio of 9.5 dBi). Therefore, the sector-like slot
antenna has a slightly better performance in terms of the maximum gain and
penetration efficiency.
The simulated and measured near-field patterns (Rff =81.7 mm at 1 GHz, r=
40 mm) of the antenna are shown in Figure 5.15 and 5.16 respectively in the
109
15 1
0.9
0.8
0.7
10
0.6
Gain in dBi
0.5
0.4
5
0.3
0.2
Gain 0.1
0 0
1 2 3 4 5 6 7 8 9
Frequency f in GHz

sector-like slot antenna with the phantom ("r D 20, D 0 S/m).
E-plane and H-plane. Slightly asymmetrical patterns at different frequencies

are observed and it shows that the asymmetrical connection of radiator and
feed network (manufacturing inaccuracy) strongly affect the near-field pattern
compared to the double elliptical slot antenna.
The radiation patterns are more directive than those of the double elliptical
slot antenna both in the E-plane and H-plane, which is consistent with the
simulated maximum gain. The main beam direction in the H-plane is quite
constant in the whole frequency band. In the E-plane, the maximum of the
pattern at high frequencies is skewed. Significant sidelobes are observed at 4
GHz [Yan11].
Compared to the double-elliptical slot antenna, the lowest operational fre-
quency of the sector-like slot antenna was decreased at the expense of a
slightly larger slot size based on the simulated results. However, the mea-
sured results exhibited only a slight improvement of the lowest operational
frequency, which is strongly affected by the dielectric properties of the tissue-
simulating liquid. The final design of the antenna has a size of 353552 mm3 .
110
0° 0°
30° −30° 30° −30°
60° −60° 60° −60°

90° −90° 90° −90°
−30 −20 −10 0 −30 −20 −10 0
Figure 5.15.: Simulated and measured near-field pattern of the sector-like

slot antenna (r= 40 mm), E-plane (yz-plane) for co-polarization
(normalized in dB).
0° 0°
30° −30° 30° −30°
60° −60° 60° −60°

90° −90° 90° −90°
−30 −20 −10 0 −30 −20 −10 0
Figure 5.16.: Simulated and measured near-field pattern of the sector-like

slot antenna (r= 40 mm), H-plane (xz-plane) for co-polarization
(normalized in dB).
5.3.3. Stepped-slot antenna

In order to further lower the lowest operational frequency fL (by increasing
the circumference yet reducing the overall size of the antenna), a stepped-
slot design concept is proposed as shown in Figure 5.17. The steps along the
whole slot increase the electrical length of the slot. The two steps near the two
monopoles allow a good electrical connection between the slot antenna and
the feed network. The angles and distances (to the center point of the antenna
structure) of the steps are optimized with respect to S11 and the radiation
pattern [3] [Yan11].
111
(a) Top view (b) fabricated an- (c) fabricated

tenna (microstrip antenna (ground
side) side)
Figure 5.17.: Layout and photos of the stepped-slot antenna (conductors are
grey, unit: mm).
The sidelobes and grating lobes at high frequencies (especially in the yz-
plane) can be minimized by reducing the size of the slot and the distance
of the two elliptically shaped monopoles in the y-direction. Furthermore,
due to the structure of the stepped-slot, the surface current is concentrated at
the edges of the steps (see in Figure 5.18 at different frequencies). Hence
the size of the ground area can be significantly reduced (reducing the dis-
tance between the slot and the edge of the ground) without influencing the
impedance matching of the antenna input port, since the current distribution
on the ground at different frequencies is not significantly changed. The final
size of the antenna is optimized to be 263552 mm3 .
The S11 in Figure 5.19 shows that the simulated antenna is matched to the
phantom of the human body from 1 to 9 GHz. The measured result in PEG-
water solution shows the S11 is under -10 dB in the frequency range from 1.07
to 9 GHz. At high frequencies, a lower S11 is observed in the measurement.
The simulated maximal gain of the antenna in contact with the phantom is
shown in Figure 5.20. The results show that the antenna has a relatively high
and constant gain (> 10 dBi) from 3 to 9 GHz. A small gain is observed at
112
(a) 1.15 GHz (b) 7 GHz
Figure 5.18.: Surface current distribution of the stepped-slot antenna.
−5
−10
in dB
−15
11
S
−20
simulated in PEG
measured on skin
−30
1 2 3 4 5 6 7 8 9
Frequency f in GHz
Figure 5.19.: Simulated and measured S11 of the stepped-slot antenna over
frequency.
the lower frequencies due to the antenna being electrically small in terms of
wavelength at the lower frequencies. Moreover, the simulated penetration
efficiency confirms that more than 80% of the radiated energy can penetrate
into the phantom (human body).
The simulated and measured near-field patterns at a distance of 40 mm from
the antenna at different frequencies are shown in Figure 5.21 and 5.22. Sym-
metrical patterns both in the E-plane and H-plane are observed with a stable
113
15 1
0.9
0.8
0.7
10
0.6
Gain in dBi
0.5
0.4
5
0.3
0.2
Gain 0.1
0 0
1 2 3 4 5 6 7 8 9
Frequency f in GHz

stepped-slot antenna with the phantom ("r D 20, D 0 S/m).
main beam direction due to the phase center being constant over the whole
frequency range. It can be seen that the pattern in the E-plane at 3 and 4
GHz features lower sidelobes compared to that of the sector-like slot antenna
in Figure 5.15. The reason is that the distance of the two steps (y-direction)
near the two monopoles is reduced but the lower frequency is maintained
from the miniaturization technique used in this design.
The simulated and experimental results have shown that the stepped-slot an-
tenna is able to radiate in the frequency range from 1 GHz to at least 7
GHz when placed on the human body. The lowest operational frequency
of 1.07 GHz is achieved while the antenna size of 263552 mm3 maintains
compared to the double-elliptical slot antenna.
5.3.4. Stepped-slot antenna with slotline feed network

After the optimization of the ground of the radiator (the stepped slot), the
size of the radiator of 2635 mm2 is obtained. To achieve an even more
compact antenna together with a feed network, the feed network must also be
miniaturized.
114
0° 0°
30° −30° 30° −30°
60° −60° 60° −60°

90° −90° 90° −90°
−30 −20 −10 0 −30 −20 −10 0
Figure 5.21.: Simulated and measured near-field pattern of the stepped-slot

antenna (r= 40 mm), E-plane (yz-plane) for co-polarization
(normalized in dB).
0° 0°
30° −30° 30° −30°
60° −60° 60° −60°

90° −90° 90° −90°
−30 −20 −10 0 −30 −20 −10 0
Figure 5.22.: Simulated and measured near-field pattern of the stepped-slot

antenna (r= 40 mm), H-plane (xz-plane) for co-polarization
(normalized in dB).
A further size reduction of the feed network using microstrip lines (see in
Figure 5.4) is limited by the /4 impedance transformer and tapered ground.
Moreover, the connection between the feed network and the radiator intro-
duces further impedance mismatch and loss. Therefore, a new concept of
using aperture coupling and a slotline is developed as shown in Figure 5.23.
The signal is fed to the microstrip line and transmitted to the slotline through
an aperture coupling. The orientation of the E-field along the slotline illus-
trated in Figure 5.23 (a) causes a differential feed to the monopoles of the
radiator. The antenna using this concept is shown in Figure 5.23 (b) and (c).
115
The Koch-shape (fractal structure) [KGAC 09] on the ground plane of the
feed network is applied to prevent the radiation of the feed network at the
lower frequencies. Since the feed network is connected to the stepped slot
radiator, the Koch-shape between the slotlines and monopoles become a loop,
which excites unwanted radiation. The direction of this unwanted radiation
is normal to the surface of the feed network and the excited frequency is
dependent on the circumference of the slot edge. The Koch-shape enables
the increase of the circumference within a small area of the ground and hence
decrease its resonance frequency down to 1 GHz. A small l2 and hence small-
sized feed network can be achieved.
(a) E-field distribution (b) Photo of the antenna (c) Photo of the antenna
along the slotline (left side) (right side)
Figure 5.23.: Design of the stepped slot antenna with the slotline feed network
((a): top and bottom sides superimposed).
The layout of the slotline feed network is shown in Figure 5.24. The rect-
angular shapes (with dimensions of w3 and l3 ) at the right upper and lower
corner on the top side must be soldered together with the slotted ground of
the antenna. In this way, the feed network can be fixed stably together with
the stepped slot antenna. The width of the slotline on the boundary of the
feed network is exactly the same as the distance between monopoles and the
slotted ground in the middle of the radiator structure. Therefore, this con-
cept allows for a stable connection between the feed network (slotline) and
radiator. The important design parameters are given in Table 5.2.
116
Figure 5.24.: Layout of the slotline feed network (top and bottom sides super-
imposed, conductor is grey).
Table 5.2.: Parameters of the slotline feed network.

Parameters w1 l1 ws ls w2 l2 w3 l3
Values in mm 26 36.5 0.57 8.79 16.5 20.5 2 1
w4 wo lo ws1 ws2 sc1 sc2 b1 bo
12 2.5 0.5 0.55 0.2 3.5 2.5 17.35 1.93
From the simulated and measured S11 as shown in Figure 5.25, it can be ob-
served that a good impedance matching is achieved from 1 to 7 GHz. The
measured result in PEG-water solution shows that the lowest operational fre-
quency is 1.25 GHz, while that of the measurement on the skin is 1.1 GHz.
The impedance matching around 7 GHz is about -7.8 dB (measured in PEG-
water solution), affected by the impedance matching of the slotline feed net-
work. This is because the slotline feed network exhibits a relatively smaller
bandwidth compared with the microstrip feed network (matched from 1 to
10 GHz) [Shu88]. However, the impedance matching of this antenna up to
7 GHz has fulfilled the requirement for microwave medical diagnosis based
on the analysis mentioned in chapter 2.
117
−5
−10
in dB
11 −15
S
−20
simulated in PEG
measured on skin
−30
1 2 3 4 5 6 7
Frequency f in GHz
Figure 5.25.: Simulated and measured S11 of the stepped slot antenna with
slotline feed network over frequency.
Regarding the simulated maximum gain and penetration efficiency shown in

Figure 5.26, the antenna features very high gain at the higher frequencies and
this drops down significantly at the lower frequencies (about 0 dBi at 1 GHz
due to the impedance mismatch). However, the maximum gain at 1.07 GHz is
still 2.8 dBi. Similar behavior can be observed with respect to the penetration
efficiency. A high front-to-back ratio can be achieved from 3 to 6 GHz, since
the penetration depth is larger than 0.9. A minimum of the penetration effi-
ciency (0.65) can be seen at 1.6 GHz. The reason is that the slotline feed net-
work contributes to the radiation into the free space region in this frequency
range (from 1.2 to 2.3 GHz). It can be concluded that this antenna has a very
high radiation performance in the frequency range from 3 to 7 GHz. The ra-
diation gain and penetration efficiency from 1 to 2 GHz decreased slightly
compared to the differentially-fed slot antenna with microstrip feed network.
Furthermore, the radiated near-field pattern is verified by comparing the sim-
ulated and measured pattern both in the E-plane and H-plane as shown in
Figure 5.27 and 5.28, respectively. Compared to the radiation pattern of
the stepped-slot antenna with microstrip feed network, the patterns exhibit
slightly narrower beamwidth in the E-plane and H-plane.
118
15 1
0.9
0.8
0.7
10
0.6
Gain in dBi
0.5
0.4
5
0.3
0.2
Gain 0.1
0 0
1 2 3 4 5 6 7
Frequency f in GHz

stepped slot antenna with slotline feed network ( phantom:
"r D 20, D 0 S/m).
0° 0°
30° −30° 30° −30°
60° −60° 60° −60°

90° −90° 90° −90°
−30 −20 −10 0 −30 −20 −10 0
Figure 5.27.: Simulated and measured near-field pattern of the stepped slot
antenna with slotline feed network (r= 40 mm), E-plane (yz-
plane) for co-polarization (normalized in dB).
In conclusion, by introducing the new slotline feed network, the antenna size
is reduced to 263536.5 mm3 compared to the antenna with microstrip feed
network (263552 mm2 ). The characteristics of the new antenna maintain
the desirable operational frequency range from 1 to 7 GHz.
119
0° 0°
30° −30° 30° −30°
60° −60° 60° −60°

90° −90° 90° −90°
−30 −20 −10 0 −30 −20 −10 0
Figure 5.28.: Simulated and measured near-field pattern of the stepped slot
antenna with slotline feed network (r= 40 mm), H-plane (xz-
plane) for co-polarization (normalized in dB).
5.3.5. Summary of the differentially-fed slot antennas

Table 5.3 provides a comparison of the four differentially-fed slot antennas
with respect to the size of the antenna together with the feed network, the
lowest (fL ) and highest (fH ) operational frequency and the relative band-
width. Since the impedance matching at the higher frequencies can easily be
achieved (even higher than 9 GHz), a stable pattern with low sidelobes is set
to be the criterion to determine the highest operational frequency. The upper
frequency is based on the simulation results, since the measured pattern is
limited up to 4 GHz due to high signal attenuation in tissue-simulating liquid.
The lowest operational frequency is determined based on measured S11 .
The common ground of the four antennas is that they all have a very stable
phase center and a constant main beam direction over the whole operation
frequency due to the symmetrical antenna structure. This feature leads to a
high applicability of the antenna for pulse-based wideband systems for med-
ical diagnosis. Second, a very high penetration efficiency and front-to-back
ratio can be achieved by directly placing the antenna between air and human
tissues. This results in a high efficiency of the radiation into the human body
and also minimizes the coupling to the environment.
Regarding the size of the radiator, a significant size reduction (26% in one
dimension) was successfully achieved by using a stepped slot compared to
the sector-like slot. However, increasing the circumference of the slot cannot
120
5.4. On-body matched antennas with a lowest operational frequency of 0.5 GHz
infinitely decrease the lowest operational frequency. Enlarging the size of

the slot can lower the fL very effectively, but introduces strong sidelobes and
grating lobes at high frequencies due to the large size of the slot compared to
the wavelength.
The slotline feed network contributed significantly to the size reduction of
the overall antenna with feed network. This shows the applicability of the an-
tenna for the construction of antenna arrays within a limited space for medical
applications.
The fL and the operational frequency band of the four antennas show only
slight differences. however, the stepped-slot antenna with slotline feed net-
work exhibits the best performance with regard to the impedance matching,
the radiation pattern, the size and the robustness.
Table 5.3.: Comparison of the developed differentially-fed slot antennas.

Slot antenna model Size (mm3 ) fL (GHz) fH (GHz) Rel. BW
Double-elliptical 263552 1.31 7 137%
Sector-like 353552 1.07 7 147%
Stepped-slot 263552 1.07 7 147%
Stepped-slot with 263536.5 1.1 7 145%
slotline feed network
5.4. On-body matched antennas with a lowest

operational frequency of 0.5 GHz
In the previous section, different wideband antennas characterized from 1 to
7 GHz were developed, which can be used in radar imaging for medical diag-
nosis due to the large bandwidth and very low operational frequency. More-
over, the antennas exhibit a robust near-field performance in tissue-simulating
liquid. However, to guarantee the deep penetration of the signal inside the
head tissues, even lower frequencies (< 1 GHz) are desirable for medical di-
agnosis (e.g. brain imaging) [17].
121
In the literature, there are only a few contributions about antennas with very
low operational frequency (< 1 GHz) for medical diagnosis. In [TP08], a tri-
angular microstrip antenna with a resonance frequency at 0.7 GHz was pro-
posed for stroke detection using an electromagnetic time-domain inversion al-
gorithm. However, the antenna has a large size of 4046 mm2 . A small-sized
antenna is regarded as a major challenge for very low operational frequency,
since a large number of elements in an array for improving the imaging result
are required.
Therefore, the goal of the further antenna development is to design anten-
nas having the lowest operational frequency of 0.5 GHz for a deep penetra-
tion into human tissues (e.g. brain or heart) and with an antenna size of not
larger than 3535 mm2 (the largest dimensions of the differentially-fed slot
antennas). For that reason, two antenna concepts are introduced in the follow-
ing sections. First, a dual-band slotted Bowtie antenna with microstrip feed
and aperture-coupling is described. It is followed by a Bowtie antenna using
folded structures with meandered lines and a coaxial-feed. Both of these two
antennas provide a lowest operational frequency of about 0.5 GHz.
5.4.1. Dual-band aperture-coupled Bowtie antenna

The dual-band aperture-coupled Bowtie antenna has two operational frequency
bands, with the first operational band from 0.5 to 0.7 GHz (lower frequencies)
and the second from 1.3 up to 4 GHz (higher frequencies). This antenna is
especially suitable for microwave tomography using mono-frequency opera-
tion, since the operational frequency can be chosen in different bands.
As shown in Figure 5.29, the dual-band Bowtie antenna consists of a mi-
crostrip line as feed on the top side of the substrate and a slotted Bowtie
ground plane on the bottom side. The signal is fed by the microstrip line with
a characteristic impedance of 50 . The signal is then coupled to the slotline
on the bottom side using an aperture-coupled transition. To achieve a broad-
band transition from microstrip to slotline, the microstrip on the top side is
connected to a virtual shorted circular stub [Shu88]. The bandwidth of this
transition is determined by the diameter of the circular stub dst , the width of
the slot ds and the relative position to the slotline lst . In addition, the photos
of the fabricated antenna are shown in Figure 5.30.
122
Figure 5.29.: Layout of the dual-band aperture-coupled Bowtie antenna (mi-

crostrip line on the top side and slotted Bowtie structure on the
bottom side).
(a) Top view (b) Bottom view
Figure 5.30.: Photos of the fabricated dual-band aperture-coupled Bowtie an-

tenna.
Considering the radiation mechanisms, the EM waves are coupled from the
microstrip line to the slotted Bowtie and are radiated. To achieve a broadband
impedance matching, the spline curve and the tapered section of the slotted
Bowtie are used in addition. Moreover, in Figure 5.31, the E-field distribution
123
between the slotted Bowtie is illustrated, which also indicates the polarization
of the radiated waves.
Figure 5.31.: Schematic representation of the E-field distribution between

slotted Bowtie of dual-band aperture-coupled Bowtie antenna
(dark color denotes the ground of the slot).
Regarding the first operational band, a resonance of the surface current along
the slotted Bowtie is excited, which enables the radiation of the E-fields. The
central frequency of this band is determined by the circumference of the slot:
c0
fL D p ; (5.4)
Cs "r;eff
where Cs is the circumference of the slot and is determined as:
(5.5)

Cs D 2 ds1 C 2lx1 C 2lsp :
At the second band (high frequencies), the radiation of the traveling waves
along the slotted Bowtie occurs. The EM waves propagate from the center
point of the slotted Bowtie to the left and right side, respectively. Therefore,
the radiations along the slot at the both sides have to be taken into account.
With a large aperture of the slot in the x direction, grating lobes at the higher
frequencies can be observed. The main pattern in the xz-plane is split into
different maximums. This can be minimized by reducing the size of the slot
in x-direction. In the optimization procedure, the slot widths (ds1 and ds2 )
are increased in y-direction so that the same circumference for the first band
is maintained. In this way, the grating lobes are suppressed at the higher
frequencies. The substrate Rogers RT 6010 and thickness of 0.635 mm is
chosen for the antenna design and fabrication.
124
The antenna element is in contact with a phantom material ("r D 20 (due to

the highest operational frequency up to 4 GHz), D 0 S/m) similar to the
different-fed slot antennas in the previous sections. With this configuration,
the signals are radiated perpendicularly (Cz direction) to the surface of the
bottom side of the antenna into the human body. The top side of the antenna
is isolated by using an absorber material.
In the parameter optimization, it can be noted that the aperture-coupled tran-
sition strongly influences the impedance matching at the two bands, which
can be seen in the measured results with varied dst (see Figure 5.32). By tun-
ing dst , an optimal impedance matching can be achieved. The final design
parameters of the antenna are given in Table 5.4.
The simulated and measured S11 of the slotted Bowtie antenna are shown in
Figure 5.33. In the measured S11 on skin, the two radiation bands as dis-
cussed before can be clearly seen. The measured S11 on skin and in PEG-
water solution show very similar resonance around 0.55 GHz. However, a
slight difference on the whole band is observed due to the multilayer model
(skin, fat, muscle), which causes multiple reflections in the measurement on
skin. The different resonant frequencies of the measured and simulated S11
in PEG-water solution are due to different boundary conditions between sim-
ulation and measurement.
−5
−10
S11 in dB
−15
−20 dst= 4.4 mm

dst= 5.0 mm
−25
d = 5.6 mm
st
−30
0 1 2 3 4 5
Frequency f in GHz
Figure 5.32.: Measured S11 of dual-band aperture-coupled Bowtie antenna

with different dst on skin.
125
Table 5.4.: Design parameters of the dual-band aperture-coupled Bowtie an-

tenna.
Parameter lx ly ds1 ds2 ds lx1
Value (mm) 35 26 20 14:8 2:8 5:6
Parameter lx2 lsp w1 d dst lst
Value (mm) 2 11:8 0:52 8:28 5:6 2:1
0
measured on skin
simulated in PEG
−10
in dB
−15
11
S
−20
−25
−30
0 1 2 3 4 5
Frequency f in GHz
Figure 5.33.: Simulated and measured S11 of the dual-band aperture-coupled

Bowtie antenna in PEG-water solution and on the skin.
The radiation pattern of the antenna is measured in PEG-water solution at

a distance of 5 cm. In the measurement of the radiation pattern, a distance
of at least of 5 cm is required to rotate the AUT in the E-plane by 180ı ,
since the feed point of the antenna is at the side of the substrate (refer to
Figure 5.30) and hence additional space is needed for the connector. This
distance (5 cm) corresponds to the radiating near-field region of the antenna
at the second band (1.3 to 4 GHz), but the far-field region at the first band
(0.5 to 0.7 GHz). The measured radiation pattern is limited to 2 GHz with a
measurement distance of 5 cm to the antenna. As shown in Figure 5.34, the
measured pattern in the E-plane is asymmetrical. This is because the current
distribution of the slotted Bowtie is asymmetrical due to the feed being on
one side of the ground. However, the main beam direction remains normal
126
to the surface of the antenna. In the H-plane, the patterns are very similar at
0.6 GHz, 1 GHz and 2 GHz.
0° 0°
30° −30° 30° −30°
60° −60° 60° −60°
0.6 GHz 0.6GHz

1 GHz 1GHz
2 GHz 2GHz
90° −90° 90° −90°
−30 −20 −10 0 −30 −20 −10 0
(a) E-plane (yz-plane) (b) H-plane (xz-plane)
Figure 5.34.: Measured radiation pattern of dual-band aperture-coupled

Bowtie antenna (r= 50 mm), in co-polarization (normalized in
dB).
In conclusion, the dual-band slotted Bowtie antenna features an extremely

small size of 3426 mm2 . The antenna can be operated in the frequency
range from 0.5 to 0.7 GHz and from 1.3 to 4 GHz.
This antenna is suitable for microwave tomography systems, in which mono-
frequency operation at different frequencies are required to be evaluated. Op-
erating from 0.5 to 4 GHz for radar imaging for medical diagnosis, the an-
tenna can introduce additional signal distortion (e.g. signal in the time do-
main), since the stop band from 0.7 to 1.3 GHz exists. In the next section,
an optimized Bowtie antenna characterized from 0.5 GHz to at least 2 GHz is
described for the radar imaging for medical diagnosis.
5.4.2. Compact double-layer folded Bowtie antenna

A regular Bowtie antenna is preferred for the design of UWB antennas in
free space, since it features wideband impedance matching and low design
complexity. Matching the regular Bowtie antenna to the human body with the
lowest operational frequency at 0.5 GHz, the Bowtie antenna will have a size
of at least 50 50 mm2 using the substrate RT 6010 (thickness of 1.27 mm).
Therefore, the miniaturization of the regular Bowtie antenna is the focus of
this section.
127
The miniaturization technique of the regular Bowtie antenna by the optimiza-

tion of its structure is used. The main idea is to increase the current path
along the antenna to fulfill the radiation requirements and maintain the an-
tenna size. Based on this, a compact double-layer folded Bowtie antenna is
developed and described in the following paragraphs.
To maintain a realistic scenario, a phantom emulating the human body is
placed on the bottom side of the antenna in the simulation as shown in Figure
5.35. The phantom ("r =35, =0 S/m) has the average relative permittivity
of skin and fat at 0.5 GHz [GGC96, GLG96a, GLG96b] and the dispersion
of the phantom material is not considered in the primary optimization of the
antenna based on the design principle discussed at the beginning of the chap-
ter. At the top side of the antenna, the free space region is maintained. The
direction of the radiation is normal to the planar surface of the antenna and
goes into the phantom.
To miniaturize the Bowtie antenna, a new concept by the introduction of a
double-layer structure and meandered lines is proposed. The antenna con-
sisting of two identical substrates is shown in Figure 5.36. Structure 1 is the
Bowtie element on the bottom side of the antenna, which is folded and con-
nected together with two symmetric meandered lines at the top side of the
antenna. With this concept, at low frequencies, the antenna acts like a closed
loop consisting of the Bowtie and the meandered lines. The electrical length
at the lower frequencies is enlarged by the meandered line hence the size of
the antenna can be significantly reduced. Looking at a superimposed image
of structure 1 and 3, the meandered structure must not overlap the triangles,
which make up the Bowtie antenna to ensure that the radiation is normal to
the antenna surface. Furthermore, the resonance between the Bowtie structure
and the meandered lines can be prevented in this way and wideband operation
of the antenna is then guaranteed. At the higher frequencies, the radiation is
mainly based on the Bowtie structure. A good impedance matching for a high
efficiency of radiation for the whole frequency range is thus required and is
done by optimizing the parameters of structure 1 and 3.
A coaxial cable (shown in Figure 5.36 (d)) with an inner diameter of 0.9 mm
and an outer diameter of 3.0 mm (characteristic impedance: 50 ) is used to
connect to the antenna structure normally to the top side of the antenna. A
128
Figure 5.35.: Arrangement of the double-layer folded Bowtie antenna to-

gether with the phantom.
balun (on structure 2 and 3) using a microstrip line is used to convert the un-
balanced coaxial feed to symmetrical (balanced) feed for the Bowtie structure.
The inner conductor of the coaxial cable is connected through the microstrip
on structure 2 to the left triangle on structure 1, while the outer conductor of
coaxial cable is connected to the right triangle. Two vias allow the connection
between different structures.
Regarding the current distribution as shown in Figure 5.37, a strong current
exists around the two feed points on the Bowtie structure and flows further
to the two sides. At 0.5 GHz (lower frequency) the current is concentrated
along the meandered line at the top side. A long electrical length at this
frequency can clearly be identified. At the higher frequencies (e.g. 1.5 GHz),
as expected in the analysis, the current flows along the Bowtie structure with
higher intensity and lower current density on the meandered line is observed,
which results in a low influence of the meandered structure with regard to the
impedance matching at high frequencies [Bar12].
Details of the parameter study regarding the impedance matching are de-
scribed in the following paragraphs. All relavant parameters are shown in
Figure 5.36. Increasing w1 , the impedance matching at high frequencies is
improved and hence the bandwidth of the antenna is increased significantly
(see in Figure 5.38 (a)). However, the S11 degrades at the higher frequen-
cies when the value of w1 becomes too large (e.g. 14.8 mm) due to the bad
impedance matching at the feed points of the Bowtie structure. Furthermore,
129
(c) Structure 3 (d) Side view
Figure 5.36.: The layout of the doubled-layered Bowtie antenna.
only slight changes of the lowest operational frequency (determined by S11

below -10 dB) of the antenna with varied w1 are observed.
The input impedance of the Bowtie structure is dependent on the position of
its feed point (w2 ), which allows the impedance matching from the coaxial
cable to the Bowtie structure. Figure 5.38 (b) shows the reflection coefficient
at the coaxial port of the antenna with different w2 . It must be noted that the
impedance matching at the lower frequencies is almost independent of w2 .
However, the S11 degrades at the higher frequencies with increasing w2 . At
w2 =14 mm, a bandwidth of 1 GHz (from 0.45 to 1.45 GHz) is observed. This
130
Figure 5.37.: Surface current distribution of the doubled-layered Bowtie an-

tenna at 0.5 GHz and 1.5 GHz(overlapping of three layers).
is because the feed points are too close to the middle of the Bowtie structure
resulting in very high input impedance between the two feed points at the
higher frequencies. The impedance mismatch is caused by the connection at
the higher frequencies from the coaxial cable to the Bowtie structure by vias.
Since the antenna is matched to the human skin and the dielectric property of
human skin differ among people, the sensitivity of the antenna to the permit-
tivity of the human skin is investigated. The simulations are performed with
4 different permittivities of the phantom (see Figure 5.35). The simulated S11
in Figure 5.39 shows that the curves shift slightly to the low frequency region
with higher permittivity of phantom since the wavelength in the phantom with
high relative permittivity becomes smaller. The frequency shift at the lowest
operating frequency is weaker than that at the highest operating frequency.
This is due to the current distribution on the meandered line, which mostly
contributes to the radiation at low frequencies, being not strongly affected
by the change of the relative permittivity of the phantom. Furthermore, the
bandwidth remains almost the same. Therefore, it can be concluded that the
impedance matching of the antenna is not degraded strongly, when the dielec-
tric property of human skin changes slightly. Therefore, the antenna shows
low sensitivity to the permittivity of human skin and is applicable for diag-
nostics on different parts of human body or different patients.
131
0 0
−5 −5
−10 −10
S11 in dB
S11 in dB
−15 −15
w1=10 mm w2=4 mm
−20 w1=13.2 mm −20 w2=8 mm
w1=14.5 mm w2=10 mm
−25 −25
w1=14.8 mm w2=14 mm
−30 −30
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
Figure 5.38.: Simulated S11 of the double-layer folded Bowtie antenna with
different widths w1 and w2 .
For the purpose of the comparison with regard to the size reduction, the
double-layer folded Bowtie antenna and a reference antenna (regular Bowtie
antenna) are shown in Figure 5.40. The designs of these antennas having the
same operational frequency band (0.5 to 2 GHz in this work) are based on the
configuration shown in Figure 5.35.
−5
−10
in dB
−15
ε =30
11
r
S
−20 ε =35
r
ε =40
−25 r
ε =45
r
−30
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
Frequency f in GHz
Figure 5.39.: Simulated S11 of the proposed antenna with different Permittiv-
ities (r ) of the phantom (=0 S/m).
132
The optimization of the antenna is performed by adjusting all parameters ac-

cordingly. Finally, the simulated S11 of the proposed antenna and the refer-
ence antenna are provided in Figure 5.41. These two antennas indicate similar
operational bandwidth, especially at the same lowest operational frequency of
0.5 GHz. However, the size of the proposed antenna (30 30 mm2 ) is reduced
by 40 % compared to the reference antenna (50 50 mm2 ).
(a) Reference antenna (b) Proposed antenna
Figure 5.40.: Layout of the reference antenna (regular Bowtie antenna) and
the doubled-layer Bowtie antenna.
After the initial optimization of the antenna with the phantom ("r =35, =0
S/m), a further optimization of the antenna is performed using realistic di-
electric properties (PEG-water solution) to include the conductivity and fre-
quency dispersion. The final optimized parameters of the antenna are given
in Table 5.5. The size of the optimized antenna (30 30 mm2 ) remains.
The fabricated antenna is shown in Figure 5.42. Figure 5.43 shows the simu-
lated S11 in PEG and the measured S11 in PEG-water solution and on human
skin. The antenna is finally optimized in PEG-water solution with S11 below
-10 dB from 0.5 to 2 GHz with a relative bandwidth of 120 %.
The simulated and measured results in PEG-water solution have very similar
run of the curves. The simulated S11 is below -10 dB in the whole frequency
range from 0.5 to 2 GHz, while the measured S11 indicates a slightly lower
operational frequency of 0.45 GHz. The difference at around 1 GHz of S11
could be caused by leaking of PEG-water solution into the free space region
at the top of the antenna. However, this problem can be avoided in a real
133
0
folded Bowtie antenna
reference antenna
−5
−10
in dB
11
−15
S
−20
−25
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
Frequency f in GHz
Figure 5.41.: Simulated S11 of the proposed antenna and the reference an-
tenna with the phantom (r =35, =0 S/m).
Table 5.5.: Parameters of the antenna

Parameter Value (mm) Parameter Value (mm)
wa 30 d2 9.56
la 30 d3 6.3
w1 14.5 d4 3.25
w2 9.56 d5 2.4
w3 8.61 d6 1
w4 7.8 w6 2.52
dv 0.9 s1 1.5
d1 1.9 s2 0.5
w5 9.56 a 3.11
d2 9.3 b 3.11
measurement scenario on the human body, since the antennas are in contact
with the human body and tissue-simulating liquid is not required.
The measured S11 on the human skin shows a better S11 in a realistic sce-
nario. Two very strong resonances can be observed due to the presence of
multi-reflections between the multilayer tissues. The S11 of -10 dB, ranging
from 0.44 GHz to at least 2 GHz, shows a bandwidth of more than 1.5 GHz.
134
(a) Front side (b) Back side (c) Antenna with cavity
Figure 5.42.: Photos of the fabricated antenna with coaxial cable and cavity
of Styrofoam.
Considering the different arrangement of the double-layer folded Bowtie an-

tenna (radiator placed between the substrate and the human body) compared
to the differentially-fed slot antenna (radiator placed between the substrate
and free space region), the penetration efficiency of the double-layer folded
Bowtie antenna is investigated. The results in Figure 5.44 show that the
penetration efficiency almost remains the same with the increase of the rel-
ative permittivity of the phantom ("r ) at the higher frequencies (from 1.25
to 2 GHz). However, at the lower frequencies, the penetration efficiency is
improved with increased "r . The reason is that, at the lower frequencies, the
current distribution of the whole antenna can be significantly altered due to
the presence of the meandered lines (at the top side of the antenna) by the
changing contrast of the permittivities between the phantom and air. In the
case of lower "r (e.g. 20), significant radiation into free space from the mean-
dered lines at the top side of the antenna occurs. With higher "r (e.g. 35), more
electric fields (more than 75% at 0.5 GHz) are radiated into the phantom be-
cause of a longer electrical length of the Bowtie structures at the bottom side
of the antenna leading to a larger effective wavelength. Therefore, the radia-
tion from the Bowtie structures into the phantom increases. In Figure 5.45, it
can be seen that a p of 0.9 can be achieved in the frequency range from 1 to
2 GHz. This corresponds to a extremely high front-to-back ratio (9.5 dB).
The simulated maximal gain of the antenna in the main beam direction with
the phantom ("r =35, =0 S/m) is shown in Figure 5.45. The results show that
the antenna has a relatively high and constant gain ( 6 dBi) from 1 to 2 GHz.
A small gain is observed at the lower frequencies compared to the gain at the
135
0
simulated in PEG
measured on skin
in dB −10
−15
11
S
−20
−25
−30
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
Frequency f in GHz
Figure 5.43.: Simulated and measured S11 of the double-layer folded Bowtie
antenna.
50 1
Relative permittivity of phantom εr
0.9
40
0.8
0.7
30
0.6
20 ηp 0.5
0.5 0.75 1 1.25 1.5 1.75 2
Frequecy f in GHz
Figure 5.44.: Simulated penetration efficiency p (in linear scale) of the

double-layer folded Bowtie antenna over frequency with varied
relative permittivity "r of phantom ( =0 S/m).
136
high frequencies. At 2 GHz, a decrease of the gain is observed. It is because

that the balun for the connection of the cable and Bowtie structure causes loss
at the higher frequencies due to the presence of surface waves and high mode
of the propagating waves. It can be still concluded that the antenna features
a very high radiation gain and radiation efficiency (better than 91% on the
whole operational band neglecting the conductivity of the phantom).
8 1
0.9
0.8
6
0.7
0.6
Gain in dBi
4 0.5
0.4
0.3
2
0.2
Penetration efficiency 0.1
Gain
0 0
0.5 0.75 1 1.25 1.5 1.75 2
Frequency f in GHz
Figure 5.45.: Simulated gain and penetration efficiency of the double-layer

folded Bowtie antenna ("r D 35, =0 S/m).
The simulated and measured radiation pattern at different frequencies (0.5, 1,

1.5 and 2 GHz) are shown in Figure 5.46 and 5.47 in the E-plane and H-plane,
respectively. The measurements are performed in the PEG-water solution
and the E-fields of the radiation are captured at a distance of 50 mm from
the antenna. As discussed in dual-band aperture-coupled Bowtie antenna, it
must be emphasized that the near-field region cannot be maintained for the
whole frequency range due to the geometric limitation in the measurement
(rotation of the AUT required). The measurement distance of 50 mm is far-
field region at 0.5 GHz (Rff =71 mm), which the radiating near-field pattern
can be measured at this distance at 1 GHz (Rff =35.5 mm). A symmetrical
pattern at different frequencies is observed due to the symmetrical antenna
137
0° 0°
30° −30° 30° −30°
60° −60° 60° −60°
0.5 GHz simulated 1.5 GHz simulated

1 GHz simulated 2 GHz simulated
0.5 GHz measured 1.5 GHz measured
1 GHz measured 2 GHz measured
90° −90° 90° −90°
−30 −20 −10 0 −30 −20 −10 0
(a) 0.5 and 1 GHz (b) 1.5 and 2 GHz
Figure 5.46.: Simulated and measured radiation pattern of the double-layer

folded Bowtie antenna (r= 50 mm), E-plane (yz-plane) for co-
and feeding structures. In the E-plane, minor sidelobes can be seen. The
beamwidth in the H-plane is wider than in the E-plane and the pattern are
very similar at different frequencies.
0° 0°
30° −30° 30° −30°
60° −60° 60° −60°
0.5 GHz simulated 1.5 GHz simulated

1 GHz simulated 2 GHz simulated
0.5 GHz measured 1.5 GHz measured
1 GHz measured 2 GHz measured
90° −90° 90° −90°
−30 −20 −10 0 −30 −20 −10 0
(a) 0.5 and 1 GHz (b) 1.5 and 2 GHz
Figure 5.47.: Simulated and measured radiation pattern of the double-layer

folded Bowtie antenna (r= 50 mm), H-plane (xz-plane) for co-
The SAR value is investigated in simulation of the antenna together with a

head model (shown in Figure 5.48). The frequency dependent dielectric prop-
erties including conductivity of the tissues are taken into account (according
to [GLG96a][GLG96b]). The head model consists of skin, skull, grey mat-
ter and white matter with a dimension of 60 60 mm2 in the cross section
(xy-plane).
138
The radiated E-fields in different tissues in the simulation are recorded and the
corresponding SAR values are calculated by (3.11). The density of different
tissues according to [BBH70] are used. As results, the 10 g averaged SAR
distribution of the cross section in skin and grey matter are depicted in Figure
5.49 and 5.50 at different frequencies, respectively. The input power of 1 mW
is used to calculate the SAR.
As can be seen in these results, the distributions at different frequencies
are similar but however with different intensity. At the higher frequencies
(e.g. 2 GHz), the SAR values are higher than at the lower frequencies. This
is due to the higher penetration efficiency and gain at the higher frequen-
cies, which result in stronger E-fields in tissues. Moreover, a large con-
ductivity also contributes to higher SAR values at the higher frequencies.
In the grey matter, the SAR values decrease compared to in the skin layer,
since the E-fields are attenuated during the propagation from the skin to grey
matter. Overall, the maximum value in the model is 0.06 W/kg (10 g tis-
sue), which fulfills the requirements for general public exposure according to
[oNiRPI98, Com01, CLC95].
Figure 5.48.: Simulation model of the human head with the double-layer
folded Bowtie antenna for evaluation of the SAR.
It can be concluded that the design of the double-layer on-body matched

antenna using the folded-Bowtie and meandered lines allows a significant
miniaturization of the antenna size. The radiation mechanism is validated
139
(a) 0.5 GHz (b) 1 GHz
(c) 1.5 GHz (d) 2 GHz
Figure 5.49.: Simulated SAR distributions in the human head for an input
power of 1 mW: slices of SAR at z D 0:5 mm in the skin layer.
by observing the current distribution at different frequencies. The param-

eter study showed that the impedance matching at the lower frequencies is
strongly dependent on the meandered structure at the top side, while the size
of the Bowtie structure and the feeding position affect the performance at the
higher frequencies. The final design after parameter optimization is character-
ized from 0.5 to 2 GHz and enables a size reduction of 40% compared to the
reference antenna (regular Bowtie antenna). The simulation results are val-
idated with a special measurement system using the tissue-simulating liquid
(PEG-water solution). The measured results agree well with the simulation
results. Moreover, the investigation of the SAR based on a multilayer model
of the human head showed the applicability of the proposed antenna for med-
ical diagnosis (e.g. stroke detection). The antenna characteristics such as
140
5.5. Summary
(a) 0.5 GHz (b) 1 GHz
(c) 1.5 GHz (d) 2 GHz
Figure 5.50.: Simulated SAR distributions in the human head for an input
power of 1 mW: slices of SAR at z D 5 mm in the grey matter
layer.
high gain, efficiency and front-to-back ratio will improve the SNR of radar
imaging system.
5.5. Summary
In reviewing the design challenges of on-body matched antennas, the general
design principles of these antennas were established with the aid of promoting
the optimization procedures and guaranteeing the simulation accuracy. The
efficiency of the proposed design procedure for on-body matched antenna is
reflected significantly in the development of the proposed antennas in this
141
Table 5.6.: Overview of the developed on-body matched antennas in this

chapter (measured results).
Antenna model Size (mm2 ) Frequency band (GHz) Rel. BW
Double-elliptical slot 2635 1.31-7 137%
Sector-like slot 3535 1.07-7 147%
Stepped-slot 2635 1.07-7 147%
Stepped-slot with 2635 1.1-7 145%
slotline feed network
Dual-band Bowtie 2635 0.5-0.7 and 1.3-4 -
Double-layer Bowtie 3030 0.44 - 2 128%
chapter.
A short overview of the developed on-body matched antennas are given in Ta-
ble 5.6 with respect to the different characteristics in simulation and measure-
ment. Regarding the operational frequency band, the developed antennas can
be divided into two categories. The differentially-fed slot antennas belong to
the first category, which exhibits impedance matching in the frequency range
from around 1 to 7 GHz. These antennas can be used for microwave medical
imaging, where high range resolution and hence a large bandwidth are desir-
able. The studies of different concepts of slotted ground and feed networks
showed that the lowest operational frequency was improved slightly. How-
ever, the size reduction of the antenna was of great significance for the overall
size and performance of the imaging system based on these antennas.
All of these four slot antennas indicated a stable phase center over the whole
operated frequency range and a high front-to-back ratio. Among these, the
stepped-slot antenna with slotline feed network features the smallest antenna
size (263536.5 mm3 ) and lowest operational frequency of 1.1 GHz, which
shows a high potential for medical imaging systems (e.g. radar imaging)
for the detection of breast cancer, stroke in human brain and other relevant
medical diagnosis.
For a deeper penetration of EM waves into human tissues (e.g. brain or
heart), two modified Bowtie antennas with a lowest operational frequency
of 0.5 GHz were designed. The dual-band aperture coupled Bowtie antenna
142
5.5. Summary
allows to perform microwave imaging diagnosis in different frequency bands

using microwave tomography. For the usage of the whole frequency band
from 0.5 GHz to at least 2 GHz, a double-layer folded Bowtie antenna can be
used.
All of the antennas proposed in this chapter were significantly miniaturized
based on different techniques. In this way, more antenna elements can be
implemented in an array within a limited area/space for medical diagnosis
and therefore, more data can be gathered from medical scenarios. This small-
sized antenna, will contribute positively to the improvement of the resolution
and signal-to-clutter ratio (SCR) of microwave imaging systems for medical
diagnosis such as stroke and breast cancer detection.
To demonstrate the applicability of the proposed antennas, an antenna array
with the stepped-slot antenna with slotline feed network is constructed and a
microwave imaging system together with a brain phantom will be discussed
in the following chapter.
143
6. Body-matched antennas based
microwave medical imaging
In the previous chapters, various designs of body-matched antennas were de-

veloped for different operational frequency bands and radiation properties.
To demonstrate the performance enhancement of microwave medical systems
utilizing the designed body-matched antennas, this chapter provides a feasi-
bility study of the microwave medical diagnosis for stroke detection by using
a realistic measurement setup.
The measurement setup using the on-body matched antennas is demonstrated
for a comparison with the system using the off-body antennas. Investigations
by the author in [20] [Riv11] showed that the detection of water accumulation
in an oil-phantom with Vivaldi-antennas (matched in free space) was not suc-
cessful. The reason is that there is a strong reflection from the air-phantom in-
terface and strong antenna coupling, which cannot be strictly calibrated from
the received signals. Thus, the detection capability and imaging results of the
microwave medical imaging system is significantly degraded. Therefore, the
goal of this chapter is to show the impact of the developed on-body matched
antennas on the system performance. These antennas feature a very low oper-
ational frequency (about 1 GHz), stable radiation pattern and compact size of
a single antenna, which improves the penetration capability and image qual-
ity (contrast, resolution and signal-to-clutter ratio (SCR)) of the microwave
medical imaging system.
In the first section, the measurement setup together with a brain phantom for
the hemorrhagic stroke detection is introduced. Next, the calibration methods
for removing the background reflection and antenna coupling as well as the
applied imaging algorithm (wideband beamforming algorithm based on time-
domain signals) are presented. The received signals are then analyzed and
the results of the two-dimensional (2D) microwave imaging are provided.
145
6. Body-matched antennas based microwave medical imaging
Finally, the improvements of the on-body matched antennas for microwave

medical imaging are concluded.
6.1. Measurement setup for microwave

medical imaging
The measurement setup is constructed for the detection of hemorrhagic stroke.
First, the construction of the brain phantom including the realistic dielectric
properties is briefly discussed. Based on the shape of the brain phantom, an
antenna array is put directly in contact with the brain phantom. After that,
the measurement setup of the overall medical diagnosis system is described
together with the measurement procedure and data acquisition.
6.1.1. Brain phantom for stroke detection

The demonstrator for stroke detection using microwaves is based on a tissue-
simulating phantom, which is a standard norm before a clinical trial is per-
formed. Thus, a brain phantom is required to be constructed for the stroke
detection and will be briefly described. Moreover, to better evaluate the im-
provements of the on-body matched antenna for the medical imaging sys-
tem without the significant influence of a complicated brain model, the brain
phantom for the detection of hemorrhagic stroke is simplified to be con-
sisting of the blood and an averaged brain tissue (averaged permittivity of
skull, grey matter and white matter). The goal is to detect the existence and
position of the bleeding in the brain. In the following section, a concept of
constructing a brain phantom is investigated in terms of its dielectric property
and stability. Then, the geometric arrangement of the brain phantom with the
blood and averaged brain tissue is provided.
The brain phantom must be constructed to be of a fixed shape according to
the realistic scenario. Considering the cost, availability and toxicity, gelatin
(bloom 180), sugar and distilled water are chosen to be the ingredients of the
phantom. Gelatin helps to fix the shape of the phantom. Sugar and gelatin
are soluble in water, which enables a homogeneous mixture. The achieved
relative permittivity can be altered by the weight ratio of sugar in the mixture.
146
6.1. Measurement setup for microwave medical imaging
The phantom can be produced in three steps: in the first step, the water and
sugar are mixed in a pan at room temperature and heated to 60ı C. Then, the
gelatin is added into the mixture. The mixture must be stirred for it to be ho-
mogeneous at around 60ı C. In the final step, the mixture is carefully poured
into a container (to avoid/remove any air bubbles in the mixture or on the sur-
face) and quickly cooled to 8ı C in a refrigerator to avoid any inhomogeneity
that might be caused during the cooling process.
Since the measurements with phantom can take a long time and must be re-
peatable (after a few days), the mixture must be stable with regard to the
dielectric properties for a long period. For this purpose, a test phantom (sugar
45%, gelatin 10%, water 45%) was measured three times after different time
periods and its dielectric property is plotted in Figure 6.1. It can be observed
that a stable dielectric porperty is guaranteed during a period of one week
and the conductivity decreases slightly after 3 weeks due to the reduced water
content in the mixture. In conclusion, the dielectric property of the phantom
based on this concept is very stable for the period of one week. Furthermore,
the homogeneity and fixed shape of the whole phantom can be guaranteed.
Based on this concept, the brain phantom including blood and averaged brain
tissue can be constructed. The ingredients with weight ratio for the blood and
averaged brain tissues, based on the concept discussed before, are provided
in Table 6.1 [Gee12].
The shape of the blood and brain phantoms are shown in Figure 6.2. The
blood phantom is enclosed within a thin latex bag with a length of 3 cm and
radius of 0.5 cm. The averaged brain tissue consists of a cylindrical section
(length of 20 cm and radius of 7.5 cm) and a half spherical section as shown
in Figure 6.2 (c). The blood phantom is then placed inside the averaged brain
tissue. The exact location of the blood phantom will be given later. The
presence of the latex between the blood and averaged brain tissue affects the
reflection behavior only very slightly thanks to the low thickness of the latex
(less than 0.1 mm) compared to the wavelength in the operation frequency
range.
Figure 6.3 provides the measured relative permittivities and conductivities of
the blood and averaged brain tissue of the constructed phantom. In compar-
ison, the references of the dielectric properties of blood and averaged brain
147
60 9
on the first day 8 on the first day
50 after one week after one week
7
after three weeks
after three weeks

40 6
5
30
4
20 3
2
10
1
0 0
1 2 3 4 5 6 7 8 1 2 3 4 5 6 7 8
(a) Relative permittivity (b) Conductivity
Figure 6.1.: Comparison of the relative permittivity (a) and (b) conductivity
of the test phantom, measured on the first day, after one week and
three weeks.
Table 6.1.: Ingredients of blood and averaged brain tissue for the brain phan-
tom with their weight ratios.
Tissue Gelatin in % Sugar in % Distilled water in %
Blood 11 15 74
Averaged brain tissue 6 47 47
(a) Phantom of blood (b) Top view of whole phan- (c) Side view of whole
tom phantom
Figure 6.2.: Photos of the brain phantom for hemorrhagic stroke detection.
148
tissue according to the Cole-Cole equation are also plotted. The reference
of averaged brain tissue is the average of the permittivity of the skin, skull,
grey and white matter. Both the relative permittivity and conductivity of the
blood phantom agree very well with the reference. The relative permittiv-
ity and conductivity of the averaged brain tissue indicate a slight difference
compared to its reference. The conductivity of the averaged brain tissue is
larger than that of the reference in almost the whole frequency range from 1
to 7 GHz. It increases the signal attenuation and makes the detection more
difficult (considered as a worse case). The large difference of the relative per-
mittivity between blood and averaged brain tissue can be clearly seen and can
be utilized for the detection of the blood phantom due to the large contrast of
the dielectric properties at their boundary.
70 12
blood phantom
60 blood reference
10
phantom (averaged brain tissue)
reference (averaged brain tissue)

50
8
40
6
30
4
20 blood phantom
blood reference
10 2
phantom (averaged brain tissue)
reference (averaged brain tissue)
0 0
1 2 3 4 5 6 7 8 1 2 3 4 5 6 7 8
(a) Relative permittivity (b) Conducttivity
Figure 6.3.: Relative permittivities and conductivities of the blood and aver-
aged brain tissue compared with references from the Cole-Cole
equation (averaged permittivity of skull, grey matter and white
matter).
To detect the existence and position of the blood within the brain tissue, an
antenna array is required to be placed around the phantom to capture the
reflections, which will be introduced in the following section.
149
6.1.2. Array with on-body matched antennas

Since the brain phantom was simplified to be of a nearly cylindrical shape,
the antenna elements are arranged in a circular configuration so that each
element in the constructed antenna array can be directly in contact with the
brain phantom. Moreover, the imaging algorithm can also be simplified, since
the captured reflections are based on a circular scan.
To investigate reflections from the target in a wide frequency range in the
radar imaging, the differentially-fed slot antennas optimized from 1 to 7 GHz
are chosen to construct the antenna array. To allow more antennas to be placed
in the circular array, the small-sized stepped-slot antenna with slotline feed
network (263536.5 mm3 ) is used and an array of 10 antenna elements is
constructed as shown in Figure 6.4. Moreover, this antenna exhibits high
maximum gain and penetration efficiency at high frequencies compared to the
other differentially-fed slot antennas. The shortest dimension of this antenna
is 26 mm on the horizontal plane. The main direction of radiation of each
antenna points to the center point of the array and the distance between this
center point to the surface of the radiator is 75 mm, which corresponds to the
radius of the brain phantom. All of the antennas are fixed between two plastic
rings with a width of 20 mm and radius of 90 mm.
(a) Top view (b) Perspective view
Figure 6.4.: Array with 10 single antennas in a circular arrangement (unit:

mm).
150
The single antenna configuration is shown in Figure 6.5. To reduce the cou-
pling between adjacent antennas, each antenna is wrapped with absorber ma-
terial with Styrofoam in between. Vertical polarization is used. The charac-
teristics of the antenna element can be seen in 5.3.4.
(a) Radiator, top and bottom (b) Slotline feed network (c) Antenna (d) Antenna
side with feed wrapped with
network absorber
Figure 6.5.: The fabricated stepped-slot antenna with slotline feed network
for the construction of the array.
6.1.3. Measurement setup with brain phantom

With the brain phantom and the antenna array, the overall measurement setup
for hemorrhagic stroke detection will be introduced.
The frequency sweep approach using a VNA is applied for the medical imag-
ing to achieve a high system dynamic range. The measurement is performed
with a frequency sweep from fmin to fmax in frequency interval of f . The
transfer function of channel Hc .f / can be obtained from the step-frequency-
mode of the VNA. Then the channel impulse response (CIR) hc .t/ in the time
domain is obtained through an Inverse Fast Fourier Transformation (IFFT) in
(6.1). Furthermore, a window function (Hamming window) with very high
side lobe suppression is used before the IFFT to reduce the side lobes in the
time domain.
hc .t/ D IFFT .Hc .f // : (6.1)

Though the antennas used are matched in the frequency range from around 1
to 7 GHz, the measured frequency range is set to be from 0.5 to 5 GHz (801
151
points). It is because the signal attenuation in the brain phantom is too high
above 5 GHz and thus the reflections will be too weak to be detected.
The frequency band is divided into many frequency points and a small IF
bandwidth is used, which leads to a better SNR. Due to the high SNR, a high
measurement accuracy is achievable by using this frequency sweep approach.
Otherwise, the decrease of the IF bandwidth increases the whole sweep time
and a real time measurement requires large IF bandwidth. Therefore, a com-
promise between sweep time and SNR must be found for different applica-
tions. In the following measurement, an IF bandwidth of 150 Hz is used and
the measured time of a whole sweep is in the millisecond range, which is still
acceptable for the real time process. The input power of the NWA is set to be
5 dBm and a power amplifier (discussed in 3.2.1) is used.
The overall measurement setup is illustrated in Figure 6.6. The signal is radi-
ated by one antenna (Tx) and the other antennas (Rx) receive the reflections.
This is repeated with different antennas as the Tx. With 10 antennas in an
array, 10 .10 1/=2 unique pairs of antennas and hence 45 signals can be
obtained (Tx and Rx are reciprocal). The S21 of the VNA is recorded for each
transmission pair. For the signal analysis, Si;j is used to denote the measured
S21 with the i th antenna as Tx and the j th antenna as Rx. In the demon-
strator, the response (S21 ) of the different channels is achieved by manually
switching the different antennas.
The signals are recorded for further image processing. For successful image
reconstruction, the exact location of each antenna has to be known to deter-
mine the corresponding time delay of the useful reflection. Then, based on
the recorded data and geometrical arrangement of the antenna array, the mi-
crowave image of the detection scenario can be generated using a microwave
reconstruction algorithm. With regard to the microwave reconstruction algo-
rithm, a beamforming algorithm will be described in the following sections.
152
6.2. Image reconstruction with beamforming algorithm
Figure 6.6.: Measurement setup of the medical imaging system for hemor-
rhagic stroke detection.
6.2. Image reconstruction with beamforming

algorithm
For radar imaging, several adaptive/non-adaptive processing approaches have
been developed and applied to wideband medical imaging in the literature.
To obtain a high resolution and low sidelobes of imaging results, an adaptive
technique, which is a time-domain multistatic adaptive microwave imaging
(MAMI), was reported in [XGXC 06]. In [OJG10], the data-independent mi-
crowave imaging via space-time (MIST) beamforming technique based on
finite impulse response (FIR) filters designed in the frequency domain was
extended and applied to breast cancer detection. The different improvements
of various algorithms will not be discussed further. For the image recon-
struction in this thesis, a general beamformer (Delay-and-Sum beamformer)
is used [FLHS02] [19].
Since the proposed antennas are matched directly to the human body, the
strong reflection from the skin in the received signals is inherently eliminated.
However, the recorded signals still contain unwanted responses such as back-
ground reflections and antenna coupling, which are of a much higher ampli-
153
tude than the reflections from the target. Thus, before the imaging recon-
struction algorithm is performed, calibration methods to remove unwanted
responses are first discussed to enhance the response from the target.
6.2.1. Calibration methods
The goal of the calibration is to remove the residual antenna reverberation,

background reflections and strong antenna coupling. Due to the small dis-
tance between the antenna elements, a strong direct coupling between Tx and
Rx has to be taken into account. The distance between the antenna and the tar-
get is also comparable to the space between antennas. Therefore, the antenna
coupling cannot be separated and eliminated in the time domain. Moreover,
the reflection from the target is very weak due to the high signal attenuation
in the brain phantom. Therefore, the useful reflection is overlapped by the an-
tenna coupling, which cannot be separated from the desired reflection. Thus,
the antenna coupling has to be calibrated out.
Based on the proposed antenna array, the coupling between different antenna
elements in the presence of the brain phantom are measured. The S21 of
the different channels is measured directly by the NWA. The results are then
evaluated in the time domain, where the strongest peak values with regard to
the channel impulse response are considered to be the magnitude of the an-
tenna coupling. The coupling between different antenna elements is shown
in Figure 6.7. The minimum of the peak is set to be -80 dB, which is lo-
cated diagonally (no coupling exists for the antenna itself). The results show
that the adjacent antennas have the strongest coupling. The antenna coupling
decreases significantly with the increasing spacing between the antennas. It
has been shown that the antenna coupling is especially dominant when it is
adjacent to the Tx. Therefore, the calibration of the antenna coupling is of
significant importance for the quality and reliability of the imaging results.
Regarding the aforementioned issue, the reference signal that is the average
of the signals from all channels is often used for the calibration. The perfor-
mance of the averaging calibration is investigated in the following section.
154
−45
10
9 −50
8
−55
7
Antenna index
6 −60
5
−65
4
−70
3
2 −75
1
−80
1 2 3 4 5 6 7 8 9 10
Antenna index
Figure 6.7.: Antenna coupling between different antenna elements of the ar-
ray (peak value of the CIR in dB).
(a) Original configuration (b) Configuration after rotation
Figure 6.8.: Illustration of the antenna array configuration together with blood
phantom before and after rotation.
155
Averaging calibration
The antenna array configuration with the antenna index and the exact position
of the blood phantom is shown in Figure 6.8 (a). The blood phantom is placed
in front of antenna 4 and 5 at a distance of around 20 mm. Figure 6.9 shows
the photos of the measurement setup of the brain phantom and antenna array.
(a) Side view (b) Top view
Figure 6.9.: Measurement setup with the antenna array and the gelatin phan-
tom.
Before the averaging calibration is performed, the signals (Si;j ) are trans-
formed into the time domain using the IFFT to obtain xij .t/, which denotes
a signal transmitted from the antenna j and received by the antenna i . The
signal is measured from 0.5 to 5 GHz and then zero-padding is applied in the
frequency domain to increases the time resolution of the signal in the time
domain.
The calibration can be mathematically expressed as
xij ;cal .t/ D xij .t/ xij ;ref .t/ (6.2)
where xij ;cal .t/ and xij ;ref .t/ are the calibrated and reference signal, respec-
tively.
The reference signal xij ;ref .t/ is generated by averaging all signals separately
for the channels with the same space between Tx and Rx, since their coupling
is very similar. Therefore, 5 different groups of reference signals are needed
for the calibration of all signals ( 0, 1, 2, 3 or 4 antennas between Tx and Rx).
156
After that the reference signals for 5 groups are obtained, each signal xij .t/
is subtracted by its reference signal.
In comparison, a rotation calibration will be introduced in the following sec-
tion.
Rotation calibration
The calibrated signal using rotation calibration can be written as
xij ;cal .t/ D xij .t/ xij ;rot .t/ ; (6.3)
where xij ;rot .t/ are the recorded reference signals after rotation.
For the rotation calibration, the overall array is rotated 15ı anti-clockwise
around the center point of the array as shown in Figure 6.8 (b). Altogether two
measurements must be performed based on the original arrangement (origi-
nal signals xij .t/) and the arrangement after the rotation (reference signals
xij ;rot .t/). For each channel, the background reflection and antenna coupling
are calibrated with its reference signal (the same transmission pair after ro-
tation). Since the antennas and antenna distance remain the same, the back-
ground reflection and antenna coupling can be successfully calibrated. Oth-
erwise, the response from the target changes slightly with array arrangement
before and after rotation in terms of time delay and amplitude and hence the
calibrated signal still represents the response from the target.
Though the rotation calibration requires two measurements, the antenna cou-
pling can be completely calibrated, since the reference signal is obtained for
each channel, respectively.
6.2.2. Delay and Sum beamformer

The Delay and Sum (DAS) beamformer [FLHS02] is a passive process, in
which the backscattered signals are first assigned with appropriate time delays
so that the beamformer focuses on a specific point inside the target region.
Afterwards the time-aligned signals are added and the energy of the resulting
signal which corresponds to the intensity level of a focal point within the
imaging area is estimated. In this thesis, the DAS beamformer is based on the
157
time domain signals, which are obtained from the wideband measured data in
the frequency domain (using IFFT).
Considering that the signal radiated from the Tx (antenna index i ), propagates
to the target, the response from the target can be received by the Rx (antenna
index j ) (i and j could be any single antenna of the array), the output of a
DAS beamformer is given by:
N N N N
X X X X dij
y.t/ D wij xij .t ij / D wij xij .t /; (6.4)
cm
j Di C1 i D1 j Di C1 i D1
where ij is the time delay of the propagation, which is determined by the
total distance dij from Tx to target and from target to Rx with regard to the
propagation velocity of the waves in medium cm . Including ij in the beam-
forming, the different time delays are then compensated and hence the signals
can be coherently added to reconstruct the target as shown in Figure 6.10.
wij is the beamformer apodization weights (window function), which is of
significance for the optimization of the shape of the peak and the suppres-
sion of the sidelobes of the signals. A Hamming window is also used in the
beamformer.
The DAS beamformer described above has several benefits and drawbacks.
From (6.4), it can be observed that the DAS beamformer is carried out on
each received signal independently. Moreover, the DAS beamformer features
simplicity and robustness of the signal processing. However, its performance
is limited in terms of the artifact removal and the resolution of the image [19].
To improve the artifact removal effect and the resolution of the image, a co-
herence factor (CF) is introduced. The coherence factor coefficients are de-
fined as the ratio of the coherent sum of the signals and the non-coherent sum
of the signals. This can be written as
ˇP ˇ2
2ˇ N
PN
i D1 xij .t ij /ˇ
ˇ ˇ
j Di C1
CF .t/ D ˇ2 : (6.5)
N.N 1/ N
P PN ˇ
j Di C1 i D1 xij .t ij /ˇ
ˇ
The CF coefficients are multiplied directly to the output signal y.t/ of the
DAS beamformer. By taking these coefficients into account, the directivity of
158
6.3. Analysis of calibrated signals and 2D imaging results
Figure 6.10.: Block diagram of the Delay-and-Sum beamformer.
the beamformer can be considerably increased, which increases the dynamic

range of the imaging result and improves significantly the resolution of the
microwave image.
6.3. Analysis of calibrated signals and 2D

imaging results
Having the DAS beamforming in mind, the calibrated signals obtained from
the measured data as well as the processed two-dimensional (2D) microwave
images will be analyzed and discussed in terms of dynamic range of im-
age (range of peak level and noise level of the image), signal-to-clutter ratio
(SCR) and image resolution.
Figure 6.11 shows the received signal from the adjacent antennas (antenna in-
dex 1, 2, 3, 4 and 5) based on the configuration in Figure 6.8 (a). The antenna
coupling is dominant before the calibration is performed. In the signals after
calibration using the averaging method, the reflection from the blood phan-
tom can be clearly identified especially in x45 . Therefore, with very careful
measurements, the reflection from the target (blood phantom) is visible in the
time domain. The strongest reflection occurs around at t D3.6 ns.
159
4 2
x x
3 12 1.5 12
x x
23 23
2 1
x x
Amplitdue in mV
Amplitude in mV
34 34
1 x 0.5 x
45 45
0 0
−1 −0.5
−2 −1
−3 −1.5
−4 −2
0 1 2 3 4 5 6 7 0 1 2 3 4 5 6 7
Time t in ns Time t in ns
(a) Before calibration (b) After calibration
Figure 6.11.: Received signals before and after averaging calibration of the
channels of adjacent antennas (antenna index: 1, 2, 3, 4, and 5)
in the time domain.
The width of the pulse refers to the signal bandwidth, which is determined
by the bandwidth of the antennas. Weak ringing effects can also be observed,
which are caused by the non-linearity of the antenna and additional multi-
reflections in the phantom. Using these results in the time domain, the signals
are processed using the beamforming algorithm.
The processed microwave image (xy-plane refer to Figure 6.8) is shown in
Figure 6.12(a). For a better comparison with the real setup, the microwave
image together with the antenna array is shown in Figure 6.12(b). The posi-
tion of the blood can be clearly identified, which agrees with the real location
of the blood in the brain phantom. However, different clutters are also visible
in the microwave image and can be regarded as ghost targets caused by the
averaging calibration. A very strong reflection is observed in the calibrated
x78 in the time domain, which results in the strong clutter in the image in
front of antenna 7 and 8.
160
(a) Processed microwave image
(b) Processed microwave image together with antenna ar-

ray
Figure 6.12.: Processed microwave image using DAS-algorithm (averaging

calibration).
161
From the imaging result (see Figure 6.12(a)), it can be observed that the aver-
aging calibration has a low SCR level, since strong clutter signals are caused
by the nonideal calibration.
To perform a better calibration of the background reflection and antenna cou-
pling, the rotation calibration is introduced based on an additional measure-
ment with a rotation (15ı ) of the antenna array (see Figure 6.8 (b)). The
calibrated signals of some channels in Figure 6.13 using rotation calibration
show that reflections in x57 and x58 contribute to the final imaging results.
The processed microwave image with the rotation calibration is shown in
Figure 6.14(a). A strong reflection is observed at (x D 2 cm, y D -3.5 cm),
which is estimated to be the blood phantom. The blood phantom appears in
the image with better contrast and its position is estimated with good accuracy
compared to the result using averaging calibration. The image together with
the array configuration in Figure 6.14(b) demonstrates that the blood phantom
is clearly represented in the right location in the brain phantom. Otherwise,
some weak clutters are still identified. The SCR of the image is estimated to
be 6 dB, which allows the detection of the target (the blood).
It can be seen that the SCR of the processed image using the rotation calibra-
tion is improved significantly, compared to the result using averaging calibra-
tion. The image result indicates also lower sensitivity to the position error of
the antennas between measured signals and reference signals.
The benefit of the averaging calibration procedure is that only one measure-
ment is required compared to the rotation calibration. However, it must be
assumed that the signals recorded from different channels in the same group
exhibit similar background reflection and antenna coupling. In practice, how-
ever, this cannot be achieved and a slight difference of the coupling between
adjacent antennas introduces clutter after calibration, which results in the in-
accuracy of this calibration method. The reason for the different coupling
is that the fabricated antennas cannot feature exactly the same impedance
matching and radiation properties. On the other hand, a not 100% uniform
arrangement of the antenna in a circular configuration also causes inaccu-
racy of the antenna position and slight differences of the antenna spacing.
This results in different antenna coupling in the same group and hence the
difference is averaged and included in each reference signal. Due to this dif-
162
0.15 0.15
x x
24 58
0.1 x 0.1 x
35 69
x x
Amplitude in mV
Amplitude in mV
0.05 46 0.05 7,10
x x
57 81
0 0
−0.05 −0.05
−0.1 −0.1
−0.15 −0.15
4.5 5 5.5 6 6.5 7 4.5 5 5.5 6 6.5 7
Time t in ns Time t in ns
(a) With one antenna between Tx and Rx (b) With two antennas between Tx and Rx
Figure 6.13.: Received signals calibrated with rotation calibration in the time
domain.
ference, ghost reflections are caused in the calibrated signal and cannot be
distinguished from the realistic reflections. To reduce this effect, the anten-
nas should be fabricated very carefully to assure almost the same performance
and the antennas should be placed uniformly.
Due to the better performance of the rotation calibration, this calibration
method will be used for further investigation of the microwave imaging sys-
tem.
To demonstrate the influence of the signal bandwidth on the microwave im-
age, the bandwidth of the received signals using the rotation calibration are
limited to 2.5 GHz (0.5-3 GHz). From the processed imaging result in Figure
6.15, the degradation of the imaging result in terms of SCR and resolution
can be clearly identified. This result confirms that the contribution of higher
frequencies (between 3 to 5 GHz) to the imaging result is significant, though
the signal attenuation in lossy medium (brain phantom) at the higher frequen-
cies is considerably higher compared to lower frequencies. Therefore, the
designed on-body matched antennas must feature, besides very low opera-
tional frequency, a large operational bandwidth, which has been discussed in
chapter 5.
163
(a) Processed microwave image
(b) Processed microwave image together with antenna ar-

ray
Figure 6.14.: Processed microwave image together with antenna array (rota-
tion calibration).
164
For microwave medical imaging, a large number of antennas in an array are

desirable to obtain more scattered information from the target. The effect of
the number of antennas on the imaging result is investigated based on the re-
ceived signals using the rotation calibration. To compare the imaging result
with 10 antennas, only 5 antennas (antenna index 1, 3, 5, 7, and 9 based on the
configuration in Figure 6.8) are now used to transmit or receive signals and
5(5-1)/2=10 signals are available for the imaging algorithm. The imaging
result (see Figure 6.16) shows that the SCR is strongly degraded and the clut-
ters are stronger than the response from the target. It can be seen that the large
number of antennas in the array is of significance to the detection capability
of the microwave medical imaging system. This confirms that the on-body
matched antennas are required to be miniaturized. It is due to the fact that the
small-sized antennas enable a large number of antennas to be arranged in an
array within a limited space for the scenario in medical diagnosis. In this way,
more reflections from the target can be captured by the antenna array. Since
the antennas are required to have very low operational frequency, the minia-
turization of the on-body matched antennas (discussed in chapter 5) becomes
more challenging due to the increased wavelength.
In terms of dynamic range of the processed images, the values based on
the mentioned different procedures (calibration method, frequency range and
number of used antennas) are summarized in Table 6.2. A slightly higher
dynamic range is achieved using averaged calibration, since the signals are
averaged and hence the noise level is suppressed. The three procedures based
on rotation calibration have only slight differences in terms of dynamic range
of the processed images. The results agrees with the analysis that the dy-
namic range is strongly dependent on the used radar approach in terms of
signal modulations (PN, IR-UWB, frequency sweep approach) and the trans-
mitted power.
6.3.1. Summary
A feasibility study of the detection of the human hemorrhagic stroke using
microwave imaging is provided, which is based on an on-body matched an-
tenna array characterized from 1 to 7 GHz and a brain phantom (consisting
averaged brain tissue and blood phantom). In the image processing, the DAS
165
Figure 6.15.: Microwave image processed by DAS-algorithm: rotation cal-

ibration, reduced bandwidth effect (frequency range from 0.5
and 3 GHz).
Figure 6.16.: Microwave image processed by DAS-algorithm: rotation cali-

bration, reduced antenna elements in the array (5 antennas in
the array used).
Beamformer is applied to reconstruct the target in the measurement scenario.

From the imaging results, it has to be emphasized that the blood phantom is
166
Table 6.2.: The dynamic range of the images based on different process pro-
cedures.
Process procedure Antenna number Frequency range Dynamic range
Averaging calibration 10 0.5 to 5 GHz 36 dB
Rotation calibration 10 0.5 to 5 GHz 31 dB
Rotation calibration 5 0.5 to 5 GHz 29.1 dB
Rotation calibration 10 0.5 to 3 GHz 28.7 dB
successfully reconstructed in the right location in the image compared to the

imaging results with the antennas in free space, in which the detection was
only feasible in the simulation and no promising results can be verified in the
measurement [20] [Riv11]. The reason is that, in case of the off-body antenna,
the antenna coupling and the strong reflection from the air-skin boundary as
well as the impedance mismatching of the antennas caused by placing them
closely to the phantom, degrade the received signals. The useful reflections
cannot be reconstructed using the calibration method. Therefore, it is of sig-
nificance to match the antennas to the human body (phantom), since the signal
level of reflection is strongly improved benefiting from the elimination of the
reflection from the air-skin boundary as well as the achieved good impedance
matching of the antennas in contact with the human body.
It has been shown that small-sized antennas with a stable phase front im-
proved the location accuracy in the DAS beamformer. This is because the
operational range is comparable to the antenna size. Thus, the determination
of the phase front of the antenna is of significance for the image processing.
The small size and the stable phase front over frequency help to obtain a cor-
rect delay time of the received signals. Moreover, a significant improvement
of the microwave images is achieved in terms of SCR and contrast by increas-
ing the number of antenna elements in the array. The clutter in the image with
10 antennas was suppressed compared to the result with 5 antennas. There-
fore, the miniaturization of the single antenna contributes considerably to the
microwave imaging in terms of enabling more antenna elements to be used
within a limited space.
The bandwidth of the antenna also strongly affects the image quality. The
167
wideband operation using the proposed on-body matched antenna enables

the wideband medical imaging, in which significant improvements in resolu-
tion and artifact reduction have been achieved. It can be further emphasized
that even better image resolution and SCR can be achieved by optimizing the
imaging algorithm. However, further improvement of the algorithm is not
within the framework of this thesis.
The successful detection of the blood phantom embedded in averaged brain
tissue confirmed the applicability of the on-body matched antenna for the
microwave medical imaging. The recorded signals by the on-body matched
antennas feature high SNR and contain more useful information than that of
antenna matched in free space.
168
7. Summary and conclusions
In this thesis, new concepts of body-matched antennas for microwave med-

ical applications were developed, where the antennas are directly in contact
with the human body or implanted in the human body. The antenna designs
serve mainly to conduct microwave medical diagnosis and data telemetry.
Therefore, two categories of antennas were designed for microwave medi-
cal diagnosis (on-body matched antennas) and data telemetry (implantable
antennas). With these, a new future telemedicine home system can be en-
visioned, in which the vital signs of the patients can be monitored in real
time and hence their mobility as well as their quality of life can be improved.
Concepts were developed with antenna miniaturization techniques to result
in a compact size for the intended medical devices and to maintain the high
radiation performance in terms of the impedance matching, efficiency and ra-
diation pattern. Application-wise, a high directivity and front-to-back ratio
are desirable for the on-body matched antennas for medical diagnosis, while
an omni-directional pattern is required for implantable antennas. With these
challenges in mind, the contributions of each chapter are summarized in the
following paragraphs.
Chapter 2 provided a preliminary study for the design of the body-matched
antennas. Through the quantitative investigation of human tissues in terms of
dielectric properties, the propagation behavior of EM-waves in human tissues
was evaluated from the system point of view. Based on the high dielectric
contrast between the target (tumor, urine, etc.) and surrounding tissues, the
reflection from their boundaries can be detected for diagnostic purposes. The
results showed that a very high system dynamic range is required to capture
the reflected weak signal. With regard to the operational frequency range,
the lower frequencies enable a high penetration into the human body and the
large frequency range contributes to a higher bandwidth and hence a finer
range resolution. Depending on the applications (diagnosis with wideband
169
signal or data telemetry with narrow band signal) and location of sensors in
the human body, an optimal operational frequency can be arranged in the
consideration of the SNR of the system and the radiation regulations. Fur-
thermore, with this knowledge, the design requirements and specifications
of on-body matched and implantable antennas have been specified for mi-
crowave medical systems.
In chapter 3, characterization methods were introduced to evaluate the perfor-
mance of body-matched antennas. For the verification, two antenna measure-
ment systems with tissue-simulating liquid for the characterization of body-
matched antennas are provided, whose characterization of the radiation can-
not be performed in an anechoic chamber due to the presence of the lossy
human body. The E-field probe based measurement system allows the direct
measurement of the antenna pattern at a short distance (40-60 mm), while the
planar near-field system is designed to obtain the far-field antenna pattern by
applying the NF-FF transformation. High measurement accuracy can be as-
sured by the use of a precise positioning system with a high precision stepper
motor. Such antenna measurement systems allow the validation of the sim-
ulation results of on-body matched and implantable antennas with respect to
the impedance matching, near-field and far-field pattern.
The body-matched antenna designs for data transmission of IMDs and diag-
nosis are provided respectively in chapter 4 and 5. All of the antennas are
matched to different human tissues depending on the location of the devices
and optimized together with the tissue model for the highest performance
with respect to impedance matching, radiation efficiency and radiation pat-
tern. Miniaturization techniques contribute significantly to the reduction of
the antenna size. Furthermore, the design was focused on optimizing the an-
tenna structure and the current distribution at different frequencies.
In chapter 4, three stripline-fed implantable slot antennas at the ISM band
were designed. The antennas have an omni-directional pattern especially in
the H-plane, that allows a robust wireless connection between implants and
external devices, independent of the location and orientation of the patients to
the base station. Based on optimization of the antenna structure, the antenna
(M3-3) was miniaturized to a size of 10.45 12.9 mm2 , with a size reduction
of more than 40% compared to the original one (M3-1), allowing a signifi-
170
cant size reduction of the IMDs. Due to the stripline-fed configuration and
good impedance matching (under -10 dB) in muscle tissues, the implantable
antenna features a high radiation efficiency (94.2%) and gain (4.3 dBi), which
enable data telemetry between implants and external devices with high effi-
ciency.
For medical diagnosis, different on-body matched antennas operating at dif-
ferent frequency bands were proposed in chapter 5. By matching the antennas
directly to the human skin, very high front-to-back ratio of the radiation pat-
tern is achieved, which significantly improves the SNR of the antenna system.
The small-sized differentially-fed slot antenna characterized from 1 to 7 GHz
enables the construction of a microwave medical imaging system with high
resolution for the detection of e.g. breast cancer. To achieve deeper penetra-
tion of the microwave signals into the human body, a dual-band slotted Bowtie
antenna and a wideband folded double-layer Bowtie antenna were designed.
By applying a folded structure and meandered microstrip lines at the bottom
of the Bowtie antenna, a small size of 30 30 mm2 with a size reduction of
40% is achieved, compared to the reference antenna (regular Bowtie antenna
at 50 50 mm2 ). These antennas can either be used in radar-based imaging or
tomography. The small-sized antennas with the lowest operational frequency
of 0.5 GHz, high radiation efficiency, high front-to-back ratio and relatively
high gain improve the penetration depth and the SNR of received signals,
which showed a high potential for microwave medical diagnosis.
In chapter 6, the applicability of the on-body matched antenna for medical
diagnosis was confirmed. The measurement demonstrator using a tissue-
simulating phantom and an array of 10 stepped slot antennas show the high
detection capability for stroke thanks to the low operational frequency, high
front-to-back ratio and small antenna size. The blood phantom was success-
fully detected with good SCR, although very weak reflected signals were re-
ceived due to a strong signal attenuation in the brain phantom during the prop-
agation. Without the aforementioned features, the off-body antenna cannot
detect the weak reflection. The results confirmed that the on-body matched
antenna significantly improved the SNR of the reflections compared to the
off-body antenna. Furthermore, the small-sized differentially-fed slot antenna
allows for a large number of elements for the array in a limited space, which
171
also improves the image quality with respect to the resolution and SCR. It
has been confirmed that the miniaturization techniques, which have been ap-
plied to the body-matched antenna design, contributed very positively to the
performance of the microwave medical imaging.
In conclusion, the following considerable contributions of this thesis can be
emphasized:
The quantitative investigation of the wave propagation in human tissues
was studied, which allows the specification of the required antennas in
terms of the lowest operational frequency, bandwidth, directivity and
etc, depending on the applications and the location on/in the human
body.
For the first time in the literature, two antenna measurement systems
were developed, which facilitate the validation of both near-field and
far-field patterns of the on-body matched and implantable antennas im-
mersed in a tissue-simulating liquid.
Miniaturization techniques of the antennas based on extending the elec-
trical length and regulating the current flow in the antenna structure
at different frequencies were developed. Multiband resonances, inde-
pendent surface current regulation at lower and higher frequencies for
wideband antenna as well as fractal structure for miniaturization also
for suppression of unwanted radiation are proposed, by which signifi-
cant size reductions of the antenna operated at 1 GHz and even 0.5 GHz
were achieved.
Design methods (such as introducing slotted ground, stripline-feed and
aperture-coupling) for the prevention of radiation loss caused by sur-
face and leaky waves due to the surrounded lossy tissues. High radia-
tion efficiency and a stable pattern have been achieved.
The applicability of the on-body matched antennas is successfully veri-
fied with a microwave imaging system for the detection of hemorrhagic
stroke, where a deep penetration of the microwaves are required.
The proposed design principles of the various antennas as well as the minia-
turization techniques can be extended to the general design of antennas for
172
medical applications in the presence of the human body, or even for anten-
nas in free space. The achieved small-sized antennas at different frequency
bands contribute directly to compact medical sensors (i.g. portable sensors
for diagnosis or IMDs). Moreover, the promising results of the detection of
hemorrhagic stroke using a simplified brain phantom with the proposed an-
tenna array encouraged further research to be conducted including the use
of even lower frequencies for high penetration ability and advanced imag-
ing algorithms. All these contributions enable an extension of the medical
applications using microwaves with the objective to obtain more advanced
healthcare systems.
173
A. Parameters of tissues based on
Cole-Cole equation
Tissue "1 "1 1 (ps) ˛1 "2 2 (ns) ˛2

muscle 4.0 50.0 7.23 0.10 7000 353.6 0.10
skin 4.0 39.0 7.96 0.10 280 79.58 0.00
w. matter 36.71 0.29 8.04 0.24 - - -
g. matter 55 0.5 7.76 0.12 - - -
blood 4.0 56.0 8.38 0.10 5200 132.6 0.10
bone 2.5 18.0 13.26 0.22 300 79.58 0.25
fat 2.5 3.0 7.96 0.20 15 15.92 0.10
skull 2.5 10.0 13.26 0.22 180 79.58 0.20
Tissue "3 3 (s) ˛3 "4 4 (ms) ˛4
muscle 1:2 106 318.31 0.10 2:5 107 2.274 0.00 0.20
skin 3:0 104 1.59 0.16 3:0 104 1.592 0.20 0.0004
w. matter - - - - - - 0.47
g. matter - - - - - - 1.03
blood 0:0 0.0 0.70
bone 2:0 104 159.15 0.20 2:0 107 15.91 0.00 0.70
fat 3:3 104 159.15 0.05 1:0 107 7.958 0.01 0.01
skull 5:0 103 159.15 0.20 1:0 105 15.91 0.00 0.20
Table A.1.: Parameters of Cole-Cole equation used to predict the dielectric

properties of different tissues.
175
Bibliography
[ABWZ09] G. Adamiuk, S. Beer, W. Wiesbeck, and T. Zwick. Dual-

Orthogonal Polarized Antenna for UWB-IR Technology.
IEEE Antennas and Wireless Propagation Letters, 8:981–984,
2009.
[Ada10] G. Adamiuk. Methoden zur Realisierung von dual-

orthogonal, linear polarisierten Antennen für die UWB-
Technik. Phd thesis, Karlsruher Institut für Technologie (KIT),
Karlsruhe, 2010.
[AKS13] S. Ashok Kumar and T. Shanmuganantham. Implantable CPW

Fed Circular Slot Antenna for ISM Band. In National Confer-
ence on Communications (NCC), Feburary 2013.
[ARNT11] R. K. Amineh, M. Ravan, N. K. Nikolova, and A. K. Tpour.

Three-Dimensional Near-Field Microwave Holography Using
Reflected and Transmitted Signals. IEEE Transactions on An-
tennas and Propagation, 59(12):4777–4789, December 2011.
[ARTN11] R. K. Amineh, M. Ravan, A. Trehan, and N. K. Nikolova.

Near-Field Microwave Imaging Based on Aperture Raster
Scanning with TEM Horn Antennas. IEEE Transactions on
Antennas and Propagation, 59(3):928–940, March 2011.
[Bal05] C. A. Balanis. Antenna Theory: Analysis and Design. Wiley-

Interscience, New York, 3rd edition, 2005.
[Bar12] A. Barouni. Weiterentwicklung einer am Körper angepassten

UWB-Antenne für die Schlaganfalldiagnostik. Bachelor the-
sis, Institut für Hochfrequenztechnik und Elektronik, Karl-
177
Bibliography
sruher Institut für Technologie (KIT), Karlsruhe, November

2012.
[BBH70] T. W. Barber, J. A. Brockway, and L. S. Higgins. The Density

of Tissues in and about the Head. Acta Neurologica Scandi-
navica, 46(1):85–92, March 1970.
[BLM06] D. Baudry, A. Louis, and B. Mazari. Characterization of

the Open Ended Coaxial Probe Used for Near-Field Measure-
ments in EMC Applications. PIER, Progr. Electromagn. Res.,
60:311–333, 2006.
[BSAGC 00] S. Boyer, M. Sawan, M. Abdel-Gawad, S. Robin, and M. M.

Elhilali. Implantable Selective Stimulator to Improve Bladder
Voiding: Design and Chronic Experiments in Dogs. IEEE
Transactions on Rehabilitation Engineering, 8(4):464–470,
December 2000.
[CBHVV04] M. Converse, E. J. Bond, S. C. Hagness, and B. D. Van Veen.

Ultrawide-Band Microwave Space-Time Beamforming for
Hyperthermia Treatment of Breast Cancer: A Computational
Feasibility Study. IEEE Transactions on Microwave Theory
and Techniques, 52(8):1876–1889, August 2004.
[CKK04] A. Christ, A. Klingenböck, and N. Kuster. Exposition

durch Körpernahe Sender im Rumpfbereich, Arbeitspaket I:
Bestandsaufnahme. Foundation for Research on Informa-
tion Technologies in Society, Swiss Federal of Technology -
ETHZ, December 2004.
[CLC95] CENELEC CLC/TC111B. Human Exposure to Electromag-

netic Fields: High Frequency (10 KHz to 300 GHz). Techni-
cal report, European Prestandard (pr ENV 50166-2), Brussels,
1995.
[Com01] Federal Communications Commission. Evaluating Compli-

ance with FCC Guidelines for Human Exposure to Radiofre-
178
Bibliography
quency Electromagnetic Fields. Supplement C to OET Bul-

letin 65, June 2001.
[dCHSC 11] F. S. di Clemente, M. Helbig, J. Sachs, U. Schwarz,

R. Stephan, and M. A. Hein. Permittivity-Matched Compact
Ceramic Ultra-Wideband Horn Antennas for Biomedical Di-
agnostics. In 5th European Conference on Antennas and Prop-
agation (EUCAP), April 2011.
[Deb60] P. Debye. Polar Molecules. Dover Publications, New York,

1960.
[DFGC 08] F. D’Agostino, F. Ferrara, C. Gennarelli, R. Guerriero, and

M. Migliozzi. An Effective NF-FF Transformation Technique
with Planar Spiral Scanning Tailored for Quasi-Planar An-
tennas. IEEE Transactions on Antennas and Propagation,
56(9):2981–2987, Sepember 2008.
[ESSK00] T. Endo, Y. Sunahara, S. Satoh, and T. Katagi. Resonant Fre-

quency and Radiation Efficiency of Meander Line Antennas.
Electronics and Communications in Japan (Part II: Electron-
ics), 83(1):52–58, January 2000.
[FABC 07] J. Ferlay, P. Autier, M. Boniol, M. Heanue, M. Colombet, and

P. Boyle. Estimates of the Cancer Incidence and Mortality in
Europe in 2006. Ann Oncol., 18(3):581–92, 2007.
[FCC02] Revision of Part 15 of the Commission’s Rule Regarding

Ultra-Wideband Transmission Systems. First Report and
Order, Federal Communications Commision (FCC), Februar
2002.
[FG95] J. Fromm, R. E. Varon and L. R. Gibbs. Congestive Heart

Failure and Pulmonary Edema for the Emergency Physician.
Journal of Emergency Medicine, 13(1):71–87, 1995.
[FLHS02] E. C. Fear, X. Li, S. C. Hagness, and M. A. Stuchly. Confocal

Microwave Imaging for Breast Cancer Detection: Localiza-
179
Bibliography
tion of Tumors in Three Dimensions. IEEE Transactions on

Biomedical Engineering, 49(8):812–822, August 2002.
[Gee12] D. Geenen. Aufbau und Optimierung eines körperangepassten

UWB-Antennenarrays für die Schlaganfalldiagnostik. Bach-
elor thesis, Institut für Hochfrequenztechnik und Elektronik,
Karlsruher Institut für Technologie (KIT), October 2012.
[GGC96] C. Gabriel, S. Gabriel, and E. Corthout. The Dielectric Prop-

erties of Biological Tissues: I. Literature Survey. Phys. Med.
Biol., 41(11):2231–2249, November 1996.
[GJR10] M. Guardiola, L. Jofre, and J. Romeu. 3D UWB Tomog-

raphy for Medical Imaging Applications. In IEEE Anten-
nas and Propagation Society International Symposium (AP-
SURSI), July 2010.
[GLG96a] S. Gabriel, R. W. Lau, and C. Gabriel. The Dielectric Proper-

ties of Biological Tissues: II. Measurements in the Frequency
Range 10 Hz to 20 GHz. Phys. Med. Biol., 41(11):2251–2269,
November 1996.
[GLG96b] S. Gabriel, R. W. Lau, and C. Gabriel. The Dielectric Proper-

ties of Biological Tissues: III. Parametric Models for the Di-
electric Spectrum of Tissues. Phys. Med. Biol., 41(11):2271–
2293, November 1996.
[GMP07] S. F. Gregson, J. McCormick, and C. Parini. Principles of

Planar Near-Field Antenna Measurements. Institution of En-
gineering and Technology, 2007.
[GMZC 10] C. Gilmore, P. Mojabi, A. Zakaria, M. Ostadrahimi, C. Kaye,

S. Noghanian, L. Shafai, S. Pistorius, and J. LoVetri. A Wide-
band Microwave Tomography System with a Novel Frequency
Selection Procedure. IEEE Transactions on Biomedical Engi-
neering, 57(4):894–904, April 2010.
180
Bibliography
[Guy84] A. W. Guy. History of Biological Effects and Medical Ap-

plications of Microwave Energy. IEEE Transactions on Mi-
crowave Theory and Techniques, 32(9):1182–1200, Septem-
ber 1984.
[HLCC 11] F. Huang, C. Lee, C. Chang, L. Chen, T. Yo, and C. Luo.

Rectenna Application of Miniaturized Implantable Antenna
Design for Triple-Band Biotelemetry Communication. IEEE
Transactions on Antennas and Propagation, 59(7):2646–
2653, July 2011.
[Hol38] H. E. Hollmann. Das Problem der Behandlung biologis-

cher Körper im Ultrakurz-Wellen-Strahlungsfeld. G. Thieme,
Leipzig, 1938.
[Hur85] W. D. Hurt. Multiterm Debye Dispersion Relations for Per-

mittivity of Muscle. IEEE Transactions on Biomedical Engi-
neering, BME-32(1):60–64, January 1985.
[ITU01] Sharing between the Meteorological Aids Service and Medical

Implant Communications Systems (MICS) Operating in the
Mobile Service in the Frequency Band 401-406 MHz. Interna-
tional Telecommunications Union Recommendation (ITU-R)
Std. ITU-R Recommendation SA 1346, 2001.
[Kap96] D. S. Kapp. Efficacy of Adjuvant Hyperthermia in the Treat-

ment of Superficial Recurrent Breast Cancer: Confirmation
and Future Directions. Int. J. Rad. Oncol. Biol. Phys., 35:117–
1121, 1996.
[Kar04] A. Karlsson. Physical Limitations of Antennas in a Lossy

Medium. IEEE Transactions on Antennas and Propagation,
52(8):2027–2033, August 2004.
[KCLC 09] M. Klemm, I. J. Craddock, J. A. Leendertz, A. Preece,

and R. Benjamin. Radar-Based Breast Cancer Detection
181
Bibliography
Using a Hemispherical Antenna Array - Experimental Re-

sults. IEEE Transactions on Antennas and Propagation,
57(6):1692–1704, July 2009.
[KGAC 09] D. D. Krishna, M. Gopikrishna, C. K. Aanandan, P. Mohanan,

and K. Vasudevan. Compact Wideband Koch Fractal Printed
Slot Antenna. IET Microwaves, Antennas & Propagation,
3(5):782–789, August 2009.
[KHT08] T. Karacolak, A. Z. Hood, and E. Topsakal. Design of a Dual-

Band Implantable Antenna and Development of Skin Mim-
icking Gels for Continuous Glucose Monitoring. IEEE Trans-
actions on Microwave Theory and Techniques, 56(4):1001–
1008, April 2008.
[KKYP01] J. Kim, K. Kim, J. Yook, and H. Park. Compact Stripline-fed

Meander Slot Antenna. Electronics Letters, 37(16):995–996,
August 2001.
[KRS04] J. Kim and Y. Rahmat-Samii. Implanted Antennas Inside a

Human Body: Simulations, Designs, and Characterizations.
IEEE Transactions on Microwave Theory and Techniques,
52(8):1934–1943, August 2004.
[KRS06] J. Kim and Y. Rahmat-Samii. Planar Inverted-F Antennas on

Implantable Medical Devices: Meandered Type versus Spiral
Type. Microwave and Optical Technology Letters, 48(3):567–
572, March 2006.
[Lan09] N. Langendorf. Ischemic & Hemorrhagic Stroke Report.

Technical report, Center for Health Information Analysis,
2009.
[LBVVH05] X. Li, E. J. Bond, B. D. Van Veen, and S. C. Hagness.

An Overview of Ultra-Wideband Microwave Imaging via
Space-Time Beamforming for Early-Stage Breast-Cancer De-
tection. IEEE Antennas and Propagation Magazine, 47(1):19–
34, February 2005.
182
Bibliography
[LHC93] H. J. Liebe, G. A. Hufford, and M. G. Cotton. Propagation

Modelling of Moist Air and Suspended Water/Ice Particles
at Frequencies below 1000 GHz. In AGARD 52nd Special
Meeting of the Panel on Electromagnetic Wave Propagation,
Adv. Group for Aerosp. Res. and Dev., Palmade Mallorca, May
1993.
[LS94] D. Lamensdorf and L. Susman. Baseband-Pulse-Antenna

Techniques. IEEE Antennas and Propagation Magazine,
36(1), February 1994.
[MBZC 11] F. Merli, L. Bolomey, J. Zurcher, G. Corradini, E. Meurville,

and A. K. Skrivervik. Design, Realization and Measurements
of a Miniature Antenna for Implantable Wireless Communi-
cation Systems. IEEE Transactions on Antennas and Propa-
gation, 59(10):3544–3555, October 2011.
[MD99] G. H. Markx and C. L. Davey. The Dielectric Properties of Bi-

ological Cells at Radiofrequencies: Applications in Biotech-
nology. Enzyme and Microbial Technology, 25(3-5):161–171,
August 1999.
[MJ13] J. Ma and A. Jemal. Breast Cancer Metastasis and Drug Re-

sistance. Springer, New York, 1st edition, 2013.
[Moo63] R. Moore. Effects of a Surrounding Conducting Medium on

Antenna Analysis. IEEE Transactions on Antennas and Prop-
agation, 11(3):216–225, May 1963.
[MV98] S. M. Metev and V. P. Veiko. Laser Assisted Microtechnology.

Springer, Berlin, Germany, 2nd edition, 1998.
[Nik11] N. K. Nikolova. Microwave Imaging for Breast Cancer. IEEE

Microwave Magazine, 12(7):78–94, December 2011.
[NMdSC 07] A. Natarajan, M. Motani, B. de Silva, K. K. Yap, and K. C.

Chua. Investigating Network Architectures for Body Sensor
Networks. In HealthNet 2007, June 2007.
183
Bibliography
[OJG10] M. O’Halloran, E. Jones, and M. Glavin. Quasi-Multistatic

MIST Beamforming for the Early Detection of Breast
Cancer. IEEE Transactions on Biomedical Engineering,
57(4):pp.830–840, April 2010.
[oNiRPI98] International Commission on Non-ionizing Radiation Protec-

tion (ICNIRP). ICNIRP Guidelines for Limiting Exposure to
Time-varying Electric, Magnetic and Electromagnetic Fields
(up to 300 GHz). Health Physics, 74(4):494–522, April 1998.
[Poz] D. M. Pozar. Microwave Engineering. Hoboken: Wiley, 3rd

edition.
[QLD03] H. Qian, P. C. Loizou, and M. F. Dorman. Phone-Assistive

Devices Based on Bluetooth Technology for Cochlear Implant
Users. IEEE Transactions on Neural Systems Rehabilitation
Engineering, 11(3):282–287, 2003.
[Rei98] J. P. Reilly. Applied Bioelectricity: From Electrical Stimula-

tion to Electropathology. Springer, 1998.
[RFFea07] W. Rosamond, K. Flegal, G. Friday, and et al. Heart Disease

and Stroke Statistics-2007 Update: a Report from the Ameri-
can Heart Association Statistics Committee and Stroke Statis-
tics Subcommittee. Circulation, 115(5):69–171, 2007.
[Riv11] J. Rivera. UWB-Imaging-System zur Detektion von Wasser-

ansammlungen im menschlichen Körper. Diploma thesis, In-
stitut für Hochfrequenztechnik und Elektronik, Karlsruher In-
stitut für Technologie (KIT), January 2011.
[Roe99] R. B. Roemer. Engineering Aspects of Hyperthermia Therapy.

Annual Review of Biomedical Engineering, 1:347–376, 1999.
[Sam12] K. Chu Sam. Aufbau eines Messsystems zur Charakter-

isierung von an Gewebe angepassten Antennen. Bache-
lor thesis, Institut für Hochfrequenztechnik und Elektronik,
184
Bibliography
Karlsruher Institute für Technologie (KIT), Karlsruhe, March

2012.
[Sch10] S. Scherr. UWB-Pseudo-Noise-Radaruntersuchungen für

medizinische Anwendungen. Diploma thesis, Institut für
Hochfrequenztechnik und Elektronik, Karlsruher Institut für
Technologie (KIT), Karlsruhe, October 2010.
[SCK07] G. Shobha, R. R. Chittal, and K. Kumar. Medical Applications

of Wireless Networks. In Second International Conference
on Systems and Networks Communications (ICSNC), August
2007.
[SDMS05] C. J. Simon, D. E. Dupuy, and W. W. Mayo-Smith. Microwave

Ablation: Principles and Applications. Radiographics, 25:69–
83, October 2005.
[Ser09] Medical Device Radiocommunications Service. Federal Com-

munication Commission Std. CFR, Part 95.601-673 Subpart
E, Part 95.1201-1221 subpart I, 2009.
[SFC04] P. Soontornpipit, C. M. Furse, and You Chung Chung. Design

of Implantable Microstrip Antenna for Communication with
Medical Implants. IEEE Transactions on Microwave Theory
and Techniques, 52(8):1944–1951, August 2004.
[Shu88] B. Shuppert. Microstrip/Slotline Transitions: Modeling and

Experimental Investigation. IEEE Transactions on Microwave
Theory and Techniques, 36(8):1272–1282, August 1988.
[SKRC 11] M. L. Scarpello, D. Kurup, H. Rogier, D. Vande Ginste, F. Ax-

isa, J. Vanfleteren, W. Joseph, L. Martens, and G. Vermeeren.
Design of an Implantable Slot Dipole Conformal Flexible An-
tenna for Biomedical Applications. IEEE Transactions on An-
tennas and Propagation, 59(10):3556–3564, October 2011.
[Smi75] Glenn S. Smith. A Comparison of Electrically Short Bare and

Insulated Probes for Measuring the Local Radio Frequency
185
Bibliography
Electric Field in Biological Systems. IEEE Transactions

on Biomedical Engineering, BME-22(6):477–483, November
1975.
[SSC94] B. M. Steinhaus, R. E. Smith, and P. Crosby. The Role of

Telecommunications in Future Implantable Device Systems.
In the 16th Annual International Conference of the IEEE En-
gineering in Medicine and Biology Society, Baltimore, 1994.
[Ste23] H.E. Stewart. Diathermy and Its Application to Pneumonia.

P. B. Hoeber, inc., 1923.
[SW05] W. Sörgel and W. Wiesbeck. Influence of the Antennas on the

Ultra-Wideband Transmission. EURASIP Journal on Applied
Signal Processing, 2005:296–305, January 2005.
[TDGL11] K. F. Tong, A. Dufour, L. Ge, and K. M. Luk. Beamscanning

Probe Antennas for Deep Brain Stimulation. In Proceedings
of the 5th European Conference on Antennas and Propagation
(EUCAP), Rom, April 2011.
[TP08] H. Trefna and M. Persson. Antenna Array Design for Brain

Monitoring. In Antennas and Propagation Society Interna-
tional Symposium AP-S, San Diego, July 2008.
[TRVdVG95] T. N. Tulasidas, G. S. V. Raghavan, F. Van de Voort, and R. Gi-

rard. Dielectric Properties of Grapes and Sugar Solutions at
2.45 GHz. Journal of the Microwave Power and Electromag-
netic Energy, 33(2):117–123, 1995.
[USNLoM] National Institutes of Health United States National Library of

Medicine. The Visible Human Project MRI Scans.
[Vat10] E. Vatamaniuc. UWB-Systemkonzeptuntersuchungen zur De-

tektion von Wasseransammlungen im menschlichen Körper.
Diploma thesis, Institut für Hochfrequenztechnik und Elek-
tronik, Karlsruher Institut für Technologie (KIT), 2010.
186
Bibliography
[VC09] L. Venkatraghavan and V. Chinnapa. Non-Cardiac Im-

plantable Electrical Devices: Brief Review and Implications
for Anesthesiologists. Can. J. Anesth., 56:320–326, 2009.
[VRK06] A. Vander Vorst, A. Rosen, and Y. Kotsuka. RF/Microwave

Interaction with Biological Tissues. Wiley-IEEE Press, 2006.
[WDJC 96] J. A. Warren, R. D. Dreher, R. V. Jaworski, J. J. Putzke, and

R. J. Russie. Implantable Cardioverter Defibrillators. Pro-
ceedings of the IEEE, 84(3):468–479, March 1996.
[WSTI09] X. Wei, K. Saito, M. Takahashi, and K. Ito. Performances

of an Implanted Cavity Slot Antenna Embedded in the Hu-
man Arm. IEEE Transactions on Antennas and Propagation,
57(4):894–899, April 2009.
[XGXC 06] Y. Xie, B. Guo, L. Xu, J. Li, and P. Stoica. Multistatic Adap-
tive Microwave Imaging for Early Breast Cancer Detection.
IEEE Transactions on Biomedical Engineering, 53(8):1647–
1657, August 2006.
[Yan11] J. Yan. Entwurf und Charakterisierung von UWB-Antennen

zur Abstrahlung in den menschlichen Körper für medizinische
Diagnostik. Bachelor thesis, Institut für Hochfrequenztech-
nik und Elektronik, Karlsruher Institut für Technologie (KIT),
Karlsruhe, November 2011.
[You12] W. You. Development of a Miniaturized Antenna for Wire-

less Communication of Implantable Sensors. Bachelor thesis,
Institut für Hochfrequenztechnik und Elektronik, Karlsruher
Institut für Technologie (KIT), Karlsruhe, November 2012.
[ZL07] K. Zhang and D. Li. Electromagnetic Theory for Microwaves

and Optoelectronics. Springer, Tsinghua University, Beijing,
2nd edition, 2007.
187
Own publications
Journal publications
[1] X. Li, G. Adamiuk, E. Pancera, and T. Zwick. Physics-Based Propaga-
tion Characterisations of UWB Signals for the Urine Detection in Human
Bladder. Int. J. Ultra Wideband Communications and Systems, 2(2):94–
103, December 2011.
[2] X. Li, M. Jalilvand, Y. L. Sit, and T. Zwick. A Compact Double-
Layer On-Body Matched Bowtie Antenna for Medical Diagnostics. IEEE
Transactions on Antennas and Propagation (accepted), 2013.
[3] X. Li, Y. L. Sit, L. Zwirello, and T. Zwick. A Miniaturized UWB
Stepped-Slot Antenna for Medical Diagnostic Imaging. Microwave and
Optical Technology Letters, 55(1):105–109, January 2013.
Conference publications
[4] X. Li, M. Janson, G. Adamiuk, C. Heine, and T. Zwick. A 2D Ultra-
Wideband Indoor Imaging System with Dual-Orthogonal Polarized An-
tenna Array. In COST 2100, Athens, February 2009.
[5] J. Schmid, E. Pancera, X. Li, L. Niestoruk, S. Lamparth, W. Stork,
and T. Zwick. Ultra-Wideband Detection System for Water Accumu-
lations in the Human Body. In 11 th International Congress of the
IUPESM-Medical Physics and Biomedical Engineering World Congress
- WC2009, Munich, Germany, September 2009.
[6] H. Wu, X. Li, and T. Zwick. Motion Compensation of Automotive SAR
for Parking Lot Detection,. In International Radar Symposium (IRS),
Hamburg, September 2009.
189
Own publications
[7] X. Li, G. Adamiuk, M. Janson, and T. Zwick. Polarization Diversity

in Ultra-Wideband Imaging Systems. In International Conference on
Ultra Wideband- ICUWB 2010, Nanjing, September 2010.
[8] X. Li, E. Pancera, L. Zwirello, H. Wu, and T. Zwick. Detection of Water

Accumulation in the Human Bladder with Ultra Wideband Radar. In
COST 2100, Athens, February 2010.
[9] X. Li, E. Pancera, L. Zwirello, H. Wu, and T. Zwick. Ultra Wideband

Radar for Water Detection in the Human Body. In German Microwave
Conference GeMiC, Berlin, March 2010.
[10] E. Pancera, X. Li, T. Zwick, and W. Wiesbeck. Fidelity Criterion for

UWB Medical Diagnostic. In IEEE International Conference on Ultra-
Wideband, Nanjing, September 2010.
[11] E. Pancera, X. Li, T. Zwick, and W. Wiesbeck. UWB Antennas for Med-
ical Diagnostics Purposes. In IEEE Antennas and Propagation Society
International Symposium, Toronto, Juli 2010.
[12] E. Pancera, X. Li, L. Zwirello, and T. Zwick. Performance of Ul-

tra Wideband Antennas for Monitoring Water Accumulation in Hu-
man Bodies. In European Conference on Antennas and Propagation,
Barcelona, April 2010.
[13] J. Schmid, E. Pancera, L. Niestoruk, X. Li, S. Lamparth, T. Zwick, and

W. Stork. Ultra-Wideband Signals for the Detection of Water Accumu-
lations in the Human Body. In International Conference on Bio-inspired
Systems and Signal Processing BIOSIGNALS, Valencia, January 2010.
[14] H. Wu, X. Li, and T. Zwick. Motion Compensation for Landmine De-
tecting Vehicle-borne SAR. In International Radar Symposium (IRS),
Vilnius, June 2010.
[15] L. Zwirello, E. Pancera, X. Li, and T. Zwick. Ultra-Wideband Localiza-

tion Systems - Possibilities, Challenges and Solutions. In COST ic0803,
Lausanne, Switzerland, November 2010.
190
Own publications
[16] G. Adamiuk, C. Rusch, X. Li, M. Janson, and T. Zwick. Dual-Polarized

UWB Antenna for High-Resolution-Imaging-Systems. In IEEE Inter-
national Workshop on Antenna Technology (iWAT), Hong Kong, March
2011.
[17] M. Jalilvand, X. Li, T. Zwick, W. Wiesbeck, and E. Pancera. Hemor-

rhagic Stroke Detection Via UWB Medical Imaging. In Europan Con-
ference on Antennas and Propagation EuCAP, Rome, April 2011.
[18] M. Jalilvand, X. Li, T. Zwick, W. Wiesbeck, and E. Pancera. UWB

SAR Medical Imaging via Broadband Minimum Distortionless Re-
sponse MVDR Algorithm. In Europan Conference on Antennas and
Propagation (EuCAP), Rome, April 2011.
[19] M. Jalilvand, E. Pancera, X. Li, T. Zwick, and W. Wiesbeck. A Sparse

Synthetic-Aperture Based UWB Medical Imaging System. In German
Microwave Conference (GeMiC), Darmstasdt, March 2011.
[20] X. Li, M. Jalilvand, L. Zwirello, and T. Zwick. Array Configurations

of a UWB Near Field Imaging System for the Detection of Water Accu-
mulation in Human Body. In the European Radar Conference (EuRAD),
Manchester, October 2011.
[21] X. Li, M. Jalilvand, L. Zwirello, and T. Zwick. Synthetic Aperture-

Based UWB Imaging System for Detection of Urine Accumulation in
Human Bladder. In IEEE International conference on Ultra Wideband
(ICUWB), Bologna, September 2011.
[22] X. Li, S. Scherr, L. Sit, E. Pancera, and T. Zwick. Performance Analysis

of Various UWB Radar Approaches for Medical Diagnostics. In Pro-
ceedings of the IEEE International Microwave Symposium, Baltimore,
USA, June 2011.
[23] X. Li, S. Scherr, H. Wu, E. Pancera, and T. Zwick. Feasibility Study of

an Ultra Wideband Pseudo-Noise-Radar for Medical Applications. In
Proceedings of the IEEE European Conference on Antennas and Prop-
agation, Rom, Italy, April 2011.
191
Own publications
[24] E. Pancera, H. Barba, X. Li, M. Jalilvand, and T. Zwick. Uwb antennas

optimization for in-body radiation. In German Microwave Conference
(GeMiC), Darmstadt, Germany, March 2011.
[25] E. Pancera, X. Li, M. Jalilvand, T. Zwick, and W. Wiesbeck. Uwb
medical diagnostic: In-body transmission modeling and applications.
In European Conference on Antennas and Propagation (EuCAP), Rom,
April 2011.
[26] J. Pontes, G. Adamiuk, S. Beer, L. Zwirello, X. Li, and T. Zwick. Novel
Design Method for Frequency Agile Beam Scanning Antenna Arrays.
In IEEE International Workshop on Antenna Technology (iWAT), 2011.
[27] S. Scherr, X. Li, S. Ayhan, and T. Zwick. A Polarity Correlator in a
UWB-PN-Radar for the Detection of Multiple Targets. In European
Radar Conference (EuRAD), Manchester, October 2011.
[28] H. Wu, L. Zwirello, X. Li, L. Reichardt, and T. Zwick. Motion Compen-
sation with One-axis Gyroscope and Two-axis Accelerometer for Auto-
motive SAR. In German Microwave Conference GeMiC, Darmstadt,
March 2011.
[29] L. Zwirello, X. Li, E. Pancera, and T. Zwick. An Ultra-Wideband Com-
munication System with Localization Capabilities - Hardware Realiza-
tion of HF-Frontend Modules. In 6th Management Comitee/Working
Group Meeting and Workshop COSTic0803, Perugia, Italy, April 2011.
[30] M. Jalilvand, X. Li, and T. Zwick. Improved Minimum Variance Pro-
cessing for UWB Medical Imaging Applications. In European Radar
Conference, Amsterdam, October 2012.
[31] X. Li, M. Jalilvand, J. Yan, and T. Zwick. Compact Double-Elliptical
Slot-Antenna for Medical Applications. In IEEE European Conference
on Antennas and Propagation, Prague, March 2012.
[32] X. Li, K. Chu Sam, W. Wiesbeck, and T. Zwick. A Planar Near-Field
Measurement System of UWB Antennas for Medical Diagnostics. In In-
ternational Conference on Electromagnetics in Advanced Applications,
Cape Town, September 2012.
192
Own publications
[33] X. Li, Y. L. Sit, L. Zwirello, M. Jalilvand, and T. Zwick. An E-Field

Probe Based Near-Field Measurement System for On- and In-Body An-
tennas. In European Microwave Conference (EuMC), Amsterdam, Oc-
tober 2012.
[34] X. Li, L. Zwirello, M. Jalilvand, and T. Zwick. Design and Near-field
Characterization of a Planar On-body UWB Slot-Antenna for Stroke
Detection. In IEEE International Workshop on Antenna Technology
(iWAT), Tucson, USA, March 2012.
[35] L. Zwirello, C. Heine, X. Li, T. Schipper, and T. Zwick. SNR Perfor-
mance Verification of Different UWB Receiver Architectures. In Euro-
pean Microwave Conference (EuMC), Amsterdam, October 2012.
[36] L. Zwirello, L. Reichardt, X. Li, and T. Zwick. Impact of the An-
tenna Impulse Response on Accuracy of Impulse-Based Localization
Systems. In 6th European Conference on Antennas and Propagation,
Prague, March 2012.
[37] M. Jalilvand, X. Li, and T. Zwick. A Model Approach to the Analytical
Analysis of Stroke Detection using UWB Radar. In Europan Conference
on Antennas and Propagation (EuCAP), Gothenburg, April 2013.
[38] M. Jalilvand, X. Li, L. Zwirello, and T. Zwick. Sensititvity Analysis of
UWB Minimum Variance Beamformer for Medical Imaging. In Euro-
pean Microwave Conference, Nuernberg, Germany, October 2013.
[39] X. Li, M. Jalilvand, W.You, W. Wiesbeck, and T. Zwick. An Implantable
Stripline-Fed Slot Antenna for Biomedical Applications. In IET Inter-
national Radar Conference, Xi An, April 2013.
[40] X. Li, W. Wiesbeck, and T. Zwick. Design considerations for UWB
antennas. In 7th European Conference on Antennas and Propagation
(EuCAP), Gothenburg, April 2013.
[41] Y. L. Sit, X. Li, L. Reichardt, H. Liu, R. Liu, and T. Zwick. A Planar
Dual-mode UWB Antenna for an Indoor MIMO Communication Sys-
tem. In International Workshop on Antenna Technology (iWAT), Karl-
sruhe, Germany, March 2013.
193
Own publications
[42] Y. L. Sit, L. Reichardt, X. Li, H. Liu, R. Liu, and T. Zwick. Dual-

Orthogonal Mode Planar Antenna for Indoor UWB MIMO Communi-
cation Systems. In IEEE International Symposium on Antennas and
Propagation (IEEE-APS), Orlando Florida, USA, July 2013.
194
In this work new concepts of body-matched antennas

for microwave medical applications were developed. The
antennas are placed directly on the human body or in
the human body to strongly reduce the reflection at the
boundary of air and skin. Two types of antennas were de-
signed – on-body matched antennas for microwave medi-
cal diagnosis (i.e. detection of stroke, breast cancer, etc.)
and implantable antennas for data telemetry. The main
focus of this work is the optimization of the antenna radia-
tion performance for better detection of weak reflections
from the human body as well as antenna miniaturization
to allow a portable microwave medical device. Since the
antennas operate in the near-field, novel measurement
verification systems were also designed to verify the ra-
diation performance of the antennas. The applicability of
the on-body matched antennas for microwave imaging is
proved by means of a demonstrator system with an anten-
na array for the detection of hemorrhagic stroke. All these
contributions enable a portable medical diagnosis system
to complement the current medical applications with the
goal of providing more advanced healthcare systems.
Xuyang Li received his Bachelor degree from Tongji Uni-

versity in Shanghai in 2006 and his Diploma in electrical
engineering from Universität Karlsruhe (TH) in Karlsruhe
in 2009. Since July 2009 he worked toward his doctoral
degree at the Institut für Hochfrequenztechnik und Elek-
tronik at KIT in the field of ultra wideband technology. His
main research topics are the design of antennas and radar
systems for medical applications.
ISBN 978-3-7315-0147-3
ISSN 1868-4696
ISSN
ISBN XXXX-XXXX
978-3-7315-0147-3 9 783731 501473

Body Matched Antennas For Microwave Medical Applications: Xuyang Li

Uploaded by

Copyright:

Available Formats

Body Matched Antennas For Microwave Medical Applications: Xuyang Li

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Body Matched Antennas For Microwave Medical Applications: Xuyang Li

Uploaded by

Copyright:

Available Formats

Karlsruher Forschungsberichte aus dem Band

Institut für Hochfrequenztechnik und Elektronik

Body Matched Antennas for

Body Matched Antennas for

Herausgeber: Prof. Dr.-Ing. Thomas Zwick

Karlsruher Institut für Technologie (KIT)

This document – excluding the cover – is licensed under the

Print on Demand 2014

Originally motivated by the ever increasing demand on the high data

Prof. Dr.-Ing. Thomas Zwick

Band 1 Daniel Kähny

Band 12 Friedhelm Rostan

Band 24 Dirk Didascalou

Band 36 Jürgen von Hagen

Band 49 Thomas Dreyer

Herausgeber: Prof. Dr.-Ing. Thomas Zwick

Die Bände sind unter www.ksp.kit.edu als PDF frei verfügbar

Band 55 Sandra Knörzer

Band 62 Jutta Kühn

Band 63 Małgorzata Janson

Zur Erlangung des akademischen Grades eines

von der Fakultät für

Tag der mündlichen Prüfung: 28. 05. 2013

The microwave technique is a currently very attractive technique for medi-

Die vorliegende Arbeit entstand während meiner Zeit als wissenschaftlicher

List of Symbols and Acronyms v

2. Analysis of microwave propagation in a multilayer human body 11

3. Characterization and measurement systems of body-matched

3.2. Measurement techniques for the characterization of body-matched

4. Implantable antennas for wireless communication of IMDs 65

5. On-body matched antennas for medical diagnosis 91

6. Body-matched antennas based microwave medical imaging 145

7. Summary and conclusions 169

A. Parameters of tissues based on Cole-Cole equation 175

Own publications 189

MICS Medical Implant Communication Service

c0 speed of light in vacuum ( 2; 997925 108 m/s)

ksp fraction of the resonant wavelength

B Magnetic flux density

J Conduction current density

"r;eff Effective relative permittivity

Mathematical notation and symbols

r sE Rotation of vector field

1.1. State-of-the-art microwave medical

Mammography and CT scan, using ionizing radiation, are techniques with

Figure 1.1.: Three major microwave medical applications.

years. Examples of such devices are implantable cardioverter defibrillators

Medical diagnosis using microwaves

scattering problem. Recently, the microwave holography in medical near-

Medical treatment using microwaves

Medical treatment using microwaves is based on using the heat generated by

1.2. Telemedicine using microwaves and its

abnormalities are detected, the doctor is immediately informed so that neces-

ization of these antennas. Therefore, the performance of the antennas (e.g.

1.3. Goal and organization of this work

This chapter provides the investigation of the propagation characteristics of

2.1. Microwave propagation in a lossy medium

" D "0 j"00 ; (2.1)

EE .x; y; z; t/ D EE .x; y; z/ e j!t ; (2.5)

c0 speed of light in vacuum ( 2; 997925 108 m/s)

where ! D 2f is the angular frequency and f is the frequency.

An open-ended E-field probe, fabricated from a 50 semi-rigid cable (5.6