Thesis Qiuyan Lund University

Low-Complexity Multiband and Beam-Reconfigurable Antennas for Beyond 5G
Communications
Liang, Qiuyan
2023
Link to publication
Citation for published version (APA):

Liang, Q. (2023). Low-Complexity Multiband and Beam-Reconfigurable Antennas for Beyond 5G
Communications. Electrical and Information Technology, Lund University.
Total number of authors:

1
General rights
Unless other specific re-use rights are stated the following general rights apply:
Copyright and moral rights for the publications made accessible in the public portal are retained by the authors
and/or other copyright owners and it is a condition of accessing publications that users recognise and abide by the
legal requirements associated with these rights.
• Users may download and print one copy of any publication from the public portal for the purpose of private study
or research.
• You may not further distribute the material or use it for any profit-making activity or commercial gain
• You may freely distribute the URL identifying the publication in the public portal
Read more about Creative commons licenses: https://creativecommons.org/licenses/

Take down policy
If you believe that this document breaches copyright please contact us providing details, and we will remove
access to the work immediately and investigate your claim.
LUNDUNI
VERSI
TY
PO Box117
22100L und
+4646-2220000
Low-Complexity Multiband and Beam-
Reconfigurable Antennas for Beyond 5G
Communications
Qiuyan Liang
Lund 2023
© Qiuyan Liang, 2023
“Low-Complexity Multiband and Beam-Reconfigurable Antennas for Beyond 5G
Communications”
Published articles have been reprinted with permission
from the respective copyright holder.
Series of licentiate and doctoral theses

No. 153
ISSN 1654-790X153
ISBN 978-91-8039-564-9 (print)
ISBN 978-91-8039-565-6 (pdf)
This thesis is produced by using the Microsoft Word

Printed in Sweden by Tryckeriet i E-huset, Lund.
Department of Electrical and Information Technology

Lund University, Box 118, SE-221 00 Lund, Sweden
To my dear family
Popular Science
Wireless communication networks have been widely developed and utilized for the
past several decades. Nowadays the use of wireless internet has permeated many
aspects of our lives, including study, work, travel, and entertainment.
Correspondingly, the content of wireless communication has been enriched from
voice in the early days to text, pictures, videos, and live streaming today, providing
many conveniences to our lives. These ever-more-complex contents transmitted in
wireless communication networks are essentially in the form of data, and ever-
increasing data rates are needed to support the content delivery.
The communication of information requires a transmitter and a receiver. For
example, if someone transmits information by speech or body language, then the
information can be received by another person’s ears or eyes. Similarly, wireless
communication also requires a transmitter and a receiver, with an antenna being a
critical component in each. At the transmitter, the antenna converts guided signals
that contain the information to unguided signals in the form of electromagnetic
waves, which propagate towards the antenna at the receiver to be converted back to
guided signals. The received signals are then processed to recover the transmitted
information. Following the demand for high data rates, the antennas in wireless
communications are required to provide more features and higher performance.
Utilizing larger absolute frequency bandwidth for the wireless communication is
one common approach to provide higher data rate. Therefore, due to the availability
of more unoccupied frequency spectra at higher frequencies, wireless
communication is expanding towards the higher frequencies, from sub-6 gigahertz
(GHz) bands to millimeter-wave (mm-wave) bands in 5G, and in the future, to
terahertz (THz) bands. However, the smaller signal wavelengths at higher
frequencies incur higher propagation losses. In fact, this effect is already noticeable
within the sub-6GHz bands, where sub-1GHz bands are typically favoured for
providing larger coverage areas. Moreover, the electromagnetic waves in mm-wave
bands and above experience more noticeable loss due to the atmospheric absorption
and obstacles in the environment than in sub-6GHz bands. Therefore, to establish
more reliable communication, the signal energy needs to be focused in the desired
direction, which requires the antenna to generate a beampattern with higher gain in
v
vi Popular Science
this direction. However, the higher gain is achieved at the cost of a smaller
beamwidth (i.e., angular range in which the gain is increased). Since the relative
direction between transmitter and receiver is often not fixed, the smaller beamwidth
implies the need for beam steering, i.e., the beampatterns of the high gain transmitter
and receiver antennas need to be steered towards each other.
Due to higher losses in higher frequency bands, smaller cells and hence many more
base stations are needed to maintain the same coverage area of the wireless network.
Traditionally, beam-steerable high gain antennas are provided by phased array
antennas. However, an array antenna requires many antenna elements connected by
a feeding network, which is complicated and hence expensive to design and
implement. The feeding network is also bulky, and it tends to increase the loss of
energy. Therefore, beam-steerable antennas with low-complexity structures and
high space utilization are highly desirable to improve system architecture and save
installation resources. In this context, antennas based on partially reflective surface
(PRS) can provide high gain with simple structure and low cost, which have good
potential for application in future base stations. However, PRS antenna needs to be
further developed to realize its full potential, despite being proposed more than half
a century ago. This is because, until recently, it had not attracted much research
interest. The first topic of this thesis is on enhancing two aspects of PRS antenna
design, namely beam steering capability and shared-aperture antenna design.
Existing PRS antennas mainly use standard reconfigurable approaches for beam
steering, e.g., using reconfigurable loads on the PRSs’ unit cells. The beam can be
pointed towards different directions in different reconfigurable states. However,
current beam-reconfigurable PRS antennas suffer from narrow coverage range,
distorted beam shape and considerable gain variations over the beams in different
directions, which may lead to degradation of communication quality. On the other
hand, shared-aperture antenna design with PRS aims to integrate antennas working
at different frequency bands into a shared space to provide high space utilization.
However, existing design schemes suffer from inflexible frequency ratio (of the
bands) and bulky antenna structures. Therefore, the main part of this thesis deals
with the research question on how to solve the challenges encountered by PRS
antennas with respect to beam reconfigurability and shared aperture designs, while
maintaining low-complexity structures.
Besides base stations with fix installations, wireless communication networks
consist of a large number of user devices, including mobile terminals. With the
rollout of 5G, terminal antennas are being developed to cover both existing and new
5G frequency bands, spanning both sub-6GHz bands and mm-wave bands. To save
antenna implementation space, it is desirable to co-design and even co-locate these
antennas. However, existing co-design approaches suffer from complex structure of
the mm-wave antenna and low space utilization. Therefore, the other part of this
thesis is about solving the research question on how to co-design the sub-6GHz and
mm-wave antennas for mobile terminals for compactness and low-complexity.
Abstract
Antennas with large frequency bandwidth, high gain, and beam steering capability
are very importance for future wireless communication systems. However, it is very
challenging to design antennas with low-complexity structures and high space
utilization that can achieve these desired features. Base stations with fixed
installations and non-stationary mobile terminals play critical roles in wireless
communication networks. The research in this thesis focuses on the low-complexity
multiband and beam-reconfigurable antenna design for the two applications.
The first topic of this thesis is about partially reflective surfaces (PRS) antennas,
which have significant potential for application in future base stations. In the first
part (Part I) of this thesis, existing PRS antennas and challenges to be addressed in
PRS antenna design are introduced with respect to beam steering capability and
shared-aperture antenna design. To give a clearer insight into the working principle
of PRS antenna as well as several observed phenomena involving PRS, the existing
theory for PRS antenna is extended. Ray-tracing models as well as the theory of
PRS unit cells are utilized to explain the operation of PRS antennas that generate
broadside and/or deflected beams. Using these tools, two practical PRS antennas are
designed for beam reconfigurability and shared-aperture implementation,
respectively. In the first work, we compare the beam deflection capability of
different types of PRSs using ray-tracing analysis, with the aim of providing a
guideline for selecting a suitable PRS type that would yield a larger beam steering
range. In addition, the role of a feeding source in enhancing beam deflection of PRS
is explained using ray-tracing analysis, and a beam-reconfigurable feeding source
with low-complexity structure is presented. In the second work, we derive the
frequency ratio gap for traditional dual-band shared-aperture Fabry-Pérot cavity
(DS-FPC) antennas with single-layer PRS and subsequently present a shared-
aperture antenna design method with flexible frequency ratio to fill the gap.
The second topic of this thesis is on the co-design of mobile terminal antennas that
can cover a wide range of frequency bands. Such antennas are of significant current
interest due to the current trend of utilizing higher frequency spectra in wireless
communication. A co-designed millimeter-wave (mm-wave) and sub-6GHz
antenna system is conceived, where the mm-wave antennas that offer reconfigurable
vii
viii Abstract
beams for beam steering are integrated into the capacitive coupling elements (CCEs)
of the sub-6GHz antenna. Such an implementation aims to achieve a compact and
low-complexity antenna structure. In addition, several techniques have been
investigated by simulation to achieve further performance improvements in the
proposed antenna system with respect to mm-wave antenna gain and bandwidth as
well as sub-6GHz antenna tunability and bandwidth.
Preface
This doctoral thesis consists of two parts. The first part (Part I) gives an overview
of the research during the three and half years as a doctoral student at the department
of Electrical and Information Technology (EIT), Lund University, Sweden. The
second part (Part II) is composed of four papers that constitute my main scientific
work, including:
Paper I
Qiuyan Liang and Buon Kiong Lau, “Comparison of capacitive and inductive
partially reflective surface antenna using ray-tracing,” in Proc. 16th Europ. Conf.
Antennas Propag. (EuCAP’2022), Madrid, Spain, Mar. 27- Apr. 1, 2022.
Paper II
Qiuyan Liang, Buon Kiong Lau, and Gaonan Zhou, “Beam-reconfigurable antenna
with inductive partially reflective surface and parasitic elements,” IEEE Trans.
Antenna Propag., 2023, Manuscript finished.
Paper III
Qiuyan Liang, Buon Kiong Lau, and Gaonan Zhou, “Dual-band shared-aperture
antenna with single-layer partially reflecting surface,” IEEE Trans. Antenna
Propag., 2023, Submitted.
Paper IV
Qiuyan Liang, Hanieh Aliakbari, and Buon Kiong Lau, “Co-designed millimeter-
wave and sub-6GHz antenna for 5G smartphones,” IEEE Antennas Wireless Propag.
Lett., vol. 21, no. 10, pp. 1995-1999, Oct. 2022.
During my doctoral studies, I had also contributed to the following publications,

which are not included in this thesis:
ix
x Preface
Conference papers
 Qiuyan Liang and Buon Kiong Lau, “Beam manipulation using characteristic
mode analysis for switchable beam patch antenna,” in Proc. 2020 IEEE Int.
Symp. Antennas Propag., Montreal, Canada, Jul. 5-10, 2020.
 Qiuyan Liang and Buon Kiong Lau, “Beam reconfigurable reflective
metasurface for indoor wireless communications,” in Proc. 2021 IEEE Int.
Symp. Antennas Propag., Singapore, Singapore, Dec. 4-10, 2021.
 Qiuyan Liang and Buon Kiong Lau, “Analysis of partially reflective surface
antenna with different reflection magnitudes using ray-tracing,” in Proc. 2022
IEEE Int. Symp. Antennas Propag., Denver, CO, Jul. 10-15, 2022.
Journal papers
 Qiuyan Liang, Baohua Sun, and Gaonan Zhou, “Miniaturization of Rotman
lens using array port extension,” IEEE Antennas Wireless Propag. Lett., Early
Access, 2022.
 Qiuyan Liang, Baohua Sun, and Gaonan Zhou, “A dual-band shared-aperture
parasitic array radiator antenna for WLAN applications,” Int. J. RF Microw.
Comp.-Aided Eng., vol. 31, no. 11, 2021.
 Gaonan Zhou, Baohua Sun, Qiuyan Liang, Yuhang Yang, and Jianghong Lan,
“Beam-deflection short backfire antenna using phase-modulated metasurface,”
IEEE Trans. Antenna Propag., vol. 68, no. 1, pp. 546-551, Jan. 2020.
Acknowledgements
The three and a half years of my PhD study at Lund University have been the most
enriching time of my life. I have gained so much, and I am so grateful that I have
met so many wonderful people during this experience.
First and foremost, I would like to express my sincere gratitude to my supervisor
Prof. Buon Kiong Lau. Thank you for offering me the opportunity to start this
journey at Lund University, for your time and enthusiasm in the discussions, and
for your generosity in providing me with plenty of valuable guidance in my research.
Your rigorous work attitude is a valuable quality that I am and will continue to
pursue in my career. Your positive attitude and enthusiasm inspire people around
you.
I would like to thank Prof. Mats Gustafsson, my co-supervisor and Prof. Daniel
Sjöberg. Thank you for organising a variety of electromagnetics-related seminars
that have provided me with access to further knowledge and expertise in my field
of study. Thank you for your valuable advice on my individual study plan during
my PhD study.
I would like to thank my colleague Hanieh Aliakbari for collaborating with me in
some of my research work. I am fortunate to have you as a friend and thank you for
all your support in my research, courses, teaching, and daily life.
I also want to thank all my former and current colleagues at EIT for the great
working environment we have created together. More specifically, I thank all the
administrative staff for providing strong and essential support to my PhD study, and
all my colleagues in Communications Engineering Division for your company in
the division meetings, activities and “fika”.
I would also like to thank these colleagues for their help in the writing of my thesis:
Xuhong Li, Leif Wilhelmsson and Siyu Tan.
During the past three and a half years, I have made many friends and have been
fortunate to spend time with them: Hanieh Aliakbari, Liying Nie, Shang Xiang,
Sevda Özdemir, Kranti Kumar Katare, Irfan Yousaf, Kentaro Murata, Ahmed El
Yousfi, Ben Nel, Xuhong Li, Tao Qin, Yuyan Cao, Guoda Tian, Siyu Tan, Xuesong
xi
xii Acknowledgements
Cai, Yanan Wu, Haipeng Li, Jie Ding, Siyuan Cang, Xiaotong Tu, Jing Yang, Hui
Zhu, and Xiaoya Li. Your company has brought me lots of much needed warmth.
I would like to express my deepest gratefulness to my family: my parents, my sisters
and brothers, and my lovely nephews and niece, for your support and remote
companionship, and for filling the whole family with love and hope.
Most importantly, I would like to thank my husband Gaonan Zhou. You are always
warm, patient, and optimistic. Your firm companionship has been the biggest
support for me to get to where I am today. It is my greatest fortune to have you in
my life.
Finally, I thank all the strangers in the big world around me; everyone’s own world
makes up the whole world we live in. May the world be full of peace and love.
Qiuyan Liang
Lund, Feb. 2023
List of Acronyms and
Abbreviations
1-D One-Dimensional
5G Fifth-Generation
5G NR 5G New Radio
6G Sixth-Generation
AiP Antenna-in-Package
AMC Artificial Magnetic Conductor
CCE Capacitive Coupling Elements
DC Direct Current
DS-FPC Dual-Band Fabry-Pérot Cavity
FEM Finite Element Method
FPC Fabry-Pérot Cavity
GNSS Global Navigation Satellite System
HB High Band
ISM Industrial Scientific Medical
LB Low Band
LTE Long-Term Evolution
MIMO Multiple-Input Multiple-Output
NFC Near-Field Communication
PBC Periodic Boundary Condition
xiii
xiv List of Acronyms and Abbreviations
PCB Printed Circuit Board
PEC Perfect Electric Conductor
PIFA Printed Inverted-F Antennas
PRS Partially Reflective Surface
Mm-Wave Millimeter-Wave
SBA Short Backfire Antenna
SLL Sidelobe Level
UWB Ultrawide Band
Wi-Fi Wireless Fidelity
WLAN Wireless Local Area Network
Contents
Popular Science.......................................................................................................v
Abstract ................................................................................................................ vii
Preface ................................................................................................................... ix
Acknowledgements ............................................................................................... xi
List of Acronyms and Abbreviations ................................................................ xiii
Part I Introduction ......................................................................................1
1 Motivation and Outline ....................................................................................3
1.1 Overview of Antenna Development Trends ..............................................3
1.2 PRS Antennas for Base Stations and Design Challenges..........................8
1.3 Co-Designed Antennas for Mobile Terminals and Design Challenges...13
1.4 Research Questions and Thesis Goals .....................................................15
1.5 Thesis Structure .......................................................................................17
2 PRS Antenna Design for Base Stations ........................................................19
2.1 Working Principle of Conventional PRS Antenna ..................................19
2.2 PRS Properties ........................................................................................24
2.3 Beam-Reconfigurable PRS Antenna Design ...........................................31
2.4 Shared-Aperture Antenna Design Using PRS .........................................36
3 Co-Designed Antenna for Mobile Terminals ...............................................41
3.1 Working Principle of Co-Designed Antennas .........................................41
3.2 Investigation of Further Performance Improvements .............................43
3.3 Effects of Practical Design Considerations .............................................46
4 Conclusion and Outlook ................................................................................49
4.1 Research Contributions ...........................................................................49
4.2 General Conclusions ...............................................................................51
xvi Contents
4.3 Future Research .......................................................................................52
Part II Included Papers .......................................................................63
Comparison of Capacitive and Inductive Partially Reflective Surface Antenna
Using Ray-Tracing................................................................................................67
1 Introduction ...............................................................................................69
2 Derivation of Phase Delay Distribution ....................................................70
3 Phase Delay Distributions of Different PRSs ...........................................73
4 Conclusion ................................................................................................79
5 Acknowledgment ......................................................................................79
Beam-Reconfigurable Antenna with Inductive Partially Reflective Surface
and Parasitic Elements .........................................................................................83
1 Introduction ...............................................................................................85
2 Ray-Tracing Analysis of Nonuniform PRS ..............................................87
3 Beam-Reconfigurable PRS and Feeding Source .......................................96
4 Reconfigurable PRS Antenna Design .......................................................99
5 Simulated and Measured Results ............................................................105
6 Conclusion ..............................................................................................108
Dual-Band Shared-Aperture Antenna with Single-Layer Partially Reflecting
Surface .................................................................................................................115
1 Introduction .............................................................................................117
2 Frequency Ratio Analysis For DS-FPC Antennas ..................................119
3 Proposed Dual-Band Shared-Aperture Antenna .....................................123
4 Measurement Verification .......................................................................133
5 Conclusion ..............................................................................................137
Co-Designed Millimeter-Wave and Sub-6GHz Antenna for 5G Smartphones
..............................................................................................................................145
1 Introduction .............................................................................................147
2 Co-Designed Antenna System ................................................................148
3 Simulated and Measured Results ............................................................152
4 Conclusion ..............................................................................................157
5 Acknowledgement...................................................................................157
Part I
Introduction
1 Motivation and Outline
Nowadays wireless communication is ubiquitous and has become an indispensable
part of our daily lives. We have become accustomed to viewing the latest news,
watching videos, participating in online meetings, etc. on our wireless devices. With
the widespread adoption of user devices and rich multimedia content, the demand
for higher data rates is also increasing rapidly. As a critical component in wireless
communication systems, antennas play the roles of transmitting and receiving
signals. To facilitate higher data rates, the antenna systems are required to provide
more features and higher performance.
In this chapter, we first provide an overview of the development trends of antennas
in wireless communication and several prevailing antenna design approaches to
support these trends. Then, we introduce the current developments and challenges
in antenna design for applications in base stations and mobile terminals, on which
the motivation for the research in this thesis is formed. The research questions and
thesis goals are then presented, alongside with the thesis structure.
1.1 Overview of Antenna Development Trends

With the rapid growth in both the number of user devices and the data rates, more
advanced wireless communication networks are needed. The deployment of fifth-
generation (5G) networks and the increasing volume of research in future sixth-
generation (6G) networks are aimed at providing enhanced wireless communication
capabilities and novel features, which require higher data rate and higher precision
in terms of latency, synchronicity, reliability, and localization [1], [2].
As shown in Fig. 1.1, smaller cells and hence many more base stations are
indispensable components in the architecture of future wireless communication
networks that provide continuous connectivity, particularly at higher frequency
bands [3]. Consequently, a substantial number of antennas will be needed for base
3
4 Introduction
Figure 1.1: Many more base stations in future wireless communication networks.
stations in the deployment of future wireless communication networks [4].

Therefore, the antennas need to be designed using more advanced techniques to
improve system architecture and save resources. In the following, four
representative antenna development trends for future communication systems and
the corresponding antenna design techniques to meet these trends are summarized:
1. Wide/multi-band operation: The 5G New Radio (5G NR) air interface has been
developed to meet the requirements of 5G networks. The air interface covers the
wide range of frequency bands from sub-1GHz bands to millimeter-wave (mm-
wave) bands [5]. Accordingly, the antennas need to be designed for wideband
operation, to minimize the number of antennas needed for frequency coverage.
Ultrawide band (UWB) antennas can be designed to operate over a wide range of
frequency bands. Some UWB antennas, including Vivaldi antenna and bi-cone-
shaped antenna, have been designed for an ultrawide range of the 5G frequency
bands [6]. However, since the size of UWB antennas tends to be limited by the
wavelength at the lowest frequency of operation, they tend to be physically large
and unsuitable for applications with limited implementation space. In addition, a
continuous wide operating band is not needed or even allowed in many
applications. For example, 5G NR should not operate in the unlicensed industrial
scientific medical (ISM) bands (e.g., 0.9, 2.4, and 5.8GHz). Therefore, antennas
working in multiple non-contiguous frequency bands have also been proposed for
Chapter 1. Motivation and Outline 5
User devices User devices
Wide Narrow
beams beams
Base station Base station
(a) (b)
Figure 1.2: Beam distributions of base stations and user devices in (a) low and (b) high
frequency bands.
a variety of applications, such as wireless local area network (WLAN) [7], base
stations [8], and mobile phones [9].
2. Beam steering capabilities: Compared with the lower frequency bands (e.g.,
sub-1GHz), the electromagnetic waves in the higher frequency bands experience
more noticeable attenuation due to higher propagation losses; and in the case of
millimeter waves, high atmospheric and material absorptions in the environment
as well [10]. To compensate for the high propagation losses, the antennas
operating in higher frequency bands need to generate beams with higher gains1.
However, a higher gain is achieved at the cost of a narrower half-power
beamwidth 2 and thus the angular coverage range is reduced. Therefore, base
station and user device antennas in high frequency applications need to be capable
of beam steering to maintain the same coverage area as in low frequency bands,
as depicted in Fig. 1.2.
The high gain performance of an antenna is attributed to a large effective antenna
aperture and the beam steering capability typically relies on changing the phase
distributions of the signal across the antenna aperture [11]. A wide variety of
antennas have been developed for beam steering, including phased array antennas
and antennas based on passive beamforming feeding networks, reflectors, and
1 According to [77], gain (in a given direction) describes the ratio of the radiation intensity in a
given direction to the radiation intensity that would be produced if the power accepted by the antenna
were isotropically radiated.
2 According to [77], half-power beamwidth describes the angle between the two directions in which
the radiation intensity is one-half the maximum value in a radiation-pattern cut containing the direction
of the maximum of a lobe.
6 Introduction
lenses [12]. These beam steering antennas are introduced in the following, and
their characteristics in terms of beamforming flexibility, occupied space, structure
complexity, and economic cost are summarized.
Phased array, invented more than a century ago [13] is arguably the most common
type of beam steering antenna. By connecting the antenna elements with phase
shifters and other RF elements, the amplitude and phase distributions of the
phased array can be controlled flexibly for more diverse beampatterns, which is
a key technique to provide coverage in satellite communications [14]. However,
the large numbers of required phase shifters and RF components render the
antenna system highly complex, bulky, and expensive.
Passive beamforming feeding networks can also be used for antenna arrays to
achieve multiple fixed beams pre-steered to cover different angular ranges. They
can be classified into two categories according to their working principles: lens-
based [15] or circuit-based feeding networks [16]. Both feeding networks can
achieve multiple phase distributions along their output ports for antenna arrays to
generate beams in different directions without the use of active RF components.
However, the lens-based and circuit-based networks suffer from the problems of
large size and complex circuit architecture, respectively. For antenna arrays with
a large number of antenna elements, these problems are even more pronounced,
making integration into compact antenna systems difficult.
The reflector-based and lens-based antennas operate with quasi-optical principles,
where the beam steering is achieved by changing the position of the external
feeding sources relative to the reflectors or lenses [17], [18]. These two types of
antennas have rotationally symmetrical structures, which facilitate flexible beam
steering in both azimuth and elevation. However, due to the use of external
feeding sources and non-planar reflectors or lenses, these two types of antennas
tend to occupy a large implementation space, which may increase the installation
cost of the antenna systems. Recently, transmitarray/reflectarray antennas that are
based on planar metasurfaces have been proposed to replace the traditional bulky
reflector or lens [19], [20], thus simplifying the antenna structures.
3. Low complexity: Considering the increasing use of higher frequencies and the
need for more base stations for seamless connectivity, it can be anticipated that
an unprecedentedly large number of base station antennas will be deployed. To
keep the fabrication and installation costs down, antenna designs with minimal
structural complexity are desired. The requirement for low-complexity antenna
design can be addressed from several perspectives: the antenna structure, beam
steering techniques, fabrication process, and antenna installation.
The antenna structure can be simplified by minimizing the usage of feeding
networks. As mentioned earlier, the high gain performance and beam steering
capability of phased array antennas rely on sophisticated feeding networks to
provide the required amplitude and phase distributions, which brings high
structural complexity and high insertion losses. Several types of antennas based
on metasurfaces or parasitic elements can generate high gain beams without using
feeding networks, such as transmitarray and reflectarray antennas as mentioned
above, partially reflective surface (PRS) antenna [21], planar Yagi-Uda antenna
[22], and short backfire antenna [23]. By manipulating the propagation of waves
from the excited feeding source to some resonant structures (e.g., metasurfaces)
with specific transmission/reflection properties or some parasitic elements with
specific structural parameters, the phase distribution across the antenna aperture
can be rearranged. This allows the waves across the antenna aperture to achieve
in-phase superposition in a certain direction, to generate high gain beam in that
direction.
For antennas without feeding networks, beam steering capability can be achieved
by using reconfigurable approaches without altering the antenna construction. For
a metasurface-based antenna, the transmission/reflection properties of unit cells
and the amplitude and phase distributions across the antenna aperture can be
reconfigured by electrically controlled components, such as switches or diodes
[24], [25]. For an antenna array employing parasitic elements, the reconfigurable
states of the electrically controlled components also affect the equivalent
electrical dimensions of parasitic elements, providing an effect similar to phase
shifters [26].
The approaches of electrical reconfiguration and the usage of metasurfaces
provide significant flexibility for the design of low-complexity beam steering
antennas. To further simplify the antenna design, it is highly desirable for the
antennas to be fabricated and installed in simple and cost-effective ways. For
example, single-layer dielectric substrate is preferred to multi-layer dielectric
substrate in antenna design. Moreover, planar antennas are aesthetically
appealing and easy to install since the base stations are often mounted on the walls
or ceilings of buildings for indoor communication scenarios. In this context, beam
steering antennas based on single-layer metasurfaces have distinct advantages.
4. High space utilization: As more antennas will be deployed in future
communication networks for higher data rates, it can be foreseen that the
installation resources, such as the installation space, will be limited in practical
application scenarios. Moreover, antennas for multiple generations of wireless
communication systems, working in different frequency bands, need to coexist.
Therefore, these antennas need to be designed with high space utilization.
Improving antenna space utilization mostly involves two approaches. One
common approach is to minimize the volume occupied by individual antennas,
such as lowering the antenna profile (height). However, a trade-off between
compact antenna size and acceptable antenna performance is always needed. The
second, often highly regarded, approach is to integrate multiple antennas working
in different frequency bands in a shared aperture. The goal of this shared-aperture
8 Introduction
technology is to accommodate antennas of multiple frequency bands in a compact
manner while maintaining their respective performance.
Metasurface plays a significant role in both approaches for high space utilization.
Compared to a traditional perfect electric conductor (PEC) ground, the use of a
metasurface with proper reflection phase as the ground plane can enable a lower
antenna profile [27]. Moreover, because of its frequency-selective features, a
metasurface can be shared by antennas in different frequency bands through
providing the required transmission and reflection properties in each band [8].
In summary, according to the four antenna development trends, the antennas in
future wireless communication systems need to support more frequency bands and
have beam steering capabilities, especially at high frequencies, while maintaining
low-complexity structure and high space utilization. These requirements present
tough challenges to the design of such antennas. The utilization of metasurfaces as
well as the aforesaid antenna design methods, such as beam reconfiguration
approaches and shared-aperture antenna technology, provide significant potential in
addressing these requirements. This thesis focuses on antennas for base stations and
mobile terminals, which are critical parts of today’s wireless communication
networks.
1.2 PRS Antennas for Base Stations and Design

Challenges
As described in the last section, metasurfaces can be used to design planar antenna
structures with high gain and shared aperture. Among many types of metasurface-
based antennas, those that utilize PRS have attracted a lot of attention recently due
to the advantages of high gain, simple structure, and low cost. PRS antenna was first
proposed for gain enhancement in the classical paper by G. Trentini [21]. Because
the working principle is based on the Fabry-Pérot resonance condition [28], PRS
antenna is also known as Fabry-Pérot cavity (FPC) antenna.
Since its discovery, PRS antenna has been adopted for many applications, such as
CubeSat [29], WLAN [30], and mobile communications [31]. In particular, the high
gain performance of the PRS antenna is suitable for base station deployment at
higher frequencies, due to its low-cost and simple structure. The high gain of the
PRS antenna is achieved by manipulating the propagation of waves between the
feeding source (or ground plane) and the PRS. This operating principle allows the
antenna to function without a feeding network.
A PRS antenna consists of a feeding source, a PRS and a fully reflecting ground
plane. A conventional PRS antenna employs a uniform PRS placed at a specific
distance from the ground to generate a broadside beam, as shown in Fig. 1.3(a). By
(a) (b)
Figure 1.3: PRS antennas generating (a) broadside beam and (b) deflected beam.
modifying the PRS into a nonuniform phase-varying PRS, the generated beam can
be deflected from the broadside direction, as shown in Fig. 1.3(b). Recently, to
enable beam steering capability and high space utilization, beam-reconfigurable
PRS antennas and shared-aperture antennas with PRSs have been proposed,
following the development trends of antennas for wireless communication systems
as described in Section 1.1. In the first topic (main part) of this thesis, PRS antennas
addressing these two aspects are studied. The aim is to enhance the antennas
performance and solve the challenges encountered by the two types of PRS antennas.
1.2.1. Beam-reconfigurable PRS antennas

Typically, a PRS is composed of many unit cells with specific structures designed
to provide the required transmission and reflection properties. The waves from the
feeding source are reflected multiple times between the PRS and the fully reflecting
ground plane, with each incident wave at the PRS also producing a transmitted wave.
In this process, the multiple transmitted waves formed across the PRS can be
regarded as new sources re-radiating towards the upper half space. The direction
and shape of the beam generated by a PRS antenna are dependent on the amplitude
and phase distributions of the multiple transmitted waves across the PRS. Therefore,
the unit cell design of the PRS (e.g., with uniform vs. nonuniform transmission and
reflection properties) affects the phase distribution across the PRS and hence the
generated beam by the antenna, as depicted in Fig. 1.3. For a PRS antenna with
beam steering capability, different phase distributions are required for beams
pointing in different directions. However, since that the structural parameters of a
fabricated PRS antenna are fixed, a reconfigurable approach is required for the PRS
antenna to generate different phase distributions across the PRS for beam steering.
10 Introduction
(a) (b)
Figure 1.4: Beam-reconfigurable PRS antennas with two PRS parts for 1-D beam steering.
(a) A reconfigurable PRS is employed. (b) A reconfigurable PRS and a phased array as
feeding source are employed [41].
A reconfigurable PRS is mainly implemented using three approaches: mechanical

rotation, movement of liquid materials, and control of electronic elements. In [32]
and [33], the reconfiguration of phase distribution across a PRS is achieved by using
the mechanical rotation of a pair of metasurfaces above the feeding source or by
injecting the liquid metal into specific zones of the PRS. Compared to these two
reconfigurable approaches, which require external motors or an additional layer
with nested microfluidic channels for liquid materials, the reconfigurable approach
with electronic elements is simpler in structure and can offer faster beam steering.
Thus, this approach is gaining more and more interest [34]-[41].
By varying the supplied direct current (DC) biasing voltages for the electronic
elements, such as varactors or PIN diodes, which are loading the PRS unit cells, the
reconfigurable state of the PRS can be controlled. Compared to PIN diodes,
varactors allow for independent voltage control (within a continuous range) to each
PRS unit cell, which facilitates a wider range of phase distributions and finer beam
steering. However, the structural complexity of the DC biasing network is
significantly higher. Therefore, many existing reconfigurable PRS employ PIN
diodes with low-complexity DC biasing network, consisting of two or four PRS
sectors to realize one-dimensional (1-D) [37], [41] or 2-D beam steering [39], [40].
An illustration of beam-reconfigurable PRS antennas with two PRS parts for 1-D
beam steering is presented in Fig. 1.4(a). By varying the reconfigurable state of the
PRS, two symmetrical deflected beams and a broadside beam can be realized.
In addition to the beam deflection achieved by the reconfigurable PRS, it is found
from practical design examples that a beam deflected feeding source can provide
additional beam deflection to a PRS antenna [41], [42]. In [41], with a
reconfigurable feeding network consisting of phase shifters and impedance
matching networks, the phased array can act as the beam deflected feeding source
to provide additional beam deflection, as illustrated by the larger beam deflection
angles in Fig. 1.4(b) than in Fig. 1.4(a).
By introducing reconfigurability at both the PRS and the feeding source, the beam
steering capability of the PRS antenna can be enhanced. However, current beam-
reconfigurable PRS antennas suffer from some common performance limitations,
i.e., small beam steering range, high sidelobe levels1 (SLLs), and considerable gain
variations over different reconfigurable states. In the design of PRS antenna using a
phased array as the feeding source, the reconfigurable feeding network for the
phased array is still complicated and it leads to extra insertion losses. In addition,
the mechanism underlying the phenomenon of the beam deflected feeding source
further enhancing the beam deflection angle of the PRS antenna remains
unexplained. To address the performance issues, it is essential to provide an intuitive
theoretical analysis framework for beam-reconfigurable PRS antennas and utilize it
to gain insight into these antennas’ working principles. The resulting insight can
then be exploited to achieve better antenna performance. In addition, further study
is needed to lower the structural complexity of the beam-reconfigurable feeding
source.
1.2.2. Shared-aperture antennas with PRS

In the design of a PRS antenna, the Fabry-Pérot resonance condition needs to be
satisfied, where the reflection phase of PRS needs to be matched with the PRS
height such that the multiple transmitted waves through the PRS are in-phase. For a
PRS antenna with fixed structural parameters and PRS height, the Fabry-Pérot
resonance condition can be satisfied at multiple specific frequencies, which enables
the construction of a dual-band shared-aperture Fabry-Pérot cavity (DS-FPC)
antenna using a single-layer PRS. Two high gain beams can thus be generated at the
low band (LB) and the high band (HB), respectively, as depicted in Fig. 1.5(a).
However, it is observed from practical design examples that the frequency ratios of
DS-FPC antennas using single-layer PRS tend to be close to the ratio of the orders
of the utilized FPC modes in the two frequency bands, implying that this ratio is
inflexible [43]-[46].
To achieve a more flexible frequency ratio, two layers of PRS with different PRS
heights can be used to form two separate Fabry-Pérot cavities working in LB and
HB, respectively [47]-[53]. The PRS layers are designed to exhibit high
transmission in the other frequency band to minimize wave blockage, hence
1 According to [77], sidelobe level (maximum relative) describes the maximum relative directivity
of the highest sidelobe with respect to the maximum directivity of the antenna.
12 Introduction
(a) (b)
Figure 1.5: Shared-aperture PRS antennas with (a) single-layer PRS and (b) two separate
layers of PRS.
generating two high gain beams in both frequency bands, as seen in the Fig. 1.5(b).
However, since this design scheme requires two layers of PRS, the structural
complexity of the antenna is increased and the antenna profile for HB is also
increased. An artificial magnetic conductor (AMC) ground has been used in DS-
FPC antennas to lower the antenna profile by compensating the propagation phase
in one of the two bands [54]-[55]. The downside of this approach is that the antennas
tend to exhibit narrow overlapping bandwidths between the 10dB impedance
bandwidths and 3dB realized gain bandwidths.
The above DS-FPC antennas aim to construct the Fabry-Pérot cavities in both
frequency bands using single- or dual-layer PRS. However, the shared-aperture
antenna design can also be achieved based on different concepts in two frequency
bands, where the Fabry-Pérot resonance condition is satisfied in only one frequency
band. In [56], a folded transmitarray antenna working in HB and a PRS antenna
working in LB are integrated in a shared aperture using four layers of
substrate. However, it also suffers from an inflexible frequency ratio and the
structural complexity is high. Therefore, to obtain a flexible frequency ratio while
maintaining a low-complexity antenna structure, it is important to choose a suitable
type of high gain antenna to integrate with a PRS antenna for high space utilization.
In addition, no explanation has yet been provided for the observed limitations in the
frequency ratio of DS-FPC antennas using single-layer PRS.
1.2.3. Ray-tracing analysis for PRS antenna

Ray-tracing analysis has been found to be useful for the design of PRS antennas. In
the classical article that first proposed the PRS antenna [21], its working principle
and design parameters are all presented using an intuitive ray-tracing model. In this
early work, the emphasis is on the ray with normal incidence at the PRS, to simplify
the analysis of directivity in the broadside direction. In some subsequent designs of
PRS antennas, ray-tracing models accounting for oblique incidence have also been
used to achieve higher gain [57] and arbitrary beam deflection angle [58].
The following are some phenomena of PRS antennas observed from practical design
examples:
 Capacitive PRS exhibits higher gain than inductive PRS, given the same
reflection magnitude [59], [60].
 A uniform PRS with a larger reflection magnitude than that with a smaller
reflection magnitude exhibits a higher gain for an ideal, infinitely large PRS
[21].
 A beam deflected feeding source can provide additional beam deflection for a
PRS antenna [41], [42].
These observed phenomena show that the PRS type, the reflection magnitude of the
PRS, and the beam deflection capability of the feeding source influence the antenna
gain or beam direction of the PRS antenna. However, clear and intuitive
explanations have not yet been provided for these phenomena. To achieve PRS
antennas with more features and higher performance, proper guidelines for choosing
the suitable PRS types, PRS properties, and feeding source are highly desirable.
Therefore, in this thesis, ray-tracing analysis is used to explain these phenomena,
and the obtained insights are utilized to address the design challenges in two
research areas covered by this thesis: beam-reconfigurable PRS antenna and shared-
aperture antenna design using PRS.
1.3 Co-Designed Antennas for Mobile Terminals and

Design Challenges
Since the invention of the mobile phone until today, the content delivered to users
has become more and more diverse: from voice to text, pictures, videos, and live
streaming. The increasingly rich multimedia content delivery is supported by the
increasing data rates in wireless communication networks. One approach to provide
high data rates in 5G is to utilize the larger available frequency spectra in the mm-
wave bands. Therefore, mobile terminals for 5G communication are starting to be
developed for operation in both sub-6GHz and mm-wave bands [9].
14 Introduction
As mentioned in Section 1.1, to compensate for the high losses in propagation of
electromagnetic waves, mm-wave antennas are required to generate high gain and
steerable beams. Currently, the most prevalent antenna type for mm-wave antennas
is phased array antenna with feeding networks. Various antenna element types have
been employed for the phased arrays, including patch antennas [61], Vivaldi
antennas [62], slot antennas [63], and folded dipoles [64]. Typically, multiple mm-
wave antenna arrays need to be deployed in different locations of the mobile
terminal to enable beam scanning over the full angular range [62]. In a mobile
terminal, the mm-wave antennas can be designed on the printed circuit board (PCB)
or integrated on the protected metal frame, or added as a compact module using the
antenna-in-package (AiP) method [9].
The antennas of a mobile terminal are required to cover many different bands,
including the long-term evolution (LTE)/NR bands, the bands used by wireless
fidelity (Wi-Fi), the bands for near-field communication (NFC) and global
navigation satellite system (GNSS), and mm-wave bands [65]. Therefore, the sub-
6GHz band antennas and mm-wave band antennas need to coexist in the limited
space of a mobile terminal. Therefore, co-designed sub-6GHz and mm-wave
antennas for mobile terminals are attracting more and more research interest.
Many co-design schemes of sub-6GHz and mm-wave antennas for mobile terminals
have been proposed. In [61], [63], [66], and [67], the mm-wave band antennas are
placed on the PCB in separate locations to the sub-6GHz antennas which employ
chip antennas, printed inverted-F antennas (PIFA), monopoles, and patch antennas,
respectively. The performance of the antennas in the two bands can then be
optimized independently. In [64], [68], and [69], the mm-wave antennas are
integrated with the sub-6GHz antenna by sharing their antenna structures, which
enhances the space utilization of the PCB. In addition, by modifying the structure
of metal bezels used to excite the sub-6GHz antennas, the mm-wave antenna can be
embedded in the metal bezels [62], [70], [71].
The current co-designed antennas suffer from poor space utilization, which is caused
by two main factors: the large size of sub-6GHz antenna and the considerable area
occupied by the feeding networks of mm-wave phased array antennas. To design a
small radiator and avoid the usage of phase shifters, reconfigurable parasitic
elements are employed in [72] in a mm-wave antenna for beam steering. In this
design, the beam steering is realized by shorting the parasitic elements via four
transmission lines of different lengths. Even though no real switch is used in the
measurement and the transmission lines occupy considerable PCB space, the
utilization of parasitic elements facilitates a low-complexity implementation for the
beam-steering mm-wave antenna. In summary, to further enhance the space
utilization of the mobile terminals, it is crucial to choose a compact excitation
technique for the sub-6GHz antennas and propose a more compact mm-wave
antenna, while maintaining a low-complexity structure.
Figure 1.6: Simplified antenna design environment of a mobile terminal.
The complex environment of the mobile terminal also brings challenges to the co-
designed antennas. As shown in Fig. 1.6, today’s mobile terminal typically requires
a metal frame, a PCB, a battery, a screen, and some other components that interact
with the user [65]. Nowadays, users have higher and higher demands for the
appearance of mobile phones, such as slim profile and large screen-to-body ratio,
which lead to a highly integrated architecture with many metallic components.
Considering that the metal components may influence the antenna performance, the
antenna co-design scheme is required to be practical in the actual implementation
environment. For instance, the antennas should be able to function properly in the
presence of a metal screen and frame. In addition, the performance of the antennas
should be rather tolerant to a slight variation in the thickness. Moreover, while
selecting a co-design scheme, the locations of the antennas should be chosen to
minimize blockage due to the user’s hands.
1.4 Research Questions and Thesis Goals

The research work of this thesis focuses on two aspects: (1) design of low-
complexity PRS antennas for base stations; (2) co-design of sub-6GHz and mm-
wave antennas for mobile terminals. The research questions and goals considered
for each aspect are summarized as follows.
Research questions for PRS antennas:
 Clear and intuitive explanations have not yet been provided for the phenomena
observed from several design examples that utilize different PRS types,
structural parameters, and feeding sources. Such insights are essential for the
design of the PRS antennas to achieve better performance.
 Current beam-reconfigurable PRS antennas suffer from some common
performance limitations in terms of small beam steering range, high SLLs, and
16 Introduction
considerable gain variations over different reconfigurable states. The beam-
reconfigurable feeding sources for further deflection enhancement still use
complicated phased array antennas with sizeable insertion losses.
 From practical design examples, the DS-FPC antennas with single-layer PRS
cannot provide a flexible dual-band frequency ratio while retaining high
antenna gain. No explanation has yet been provided for this observed limitation
and a low-complexity dual-band shared-aperture scheme using PRS is needed
to fill the frequency ratio gap.
Research goals for PRS antennas:
 To pursue intuitive explanations for the different effects of different PRS types
and feeding sources on the antenna performance, in terms of antenna gain,
beam deflection capability, and SLLs. Ray-tracing analysis has been identified
as a promising tool for this purpose.
 To design a beam-reconfigurable PRS antenna with a low-complexity feeding
source to achieve a larger beam deflection angle, a lower SLL and a smaller
gain variation than conventional solutions. New insights on the working
principles of PRS antennas from ray-tracing analysis are expected to be
valuable input to this effort.
 To derive the dual-band frequency ratio gap for traditional DS-FPC antennas
with single-layer PRS, given a specific reflection magnitude range for gain
enhancement.
 To design a dual-band shared-aperture antenna with a more flexible frequency
ratio and a simple structure to fill the frequency ratio gap derived for traditional
DS-FPC antennas with single-layer PRS.
Research questions for co-designed antennas:
 Current co-designed antennas for mobile terminals suffer from poor space
utilization due to the large size of the sub-6GHz antennas and the considerable
area occupied by the feeding networks of mm-wave phased array antennas,
which also bring extra insertion losses.
 The complex practical environment with many metal components of the mobile
terminals brings challenges to the co-designed antennas, which needs to be
considered in the antenna design.
Research goals for co-designed antennas:
 To co-design the sub-6GHz and mm-wave antennas with high space utilization
and low-complexity structure, which will support two sub-6GHz bands and
offering beam scanning for the 28GHz band.
 To evaluate the effects of several practical design considerations on the

performance of the proposed co-designed antenna and suggest solutions to the
observed performance issues.
1.5 Thesis Structure

This thesis focuses on the design of two types of low-complexity multiband and
beam-reconfigurable antennas for beyond 5G communications, namely: (1) PRS
antennas for base stations; (2) co-designed sub-6GHz and mm-wave antennas for
mobile terminals. The rest of Part I of this thesis contains three more chapters, to
give more details on the work performed on these two topics:
 Chapter 2 first introduces the working principle of PRS antenna, and then
presents a theoretical analysis of PRS properties. The theoretical analysis
provides design guidelines for beam-reconfigurable PRS antenna and shared-
aperture antenna with PRS, which are then utilized to propose new designs with
state-of-the-art performances.
 Chapter 3 briefly introduces the proposed co-designed antenna systems for
mobile terminals. Then, it presents an initial investigation on techniques to
achieve further performance improvements. Lastly, it summarizes the effects
of practical design considerations.
 Chapter 4 provides a summary of the main research contributions of each of
the included papers and my personal contributions, as well as a few promising
future research extensions of the presented work.
Part II of the thesis contains the four original research papers that cover the research
areas described in Part I.
2 PRS Antenna Design for Base
Stations
As mentioned in the previous chapter, PRS antennas have the advantages of high
gain, simple structure, and low cost, providing great potential for application in
future base stations. However, PRS antennas encounter challenges with respect to
beam-reconfigurable and shared-aperture designs.
In the first section of this chapter, the working principle of conventional PRS
antenna is introduced based on ray-tracing analysis. In the second section, the theory
of the PRS unit cell and the effects of several PRS properties on the antenna
performance are presented. These theoretical analyses clarify the working principle
of the PRS and explain several of the observed phenomena in the design of PRS
antennas. In the third section, an intuitive ray-tracing model for the working
principle of beam deflected PRS antennas, as well as design guidelines for beam-
reconfigurable PRS antennas, are provided. The final section of this chapter explains
the frequency ratio limitations for traditional DS-FPC antennas that utilize single-
layer PRS and presents a shared-aperture antenna design scheme to fill the
frequency ratio gap.
2.1 Working Principle of Conventional PRS Antenna
2.1.1. Ray-tracing analysis considering normal incidence

As described in Section 1.2, a PRS antenna typically employs a patch antenna as
the feeding source, which is printed at the center on one side of a substrate, where
the other side of the substrate is used as the fully reflecting ground plane. The PRS
antenna utilizes the multiple reflections of the waves between the PRS and the fully
reflecting ground plane to achieve high gain performance. In the classical article
19
20 Introduction
Phase front
0 1 n n+1
Uniform PRS
φr, φt
Feeding source Ground plane
Figure 2.1: Path of ray with normal incidence and predicted phase front (only half of the
PRS antenna is illustrated due to mirror symmetry).
[21], the working principle of PRS antenna is explained with a ray-tracing model
where the propagating waves are represented by rays. Oblique incident angle α of
the ray from the source (α is the angle between the ray and broadside direction, i.e.,
α ∈ [0°, 90°]) is considered in the classical article. However, to simplify the
theoretical analysis of directivity in the broadside direction, only the normal incident
angle α = 0° is considered for the PRS height derivation. As shown in Fig. 2.1, the
vertical ray paths are illustrated with solid lines, while the horizontal ray paths are
illustrated with dash lines for the continuity of the ray paths, even though they are
neither physical nor considered in the derivation.
The PRS exhibits specific reflection and transmission properties in terms of
magnitudes and phases which are dependent on the unit cell shape and structural
parameters. Since the reflection magnitude of the PRS is less than 1, each incident
wave at the PRS is partially reflected and partially transmitted. In the process of the
waves being reflected by the PRS and the ground plane, multiple transmitted waves
(rays) are formed across the PRS, which can be regarded as new wave sources
radiating towards the upper half space. Each ray experiences a total propagation
path, consisting of multiple times of reflection on the PRS and ground plane and
one time of transmission through the PRS, which causes a relative phase shift
between the rays. To achieve the in-phase condition for the multiple transmitted
waves, the PRS height l (the distance between PRS and ground plane) needs to be
matched with the reflection phase φr of the PRS in a specific manner, which has
been derived as [21]
Chapter 2. PRS Antenna Design for Base Stations 21
φ λ λ
l   r  0.5   N , (2.1)
 2π 2 2
where λ is the wavelength in free space and N = 0, 1, 2, …. Since the PRS antenna
is also known as FPC antenna, the N in the FPC antenna’s resonance condition is
called the order of the FPC mode [43].
From (2.1), the PRS height l is related to the reflection phase φr, the order of FPC
mode N, and the wavelength λ. Given the same N and a fixed wavelength λ, different
reflection phases φr’s correspond to different PRS heights l’s, so the antenna height
(or profile) is adjustable, providing a degree of flexibility in PRS antenna design.
On the other hand, with a fixed reflection phase φr and wavelength λ, the FPC
antenna’s resonance condition can be satisfied theoretically with multiple sets of
values for N and PRS height l. Since a larger N corresponds to a higher antenna
profile, N = 1 is normally chosen to minimize l (while being a physical solution).
From the ray-tracing model depicted in Fig. 2.1, the total phase shifts of the rays
and the phase difference between adjacent rays can be expressed to predict the phase
distribution across the PRS, given a specific PRS height and the reflection and
transmission properties of PRS. The total phase shift of ray n is the sum of the phase
shifts due to the ray path, the reflections on the PRS φr and the ground plane π, as
well as the transmission through the PRS φt, expressed as
2π
φrayn    2n  1 l  nφr  nπ  φt , (2.2)
λ
where the index n (n = 0, 1, 2, …) represents that the wave has been reflected n
times before it is transmitted through the PRS. The phase difference between ray (n
+ 1) and ray n is deduced as
4π
Δφ  φray n+1  φrayn   l  φr  π . (2.3)
λ
It can be verified that with the PRS height l and reflection phase φr satisfying the
FPC antenna’s resonance condition in (2.1), the phase difference Δφ in (2.3)
becomes 2πN, where N = 1, 2, …. The phase difference Δφ takes a negative value,
representing an excess phase delay of the ray (n + 1) relative to the ray n. Therefore,
it can be concluded that the multiple transmitted rays across the PRS satisfy the in-
phase condition, as depicted in Fig. 2.1.
From (2.3), given a PRS height, when the reflection phase φr of the utilized PRS is
less than the calculated one from (2.1), more phase delay will be brought to the ray
(n + 1) by the PRS than required, thus the absolute value of phase difference |Δφ|
will be greater than 2πN and no longer a multiple of 2π. In other words, the given
vertical ray path length is too long to satisfy the FPC antenna’s resonance condition
in this case, causing more phase lag to the ray (n + 1) relative to the ray n. With a
22 Introduction
similar derivation process, it can be concluded that when the reflection phase φr of
the utilized PRS is greater than the calculated one from (2.1), the FPC antenna’s
resonance condition cannot be satisfied due to the insufficient phase lag to the ray
(n + 1). Therefore, both smaller and larger reflection phases will cause phase front
tilting as the waves are transmitted through the PRS from its center to the edges.
The tilting phase distributions based on this mechanism can be utilized for the
design of beam deflected PRS antennas as presented in Paper II, which will be
further discussed in Section 2.3.
2.1.2. Ray-tracing analysis considering oblique incidence

In practice, the waves transmitted through the off-center positions of the PRS are
from multiple oblique incident angles. As mentioned in Section 1.2, ray-tracing
models accounting for oblique incidence have been used for PRS antenna designs
to achieve higher gain [57] and arbitrary beam deflection angle [58]. To construct a
more accurate ray-tracing model, the effect of oblique incidence is analyzed in this
thesis, based on which the observed phenomena mentioned in Section 1.2.3 can be
explained and a beam deflected PRS antenna with better performance can be
designed.
As shown in Fig. 2.2, assuming that a ray with arbitrary incident angle α is
propagating between the PRS and the ground plane, the path length for each
reflection is l / cos α. From (2.3), the phase difference between ray (n + 1)′ and ray
n′ becomes
4π l
Δφ=φray  n 1 '  φrayn    φr  π . (2.4)
λ cos α
The l / cos α term is larger than the vertical distance l between the PRS and the
ground plane, so the ray (n + 1)′ has more excess phase delay relative to the ray n′
in the case considering the oblique incidence than the case considering only the
normal incidence. Therefore, the phase front in the ray-tracing model considering
the oblique incidence gradually lags the ideal uniform phase front in the simplified
model considering only normal incidence, as the rays propagate from the center of
PRS to the edges, as illustrated in Fig. 2.2. The phase delay at the edge of the PRS
can weaken the constructive interference of waves in broadside direction, thus
affecting the radiation pattern performance, e.g., decreasing the antenna gain [73].
At a particular location of the PRS, the transmitted wave is composed of rays
originating from different incident angles. From (2.4), the phase difference Δφ′ is
related to the incident angle α, so different incident angles have different effects on
the phase front. The corresponding derivation details are included in Paper I, where
it is concluded that the ray with a larger incident angle and a smaller number of
Phase front of normal incidence
0 0' 1 1' n n' n+1 (n+1)'
Uniform PRS
φr, φt
l
α
Figure 2.2: Path of ray with oblique incident angle and predicted phase front (in black),
relative to that of normal incidence (in blue).
reflections causes a larger phase delay than the normal incidence at the same
location of the PRS.
Since the larger incident angles correspond to larger phase delays than the normal
incidence, the extent of total phase delay at a particular location of PRS, which is
the sum of (complex) phasor representations of rays with multiple incident angles,
depends on the proportion of the ray with larger incident angles. This proportion is
derived in Paper I and is used to explain why different types of PRS influences the
antenna gain performance, see further details in Section 2.2.
2.1.3. Proper reflection magnitude

From the working principle of conventional PRS antenna described above, multiple
transmitted waves are formed in the process of multiple reflections between the PRS
and reflecting ground plane. In addition to the requirement for the reflection phase
in (2.1), the reflection magnitude of the PRS needs to take a proper value,
considering that the size of the PRS is finite in practical design.
If the PRS reflection magnitude is too small, most of the waves radiated from the
feeding source will be transmitted through the center part of the PRS due to
insufficient reflections between the PRS and reflecting ground plane. In this case,
the aperture of PRS antenna cannot be fully utilized to achieve high gain. In contrast,
if the reflection magnitude of PRS is too large, the proportion of wave transmitted
24 Introduction
through the PRS (at each wave incidence at the PRS) will be very small. The
amplitude of the transmitted waves across the PRS will thus be small, and a
substantial proportion of the waves will leak from the sides of the cavity formed by
the PRS and the ground plane, resulting in low gain and high SLLs. According to
the classical article [21], a reasonable reflection magnitude is between 0.7 to 0.9 for
a uniform PRS that facilitates a broadside beam.
Therefore, the two reflection properties of a PRS, i.e., the reflection magnitude and
the reflection phase, are critical parameters for the performance of PRS antennas.
These PRS properties will be analyzed in detail in the next section.
2.2 PRS Properties

PRS is a type of metasurface, which is typically composed of unit cells with various
metallic patterns periodically distributed on a dielectric substrate, exhibiting varying
reflection and transmission properties with frequency. In this section, the PRS types
and the reflection and transmission coefficients of the unit cell of a single-layer PRS
are introduced, based on which the effects of PRS type on the PRS antenna
performance are analyzed.
2.2.1. PRS unit cell types

The unit cell of a single-layer PRS can be considered as a two-port system regardless
of its shape [47], as illustrated in Fig. 2.3. The incident and reflected plane waves
on both sides are complex quantities and are related to each other as
 E1   S11 S12   E1 
     , (2.5)
 E2   S21 S22   E2 
where the elements in the scattering matrix [S] are four scattering parameters (S-
parameters) of the two-port system, corresponding to the two reflection coefficients
(S11 and S22) and two transmission coefficients (S21 and S12) [74]. Each complex S-
parameter contains magnitude (i.e., |S11| is reflection magnitude) and phase (i.e., ∠
S11 is reflection phase) information.
Given the specific metallic pattern of the unit cell with certain equivalent inductance
and capacitance, the PRS unit cell resonates at specific frequencies with maximum
reflection or transmission [75]. At the resonant frequency with maximum reflection,
the unit cell is equivalent to a series LC circuit (acting essentially as a short circuit),
whereas at the resonant frequency with maximum transmission, the unit cell is
equivalent to a parallel LC circuit (acting essentially as an open circuit), as shown
in Fig. 2.4.
 
E1 E2
 
E1 E2
Single layer of PRS
Figure 2.3: Illustration of waves on both sides of a single-layer PRS [47].
Leq
Z0 Z0 Z0 Leq Ceq Z0
Ceq
(a) (b)
Figure 2.4: Equivalent circuits of PRS unit cell at resonant frequencies with maximum (a)
reflection and (b) transmission.
PRS has been classified into two types, according to the equivalent reactance of the
periodic structure: capacitive PRS with reactance less than 0, and inductive PRS
with reactance larger than 0 [76]. Since the metallic pattern of the unit cell has
varying equivalent inductance Leq and capacitance Ceq with frequency, the
equivalent reactance and the PRS type of the unit cell also vary with frequency.
From Fig. 2.4(a), the equivalent reactance X of the unit cell near the resonant
frequency with maximum reflection is
1 ω2 Leq Ceq  1
X  jωLeq   j (2.6)
jωCeq ωCeq
The resonant frequency is f r  ωr 2π  1 2π Leq Ceq , where the equivalent

reactance X is 0. It can be derived from (2.6) that X takes a negative value and
increases monotonically to 0 as the frequency increases from f r  Δf to f r , while
it takes a positive value and increases monotonically from 0  as the frequency
increases from f r to f r  Δf , where Δf is positive and small. According to the
frequency-varying trend in the equivalent reactance, it can be deduced that the PRS
26 Introduction
+∞
1 Inductive Capacitive Inductive
Reflection Equivalent
magnitude reactance
(Ω)
0 0
f
ft fr
-∞
Figure 2.5: Illustration of trends in frequency-varying reflection magnitude and PRS type.
type changes from being a capacitive one to an inductive one as the frequency varies
from f r  Δf to f r  Δf near the resonant frequency with maximum reflection, as
illustrated in Fig. 2.5.
From Fig. 2.4(b), the equivalent reactance X of the unit cell near the resonant
frequency with maximum transmission is
1 ωLeq
X  jωLeq // j (2.7)
jωCeq 1  ω2 Leq Ceq
The resonant frequency is ft  ωt 2π  1 2π Leq Ceq , where the equivalent

reactance X is infinite. It can be derived from (2.7) that the equivalent reactance X
takes a positive value and increases monotonically to +∞ as the frequency increases
from ft  Δf to ft  , whereas it takes a negative value and increases monotonically
from -∞ as the frequency increases from ft  to ft  Δf . Therefore, near the resonant
frequency with maximum transmission, the PRS type changes from being an
inductive one to a capacitive one as the frequency varies from ft  Δf to ft  Δf , as
illustrated in Fig. 2.5.
In this section, the unit cell of single-layer PRS is analyzed based on a two-port
system with equivalent circuits. The summary is as follows:
 The metallic pattern of a PRS unit cell corresponds to specific equivalent
inductance and capacitance, which vary with frequency.
 The equivalent reactance of the unit cell consists of the equivalent inductance
and capacitance that vary with frequency, leading to a specific PRS type at a
given frequency.
2.2.2. Reflection and transmission coefficients of PRS

Based on the two-port system in Fig. 2.3, assuming that the single-layer PRS is
lossless, symmetrical, and reciprocal, and neglecting the higher order harmonics of
the PRS layer, the relationships between the reflection and transmission coefficients
are [47]
S11  S22 and S12  S21 , (2.8)
S 21  1  S11 , (2.9)
π
S11  S21   . (2.10)
2
From the relationships of the S-parameters and the conversions between two-port
system parameters [74], the impedance parameters (Z-parameters) Z11 and Z21 can
be derived as
S 21
Z11  Z 21   . (2.11)
2 S11
Equation (2.10) gives the relationship between the reflection phase and transmission
phase regardless of the reflection magnitude. With the positive sign in (2.10), (2.11)
can be expressed as
S 21
Z11  Z 21  j , (2.12)
2 S11
corresponding to the inductive PRS unit cell with reactance larger than 0. With the
negative sign in (2.10), (2.11) can be expressed as
S21
Z11  Z 21   j , (2.13)
2 S11
corresponding to the capacitive PRS unit cell with reactance less than 0. It is
observed from (2.12) and (2.13) that Z11  Z 21  X11  X 21 , which are represented
with reactance X in the following.
From the above derived conclusions and the conversions between two-port system
parameters [74], the S-parameter S11 can be expressed as
28 Introduction
Z0
S11   , (2.14)
Z0  j 2 X
and S21 can be expressed as
2X
S21  . (2.15)
Z0  j 2 X
The magnitudes and phases of the S-parameters vary with the reactance X, given the
fixed characteristic impedance of the surrounding media Z0.
Given the trends in the reactance and PRS type variation with frequency as shown
in Fig. 2.5, the trend in the reflection phase ∠S11 variation can be deduced from
(2.14), as shown in Fig. 2.6. It can be derived that the reflection phase ranges of
capacitive and inductive PRSs are  π ,3π 2  and  π 2 , π  , respectively, with the
assumptions for the PRS from [47] as mentioned earlier in this section.
Based on the phase difference between the reflection phase and transmission phase
in (2.10) and the trend in the reflection phase variation shown in Fig. 2.6, the trend
in transmission phase variation can be deduced, as shown in Fig. 2.7. It can be
observed that the transmission phase is discontinuous between the two regions
separated by a maximum reflection resonance point, while the reflection phase is
discontinuous between the two regions separated by a maximum transmission
resonance point.
The relationship between the reflection magnitude and transmission phase has been
derived as [47]
S11   sin   S 21  . (2.16)
Inserting (2.10) into (2.16), the relationship between the reflection magnitude and
reflection phase can be expressed as
S11   cos   S11  . (2.17)
Since the reflection phase ranges of capacitive and inductive PRSs are  π ,3π 2 
and  π 2 , π  , the two types of PRS have opposite trends in the reflection magnitude
variation as the reflection phase decreases. For the capacitive PRS unit cell, when
the reflection phase ∠S11 decreases from 3π/2 to π, the reflection magnitude |S11|
increases from 0 to 1. For the inductive PRS unit cell, when the reflection phase
∠S11 decreases from π to π/2, the reflection magnitude |S11| decreases from 1 to 0.
These trends derived from (2.17) agree with the trends observed from Fig. 2.5 and
Fig. 2.6. This difference between the two types of PRS leads to the different beam
+∞
3π/2 Inductive Capacitive Inductive
Equivalent
Reflection
reactance
phase (deg)
(Ω)
π 0
ft fr
π/2
-∞
Figure 2.6: Illustration of variation trends in reflection phase and PRS type with frequency.
+∞
π/2 Inductive Capacitive Inductive
Equivalent
Transmission
reactance
phase (deg)
(Ω)
0 0
ft fr
-π/2
-∞
Figure 2.7: Illustration of variation trends in transmission phase and PRS type with
frequency.
deflection capability for beam deflected PRS antenna, which will be discussed
further in Section 2.3.
In this section, the reflection and transmission coefficients of single-layer PRS are
analyzed assuming that the PRS is lossless, symmetrical, and reciprocal, and
neglecting the higher order harmonics of the PRS layer. The summary is as follows:
 The reflection phase and the transmission phase have a phase difference of π/2
for an inductive PRS unit cell with reactance larger than 0 and π/2 for a
capacitive PRS with reactance less than 0.
30 Introduction
Port 1
PBC PBC
Port 2
Figure 2.8: Illustration of a simulated unit cell model.
 The magnitudes and phases of the S-parameters vary with equivalent reactance,
which explains why the unit cells with different metallic patterns or structural
parameters exhibit different frequency responses.
 The reflection phase ranges of capacitive and inductive PRS unit cells are
different, due to the different ranges of the equivalent reactance.
 The reflection magnitude of a PRS unit cell is related to the reflection phase.
The capacitive and inductive PRSs have opposite trends in reflection
magnitude for the same trend in the reflection phase variation.
In practical PRS antenna design, the reflection coefficients of a PRS can be
evaluated using its unit cell with two Floquet ports and periodic boundary conditions
(PBCs), as illustrated in Fig. 2.8. The PBCs on the four side surfaces are set to model
an infinitely large PRS composed of the unit cells. In this thesis, the unit cell model
is simulated using the frequency domain finite element method (FEM) solver in
ANSYS HFSS.
2.2.3. Effect of PRS properties on broadside beam

From the analysis in the previous section, the capacitive and inductive PRSs exhibit
different properties in terms of the reflection phase range, transmission phase range,
and trends in the reflection magnitude and phase variations. Since the PRS plays an
important role in the working principle of the PRS antenna, the performance of the
PRS antenna is affected significantly by the PRS type. As mentioned in Section
1.2.3, it has been observed from practical design examples that the capacitive PRS
exhibits higher gain than the inductive PRS for broadside beams, given the same
reflection magnitude. This phenomenon can be explained based on the theoretical
analysis of PRS properties in the previous section.
From (2.17), the capacitive and inductive PRSs exhibit different reflection phases
for the same reflection magnitude. As can be seen from (2.1), the PRS height is
determined by the reflection phase at a specific frequency, given the same order of
the FPC mode N. Therefore, given the same reflection magnitude, the capacitive
and inductive PRSs correspond to different PRS heights, which affect the magnitude
and phase distribution across the PRS, as described in detail in Paper I.
First, based on a ray-tracing model accounting for oblique incident angles, it is
concluded that a larger oblique incident angle causes more phase delay than a
smaller one at the same location on the PRS. Then, it is proven that a larger PRS
height leads to a larger proportion of power in the ray with the larger oblique
incident angle, based on which the phase and magnitude distribution trends along
the two types of PRSs are predicted. The derivations in Paper I show that the
capacitive PRS can achieve less phase delay and more uniform magnitude
distribution over the PRS, which results in higher antenna gain.
In addition to the PRS type, the reflection magnitude of the PRS also has an effect
on the antenna gain for an ideal, infinitely large PRS, as mentioned in Section 1.2.3.
This phenomenon can also be explained by the ray-tracing model proposed in Paper
I. It is derived in Paper I that the proportion of power in the ray with the larger
incident angle is inversely related to the reflection magnitude. Therefore, a larger
reflection magnitude will result in a smaller proportion of power in the ray with the
larger incident angle and hence a smaller phase delay at the same location on the
PRS, leading to a flatter phase front. For the PRS with a larger reflection magnitude,
a more uniform magnitude distribution is formed across the PRS. Therefore, the
flatter phase front and the more uniform magnitude distribution lead to a superior
gain.
2.3 Beam-Reconfigurable PRS Antenna Design

As mentioned in Section 1.2.1, PRS antennas can achieve beam steering by
introducing reconfigurability into both the PRS and the feeding source. A
reconfigurable PRS does not require a sophisticated feeding network as a
conventional phased array, and a beam deflected feeding source can further enhance
the beam deflection angle of the PRS antenna. A reconfigurable PRS consisting of
several PRS sectors formed by different PIN diode states provides the advantage of
low-complexity DC biasing network, when compared to other reconfigurable
approaches. However, the existing reconfigurable PRS antennas suffer from some
common performance limitations, i.e., small deflection angles, high SLLs, and
considerable gain variations over different reconfigurable states. In addition, the
role of a phased array in enhancing the beam deflection of a PRS antenna has not
been explained analytically. Furthermore, the existing reconfigurable feeding
32 Introduction
Phase front
PRS Part 1, φr1, φt1 PRS Part2, φr2, φt2
l1
(a)
Phase front
PRS Part1, φr1, φt1 PRS Part2, φr2, φt2
l2

(b)
PRS Part1, φr1, φt1 PRS Part2, φr2, φt2

(c)
Figure 2.9: Ray-tracing models for three PRS antennas with the same two-part PRS but
different PRS heights of (a) l1, (b) l2 and (c) l.
sources are still complicated, and they lead to extra insertion losses.
In this section, we first introduce the working principle of beam deflected PRS
antenna. Next, we compare the effect of the PRS type on the beam deflection, based
on the theory of PRS analyzed in Sections 2.1 and 2.2. Then, we provide a summary
of the metrics used to evaluate the performance of a reconfigurable PRS antenna, as
well as the trade-off between the metrics. As a supplement, we briefly explain the
beam deflection enhancement effect of the feeding source and present a beam-
reconfigurable feeding source with a simple structure.
2.3.1. Working principle of beam deflected PRS

The working principle of uniform PRS for the gain enhancement of broadside beam
has been described in Section 2.1.1. To satisfy the FPC antenna’s resonance
condition, the PRS height needs to be matched to the reflection phase of the PRS as
determined by (2.1). It has also been derived that a smaller or larger reflection phase
of the PRS than the one calculated from (2.1) will cause phase front tilting along the
PRS. Based on this mechanism, the PRS composed of two PRS parts exhibiting two
different reflection phases can be used to generate tilting phase front for 1-D beam
steering. The details of this beam steering effect are presented in Paper II.
Given the same PRS composed of two PRS parts, different PRS heights lead to
different phase distributions and beam performances. Three cases of PRS antennas
with the same PRS but different PRS heights are illustrated with three simplified
ray-tracing models in Fig. 2.9, where only the normal incidence is considered. It is
assumed that PRS Part 1 exhibits less reflection phase than PRS Part 2 (φt1 < φr2),
so the required PRS height l1 for PRS Part 1 is less than the required PRS height l2
for PRS Part 2, as calculated from (2.1).
In Case 1, where the PRS height is set to l1, the transmitted rays are in-phase across
PRS Part 1. However, since the PRS height is less than the required l2 for PRS Part
2 to satisfy the in-phase condition, the vertical path is shorter than needed, resulting
in the phase on the outer edge of PRS Part 2 leading the phase at the center, as
illustrated in Fig. 2.9(a). On the contrary, in Case 2 where the PRS height is set to
l2, the transmitted rays are in-phase across PRS Part 2, whereas the vertical path is
longer than needed, resulting in the phase on the outer edge of PRS Part 1 lagging
the phase at the center, as shown in Fig. 2.9(b). In both cases, the PRS height is
determined by one of the two PRS parts from (2.1), resulting in half of the phase
front being parallel to the PRS and another half being tilted, in theory. Consequently,
the phase front does not vary smoothly across the entire PRS, which can degrade
the beam performance. In Case 3, the PRS height l is set using the average value of
the reflection phases of the two PRS parts, which is less than the required value for
PRS Part 1 and greater than the one for PRS Part 2. The phase on the outer edges of
the PRS lags and leads by the same extent compared to that at the center, which
facilitates a smoothly tilting phase front across the entire PRS, as illustrated in Fig.
2.9(c). Therefore, the PRS height for the two-part PRS should be set according to
the average value of the two reflection phases. In addition, it can be predicted that
the larger the phase difference between PRS Part 1 and Part 2, the more the phase
front will be tilted, resulting in a larger beam deflection angle.
2.3.2. Effect of PRS type on beam deflection capability

From the ray-tracing analysis that only considers normal incidence in Section 2.3.1,
the achieved beam deflection angle should be the same theoretically given the same
34 Introduction
phase difference between the two PRS parts. However, when oblique incidence is
considered, the beam deflection capability of the PRS antenna is also influenced by
the reflection magnitudes of the two PRS parts, as will be demonstrated below.
As mentioned in Section 2.2.2, the reflection magnitude of a PRS unit cell is related
to the reflection phase as expressed in (2.17). Therefore, the two PRS parts with
different reflection phases also exhibit different reflection magnitudes. From the
analysis of the effects of PRS properties on broadside beam in Section 2.2.3, the
PRS with a smaller reflection magnitude leads to more phase delay. Therefore, for
the two-part PRS, if the PRS part with a smaller reflection phase that causes phase
lagging (i.e., PRS Part 1 in Fig. 2.9(c)) exhibits a smaller reflection magnitude, the
phase delay at the outer edge of this PRS part resulting from an oblique incidence
will increase. On the other hand, if the PRS part with a larger reflection phase that
causes phase leading (i.e., PRS Part 2 in Fig. 2.9(c)) exhibits a larger reflection
magnitude, the phase delay at the outer edge of this PRS part resulting from an
oblique incidence will decrease. Thus, to further enhance the beam deflection effect
of a nonuniform two-part PRS, it is preferred that the trend of reflection magnitude
aligns with the trend of reflection phase. The details of this derivation are included
in Paper II.
As concluded in Section 2.2.2, the inductive and capacitive PRSs have opposite
trends in reflection magnitude for the same trend in the reflection phase. The
inductive PRS has the same trend in the reflection magnitude and phase, so it can
inherently leverage a larger beam deflection than the capacitive PRS due to effect
of oblique incidence, given the same reflection phase difference between the two
PRS parts. The above analysis provides a guideline for choosing a suitable type of
PRS for better beam deflection performance.
2.3.3. Performance metrics for beam-reconfigurable PRS antenna

Beam steering capability for a PRS antenna requires multiple phase distributions
realized with reconfigurable approaches. The phase distribution over the PRS and
the direction of the generated beam are dependent on the reflection phase
distribution of the unit cells. The analysis in Section 2.2 indicates that the reflection
coefficient (i.e., magnitude and phase) of a PRS unit cell is related to its equivalent
reactance.
Given different DC excitations, a PIN diode has different equivalent models in the
forward (ON) state and the zero or reverse (OFF) state, as shown in Fig. 2.10. In the
forward state, the PIN diode is modelled as a series LC circuit with a forward series
resistance RS and a parasitic inductance LP. In the zero or reverse state, the PIN diode
is modelled as a series circuit consisting of a parasitic inductance LP connected in
series with a parallel circuit formed by a parallel resistance RP and a total
capacitance CT. The difference in the equivalent models enables the PIN diode to be
CT
LP LP
RS
RP
(a) (b)
Figure 2.10: Equivalent circuits of PIN diode in (a) forward (ON) state and (b) zero or
reverse (OFF) states.
used as a reconfigurable reactance element in the PRS unit cells. For a PRS unit cell
loaded with PIN diodes, its equivalent reactance can be controlled to achieve
reconfigurability in its reflection coefficient.
A beam-reconfigurable PRS antenna typically has one port for the feeding source
and it can generate one of multiple beams in both broadside and deflected directions
to cover a wide angular range. The metrics to evaluate the antenna performance
mainly address two aspects: the radiation patterns and the reflection coefficients of
the port over different reconfigurable states [77]. Concerning the radiation patterns,
the key metrics include the maximum deflection angle, the beamwidths, the
maximum directivities for different reconfigurable states, and the SLLs. These
metrics can be used to evaluate whether the PRS antenna can focus the wave energy
in the desired directions and provide high gain beam steering uniformly over a wide
angular range. The reflection coefficients indicate the impedance matching
performance of the antenna port in different reconfigurable states, which determine
if the required input power can be accepted by the PRS antenna. Since the PRS
properties vary for different beam directions, the corresponding reflection
coefficient of the port also vary over different reconfigurable states, except between
the symmetric ones. Therefore, the overlapping impedance bandwidth over the
impedance bandwidths of different reconfigurable states is also an important metric,
indicating if the beam steering can be realized in a specific frequency bandwidth.
From the analysis in Section 2.3.1, the PRS height for a two-part reconfigurable
PRS for 1-D beam steering is determined by the average value of the reflection
phases of the PRS unit cell in the ON and OFF states. However, this PRS height
does not match with the reflection phase required in the FPC antenna’s resonance
condition when either state is selected for a uniform PRS (i.e., ON-ON or OFF-
OFF), thus affecting the antenna gain and radiation pattern performance of the
broadside beam. The performance degradation of the broadside beam state is
evaluated in Paper II, which shows that a trade-off is needed between the broadside
beam performance and the maximum deflection angle.
36 Introduction
2.3.4. Beam deflected feeding source

Given the aforesaid design trade-off in obtaining desired properties for both
broadside and deflected beams, a beam deflected feeding source can be used to
deflect the beam further while retaining good broadside beam. The role of a phased
array in enhancing the beam deflection in a PRS antenna is explained analytically
based on a ray-tracing model in Paper II. Since each array element contributes more
to the incident rays in the PRS part just above it, the phase deflected feeding source
can further increase the total phase difference across the PRS, thus adding
constructively to the beam deflection effect of the PRS.
As mentioned in Section 1.2.1, the existing beam deflected feeding sources that
employ phased arrays have the drawbacks of complicated feeding networks and
extra insertion losses. To overcome these drawbacks, the beam deflected capability
of the feeding source can also be realized with parasitic elements (i.e., planar
parasitic patch elements), as presented in Paper II, based on the concept of Yagi-
Uda antenna [78]. The parasitic elements are loaded with PIN diodes to control its
equivalent electric size. By reconfiguring the states of the parasitic elements, the
capacitive coupling to the active element can be controlled, thus providing the
desired phase distributions and beam reconfigurability. This beam-reconfigurable
feeding source does not require a feeding network with phase shifters, leading to
lower complexity. In addition, the parasitic element can introduce a second
resonance, facilitating better impedance matching and/or impedance bandwidth in
the PRS antenna. The details of the parasitic-element based reconfigurable feeding
source are provided in Paper II.
2.4 Shared-Aperture Antenna Design Using PRS

As mentioned in Section 1.2.2, a shared single-layer PRS can be used to construct a
DS-FPC antenna for high space utilization. However, it is observed from practical
design examples that DS-FPC antennas with single-layer PRS suffer from
limitations in the dual-band’s frequency ratio. Therefore, in this section, we first
analyze the feasible range of frequency ratio for traditional DS-FPC antennas with
single-layer PRS. Then, we present a new dual-band shared-aperture antenna to fill
the frequency ratio gap. Lastly, a summary of the metrics used to evaluate the
performance of a shared-aperture antenna that utilizes PRS is provided.
2.4.1. Frequency ratio range analysis for DS-FPC antennas

For a DS-FPC antenna with a single-layer PRS, the shared PRS height l and the
reflection phases φrL and φrH need to satisfy the FPC antenna’s resonance condition
in (2.1) in both LB and HB, which can be expressed as
φ λ λ
l   rL  0.5  L  N L L , (2.18)
 2π  2 2
φ λ λ
l   rH  0.5  H  N H H , (2.19)
 2π  2 2
where λL and λH represent the wavelengths in free space in LB and HB, respectively.
Combining (2.18) and (2.19), the frequency ratio between the HB frequency fH and
the LB frequency fL can be expressed as
φrH
 0.5  N H
fH
 2π . (2.20)
fL φr L
 0.5  N L
2π
From the working principle of FPC antenna, sufficiently reflected waves are
required between the PRS and the fully reflecting ground plane to achieve the high
gain performance. This implies that the PRS should exhibit a sufficiently high
reflection magnitude within a specific range. From the theoretical analysis of the
PRS properties in Section 2.2, the reflection magnitude is related to the reflection
phase as expressed in (2.17), which results in the reflection phase of the PRS being
also constrained within a specific range. For a traditional DS-FPC antenna using a
shared single-layer PRS, the PRS should exhibit high reflection magnitudes in both
bands, where the reflection phases in both bands are all constrained to specific
ranges.
From (2.20), assuming that the PRS exhibits the reflection phases of φrL and φrH in
specific ranges with upper and lower boundaries, the frequency ratio is thus related
to NL and NH, the orders of the FPC modes in the two frequency bands. This explains
the observed phenomenon in the practical design examples that the achieved
frequency ratio tends to be close to the ratio of the FPC mode orders in the two
frequency bands, as mentioned in Section 1.2.2. Since the PRS height corresponds
to a smaller wavelength in LB than in HB, the order of the FPC mode in LB is
normally chosen as NL = 1 to minimize the cavity height.
In this thesis, the frequency ratio range of interest is (1, 2), to cater for dual-band
application combining some commonly used frequency bands, such as S-band (2-
4GHz), C-band (4-8GHz), and X-band (8-12GHz). The range of feasible frequency
ratio for a traditional DS-FPC antenna with single-layer PRS with NH = 1 and NH =
2 are calculated in Paper III. The frequency ratio gap for the two cases of NH within
the range of interest is [1.29, 1.67], given the range of reflection magnitude ρ 
(0.707, 1), where there is more power in the reflected waves than the transmitted
ones to facilitate gain enhancement (i.e., 0.707 corresponds to 3dB). Within the
frequency ratio gap, a DS-FPC antenna with single-layer PRS is predicted to exhibit
low gains due to the lack of reflection. It is also concluded that the upper boundary
38 Introduction
Subreflector Ray1 Ray2

φr = π Ray1 Ray2
≈π Subreflector
Path φr = π
Feeding <π
phase source Path Feeding
phase source
Main reflector Main reflector
(a) (b)
Figure 2.11: Propagation of waves in traditional SBA with (a) typical profile height and (b)
reduced profile height.
of this frequency ratio gap is more challenging to achieve for a DS-FPC antenna
with NH = 1, which leads to a smaller antenna profile (i.e., a more compact
implementation).
2.4.2. Dual-band shared-aperture antenna design using PRS

A dual-band shared-aperture design is proposed in Paper III, which merges a FPC
antenna with a short backfire antenna (SBA) with different working principles to
fill the frequency ratio gap while offering a simple structure. A PRS is designed
with different reflection coefficient distributions in two frequency bands to enable
different functions. The chosen frequency ratio of 1.64 is significantly larger than
the upper boundary of the frequency ratio limit for NH = 1 (1.27). The shared PRS
is designed to be capacitive for the HB FPC antenna to achieve high gain [76], and
the SBA is designed to work in LB by utilizing a parasitic element and the shared
PRS. The HB FPC antenna works with the first-order FPC mode, facilitating a lower
overall antenna height.
The PRS height as determined by the HB FPC condition is smaller than the typical
0.5 wavelength used in the traditional SBA [79], which could result in gain
degradation for the LB antenna. As illustrated in Fig. 2.11, a traditional SBA
consists of a main reflector, a small subreflector, and a feeding source between two
reflectors. In the SBA, the waves from the feeding source are reflected by one or
both reflectors and the wave components along the different paths satisfy an in-
phase condition. The different path lengths of the two rays are mainly attributed to
the initial upward path of Ray1 from the feeding source to the subreflector, the
reflection at the metallic surface of the subreflector (giving an equivalent path phase
of π), part of the continuing downward path of Ray1 from the subreflector until it
reaches the height of the feeding source. However, as the vertical distance between
the subreflector and the feeding source is reduced, the in-phase condition between
waves in the different paths is no longer satisfied. The nonuniform PRS proposed in
Paper III can improve the gain degradation of the LB SBA by utilizing the outer
PRS unit cells that are designed to offer positive and increasing transmission phase
as the incident angle to the PRS increases. The details of the design methodology
are included in Paper III.
2.4.3. Performance metrics for shared-aperture antenna using PRS

In this thesis, the shared-aperture antenna design with single-layer PRS focuses on
generating beams in the broadside direction in two frequency bands with their
respective feeding sources. The S-parameters evaluation metrics include the
impedance matching bandwidths of the two feeding ports and the isolations between
the LB and HB ports. The evaluation metrics of radiation pattern performance
include gains and 3dB gain bandwidths of the antennas in both bands. Typically,
higher gain is more desirable in HB than in LB to compensate for the higher
propagation loss in HB, as mentioned in Section 1.1. The overlapping bandwidth
between the 10dB impedance bandwidth and 3dB realized gain bandwidth is also
an essential metric to evaluate if the dual-band antenna can meet the application
requirements in terms of both the S-parameters and radiation patterns in both bands.
The flexibility of the achievable frequency ratio enables a shared-aperture design
scheme to be used for more dual-band applications. In Paper III, the frequency ratio
flexibility of the proposed antenna is investigated. It is concluded that by choosing
proper structure parameters, the proposed antenna can realize a flexible frequency
ratio to fill the frequency ratio gap of traditional DS-FPC antennas with single-layer
PRS, while maintaining a low-complexity structure. A comparison of recent dual-
band shared-aperture designs in terms of these metrics is given in Paper III, which
indicates the advantages of the proposed antenna in terms of flexible frequency ratio,
simple structure, relatively low antenna profile and high space utilization.
3 Co-Designed Antenna for Mobile
Terminals
As described in Chapter 1, the existing terminal antenna co-design schemes still
suffer from poor space utilization due to the large size of the sub-6GHz antennas
and the considerable area occupied by the feeding networks of mm-wave phased
array antennas, which also incur extra insertion losses. In this chapter, we present a
co-designed sub-6GHz and mm-wave antenna system for mobile terminals with
high space utilization. First, the working principles of the sub-6GHz and mm-wave
antennas are introduced, respectively, after which the co-design scheme is briefly
described. Then, an initial investigation is provided on techniques to achieve further
performance improvements in both bands. Finally, the feasibility of the proposed
co-design scheme is further evaluated by taking into account several practical design
considerations.
3.1 Working Principle of Co-Designed Antennas

In the sub-6GHz bands, the terminal chassis is utilized as the main radiator and
hence it plays a dominant role in the radiation properties of the antennas. Instead of
using self-resonant elements, non-resonant capacitive coupling elements (CCEs)
with proper matching networks have been proposed to excite the terminal chassis
[81]. Non-resonant coupling elements are becoming popular to realize low band
antennas due to their compactness and simple structure (see [82] and references
therein). Due to their ability to couple power into the electric near-field, CCEs
should be located at the region(s) where the electric field strength is maximum for
the mode(s) of interest. In Paper IV of this thesis, two CCEs are utilized to excite
the same wideband fundamental dipole mode of the chassis. However, different
impedance matching networks are used to match the CCE ports to 50 to cover two
non-overlapping sub-6GHz bands: 1) 0.75-1.13GHz, covering the LTE800/850/ 900
41
42 Introduction
bands; 2) 1.70-5.30GHz, covering the LTE1700-2600 and 5G NR n77-79 bands.
Since the electric field strength is maximum at the two shorter ends of the chassis
for this mode, the CCEs are placed at the diagonally opposite corners of the chassis
[83]. The detailed structure is presented in Paper IV.
Both the CCE size and the chassis dimensions influence the sub-6GHz antennas
performance. A larger CCE size facilitates a wider bandwidth for the sub-6GHz
band, so the CCE parameters are chosen considering the trade-off between the
bandwidth and size [84]. The resonant frequencies of the chassis modes (especially
the wideband fundamental dipole mode) of the terminal antennas in the sub-6GHz
bands are affected by the chassis dimensions. If the chassis dimensions are increased,
the resonant frequencies of the affected modes will decrease, which will likely
reduce the resonant frequencies of the two sub-6GHz ports. However, suitable
changes in the impedance matching circuits (which can be performed with the
matching software BetaMatch [85], for example) can be used to restore the required
bands.
A reconfigurable mm-wave antenna is proposed in Paper IV, which employs a
similar structure as the feeding source of the beam-reconfigurable PRS antenna
mentioned in Section 2.3, based on the concept of a Yagi-Uda antenna. The mm-
wave antenna utilizes two parasitic elements loaded with PIN diodes for beam
switching in two symmetrical deflected directions. Compared with the conventional
phased array, the proposed beam-reconfigurable array features a low-complexity
structure and occupies less space. Each mm-wave array can realize a 90-degree
beam scanning range with two states of the PIN diodes. In Paper IV, the insertion
losses of the PIN diode in the ON and OFF states are modelled by a 5.2Ω and a
10kΩ resistor, respectively. Instead of using the equivalent circuits, one may use the
measured scattering parameters (s2p files) to model the diode properties more
accurately, including its losses. The simulation radiation pattern results obtained by
using the s2p file show no appreciable change to the pattern shape, but the realized
gain is marginally reduced by 0.6dB due to the more accurate representation of the
diode losses.
In Paper IV, the compact beam-reconfigurable mm-wave arrays are integrated onto
the two corner CCEs of the sub-6GHz antennas for co-design purposes. The metal
ground planes of the mm-wave arrays are shared by the corner CCEs. Choosing the
diagonally displaced corners as excitation locations of the CCEs enables the four
mm-wave antennas on the two CCEs to cover the complete (360-degree) field of
view in a convenient manner. Furthermore, the corner placements are also intended
to mitigate blockage from the user’s hand(s).
In the co-designed system, due to the need of measuring the mm-wave antennas in
the prototype to validate simulation results, the loading effect of the mm-wave
connectors needs to be considered. Compared to the case of using an ideal port for
mm-wave antenna, the real connector causes the sub-6GHz bands of the prototype
Chapter 3. Co-Designed Antenna for Mobile Terminals 43
Figure 3.1: Radiation pattern of the mm-wave array with one more parasitic element,
relative to the original one (“pos” indicates positive bias voltage for switching the four PIN
diodes at port 4).
to shift towards lower frequencies. However, the loading of the real connector does
not fundamentally change the radiation properties of the chassis, and the required
bands can be restored with suitable changes in impedance matching circuits, as
accomplished in Paper IV.
3.2 Investigation of Further Performance Improvements

Paper IV provides a proof-of-concept to the proposed co-design scheme with high
space utilization and low-complexity structure. To explore the application potential
of the proposed antenna for future mobile terminals, an initial investigation of
further performance improvements is provided in this section, in terms of the
impedance bandwidth and radiation pattern performance of the mm-wave array, as
well as the impedance bandwidths of the two sub-6GHz antennas.
3.2.1. Investigation of mm-wave array

From the existing papers on mm-wave antennas for mobile terminal, the achieved
mm-wave antenna gain is found to vary from 7 to 12.5dBi, depending on the number
of antenna elements used. Typically, 8 elements would enable a 12.5dBi gain and 3
or 4 elements would offer a 7-10dBi gain. In Paper IV, the proposed mm-wave
44 Introduction
(a)
(b) (c)
Figure 3.2: (a) Mm-wave antenna with a stacked patch antenna as the active element. (b)
Re-simulated return loss in mm-wave band, relative to the original one. (c) Re-simulated
radiation patterns of mm-wave array in different frequencies.
array is less capable in achieving high gain relative to a normal phased array with
many (active) array elements and a feeding network. This is because only three
antenna elements are utilized in the proposed mm-wave array and two of them are
parasitic elements with the energy being coupled from the active element. The
simulated realized gain of the mm-wave antenna at 28GHz is 9.1dBi. The measured
gain is 7.9dBi, due to the losses in the PIN diodes and the slight discrepancy in the
pattern shape, which are common issues in mm-wave bands. With one more
parasitic element added in the proposed mm-wave antenna as shown in Fig. 3.1, the
realized gain can be increased to 10.2dBi. However, the beamwidth will also
decrease with higher gain, which will then require more reconfigurable states to
cover the original angular range. To design such a mm-wave antenna with higher
gain and more reconfigurable states is a promising direction for future work.
The mm-wave antenna proposed in Paper IV uses patch antenna elements, and the
function of the parasitic elements depends on its equivalent electrical size, which
makes the antenna exhibit narrowband characteristics. The simulated 10dB
(a) (b)
Figure 3.3: Simulated results of return losses of (a) LB antenna with a tuner and (b) HB
antenna with updated matching network.
impedance bandwidth of the mm-wave antenna is around 2GHz (26.56-28.54GHz).

The measured impedance bandwidth is narrower than 2GHz, due to the tolerance in
the soldering of the mm-wave cables. As an attempt to show the feasibility of
enhancing the bandwidth of the proposed mm-wave antenna, the regular patch
antenna is replaced with a stacked patch antenna as the active element (see Fig.
3.2(a)). From the re-simulated results, the obtained impedance band is then 26.42-
29.42GHz, giving around 3GHz of 10dB impedance bandwidth (see Fig. 3.2(b)).
The radiation pattern is also found to be stable over the operating band (see Fig.
3.2(c)). These results shows that there is a good potential for enhancing the
bandwidth of the mm-wave antenna in future work.
3.2.2. Investigation of sub-6GHz antennas

In Paper IV, the two sub-6GHz antennas cover the low band (LB: 0.79-0.96GHz)
and the high band (HB: 1.7-5GHz) with two ports, respectively. More effort can be
made to achieve wider bandwidths in the two sub-6GHz bands by using more
matching elements or a tuner. For example, the LB from 0.69 to 0.96 GHz can be
covered by using a tuneable inductor for L1 with four different inductance values as
shown in Fig. 3.3(a). The HB from 1.37 to 6.71GHz can be covered by updating the
matching network with a five-element matching network in Fig. 3.3(b). These
results are obtained by using the matching software BetaMatch [85].
46 Introduction
6mm
1mm
(a) (b)
Figure 3.4: (a) CCE structure with folded edges. (b) Radiation patterns of mm-wave array
(“pos” indicates positive bias voltage for switching the four PIN diodes at port 4).
3.3 Effects of Practical Design Considerations

As mentioned in Section 1.3, many metal components exist in the practical
environment of a mobile terminal, which may influence the performance of the
antennas. The effects of several practical design considerations on the performance
of the antennas proposed in Paper IV are evaluated. These considerations include
the presence of the metal frame and the touch screen, and the thickness of the mobile
terminal (which is related to the antenna height). Since the details on the effects of
the metal frame have been included in Paper IV, the effects of the touch screen and
the terminal thickness are provided in this section as a supplement.
3.3.1. Effects of frame height

The height of the frame (thickness of the mobile terminal) is chosen to be 8mm in
Paper IV, which is representative of the thickness of the latest smartphones on the
market. For example, based on keyword search in Google, Samsung S20 Ultra is
8.8mm thick and iPhone 13 is 7.65mm thick. When the height of the frame needs to
be reduced, the top and bottom edges of the CCEs can be folded, as shown in Fig.
3.4(a). Following this approach, the frame height and the folded edges are chosen
to be 6mm and 1mm, respectively, and the antennas are re-simulated. The
simulation results in Fig. 3.4(b) show that the mm-wave antenna with the folded
ground, which also serves as CCE, can still achieve the original radiation properties.
(a) (b)
Figure 3.5: (a) Simulated results of |S11|. (b) Simulated results of |S22| and updated matching
network (inset).
As shown in Fig. 3.5(a), the folded CCE causes almost no change in the matching
performance of the LB (0.79-0.96GHz) with the original matching network 1
proposed in Paper IV. In the HB (1.7-5GHz), the impedance matching is slightly
degraded at the lower band edge with the original matching network. However, it
still covers a wide HB from 1.85 to 5GHz (LTE1900-2600 and 5G NR n77-79
bands). Furthermore, the original working band in HB (1.5-5.7GHz) can be
recovered by updating the matching network with the one shown in the inset of Fig.
3.5(b).
3.3.2. Effects of touch screen

A touch screen typically consists of multiple substrate layers and a grounded
protective metal plate. For simplicity, it can be modelled by a metal plate due to the
limited influence from the thin dielectric layers (see [86] and references therein).
Therefore, to take the touch screen into account, it can be modeled by a metal plate
that is of the same size as the chassis, and it is placed 4mm above the chassis (see
Fig. 3.6). The metal plate is grounded to the chassis through a vertical shorting pin
located at the center position.
The terminal antenna was re-simulated with this updated chassis-screen structure.
Consistent to the findings of [86], the two-plate model does not affect the
fundamental dipole mode of the chassis, which is the primary mode of interest in
Paper IV. However, the impedance matching of the sub-6GHz antennas is affected
by this addition, which can be restored by updating the matching networks as shown
in Fig. 3.7. The performance of the mm-wave antenna is unaffected by the addition
48 Introduction
Figure 3.6: Terminal antenna with the metal plate.
(a) (b)
Figure 3.7: Simulated results of reflection coefficients in (a) LB (port 1) and (b) HB (port
2) with metal plate and updated matching networks.
of the metal plate, due to the CCEs acting as the ground plane, shielding the mm-
wave patch elements from any loading effect of the metal plate.
4 Conclusion and Outlook
This chapter briefly summarizes the work performed in each of the research papers
included in this thesis, as well as a few promising future research directions. For
each paper, the research contributions and my personal contributions as a co-author
are highlighted.
4.1 Research Contributions

Paper I
Comparison of Capacitive and Inductive Partially Reflective Surface Antenna
Using Ray-tracing
Research Contributions: In this paper, we provide an intuitive explanation for the
phenomenon that the capacitive PRS facilitates a larger antenna gain for a broadside
beam than the inductive PRS. We extend the classical ray-tracing approach to derive
analytically that the superior gain is primarily due to the greater height of the
capacitive PRS from the ground plane. The trends of the phase and magnitude
distributions across the PRS are predicted from ray-tracing analysis. To verify the
analytical study, the simulated phase and magnitude distributions of a capacitive
and an inductive PRS antenna are produced and shown to agree with the ray-tracing
analysis. This insight into the effect of PRS height on phase delay gives more clarity
to the operating principle of a PRS antenna.
Personal Contributions: I am the main contributor of this paper. I performed all
theoretical ray tracing derivation and ran all full-wave simulations. I took the lead
in writing the paper, with guidance and support of Buon Kiong Lau.
49
50 Introduction
Paper II
Beam-Reconfigurable Antenna with Inductive Partially Reflective Surface and
Parasitic Elements
Research Contributions: In this paper, a simple and intuitive ray-tracing model is
utilized to analyze the working principle of beam deflected PRS antenna with a
nonuniform two-part PRS, based on which the beam deflection capability of
different types of PRS is compared for the first time. The better beam deflection
capability of the inductive PRS than that of the capacitive PRS is derived from the
ray-tracing analysis and validated with full-wave simulation. The phenomenon of
the beam deflected feeding source further enhancing the beam deflection angle of
the PRS is also explained using ray-tracing analysis. Using the useful guidelines
from the theoretical analysis, a new beam-reconfigurable PRS antenna consisting of
a reconfigurable inductive PRS and a feeding source with parasitic elements is
proposed, where the use of multi-element antenna and lossy feeding network is
avoided in the feeding source. The proposed PRS antenna achieves a larger beam
deflection angle, lower SLLs, and a smaller gain variation than the existing
reconfigurable PRS antennas, while maintaining a low-complexity structure.
Personal Contributions: I am the main contributor of this paper. I conceived and
developed the idea. I performed all theoretical ray-tracing derivation and ran all full-
wave simulations. I planned the measurement work and the measurements were
conducted by Gaonan Zhou. I took the lead in writing the paper, with guidance and
support of Buon Kiong Lau.
Paper III
Dual-Band Shared-Aperture Antenna with Single-Layer Partially Reflecting
Surface
Research Contributions: In this paper, the feasible frequency ratio range of
traditional DS-FPC antennas with single-layer PRS is analyzed and a frequency
ratio gap is derived, given a specific reflection magnitude range for gain
enhancement. A dual-band shared-aperture antenna integrating a LB SBA and a HB
FPC antenna with a shared single-layer PRS is then proposed to fill the gap. The
PRS utilizes different unit cells for the inner and outer parts to realize different
reflection coefficient distributions for different functions in LB and HB. By using
the PRS and a parasitic element, a low-profile SBA working in the LB is achieved.
In HB, the FPC antenna works with the first-order FPC mode. Without relying on a
traditional dual-band resonance cavity, the proposed antenna achieves more flexible
frequency ratio with a simple structure.
Personal Contributions: I am the main contributor of this paper. I performed all
theoretical derivation and conceived the idea of antenna design. I ran all full-wave
simulations. I planned the measurement work and the measurements were
Chapter 4. Conclusion and Outlook 51
conducted by Gaonan Zhou. I took the lead in writing the paper, with guidance and
support of Buon Kiong Lau.
Paper IV
Co-Designed Millimeter-Wave and Sub-6GHz Antenna for 5G Smartphones
Research Contributions: In this paper, a co-designed mm-wave and sub-6GHz
antenna system is proposed for 5G smartphone application. This antenna system
utilizes a shared-aperture configuration to accommodate four mm-wave arrays and
two sub-6GHz antennas in a compact space. The mm-wave array employs parasitic
elements loaded with PIN diodes, providing beam steering capability without lossy
feeding network. The CCEs that excite the sub-6GHz bands are shared by the mm-
wave arrays as ground plane, facilitating high space utilization and low complexity.
Personal Contributions: I am the main contributor of this paper. I conceived the
structure of mm-wave antenna and ran all simulation for the co-designed antenna
system. This work is done in collaboration with Hanieh Aliakbari, who conceived
the structure of sub-6GHz antenna and designed the initial matching networks. I
conducted all the measurements. I took the lead in writing the paper, with guidance
and support of the other co-authors.
4.2 General Conclusions

This thesis contains the results obtained from research on low-complexity PRS
antennas for base stations and co-designed antenna for mobile terminals. The
following are a few general conclusions and observations drawn from the research
work in the two research areas.
PRS antennas:
 The superior gain of the capacitive PRS compared to the inductive PRS for a
given reflection magnitude can be explained with a ray-tracing model that
accounts for the oblique incidence. The capacitive PRS and the inductive PRS
exhibit different phases, which correspond to different PRS heights, given the
same reflection magnitude. Therefore, the capacitive PRS can achieve less
phase delay and more uniform magnitude distribution over the PRS, which
result in higher antenna gain.
 A ray-tracing analysis revealed that an inductive nonuniform PRS facilitates a
larger beam deflection and a lower SLL than a capacitive one, given the same
reflection phase difference. This is because the inductive and capacitive PRSs
have opposite trends in the reflection magnitude for the same trend in the
reflection phase, and the inductive PRS offers better beam performance due to
52 Introduction
the same trend in the reflection magnitude and phase. Additional beam
deflection of the PRS antenna can be provided by a beam deflected feeding
source because the phase difference created by the feeding source adds
constructively with that provided by the PRS.
 The reflection magnitude of the single-layer PRS is related to its reflection
phase, based on the analysis of an equivalent two-ports system. The gain
enhancement effect of the PRS requires high reflection magnitude, which leads
to the reflection phase of the PRS being constrained to a specific range. As a
result, the frequency ratio of a traditional DS-FPC antenna with single-layer
PRS is also constrained to a specific range, as determined by the FPC modes.
By choosing proper structure parameters, an SBA can be integrated with a FPC
antenna to facilitate flexible frequency ratio to fill the frequency ratio gap,
while maintaining a low profile and a simple structure.
Co-designed terminal antennas:
 It has been validated with measurements that mm-wave antennas with parasitic
elements can provide adequate beam steering capability for mobile terminals.
The CCEs that excite the sub-6GHz antennas can be shared by the mm-wave
antennas as their ground planes, without changing the fundamental radiation
modes of the chassis. The co-design scheme can still function properly after
taking several practical design considerations into account, such as the presence
of a metal screen and a frame, and the potential need for a thinner mobile
terminal.
4.3 Future Research

Many aspects within and beyond the studied topics could be investigated further.
The following is a selection of the most interesting and promising directions for
future research:
PRS antennas:
 To further increase the coverage range, 2-D beam steering PRS antenna may
be designed using a reconfigurable inductive PRS that consists of more parts
and a 2-D beam-reconfigurable feeding source.
 Based on the existing PRS antennas in this thesis, the dual-polarization feature
could be implemented to provide polarization diversity and higher data rate.
 With the clear insight into the working principle of PRS antennas as well as the
PRS properties, a promising direction is to design the PRS antennas with
diverse beam shapes by reconfiguring the reflection coefficient distributions
across the PRS, to cater for more application scenarios.
Chapter 4. Conclusion and Outlook 53
Co-designed terminal antennas:
 A possible future work for the mm-wave antennas is to provide more refined
beam steering with higher gain, by using more parasitic elements with more
reconfigurable states. Multiple-input multiple-output (MIMO) operation could
be added in the sub-6GHz bands of the co-designed system for higher data rate,
by designing more CCEs on the terminal chassis that can excite other modes
without affecting the fundamental mode.
 Based on the concept of the PRS antenna, the reconfigurable PRS could be
employed in the mm-wave antenna to provide high gain beam steering, as a
low-complexity alternative to phased array. However, due to the relatively
large wavelength of the 28GHz band, such a concept may be more attractive
and feasible for even higher bands such as the D-band, which is being
considered for 6G.
Bibliography
[1] J. Navarro Ortiz, P. Romero Diaz, S. Sendra, P. Ameigeiras, J. J. Ramos-
Munoz, and J. M. Lopez Soler, “A survey on 5G usage scenarios and traffic
models,” IEEE Commun. Surv. Tut., vol. 22, no. 2, pp. 905- 929, 2020.
[2] C. Yeh, G. D. Jo, Y. J. Ko, and H. K. Chung, “Perspectives on 6G wireless
communications,” ICT Express, in press.
[3] B. T. Jijo, S. R. M. Zeebaree, R. R. Zebari, M. A. M. Sadeeq, A. B. Sallow, S.
Mohsin, and Z. S. Ageed, “A comprehensive survey of 5G mm-wave
technology design challenges,” Asian J. Comp. Sci. Inf. Technol., vol. 8, no. 1,
pp. 1-20, 2021.
[4] M. Z. Chowdhury, M. Shahjalal, S. Ahmed, and Y. M. Jang, “6G wireless
communication systems: applications, requirements, technologies, challenges,
and research directions,” IEEE Open J. Commun. Society, vol. 1, pp. 957-975,
2020.
[5] X. Lin, et al., “5G new radio: unveiling the essentials of the next generation
wireless access technology,” IEEE Commun. Stand. Mag., vol. 3, no. 3, pp. 30-
37, Sept. 2019.
[6] P. Tiwari and P. K. Malik, “Design of UWB antenna for the 5G mobile
communication applications: a review,” in Proc. 2020 Int. Conf, Comp. Autom.
Knowl. Manag. (ICCAKM), 2020, pp. 24-30.
[7] K. L. Wong, H. J. Chang, C. Y. Wang, and S. Y. Wang, “Very-low-profile
grounded coplanar waveguide-fed dual-band WLAN slot antenna for on-body
antenna application,” IEEE Antennas Wireless Propag. Lett., vol. 19, no. 1, pp.
213-217, Jan. 2020.
[8] G. N. Zhou, B. H. Sun, Q. Y. Liang, S. T. Wu, Y. H. Yang, and Y. M. Cai,
“Triband dual-polarized shared-aperture antenna for 2G/3G/4G/5G base
station applications,” IEEE Trans. Antennas Propag., vol. 69, no. 1, pp. 97-108,
Jan. 2021.
55
56 Introduction
[9] H. C. Huang and J. Lu, “Retrospect and prospect on integrations of millimeter-
wave antennas and non-millimeter-wave antennas to mobile phones,” IEEE
Access, vol. 10, pp. 48904-48912, 2022.
[10] M. Shafi, et al., “5G: a tutorial overview of standards, trials, challenges,
deployment, and practice,” IEEE J. Sel. Areas Commun., vol. 35, no. 6, pp.
1201-1221, June 2017.
[11] J. Helander, “Millimeter wave imaging and phased array antennas,” PhD thesis,
Lund University, Sweden, 2019.
[12] W. Hong, et al., “Multibeam antenna technologies for 5G wireless
communications,” IEEE Trans. Antennas Propag., vol. 65, no. 12, pp. 6231-
6249, Dec. 2017.
[13] R. L. Haupt and Y. Rahmat-Samii, “Antenna array developments: a perspective
on the past, present and future,” IEEE Antennas Propaga. Mag., vol. 57, no. 1,
pp. 86-96, Feb. 2015.
[14] S. M. Moon, S. Yun, I. B. Yom, and H. L. Lee, “Phased array shaped-beam
satellite antenna with boosted-beam control,” IEEE Trans. Antennas Propag.,
vol. 67, no. 12, pp. 7633-7636, Dec. 2019.
[15] H. T. Chou and D. Torrungrueng, “Development of 2-D generalized tri-focal
rotman lens beamforming network to excite conformal phased arrays of
antennas for general near/far-field multi-beam radiations,” IEEE Access, vol.9,
pp. 49176-49188, 2021.
[16] C. C. Chang, R. H. Lee, and T. Y. Shih, “Design of a beam switching/steering
butler matrix for phased array system,” IEEE Trans. Antennas Propag., vol. 58,
no. 2, pp. 367-374, Feb. 2010.
[17] A. Rudge, “Multiple-beam antennas: offset reflectors with offset feeds,” IEEE
Trans. Antennas Propag., vol. 23, no. 3, pp. 317-322, May 1975.
[18] B. Fuchs, et al., “Comparative design and analysis of Luneburg and half
Maxwell fish-eye lens antennas,” IEEE Trans. Antennas Propag., vol. 56, no.
9, pp. 3058-3062, Sept. 2008.
[19] M. Jiang, Z. N. Chen, Y. Zhang, W. Hong, and X. Xuan, “Metamaterial-based
thin planar lens antenna for spatial beamforming and multibeam massive
MIMO,” IEEE Trans. Antennas Propag., vol. 65, no. 2, pp. 464-472, Feb. 2017.
[20] G. B. Wu, S. W. Qu, S. Yang, and C. H. Chan, “Low-cost 1-D beam-steering
reflectarray with ±70° scan coverage,” IEEE Trans. Antennas Propag., vol. 68,
no. 6, pp. 5009-5014, June 2020.
[21] G. V. Trentini, “Partially reflecting sheet arrays,” IEEE Trans. Antennas
Propag., vol. 4, no. 4, pp. 666-671, Oct. 1956.
Bibliography 57
[22] R. A. Alhalabi and G. M. Rebeiz, “High-gain Yagi-Uda antennas for
millimeter-wave switched-beam systems,” IEEE Trans. Antennas Propag., vol.
57, no. 11, pp. 3672-3676, Nov. 2009.
[23] R. Li, D. Thompson, M. M. Tentzeris, J. Laskar, and J. Papapolymerou,
“Development of a wide-band short backfire antenna excited by an unbalance-
fed H-shaped slot,” IEEE Trans. Antennas Propag., vol. 53, no. 2, pp. 662-671,
Feb. 2005.
[24] S. V. Hum and J. Perruisseau-Carrier, “Reconfigurable reflectarrays and array
lenses for dynamic antenna beam control: a review,” IEEE Trans. Antennas
Propag., vol. 62, no. 1, pp. 183-198, Jan. 2014.
[25] M. I. Abbasi, M. Y. Ismail, and M. R. Kamarudin, “Development of a PIN
diode-based beam-switching single-layer reflectarray antenna,” Int. J.
Antennas Propag., vol. 2020, no. 8891759, 2020.
[26] M. A. Towfiq, I. Bahceci, S. Blanch, J. Romeu, L. Jofre, and B. A. Cetiner, “A
reconfigurable antenna with beam steering and beamwidth variability for
wireless communications,” IEEE Trans. Antennas Propag., vol. 66, no. 10, pp.
5052-5063, Oct. 2018.
[27] H. Zhai, K. Zhang, S. Yang, and D. Feng, “A low-profile dual-band dual-
polarized antenna with an AMC surface for WLAN applications,” IEEE
Antennas Wireless Propag. Lett., vol. 16, pp. 2692-2695, 2017.
[28] B. Barakali, “Pattern and polarization reconfigurable antennas for gain
enhancement,” PhD thesis, University of Sheffield, UK, 2019.
[29] L. Leszkowska, M. Rzymowski, K. Nyka, and L. Kulas, “High-gain compact
circularly polarized X-band superstrate antenna for cubesat applications,”
IEEE Antennas Wireless Propag. Lett., vol. 20, no. 11, pp. 2090-2094, Nov.
2021.
[30] M. L. Abdelghani, H. Attia, and T. A. Denidni, “Dual- and wideband Fabry–
Pérot resonator antenna for WLAN applications,” IEEE Antennas Wireless
Propag. Lett., vol. 16, pp. 473-476, 2017.
[31] Y. Qin, R. Li, Q. Xue, X. Zhang, and Y. Cui, “Aperture-shared dual-band
antennas with partially reflecting surfaces for base-station applications,” IEEE
Trans. Antennas Propag, vol. 70, no. 5, pp. 3195-3207, May 2022.
[32] M. U. Afzal and K. P. Esselle, “Steering the beam of medium-to-high gain
antennas using near-field phase transformation,” IEEE Trans. Antennas.
Propag., vol. 65, no. 4, pp. 1680-1690, Apr. 2017.
[33] X. Yang, Y. Liu, H. Lei, Y. Jia, P. Zhu, and Z. Zhou, “A radiation pattern
reconfigurable Fabry-Pérot antenna based on liquid metal,” IEEE Trans.
Antennas Propag., vol. 68, no. 11, pp. 7658-7663, Nov. 2020.
58 Introduction
[34] R. Guzman Quiros, J. L. Gomez Tornero, A. R. Weily, and Y. J. Guo,
“Electronically steerable 1-D Fabry-Pérot leaky-wave antenna employing a
tunable high impedance surface,” IEEE Trans. Antennas Propag., vol. 60, no.
11, pp. 5046-5055, Nov. 2012.
“Electronic full-space scanning with 1-D Fabry-Pérot LWA using
electromagnetic band-gap,” IEEE Antennas Wireless Propag. Lett., vol. 11, pp.
1426-1429, Nov. 2012.
[36] A. Ourir, S. N. Burokur, and A. de Lustrac, “Electronic beam steering of an
active metamaterial-based directive subwavelength cavity,” in Proc. 2nd Eur.
Conf. Antennas Propag. (EuCAP), Edinburgh, U.K., Nov. 2007, pp. 1-4.
[37] L. Ji, P. Qin, J. Li, and L. Zhang, “1-D electronic beam-steering partially
reflective surface antenna,” IEEE Access, vol. 7, pp. 115959-115965, 2019.
[38] R. Guzman Quiros, A. R. Weily, J. L. Gomez Tornero, and Y. J. Guo, “A
Fabry-Pérot antenna with two-dimensional electronic beam scanning,” IEEE
Trans. Antennas Propag., vol. 64, no. 4, pp. 1536-1541, Apr. 2016.
[39] P. Xie, G. Wang, H. Li, and J. Liang, “A dual-polarized two-dimensional beam-
steering Fabry-Pérot cavity antenna with a reconfigurable partially reflecting
surface,” IEEE Antennas Wireless Propag. Lett., vol. 16, pp. 2370-2374, 2017.
[40] L. Ji, Z. Y. Zhang, and N. W. Liu, “A two-dimensional beam-steering partially
reflective surface (PRS) antenna using a reconfigurable FSS structure,” IEEE
Antennas Wireless Propag. Lett., vol. 18, no. 6, pp. 1076-1080, Jun. 2019.
[41] L. Ji, Y. J. Guo, P. Y. Qin, S. X. Gong, and R. Mittra, “A reconfigurable
partially reflective surface (PRS) antenna for beam steering,” IEEE Trans.
Antennas Propag., vol. 63, no. 6, pp. 2387-2395, Jun. 2015.
[42] T. Debogovic and J. Perruisseau-Carrier, “Array-fed partially reflective surface
antenna with independent scanning and beamwidth dynamic control,” IEEE
Trans. Antennas Propag., vol. 62, no. 1, pp. 446-449, Jan. 2014.
[43] F. Meng and S. K. Sharma, “A dual-band high-gain resonant cavity antenna
with a single layer superstrate,” IEEE Trans. Antennas Propag., vol. 63, no. 5,
pp. 2320-2325, May 2015.
[44] Y. Lv, X. Ding, and B. Wang, “Dual-wideband high-gain Fabry-Perot cavity
antenna,” IEEE Access, vol. 8, pp. 4754-4760, Dec. 2020.
[45] J. Chen, Y. Zhao, Y. Ge, and L. Xing, “Dual-band high-gain Fabry-Perot cavity
antenna with a shared-aperture FSS layer,” IET Microw. Antennas Propag., vol.
12, no. 13, pp. 2007-2011, Feb. 2018.
Bibliography 59
[46] C. Chen, Z. G. Liu, H. Wang, and Y. Guo, “Metamaterial-inspired self-
polarizing dual-band dual-orthogonal circularly polarized Fabry-Pérot
resonator antennas,” IEEE Trans. Antennas Propag., vol. 67, no. 2, pp. 1329-
1334, Feb. 2019.
[47] A. H. Abdelrahman, F. Yang, A. Z. Elsherbeni, and P. Nayeri. Analysis and
Design of Transmitarray Antennas, Morgan & Claypool, 2017.
[48] H. Moghadas, M. Daneshmand, and P. Mousavi, “A dual-band high-gain
resonant cavity antenna with orthogonal polarizations,” IEEE Antennas
Wireless Propag. Lett., vol. 10, pp. 1220-1223, 2011.
[49] B. A. Zeb, Y. Ge, K. P. Esselle, Z. Sun, and M. E. Tobar, “A simple dual-band
electromagnetic band gap resonator antenna based on inverted reflection phase
gradient,” IEEE Trans. Antennas Propag., vol. 60, no. 10, pp. 4522-4529, Oct.
2012.
[50] B. A. Zeb, N. Nikolic, and K. P. Esselle, “A high-gain dual-band EBG
resonator antenna with circular polarization,” IEEE Antennas Wireless Propag.
Lett., vol. 14, pp. 108-111, Sep. 2015.
[51] F. Qin, et. al., “A simple low-cost shared-aperture dual-band dual-polarized
high-gain antenna for synthetic aperture radars,” IEEE Trans. Antennas
Propag., vol. 64, no. 7, pp. 2914-2922, Jul. 2016.
[52] M. L. Abdelghani, H. Attia, and T. A. Denidni, “Dual- and wideband Fabry-
Propag. Lett., vol. 16, pp. 473-476, Jan. 2017.
[53] Z. Liu, R. Yin, Z. Ying, W. Lu, and K. Tseng, “Dual-band and shared-aperture
Fabry-Perot cavity antenna,” IEEE Antennas Wireless Propag. Lett., vol. 20,
no. 9, pp. 1686-1690, Sep. 2021.
[54] A. Pirhadi, M. Hakkak, F. Keshmiri, and R. Karimzadeh Baee, “Design of
compact dual band high directive electromagnetic bandgap (EBG) resonator
antenna using atificial magnetic conductor,” IEEE Trans. Antennas Propag.,
vol. 55, no. 6, pp. 1682-1690, Jun. 2007.
[55] J. Qi, Q. Wang, F. Deng, Z. Zeng, and J. Qiu, “Low-profile uni-cavity high-
gain FPC antenna covering entire global 2.4 GHz and 5 GHz WIFI-bands using
uncorrelated dual-band PRS and phase compensation AMC,” IEEE Trans.
Antennas Propag., early access. July 2022.
[56] P. Mei, S. Zhang, and G. F. Pedersen, “A dual-polarized and high-gain X-/Ka-
band shared-aperture antenna with high aperture reuse efficiency,” IEEE Trans.
Antennas Propag, vol. 69, no. 3, pp. 1334-1344, Mar. 2021.
60 Introduction
[57] J. Wu, D. Zhou, X. Lei, J. Gao, and H. Hu, “A high gain Fabry-Perot cavity
antenna designed by modified ray tracking model,” in Proc. Int. Workshop
Antenna Technol. (iWAT), Nanjing, China, Mar. 5-7, 2018, pp. 1-4.
[58] B. Ratni, W. A. Merzouk, A. de Lustrac, S. Villers, G. P. Piau, and S. N.
Burokur, “Design of phase-modulated metasurfaces for beam steering in
Fabry-Pérot cavity antennas,” IEEE Antennas Wireless Propag. Lett., vol. 16,
pp. 1401-1404, 2017.
[59] A. Foroozesh and L. Shafai, “Investigation into the effects of the reflection
phase characteristics of highly-reflective superstrates on resonant cavity
antennas,” IEEE Trans. Antennas Propag., vol. 58, no. 10, pp. 3392-3396, Oct.
2010.
[60] A. Foroozesh and L. Shafai, “On the characteristics of the highly directive
resonant cavity antenna having metal strip grating superstrate,” IEEE Trans.
Antennas Propag., vol. 60, no. 1, pp. 78-91, Jan. 2012.
[61] C. Lee, M. K. Khattak, and S. Kahng, “Wideband 5G beamforming printed
array clutched by LTE-A 4 × 4-multiple-input-multiple-output antennas with
high isolation,” IET Microw. Antennas Propag., vol. 12, no. 8, pp. 1407-1413,
Mar. 2018.
[62] J. Kurvinen, H. Kähkönen, A. Lehtovuori, J. Ala-Laurinaho, and V. Viikari,
“Co-designed mm-wave and LTE handset antennas,” IEEE Trans. Antennas
Propag., vol. 67, no. 3, pp. 1545-1553, Mar. 2019.
[63] M. Ikram, R. Hussain, and M. S. Sharawi, “4G/5G antenna system with dual
function planar connected array,” IET Microw. Antennas Propag., vol. 11, no.
12, pp. 1760-1764, Sep. 2017.
[64] M. M. Samadi Taheri, A. Abdipour, S. Zhang, and G. F. Pedersen, “Integrated
millimeter-wave wideband end-fire 5G beam steerable array and low-
frequency 4G LTE antenna in mobile terminals,” IEEE Trans. Veh. Technol.,
vol. 68, no. 4, pp. 4042-4046, Apr. 2019.
[65] H. Wang, “Overview of future antenna design for mobile terminals,”
Engineering, vol. 11, no. 4, 12-14, 2022.
[66] R. Hussain, A. T. Alreshaid, S. K. Podilchak, and M. S. Sharawi, “Compact 4G
MIMO antenna integrated with a 5G array for current and future mobile
handsets,” IET Microw. Antennas Propag., vol. 11, no. 2, pp. 271-279, Jan.
2017.
[67] M. S. Sharawi, M. Ikram, and A. Shamim, “A two concentric slot loop based
connected array MIMO antenna system for 4G/5G terminals,” IEEE Trans.
Antennas Propag., vol. 65, no. 12, pp. 6679-6686, Dec. 2017.
Bibliography 61
[68] Y. Liu, Y. Li, L. Ge, J. Wang, and B. Ai, ‘‘A compact hepta-band mode-
composite antenna for sub (6, 28, and 38) GHz applications,’’ IEEE Trans.
Antennas Propag., vol. 68, no. 4, pp. 2593-2602, Apr. 2020.
[69] M. Ikram, E. A. Abbas, N. Nguyen-Trong, K. H. Sayidmarie, and A. Abbosh,
“Integrated frequency-reconfigurable slot antenna and connected slot antenna
array for 4G and 5G mobile handsets,” IEEE Trans. Antennas Propag., vol. 67,
no. 12, pp. 7225-7233, Dec. 2019.
[70] R. M. Moreno, et al., “Dual-polarized mm-wave endfire antenna for mobile
devices,” IEEE Trans. Antennas Propag., vol. 68, no. 8, pp. 5924-5934, Aug.
2020.
[71] R. C. Rocio, S. Zhang, K. Zhao, and G. F. Pedersen, “Mm-wave beam-steerable
endfire array embedded in a slotted metal-frame LTE antenna,” IEEE Trans.
Antennas Propag., vol. 68, no. 5, pp. 3685-3694, May 2020.
[72] S. Zhang, I. Syrytsin, and G. F. Pedersen, “Compact beam-steerable antenna
array with two passive parasitic elements for 5G mobile terminals at 28 GHz,”
IEEE Trans. Antennas Propag., vol. 66, no. 10, pp. 5193-5203, Oct. 2018.
[73] C. A. Balanis, Antenna Theory Analysis and Design, 3rd ed., John Wiley & Sons,
2005.
[74] D. M. Pozar, Microwave Engineering, 3rd ed., John Wiley & Sons, Inc., 2005.
[75] B. A. Munk, Frequency Selective Surfaces: Theory and Design, John Wiley &
Sons, 2000.
[76] L. Zhou, X. Chen, Y. Cui, and X. Duan, “Comparative effects of capacitive and
inductive superstrates on the RCA’s gain,” IET Microw. Antennas Propag., vol.
12, pp. 1834-1838, May 2018.
[77] IEEE145-2013, IEEE Standard Definition of Terms for Antennas, Antenna
Standards Committee of the IEEE Antennas and Propagation Society, Dec.
1993.
[78] J. Huang and A. C. Densmore, “Microstrip Yagi array antenna for mobile
satellite vehicle application,” IEEE Trans. Antennas Propag., vol. 39, no. 7, pp.
1024-1030, Jul. 1991.
Feb. 2005.
[80] G. Zhou, B. Sun, Q. Liang, Y. Yang, and J. Lan, “Beam-deflection short
backfire antenna using phase-modulated metasurface,” IEEE Trans. Antennas
62 Introduction
[81] J. Villanen, J. Ollikainen, O. Kivekas, and P. Vainikainen, “Coupling element
based mobile terminal antenna structures,” IEEE Trans. Antennas Propag., vol.
54, no. 7, pp. 2142-2153, Jul. 2006.
[82] H. Aliakbari and B. K. Lau, “Low-profile two-port MIMO terminal antenna for
low LTE bands with wideband multimodal excitation,” IEEE Open Journal of
Antennas and Propag., vol. 1, pp. 368-378, 2020.
[83] R. Valkonen, M. Kaltiokallio, and C. Icheln, “Capacitive coupling element
antennas for multi-standard mobile handsets,” IEEE Trans. on Antennas
Propag., vol. 61, no. 5, pp. 2783-2791, May 2013.
[84] H. Aliakbari and B. K. Lau, “Impact of capacitive coupling element design on
antenna bandwidth,” in Proc. 12th Europ. Conf. Antennas Propag. (EuCAP),
London, UK, Apr. 9-13, 2018, pp. 1-4.
[85] Betamatch. (2019). MNW Scan Pte Ltd, Version 3.7.6. Accessed: Nov. 11,
2021. [Online]. Available: http://www.mnw-scan.com
[86] H. Aliakbari, L. Y. Nie, and B. K. Lau, “Large screen enabled tri-port MIMO
handset antenna for low LTE bands,” IEEE Open J. Antennas Propag., vol. 2,
pp. 911-920, Aug. 2021.
Part II
Included Papers
Paper I
Comparison of Capacitive and Inductive
Partially Reflective Surface Antenna
Using Ray-Tracing
A partially reflective surface (PRS) antenna is a low-cost high gain antenna,

which is often classified by the capacitive or inductive nature of the PRS. It has
been shown that, for a given reflection magnitude, capacitive PRS facilitates a
larger antenna gain than inductive PRS, based on leaky-wave model and
practical design examples. In this paper, we extend the classical ray-tracing
approach to derive analytically that the superior gain is primarily due to the
greater height of the capacitive PRS from the ground plane, which leads to a
smaller phase delay and a more uniform magnitude distribution across the
PRS. To verify the analytical study, the simulated phase and magnitude
distributions of a capacitive and an inductive PRS antenna were produced and
shown to agree with the predicted trends of the phase and magnitudes
distributions from ray-tracing. The gain of the antenna with the capacitive PRS
is 16.6dBi, which is 1.8dBi higher than that of the antenna with the inductive
PRS. This insight on the effect of PRS height on phase delay gives more clarity
to the operating principle of PRS antenna.
©2022 IEEE. Reprinted, with permission, from

Qiuyan Liang and Buon Kiong Lau,
“Comparison of Capacitive and Inductive Partially Reflective Surface Antenna Using Ray-Tracing,”
in Proc. 16th Europ. Conf. Antennas Propag. (EuCAP’2022), Madrid, Spain, Mar. 27- Apr. 1, 2022.
PAPER I 69
1 Introduction
A partially reflective surface (PRS) antenna, consisting of a feeding source, a PRS
and a completely reflecting ground, can be used to provide high directivity with
simple structure and low cost [1]. According to the classical working principle of
PRS antenna, the waves from the feed are reflected multiple times between the PRS
and the ground, with each incident wave at the PRS also producing a transmitted
wave. The PRS is designed so that the multiple transmitted waves are in-phase over
the PRS, thus achieving high directivity in the broadside direction.
PRSs have been classified into two types according to the equivalent susceptance of
their periodic structures: capacitive PRS with susceptance larger than 0, and
inductive PRS with susceptance less than 0 [2], [3]. Based on the analysis of
different unit cell structures, it has been observed that the capacitive PRS exhibits
higher gain than inductive PRS, for a given reflection magnitude [4], [5]. This
observation has been explained using the leaky-wave model [3], with the PRS
antenna being modelled as a (leaky) waveguide surrounded by the PRS and ground.
It is concluded that the capacitive PRS offers a smaller propagation constant, which
leads to a bigger radiation aperture and a smaller variation of the aperture field phase,
thus resulting in a higher gain.
In the classical article that first proposed PRS antenna [1], its working principle and
design parameters are all based on an intuitive ray-tracing (or ray-tracking) model.
However, the emphasis is on the ray with normal incidence at the PRS. Ray-tracing
has also been used to in many subsequent designs of such antennas [6]-[8]. In [7]
and [8], rays from oblique angles are utilized to achieve higher gain and beam-
steering, showing the usefulness of accounting for oblique incidence.
In this work, we apply the general model with oblique incident rays to offer an
alternative and arguably more intuitive explanation for the superior antenna gain
afforded by capacitive PRS over inductive PRS, than that provided using the leaky-
wave model [3]. The phase delay distribution of rays with different oblique incident
angles, which is not considered in the classical work [1], is studied in detail using
ray-tracing. When considering capacitive PRS and inductive PRS designed with
different reflection phases but the same reflection magnitude, different phase delay
and magnitude distributions are observed over the surface, resulting in different
antenna gains. Ray-tracing reveals that the higher gain of the capacitive PRS is
attributed to its greater height above the ground than the inductive PRS, which
provides more uniform phase delay and magnitude distributions across the PRS.
This in turn facilitates more uniform wave fronts across the PRS, leading to higher
antenna gain. The capacitive and inductive PRS antennas were simulated in a full-
wave solver to verify the predicted trends in the phase delay and magnitude
distributions.
70 PAPER I
Figure 1: Path of the ray with oblique incident angle (in black), contributing to phase delays
across the PRS, relative to that of normal incidence (in blue).
2 Derivation of Phase Delay Distribution

In the following, the influence of the incident angle of a single ray on the phase
delay distribution is derived. This result is then extended to compare the phase
delays of multiple rays transmitted by the PRS at the same point. Finally, the impact
on the phase delay from a ray of a larger incident angle is demonstrated. These basic
results are then utilized in Section 3 to explain the superior antenna gain facilitated
by capacitive PRS.
2.1 Phase Delay Analysis of Rays with Oblique Incidence

The original ray-tracing analysis framework allowed for an arbitrary incident angle
α of the ray from the source (α is the angle between the ray and broadside direction,
i.e., α  [0°, 90°]) [1]. However, to simplify the theoretical analysis of directivity in
the broadside direction, only the normal incident angle α = 0° is considered.
According to this assumption, the path length of the reflected ray between the PRS
and completely reflecting ground is equal to the vertical distance l (or PRS height),
which means the ray will never go beyond the center of the PRS even after multiple
reflections. To assume otherwise would be unphysical, as illustrated by the dashed
blue (horizontal) part of the ray path in Fig. 1. The transmitted wave rays drawn in
the figure are only used to distinguish the initial angle of incidence. Each transmitted
wave does not continue to propagate along this initial angle. Instead, it can be
regarded as a new source re-radiating in the upper half space.
PAPER I 71
Figure 2: Paths of rays with two different oblique incident angles (in black and green,
respectively), giving different phase delay distributions across the PRS. The proportions of
power in the transmitted rays are also indicated, with  being the reflection magnitude of the
PRS.
In practice, the waves transmitted through the off-center positions of the PRS are
from oblique incident angles, with path lengths in odd multiples of l cos α , which
cause phase discrepancies along the PRS compared with the simplified model that
assumes α  0° . To simplify the analysis, it is assumed that the reflection and
transmission phase and magnitude of the PRS are independent of the incident angle.
The phase delay between the practical ray n′ and the simplified ray n can be
expressed as
2π  1 
Φn   2n  1 l 1  , (1)
λ  cos α 
where λ is the wavelength in free space and n = 0, 1, 2, etc. When the ray is from
the incident angle of α  0° , the phase error Φ n takes a negative value, representing
an excess phase delay of ray n′ relative to the simplified ray n. The ray with a
constant incident angle has more delay when it is reflected more times
(corresponding to a larger n), This explains why the edge position at PRS has more
phase delay than the middle position. Moreover, this result can be used to analyze
the dependence of phase delay on oblique incident angle, for a given location on the
PRS.
72 PAPER I
2.2 Phase Delays of Rays with Different Incident Angles

At a particular location of the PRS, the transmitted wave is composed of rays
originating from different angles. These rays are reflected by different numbers of
times and arrive at the same location with different ray indices. In such a condition,
their incident angles are related to each other in a specific manner. Two rays with
incident angles of α1 and α2 (α2 > α1), as depicted in Fig. 2, can be used to explain
this relationship. Ray 0 of these two rays arrive at the PRS with the distances of x1
and x2 from the center, respectively, and these distances can be related to the incident
angles and l as
x1 x
l  2 . (2)
tan α1 tan α2
If the ray 1 with incident angle of α1 meets ray 0 with incident angle of α2 at x2, then
x1 and x2 should satisfy the following equation
x2  3x1 . (3)
Similarly, if ray n1 with the incident angle of α1 meets ray n2 with the incident angle
of α2 at point M, the relationships in (2) and (3) can be generalized as follows
(2n2  1) x2  (2n1  1) x1 , (4)
2n1  1
tan α2  tan α1 . (5)
2n2  1
From (1) and (2), the phase delay of ray n1 with the incident angle of α1 and that of
ray n2 with the incident angle of α2 can be expressed as follows
2π
Φn1   2n1  1 x1G(α1 ) , (6)
λ
2π
Φn 2   2n2  1 x2G(α2 ) , (7)
λ
1  1 
G (α )  1  . (8)
tan α  cos α 
Since the function G(α) with α  0° is a monotonically decreasing function of
variable α, it can be proven using (4) that Φn2 < Φn1, which means that the ray n2
with a larger incident angle causes a larger phase delay than the ray n1 at point M.
However, the net impact of the phase delay due to the two rays transmitted through
a point M is dependent on the proportion of power in these rays, which will be
explored next.
PAPER I 73
2.3 Impact of Phase Delay from Increasing Power in a Ray with a

Larger Incident Angle
Assuming no transmission loss and the total ray at point M to be composed of only
ray n1 and ray n2, with the amplitudes of the ray n1 and ray n2 being proportional to
ρ n1 1  ρ 2 and ρ n2 1  ρ 2 at point M, then the proportion of power in ray n2
can be expressed as
ρ2 n2 (1  ρ2 ) f 2 (α2 )
β (α2 )  , (9)
ρ2 n1 (1  ρ2 ) f 2 (α1 )  ρ2 n2 (1  ρ2 ) f 2 (α2 )
where ρ is the reflection magnitude of the PRS and f 2(α) represents the power
pattern of the feeding element source at the angle α. The denominator of (9)
represents the total power of the waves from both angles α1 and α2. Dividing (9) by
ρ2n2 (1  ρ2 ) f 2 (α2 ) in the numerator and denominator, it can be seen that (9) is
dependent on  n1  n2  .
Equation (9) can be further expressed in terms of ray n2 by inserting (2) and (4) into
(9) to remove n1, resulting in
f 2 ( α2 )
β ( α2 )   tan α2 
. (10)
 2 n2 +1 -1
 tan α1 
ρ f 2 (α1 )  f 2 (α2 )
From (10), it can be deduced that, if the two rays maintain the incident angles of α1
and α2, the proportion of power in ray n2 increases with its index n2. This causes a
larger total phase delay since ray n2 has more phase delay than ray n1 at point M
according to Section 2.2. This conclusion can be proven by the addition of (complex)
phasor representations of these rays.
3 Phase Delay Distributions of Different PRSs
3.1 Phase Delay Analysis of Different PRSs

As shown in Fig. 3, a typical PRS unit cell composed of two metal loops was used
to analyze the difference between the capacitive and inductive PRS at the operating
frequency of 5.5 GHz. The substrate used for the PRS is F4BM-2 (εr = 2.55, tan δ
= 0.001) with the thickness of 1.5 mm. As the structural parameter lo changes, the
unit cell exhibits different reflection magnitudes and phases (see Fig. 4). Figure 5
presents the equivalent reflection reactance variation of the unit cell, indicating that
the unit cell becomes inductive with increasing l o . A reasonable reflection
74 PAPER I
Figure 3: Structure of the unit cell.
Table 1: Parameters of capacitive and inductive unit cells. (Unit: mm)
Unit cell lo li wo wi p
Capacitive 12.2 6 1 1 20
Inductive 16.6 6 1 1 20
(a) (b)
Figure 4: Reflection (a) magnitude and (b) phase variation with different lo.
magnitude for a uniform PRS is between 0.7 to 0.9 [1], so the reflection magnitude
of 0.8 was chosen for the capacitive and inductive PRS unit cell, based on which lo
was determined to be 12.2 mm and 16.6 mm for the capacitive and inductive unit
cells, respectively. The parameters of the unit cells are provided in Table 1.
PAPER I 75
Figure 5: Reactance variation of unit cells with different lo.
Figure 6: Reflection magnitude and phase of capacitive and inductive unit cells.
As shown in Fig. 6, at the operating frequency of 5.5 GHz, the capacitive unit cell
has the reflection phase of φrc = 216° and the inductive unit cell has the reflection
phase of φri = 144°. The capacitive unit cell always has a larger reflection phase
than the inductive one, given the same reflection magnitude, since its fully reflecting
resonant frequency due to the outer loop is higher. The required height of the PRS
76 PAPER I
Figure 7: Propagation of ray with same incident angle along PRSs of different heights.
from the ground (to achieve co-phase transmitted waves assuming α = 0°) can be
calculated using to the classical equation [1]
 φ λ λ
l   r  0.5   N , (11)
 360 2 2
where N = 0, 1, 2, etc. Choosing N = 1 to minimize the PRS height (while being a

physical solution), the capacitive PRS has a larger height (l2 = 30 mm) than the
inductive one (l1 = 24.5 mm).
For each PRS antenna of a given height, the proportion of power in the ray with
index n2 can be calculated using (10), showing that a larger index n2 gives more
proportion of power. As shown in Fig. 7, when the ray with the same incident angle
of α2 arrive at the same point M on the capacitive and inductive PRSs with different
heights, the index of ray n2l on the inductive PRS will be larger than n2c on the
capacitive one, because the inductive PRS is lower than the capacitive one (i.e., l1 <
l2). Therefore, the proportion of power in the ray with the (same) incident angle of
α2 is larger for the inductive PRS than the capacitive one. Since the ray n2 with the
incident angle of α2 causes more phase delay than the ray n1 at point M, based on
the derivation in Section 2.2 (for a given PRS height), the total phase delay is larger
on the inductive PRS for a given surface location on the two PRSs.
Similarly, the index of ray with the incident angle of α1 on the inductive PRS is also
larger than the capacitive one. With smaller n1 and n2, the capacitive PRS obtains a
larger total power of rays with the two incident angles at point M, as can be
calculated from the denominator of (9). This result indicates that the capacitive PRS
can achieve more uniform magnitude distribution than the inductive one, which
facilitates a bigger radiation aperture. The predicted behavior of the phase and
magnitude distributions for the two different PRSs based on ray-tracing agree well
with the conclusion in [3], which is obtained by a leaky wave model.
PAPER I 77
PRS 0.3λ
Patch
antenna
Ground 2.5mm
Figure 8: Structure of simulated PRS antenna.
3.2 Verification Using Full-Wave Simulations

To verify above phase distribution analysis of phase delay for different PRSs, two
PRS antennas with capacitive and inductive unit cells were simulated using the
Finite Element Method (FEM) method of ANSYS HFSS with radiation boundary.
As depicted in Fig. 8, each PRS consists of 8×8 unit cells printed on the bottom
surface of the PRS substrate. The parameters of the unit cells and the PRS heights
are provided in Section 3.1. The dimensions of the PRS substrate are 0.16  0.16 
0.0015 m3. The feeding source is a patch antenna placed on a fully perfect electric
conductor (PEC) reflecting ground. The patch antenna has the dimensions of 0.0154
 0.0154 m2. The substrate of the patch antenna is F4BM-2 and it has the same
thickness and material properties as the PRS substrate. The patch is fed from the
bottom, with an offset of 2.5 mm from the patch center. The normalized phase and
magnitude distribution results and gain results of these two PRS antennas are
simulated and compared.
The normalized phase front distributions along a line 0.3λ at 5.5GHz above the two
PRS antennas (the line is shown in Fig. 8) are shown in Fig. 9. The antenna with the
capacitive PRS achieves a flatter phase front than the inductive one, which indicates
that the capacitive PRS achieves less phase delay than the inductive one. The
normalized magnitude distributions are shown in Fig. 10. The results show that the
capacitive PRS achieves a more uniform magnitude distribution than the inductive
one. Even though the capacitive and inductive unit cells have the reflection
magnitude variation of 0.08 and -0.05, as well as the reflection phase variation of -
8° and -4°, for the incident angle || < 45°, these minor variations did not affect the
conclusions obtained with the ray-tracing model, which assumes no reflection
magnitude and angle variation with incident angle.
78 PAPER I
Figure 9: Normalized phase front distributions.
Figure 10: Normalized magnitude distributions.
The antenna gains of the two PRS antennas (assuming 100% total efficiency) are
shown in Fig. 10. The antenna with the capacitive PRS has a gain of 16.6dBi, which
is 1.8dBi higher than the antenna with the inductive PRS. The different phase and
magnitude delay distributions of the two antennas affect their gains, as expected.
PAPER I 79
Figure 11: Gains of the capacitive and inductive PRS antennas.
4 Conclusion
In this paper, different gains of the capacitive and inductive PRS antennas are
analyzed by using a ray-tracing model that accounts for oblique incident angles. The
phase and magnitude distributions over the capacitive and inductive PRSs are
compared with respect to their different heights. The derivations show that the
capacitive PRS can achieve less phase delay and more uniform magnitude
distribution over the surface, which result in higher antenna gain. These conclusions
have been verified using full-wave simulations. Therefore, ray-tracing can be
utilized to provide an intuitive explanation on the operating principle of PRSs.
5 Acknowledgment
This work is supported by Vetenskapsrådet under Grant no. 2018-04717.
References
80 PAPER I
[2] A. S. Barlevy and Y. Rahmet-Samii, “Fundamental constraints on the electrical
characteristics of frequency selective surfaces,” Electromagn., vol. 17, no. 1, pp.
41-68, Oct. 2007.
12, pp. 1834-1838, May 2018.
[4] A. Foroozesh, and L. Shafai, “Investigation into the effects of the reflection
2010.
[5] A. Foroozesh and L. Shafai, “On the characteristics of the highly directive
resonant cavity antenna having metal strip grating superstrate,” IEEE Trans.
Antennas Propag., vol. 60, no. 1, pp. 78-91, Jan. 2012.
[6] M. D. Hougs, O. S. Kim, and O. Breinbjerg, “A ray-tracing method to analyzing
modulated planar Fabry-Perot antennas,” in Proc. 9th Europ. Conf. Antennas
Propag. (EuCAP), Lisbon, Portugal, Apr. 12-17, 2015, pp. 1-4.
[8] B. Ratni, W. A. Merzouk, A. de Lustrac, S. Villers, G. Piau, and S. N. Burokur,
“Design of phase-modulated metasurfaces for beam steering in Fabry-Perot
cavity antennas,” IEEE Antennas Wireless Propag. Lett., vol. 16, pp. 1401-1404,
Dec. 2017.
Paper II
Beam-Reconfigurable Antenna with
Inductive Partially Reflective Surface and
Parasitic Elements
A reconfigurable partially reflective surface (PRS) antenna can offer low-cost

and low-complexity beam steering, traditionally implemented using a phased
array with a bulky feeding network. This paper utilizes ray-tracing to prove
analytically that an inductive nonuniform PRS facilitates larger beam
deflection and lower sidelobe level (SLL) than a capacitive one, given the same
reflection phase difference. With a suitable tradeoff in the beam deflection
angle and gain variation between the reconfigurable states, an inductive
reconfigurable PRS is designed for a ±13° steering range. To enhance the
steering range with minimal complexity, two parasitic elements loaded with
PIN diodes are added to the feeding source, resulting in the overall steering
range of ±30°. The fabricated prototype verified that the pencil-shape beam
can be steered towards 0º and ±30º with SLLs of less than -19dB at 5.5GHz.
The antenna achieved peak realized gains of 9.5-10.4dBi for the three states
with gain variation of less than 0.9dBi. The measured overlapped impedance
band covers 5.41-5.63GHz for all states.
Qiuyan Liang, Buon Kiong Lau and Gaonan Zhou,
“Beam-Reconfigurable Antenna with Inductive Partially Reflective Surface and Parasitic Elements,”
IEEE Trans. Antenna Propag., 2023, Manuscript finished.
PAPER II 85
1 Introduction
Mobile communications have experienced rapid growth in the recent decade, driven
by applications that benefit from the high data rates and improved coverage [1]-[3].
Since beam steering antennas can facilitate even higher data rates and larger
coverage areas, they have become popular [4].
The conventional approach to realize beam steering is to use a phased array antenna.
However, it requires a sophisticated transmission-line feeding network to achieve
multiple phase distributions, which can cause considerable insertion loss and occupy
a relatively large implementation space [5], [6]. Hence, pattern reconfigurable
antenna is seen as a simpler and cheaper alternative to achieve beam steering, as its
main beam can be steered electronically by loading the antenna with tunable
semiconductor devices or mechanically by rotating the radiation structure [7]-[9].
Another type of pattern configuration antennas that has attracted more attention
recently is based on partially reflective surface (PRS) antenna. PRS antennas can
achieve medium-to-high gains by relying on multiple reflections of the wave
between the PRS and a fully reflecting ground [10]-[12]. Conventionally, a PRS
antenna uses a uniform PRS of uniform phase response to generate a broadside beam.
By using a nonuniform phase-varying PRS, the radiated beam can be tilted from the
broadside direction [13]-[15]. To enable beam steering, pattern reconfigurable PRS
antennas have been proposed based on mechanical rotation [16], liquid materials
[17], or electronic elements [18]-[26]. Compared with the former two
implementations, the beam reconfiguration of the latter is realized electronically,
facilitating fast beam steering.
PRSs consisting of reconfigurable unit cells with electronic elements have been
proposed for beam steering. By varying the direct current (DC) biasing voltages, the
phase distribution along the PRS, and hence the generated beam direction, can be
controlled. A parallel-plate leaky wave-guide formed by a PRS and a varactor-
loaded tunable high impedance surface (HIS) is used to steer the beam in one-
dimensional (1-D) over 9°-30° [18] and ±25° [19], respectively. However, as the
beam tilting angle increases, the thermal loss of the HIS increases, which decreases
the total antenna efficiency (<60% for angles larger than 16° [18]). The 1-D beam
steering PRS antennas designed with varactors or PIN diodes in [20] and [21]
achieve the scanning ranges of ±7º and ±18º, respectively, and the sidelobe levels
(SLLs) are around -7dB at the center frequency.
PRS antennas with 2-D beam steering have also been achieved by dividing the PRS
into four independently controlled sectors [22]-[24]. In [22], based on a tunable HIS
with varactors, the pencil beam can be steered within the scanning range of ±23º.
However, the reduction in realized gain from the broadside to the tilted direction is
more than 10dBi and -10dB impedance matching is not achieved for all states. Using
PRSs with PIN diodes, the scanning range of ±10º and ±22º are achieved in [23] and
86 PAPER II
[24], respectively. However, when the deflection angle increases, the SLLs of the
radiation patterns increase and the pencil-shape of the beams is not preserved [24].
In short, existing reconfigurable PRS antennas with different unit cell structures can
offer different beam deflection angles, but they suffer from some common
performance degradations, such as small beam deflection angles, high SLLs, severe
gain variations between different reconfigurable states and non-pencil beam.
Therefore, it is important to conduct a theoretical analysis on the working principle
of PRS relating to beam steering, to obtain insights that can be used to improve the
performance of PRS antennas.
In addition to reconfigurable PRSs, some research efforts have also been devoted to
introducing reconfigurability at the feeding source for beam steering. Although an
intuitive theoretical analysis has not been provided, it can be seen from practical
examples that applying a phased array as the feeding source can enable beam
deflection in the PRS antenna. In [25], ±10º beam steering is achieved by a PRS
antenna fed by a phased array antenna with a reconfigurable feeding network
composed of separate phase shifter network and matching network. On the other
hand, the phase shifter network can be integrated into the matching network in a
compact aperture-feed structure [26], resulting in a ±15º total beam steering range
when combined with the ±5º steering range brought by the reconfigurable PRS.
However, these feeding networks are still complicated and they lead to extra
insertion losses. Therefore, it is desirable to understand how a reconfigurable
feeding source can extend the beam steering range of a PRS and to propose a simple
structure for such a feeding source.
In this context, this paper proposes a new beam steering antenna consisting of a
reconfigurable inductive PRS and a feeding source with parasitic elements. The
contributions are:
 A simple and intuitive ray-tracing model is utilized to analyze the working
principle for beam deflection of PRS antenna with a nonuniform two-part PRS,
based on which the beam deflection capability of different types of PRS is
compared for the first time. It is concluded that an inductive nonuniform PRS
yields larger beam deflection and lower SLL than a capacitive one, given the
same reflection phase difference. This result was validated with full-wave
simulation.
 Additional beam deflection for the PRS antenna has been provided by adding
two parasitic elements to the feeding source, hence avoiding the use of multi-
element antenna and lossy feeding network (i.e., phased array) to achieve the
same result. The associated analysis confirms that the phase difference created
by the reconfigurable feeding source adds constructively with that provided by
the PRS.
PAPER II 87
 Compromising between large beam deflection and low gain variation among
reconfigurable states, an inductive reconfigurable PRS antenna is designed to
achieve ±13º beam steering, which is enhanced to ±30º by combining it with
the beam steering capability of the feeding source.
 The proposed antenna achieves three pencil-shape beams pointing at 0º and
±30º with SLLs below -19dB at 5.5GHz. The peak realized gains are 9.5-
10.4dBi for the three states with gain variation of less than 0.9dBi. The
overlapped impedance bandwidth is 4% (5.41-5.63GHz) over all three states.
The rest of the paper is organized as follows: Section 2 provides a ray-tracing
analysis of nonuniform PRS, which is then used to compare the beam deflection
capability of inductive and capacitive PRSs. Section 3 presents the ray-tracing
analysis of the reconfigurable PRS antenna fed with a beam deflected feeding source.
Using the insights gained, a new reconfigurable PRS antenna is proposed in Section
4. Section 5 presents the measurement results of the fabricated prototype. Finally,
Section 6 provides the conclusions.
2 Ray-Tracing Analysis of Nonuniform PRS

In the classical article that first proposed PRS antenna [10], its working principle
and design parameters are all based on an intuitive ray-tracing model. In some
subsequent designs of PRS antennas, the ray-tracing method has also been used to
achieve faster optimization [27], higher gain [28], and arbitrary beam deflection
angle [15]. In [15], ray-tracing is employed to derive the required phase distributions
for a phase-modulated PRS to achieve a desired beam deflection angle. In the
simplified model, only the four middle cells situated above the feeding source are
considered for the phase derivation. The model was then used to design a fixed-
beam prototype that achieves 30° deflection angle and -7dB SLL.
Although a phase-modulated PRS can be used to achieve beam steering by
incorporating varactor diodes in the unit cells, such a design will require multiple
voltage states to form different phase distributions across the PRS for different beam
deflection angles, which increases the complexity of the DC biasing network. In
addition, varactor diodes require relatively high actuated voltage [7], [29]. Therefore,
many existing reconfigurable PRSs employ PIN diodes with simple DC biasing
network, consisting only of two or four PRS parts to realize 1-D [21], [26] or 2-D
[23], [24] beam steering. To realize symmetric 1-D deflected beams, the unit cells’
structural parameters on the two sides of the PRS need to be identical, relying only
on PIN diode switching to provide mirror symmetric nonuniform phase distributions
across the two PRS parts (or uniform phase distribution with the same diode state
[21]).
88 PAPER II
In the following, the principle underlying beam deflection using nonuniform two-
part PRS is analyzed with a ray-tracing model. The analysis facilitates the
comparison of different types of PRS in terms of beam deflection capability.
2.1 Working Principle of PRS Antenna

The ray-tracing model of a conventional PRS antenna is shown in Fig. 1(a). It
consists of a feeding source, a uniform PRS and a perfectly reflecting ground.
According to the classical working principle of PRS antenna, the waves from the
feeding source are partially reflected multiple times between the PRS and the ground,
which results in the multiple transmitted waves along the PRS functioning as new
wave sources that radiate to the upper half space. The uniform PRS is placed at a
specific distance l from the ground so that the multiple transmitted rays towards the
left ( L0 to Ln ) and the right ( R 0 to R n ) are in-phase over the PRS, thus achieving
a phase front parallel to the PRS and a broadside beam. The required distance l (or
PRS height) can be calculated using the classical equation [10]
 φ λ λ
l   r  0.5   N , (1)
 360 2 2
where λ is the wavelength in free space, φr is the reflection phase of the PRS, N = 0,
1, 2, … and N = 1 is normally chosen to minimize l (while being a physical solution).
In the original ray-tracing analysis, (1) is derived by assuming normal wave
incidence, as illustrated by the solid blue lines in Fig. 1(a). This was done to simplify
the analysis of directivity in the broadside direction [10], at the expense of
nonphysical wave propagation. The dashed blue (horizontal) line portions in Fig.
1(a) are added for continuity in the two left and right ray paths, but they are neither
physical nor considered in the derivation. According to this assumption, the path
length of the reflected ray between the PRS and ground is equal to the vertical
distance l.
The phase shift of ray Ln (or ray R n ) φrayLn (or φrayRn) is the sum of the phases due
to the ray path, the reflections on the PRS φr and the ground plane φg, as well as the
transmission through the PRS φt, expressed as
2π
φrayLn  φrayRn =   2n  1 l  nφr  nφg  φt , (2)
λ
where the index n (n = 0, 1, 2, …) represents that the wave has been reflected n
times before it is transmitted through the PRS, and φg = π [10]. With N = 1 chosen
and PRS height l satisfying (1), it can be derived from (2) that the rays with adjacent
index, such as rays Ln+1 and Ln , have the phase difference of 0° and being in-phase,
as depicted in Fig. 1(a).
PAPER II 89
Broadside beam
Phase front
Ln L1 L0 R0 R1 Rn
Uniform PRS
φ r, φ t
l

(a)
Deflected beam
Ln L1 L0 R0 R1 Rn
PRS Part 1 PRS Part 2

φr2, φt2
l φr1, φt1

(b)
Deflected beam φob2
φob1
Ln′Ln L1′L1 L0′L0 R0 R0′ R1 R1′ Rn Rn′

φr1, φt1, ρ1 φr2, φt2, ρ2
l

(c)
Figure 1: Ray-tracing models of PRS antennas for (a) uniform PRS with only normal
incidence, (b) nonuniform two-parts PRS with only normal incidence, (c) nonuniform two-
part PRS with oblique incidence.
90 PAPER II
Nonuniform PRSs with different phase-varying unit cells can be used to achieve
phase front tilting. As shown in Fig. 1(b), the PRS is composed of two parts, with
the reflection and transmission phases of φr1, φt1 and φr2, φt2, respectively. The PRS
height l is set using (1) from the average value of φr1 and φr2. As a result, the rays
with adjacent index on each PRS part have non-zero phase difference, and the in-
phase condition in (1) is not satisfied for both parts. In this scenario, the phase front
becomes tilted due to gradually increasing phase lag and phase lead to the left and
right sides of the PRS, respectively. The resulted phase differences between the rays
transmitted through the two PRS parts can be derived as follows.
From (2), the phase shifts φrayLn and φrayRn of rays Ln and R n through PRS Part 1
and PRS Part 2 (at symmetric positions about the center) can be expressed as
2π
φrayLn =   2n+1 l  nφr1  nφg  φt1 , (3)
λ
2π
φrayRn =   2n+1 l  nφr 2  nφg  φt 2 . (4)
λ
From (3) and (4), with the same PRS height l and φg, the phase difference between
the rays Ln and R n is deduced as
Δφ  n  φr1  φr 2   φt1  φt 2 . (5)
The phase difference Δφ increases linearly with the number of reflections n. This
explains why the two PRS parts have more phase difference near the edges than near
the center, resulting in the smoothly tilting phase front across the PRS and a
deflected beam, as depicted in Fig. 1(b).
However, in practice, the waves transmitted through the off-center positions of the
PRS are from multiple oblique incident angles. Due to the longer ray paths, the
oblique incidence causes additional phase delay along the PRS as compared to the
simplified model considering only the normal incidence [30]. To distinguish from
the blue lines of the simplified model, the solid black lines in Fig. 1 illustrate the
two (left and right) ray paths of an oblique incident angle.
The cumulative phase delay due to all oblique incident angles compared to normal
incidence is represented by φob, which takes a negative value, representing an excess
phase delay. It was derived that φob is related to the reflection magnitude ρ of the
PRS [31] – a larger ρ leads to a smaller phase delay. With different reflection
magnitudes ρ1 and ρ2 of the two PRS parts, the phase shifts φrayLn′, φrayRn′ of rays Ln'
and R n' , and their phase difference Δφ′ can be expressed as
2π
φrayLn    2n  1 l  nφr1  nφg  φt1  φob1 , (6)
λ
PAPER II 91
2π
φrayRn    2n  1 l  nφr 2  nφg  φt 2  φob2 , (7)
λ
Δφ  n  φr1  φr 2   φt1  φt 2  φob1  φob 2
. (8)
 nΔφr  Δφt  Δφob
Therefore, the phase difference between rays Ln' and R n' in the more practical
model depicted in Fig. 1(c) is determined by the reflection phase difference Δφr, the
transmission phase difference Δφt and the phase delay difference Δφob due to
different reflection magnitudes. The resulting phase front is illustrated as the black
dashed line.
2.2 Effect of PRS Type on Beam Deflection

PRSs have been classified into two types according to the equivalent reactance of
their periodic structures: capacitive PRS with reactance less than 0, and inductive
PRS with reactance larger than 0 [32]. The two PRS types have different effects on
PRS antenna properties, e.g., the capacitive PRS is more suitable for broadside beam,
as it facilitates higher gain [30], [33]. However, the effect of PRS type on beam
deflected PRS antenna has not been studied for existing reconfigurable PRS
antennas. In this context, the previous ray-tracing analysis is utilized here to prove
that the inductive PRS is more suitable for beam deflection.
In [34], a unit cell of a single-layer PRS is considered as a two-port system, and the
reflection and transmission magnitudes and phases are analyzed regardless of the
unit cell shape. The relationship between the reflection phase and the transmission
phase was derived as [34]
π
φr  φt   , (9)
2
where the positive and negative signs correspond to the inductive and capacitive
PRSs, respectively, according to the theory of linear two-port networks [35]. The
equation (9) also implies that when the same type of PRS unit cell is employed,
designing beam deflection for PRS antennas with the simplified normal incidence
PRS model shown in Fig. 1(b) only requires a certain reflection phase difference
Δφr to be achieved.
To investigate the difference between the capacitive and inductive PRSs more
clearly, a typical PRS unit cell shown in Fig. 2 is simulated using the frequency
domain finite element method (FEM) solver of ANSYS HFSS 2021 at 5.5GHz [12].
The unit cell is composed of an outer loop and an inner square patch, printed on the
top and bottom sides of a dielectric substrate, respectively. The substrate for the
PRS is F4BM-2 (dielectric constant εr = 2.65, loss tangent tan δ = 0.001) with the
92 PAPER II
lo li p
wo
Figure 2: A classical unit cell structure (li = 8mm, wo = 0.5mm, p = 17mm).
(a) (b)
Figure 3: (a) The PRS type regions and (b) reflection and transmission properties of the unit
cell with different variable lo.
thickness of 1.5mm. The unit cell was simulated with different values of the
structural parameter lo, and the resulting equivalent transmission reactances (criteria
of unit cell type), reflection magnitudes, as well as reflection and transmission
phases are shown in Fig. 3.
As can be seen from the transmission reactance result in Fig. 3(a), the PRS unit cell
is initially capacitive as lo increases from 8mm, but later becomes inductive after a
maximum reflection resonance point at lo = 12mm. Moreover, it is observed in Fig.
3(b) that the reflection phase φr decreases monotonically as lo increases. By
choosing different lo1 and lo2 for PRS Parts 1 and 2, the two PRS parts can exhibit
different reflection phases and form a non-zero reflection phase difference Δφr. The
transmission phase φt has the same decreasing trend as the reflection phase φr in
both the capacitive and inductive region, although φt is discontinuous between the
two regions separated by a maximum reflection resonance point, which agrees with
(9). Inserting (9) in (8), the phase difference between rays Ln' and R n' becomes
PAPER II 93
Δφ   n  1 Δφr  Δφob , (10)
which shows that if Δφr is fixed, then the total phase difference Δφ′ and the
corresponding beam deflection (see the curved phase front in Fig. 1(c)) is only
dependent on the phase delay difference Δφob due to different reflection magnitudes.
In addition, as lo increases, the reflection magnitude ρ decreases in the inductive
region, whereas it increases in the capacitive region (see Fig. 3(b)). For example,
assuming that a nonuniform PRS is composed of inductive PRS Parts 1 and 2 with
structural parameters of lo1 and lo2, respectively. If lo1 > lo2, then their reflection
magnitudes become ρ1 < ρ2, which leads to φob1 < φob2 and Δφob < 0 [31]. The
negative Δφob causes more phase difference, and providing that Δφr is kept constant,
the total phase difference Δφ′ will also increase and facilitate a larger beam
deflection angle.
Following a similar argument, a capacitive nonuniform PRS with lo1 > lo2 will lead
to a positive Δφob and hence a smaller beam deflection angle than the inductive case.
In essence, since the inductive and capacitive PRSs exhibit opposite trends in ρ as
lo increases, the resulting Δφob have opposite effects on the total phase difference.
This leads to different extents of beam deflection, with the inductive PRS inherently
leveraging large beam deflection.
2.3 Full-Wave Verification

To verify the above analysis of working principle and effect of PRS type on phase
distribution, three PRS antennas using different types of PRSs but with the same
reflection phase difference Δφr (see Table 1) were simulated to achieve beam
deflection at 5.5GHz. Based on (10), Δφr is fixed to equalize the deflection angle
for the simplified model in Fig. 1(b) and to highlight the contribution in actual beam
deflection from Δφob. As depicted in Fig. 4, the PRS consists of two PRS parts, each
with 4  8 unit cells (unit cell dimensions given in Fig. 2), printed on a F4BM-2
substrate. The dimensions of the PRS substrate are 0.136 × 0.136 × 0.0015m3. The
feeding source is a patch antenna printed on the top side of another F4BM-2
substrate with the same dimensions as the PRS substrate, with the bottom side being
a perfect electric conductor (PEC) ground plane. The patch antenna and the ground
plane are infinitely thin in the simulation models. In Case 1 and Case 2, the
nonuniform two-part PRS utilizes inductive and capacitive unit cells, respectively.
For the sake of completeness, Case 3 is a PRS that utilizes both types of unit cells.
The parameters of the three PRS antennas are provided in Table 1. Following the
PRS height setting in Section 2.1, l is taken to be the average value of φr1 and φr2,
calculated from (1) with N = 1.
The normalized phase and magnitude distributions of the electric field 0.3λ above
along the center line of the three PRS antennas (i.e., the dash blue line in Fig. 4) are
94 PAPER II
Table 1: Parameters of three PRS antennas with different types of PRSs.

Three
lo1 φr1 lo2 φr2 l
cases ρ1 ρ2
(mm) (deg) (mm) (deg) (mm)
15.5 82 0.5 13.6 111 0.8
Case 1 20.9
Inductive Inductive
10.8 191 0.8 9.3 220 0.5
Case 2 29.2
Capacitive Capacitive
12.5 136 0.97 11.6 165 0.97
Case 3 25.1
Inductive Capacitive
z y PRS Part 1 PRS Part 2
0.3λ l
Patch antenna
15.2mm
2.8mm
Ground 17.5mm
Figure 4: Structure of simulated PRS antenna fed with a patch antenna.
shown in Fig. 5. As can be seen from Fig. 5(a), the phase front for Part 1 is delayed
more in Case 1 than in Case 2 due to a smaller ρ1, whereas for Part 2 it is the opposite
due to a larger ρ2. Consequently, the inductive nonuniform PRS in Case 1 achieves
more phase front tilting than the capacitive one in Case 2. The phase front in Case
3 has severe discontinuity because the PRS is composed of two types of unit cells
with discontinuous transmission phases. Therefore, the phase distribution results
from full-wave simulation agree with the predicted trends from ray-tracing.
Furthermore, as analyzed previously, the inductive PRS and capacitive PRS have
opposite trends in reflection magnitude, given the same reflection phase variation
(see Fig. 3), which leads to different results in terms of the superposition of the
waves. Due to ρ1 < ρ2 in Case 1, the transmitted waves through Part 1 have larger
amplitudes than the ones through Part 2, which facilitate stronger superposition of
PAPER II 95
(a) (b)
Figure 5: (a) Normalized phase front distributions (line center as phase reference) and (b)
magnitude distribution of PRS antennas with three different types of nonuniform PRS at
5.5GHz.
power in the phase front contributing to the main beam lobe (tilting towards the -x
direction) than the side beam lobes (tilting towards the +x direction). In contrast,
due to the smaller ρ2 in Case 2, a considerably larger amount of the waves is
transmitted through Part 2 than that of Case 1, leading to the strong superposition
of power in the phase fronts contributing to the sidelobes, resulting in relatively high
SLLs. In Case 3, due to the high reflection magnitudes, the transmitted waves
through the PRS are of lower magnitudes. Based on these different effects of wave
superposition, the magnitude distributions for the three cases are formed (see Fig.
5(b)).
The observed magnitude and phase distributions contribute to the simulated
radiation patterns of the three antennas shown in Fig. 6. The PRS antennas in Case
1 and Case 2 achieve -14° and -5° deflection angles with SLL of -22dB and -9.8dB,
respectively. The antenna gain in Case 1 is 1.7dBi lower than that in Case 2, which
is due to the lower PRS height l [30]. In Case 3, the antenna achieves -34° deflection
angle due to the large transmission phase difference, but at the cost of the high SLL
of -3.3dB.
Therefore, based on the full-wave simulation results, the deflected beam can be
formed by giving non-zero reflection phase difference, and the type of PRS has an
impact on the effect of beam deflection, which agrees with the ray-tracing analysis.
It can be concluded that the inductive nonuniform PRS can achieve more phase front
tilting and more tapering in the magnitude distribution, which facilitate larger beam
deflection and lower SLL than the capacitive one, given the same reflection phase
difference.
96 PAPER II
Figure 6: Simulated radiation patterns of PRS antennas with three different types of
nonuniform PRS at 5.5GHz.
3 Beam-Reconfigurable PRS and Feeding Source
3.1 Design Tradeoff in Beam-Reconfigurable PRS

In the previous section, the working principle of PRS antenna was analyzed with
respect to beam deflection and then applied to show that inductive PRS is preferred
for deflecting a fixed beam. However, steerable beam is required in many
applications, e.g., mobile communications [5], [12].
To design a beam-reconfigurable PRS, the unit cells in the previous two-part PRS
can be loaded with PIN diodes. The reflection phase difference of the unit cells
between the ON and OFF states in the two parts enables phase front tilting and hence
deflected beam. By setting all the PIN diodes in the same state (ON-ON/OFF-OFF),
uniform unit cell distribution and hence broadside beam is realized. However, since
the PRS height is determined by the average value of the phases in ON and OFF
states, the height will cause phase lagging or leading of the transmitted rays when
ON or OFF state is selected for a uniform PRS, as analyzed in Section 2.1. As a
result, the in-phase condition of (1) is no longer satisfied, thus affecting the antenna
gain and radiation pattern performance of the broadside beam [21]. The degree of
performance degradation to the broadside beam state was evaluated as follows.
PAPER II 97
(a) (b)
Figure 7: (a) Simulated broadside (θ = 0°) and deflected (θ = 30°) radiation patterns at
5.5GHz. (b) Simulated reflection coefficients.
Table 2: Parameters of the modelled reconfigurable PRS antennas in three states.

Beams lo1 φr1 lo2 φr2
ρ1 ρ2
(mm) () (mm) ()
-30° beam 16.7 62 0.3 12.8 128 0.93
0° beam 1 16.7 62 0.3 16.7 62 0.3
0° beam 2 12.8 128 0.93 12.8 128 0.93
As an illustration, a nonreconfigurable (beam deflecting) two-part inductive PRS

can be formed by two different unit cells with lo of 16.7mm and 12.8mm, giving
reflection phases of 62 and 128, respectively (see Table 2). The two different unit
cells are used to model the ON and OFF states of a reconfigurable unit cell. As listed
in Table 2, three PRSs composed of the two different unit cells are used to model
three states of a reconfigurable PRS (actual PIN diode loaded unit cell will be
designed in the next section). The fixed-beam PRS antennas using the three PRSs
fed by a patch antenna (see Fig. 4) were simulated. The PRS height was set to l =
20.9mm, determined by the average reflection phase of 95. As can be seen in Fig.
7(a), -30° deflected beam and two 0° broadside beams are achieved, and the
reflection coefficients are shown in Fig. 7(b). The gain of the 0° beam 1 is 3.4dBi
lower than that of the -30° beam, whereas 0° beam 2 has a narrower beamwidth and
slightly higher gain than 0° beam 1, but the back lobe of the 0° beam 2 is as high as
-11.2dB.
98 PAPER II
Deflected beam φob2+φi2
φob1+φi1 R0 ′ R1 ′ Rn ′
Ln′ L1′ L0′

l
φi1 φi2
Feeding array Ground plane
Figure 8: Working principle of PRS antenna composed of a nonuniform PRS and phase
distributed feeding source based on a ray-tracing model.
Therefore, a tradeoff is needed between the broadside beam pattern properties and
the size of the deflection angle, with a larger angle requiring a larger reflection phase
difference for the reconfigurable PRS design. It is also observed in Fig. 7(b) that the
PRS corresponding to 0° beam 2 incurs a larger impedance mismatch than the other
two PRSs. This is due to the much higher ρ (0.93) causing more waves to be
reflected from the PRS to the feeding source.
3.2 Ray-Tracing Analysis of PRS Antenna with Beam Deflected Feeding

Source
Given the aforesaid design tradeoff in obtaining desired properties for both
broadside and deflected beams (see Fig. 7(a)), a beam deflected feeding source can
be used instead to deflect the beam further while retaining good broadside beam. As
mentioned in Section 1, phased array has been used as such a feeding source [25]-
[26]. However, the role of a phased array in extending the beam deflection in a PRS
antenna has not been explained analytically.
Using the ray-tracing analysis of the nonuniform two-part PRS in Section 2, the
working principle of the PRS antenna fed with a phased array can be explained
intuitively as follows. As illustrated in Fig. 8, the array elements in the array have
different initial phases φi1 and φi2. Since each array element contributes more to the
incident rays in the PRS part just above it, the phase difference between rays Ln' and
R n' is
PAPER II 99
Δφ=  n  1 Δφr  Δφob  Δφi , (11)
where the Δφi = φi1 - φi2 is the initial phase difference from the feeding source.
Therefore, the phase distributed feeding source can further increase the phase
difference across the PRS, thus further increasing the beam deflection angle induced
by the PRS. In other words, (11) shows that the beam deflection capability of the
feeding source adds constructively to the beam deflection effect of the PRS.
4 Reconfigurable PRS Antenna Design

The ray-tracing analyses in Sections 2 and 3 give more clarity to the working
principle of beam deflected PRS antenna, which provides useful guidelines for PRS
antenna design: 1) Given the specific reflection phase difference, the inductive PRS
is preferred to achieve larger beam deflection angle with lower SLL; 2) The beam
deflected feeding source can further enhance the deflection angle. Based on these
guidelines, a novel PRS antenna with reconfigurability in both the inductive PRS
and the feeding source is designed below.
4.1 Reconfigurable PRS Design

As shown in Fig. 9, the inductive PRS is composed of three parts, where Parts 1 and
2 are on the two sides with 3×8 reconfigurable unit cells and Part 3 in the middle
with 2×8 nonreconfigurable unit cells. Part 3 provides a reflection phase between
those of the other two parts (in opposite diode states) to reduce the number of PIN
diodes needed. The nonreconfigurable unit cells in Part 3 utilizes the structure in
Fig. 2 but with updated parameters to achieve the required reflection phase. The
reconfigurable unit cells in Part 1 and Part 2 have identical structural parameters
and were modified from the nonreconfigurable one by dividing the inner square
patch in the bottom surface into two halves with a 0.8mm slot and inserting two PIN
diodes (see Fig. 9). The equivalent circuits of the PIN diode (SMP1340-079LF [36])
for the ON and OFF states, as shown in Fig. 10, were used in the simulation model
of the PRS antenna, where the insertion loss of the PIN diode in the ON and OFF
states were modelled with 0.85Ω and 3kΩ resistors, respectively [34].
The substrate for the PRS antenna and feeding source is F4BM-2 and the dimensions
are 0.16 × 0.16 × 0.0015m3. The surface area of the substrate is slightly larger than
that of the simulated example in Section 2.3 to fit the DC biasing circuit for the PIN
diodes as well as the nylon spacers for supporting the PRS substrate above the
ground plane. The final structural parameters of the PRS unit cells were obtained
by finding a suitable tradeoff between reflection coefficients and beam-pattern
properties over all reconfigurable states. The DC biasing lines of 0.3mm width for
100 PAPER II
Part 1 Part 2 V
Inductor
lie
loe p
w1
Reconfigurable unit cell
Part 3
y lom lim p
Resistor wo
Inductor
x GND Nonreconfigurable unit cell
(a)
z y
Inductive PRS
x
l
Feeding source
Ground
(b)
Figure 9: (a) PRS structure (loe = 14mm, lie = 9.2mm, lom = 14.3mm, lim = 8.5mm, w1 = 6mm,
wo = 0.5mm, p = 17mm). (b) 3D view of the PRS antenna (l = 20mm).
Parts 1 and 2 were printed in the bottom surface of the PRS and connected to the
two halves of the inner patches to control the states of the PIN diodes. Two RF
choke inductors of 56nH and a current protection resistor are used in the biasing
network of each part.
PAPER II 101
CT =0.15pF
L =0.7nH L =0.7nH
Rf =0.85Ω
Rp =3kΩ
(a) (b)
Figure 10: Equivalent circuits of PIN diode in (a) ON state and (b) OFF state.
(a) (b)
Figure 11: (a) Reflection magnitudes and (b) phases of the unit cells.
The reflection phases and magnitudes of the unit cells are shown in Fig. 11. The
reflection magnitudes and phases of the reconfigurable unit cell in the ON and OFF
states at 5.5GHz are ρ1 = 0.37, φr1 = 66 and ρ2 = 0.69, φr2 = 102, respectively. As
explained previously, the PRS height l was calculated using (1) from the average
value of φr1 and φr2. When the PIN diodes are switched ON, the two halves of the
inner patches of the reconfigurable unit cells are connected to each other, which
results in a larger equivalent inner patch size than in the OFF state, leading to a
lagging reflection phase and a smaller reflection magnitude. Accordingly, the
nonreconfigurable unit cell was designed to provide the reflection magnitude and
phase of ρ3 = 0.59 and φr3 = 87, respectively.
4.2 Feeding Source Design

The feeding source is composed of an active element and two passive elements
located on two sides [37]. The active element is fed with a SMA connectors. Each
passive element is connected to a metal sheet on the ground plane through a shorting
102 PAPER II
z
y
x
PIN diode Inductor
Resistor
Shorting pin GND V

(a)
dp
Parasitic
Active element
element 1 Wp Ws
Wa Ls
y Parasitic
df
a dp element 2
a La
x Lp
(b)
Figure 12: Geometry of the phase distributed feeding source. (a) 3D exploded view. (b) Top
view (a = 3mm, d = 2mm, df = 4.3mm, dp = 5.3mm, La = 17.5mm, Wa = 15.8mm, Lp = 15mm,
Wp = 18.6mm, Ls = 3mm, Ws = 2.7mm).
pin. The metal sheet is loaded with six PIN diodes bonded over a square slot etched
on the ground. The shorting pins of two parasitic elements are installed in
rotationally symmetric positions to eliminate the asymmetry of the radiation pattern
with respect to the x-axis due to the off-center installation. The six PIN diodes are
installed with bias direction from the metal sheet to the ground. A RF choke inductor
of 56nH and a current protection resistor are used in the biasing network of each
parasitic element. Since the PIN diodes has different equivalent circuits in the ON
and OFF states, which affects the electrical size of the parasitic element, the parasitic
element acts as reflector/director when the diodes are in ON/OFF states and the
beam can be steered to the director based on the principle of Yagi-Uda antenna [37].
The broadside beam is obtained in the OFF/OFF states. Moreover, the switched
beam parasitic antenna can be seen as a simplified phased array, since the parasitic
elements provide phase-shifted radiation via capacitive coupling to the active
element. As mentioned in the introduction, this choice avoids the need for a feeding
network and the associated losses.
Figure 13 shows the simulated radiation patterns at 5.5GHz and the reflection
coefficients of the feeding source in the three states. In States A and B, the feeding
PAPER II 103
(a) (b)
Figure 13: Simulated (a) radiation patterns at 5.5GHz and (b) reflection coefficients of the
feeding source in three states (State A: parasitic 1 and 2 in OFF and ON states, State B:
parasitic 1 and 2 in ON and OFF states, State C: parasitic 1 and 2 in OFF and OFF states).
Figure 14: Simulated radiation patterns of the PRS antenna fed with different sources at
5.5GHz.
source achieves the deflection angle of θ = -26° and θ = 26°, respectively. From the
simulation, it was observed that the beam deflection angle and SLL of the feeding
source increase with the operating frequency. Therefore, trading between the beam
deflection angle and the SLL of the PRS antenna in the band 5.4-5.6GHz, the
deflection angle of 26° was chosen for the final feeding source. In State C, the
104 PAPER II
Figure 15: Prototypes of the proposed PRS antenna.
broadside beam is achieved with both parasitic elements in the OFF state. It can be
seen in Fig. 13(b) that the feeding source in State C has a wider impedance
bandwidth than those of the other two states. This is because the parasitic element
in the OFF state has a smaller effective electrical size, which introduces a second
resonance at a slightly higher frequency. Since State C has two parasitic elements
in the OFF state, the bandwidth is wider than the other two cases with only one
parasitic element in the OFF state. The bandwidth broadening effect due to the
parasitic elements also benefits the impedance matching of the PRS antenna, as will
be presented in Section 5. The reflection coefficients in States A and B are slightly
different, because the biasing states of the two sets of diodes are not completely
symmetric with respect to the active element in the two states.
With the PRS Parts 1 and 2 in the ON and OFF states, respectively, the PRS antennas
fed with a patch antenna (see the schematic in Fig. 4) and the beam deflected feeding
source (see Fig. 12) were simulated to validate the effect of the parasitic-loaded
feeding source on the beam deflection. Figure 14 shows that the beam deflection
angle achieved by the PRS with the simple patch antenna is -13°, and this is
increased to -30° by using the beam deflected feeding source, consistent to the
analysis in Section 3.2.
PAPER II 105
Table 3: Reconfigurable states of the PRS antenna.
PRS PRS Parasitic Parasitic Beam
States
part 1 part 2 element 1 element 2 angles
State 0 OFF OFF OFF OFF 0°
State 1 ON OFF OFF ON -30°
State 2 OFF ON ON OFF 30°
Figure 16: Simulated and measured reflection coefficients of the PRS antenna.
5 Simulated and Measured Results

A prototype of the proposed reconfigurable PRS antenna was fabricated (see Fig.
15). Nylon spacers are used to support the PRS substrate above the ground plane.
The DC biasing voltages for the two PRS parts and two parasitic elements are
supplied with four sets of batteries, which are placed at the back of the ground plane
in the measurement. The three reconfigurable states of the PRS antenna are list in
Table 3.
The simulated and measured reflection coefficients of the PRS antenna in the three
states are shown in Fig. 16. As explained, State 0, corresponding to the broadside
beam, has a wider measured impedance bandwidth (5.29-5.68GHz) than the other
two states due to both parasitic elements being in the OFF state. The measured
overlapped impedance bandwidth of 5.41-5.63GHz is achieved for all states. The
106 PAPER II
(a)
(b)
(c)
Figure 17: Simulated and measured radiation patterns of the PRS antenna. (a) 5.4GHz. (b)
5.5GHz. (c) 5.6GHz.
PAPER II 107
Figure 18: Simulated and measured realized gains of the PRS antenna.
Table 4: Comparison between the proposed and previous reconfigurable PRS antennas.
Maximum Realized Overlapped
SLL
Deflection Gain Volume/λ3 Impedance
(dB)
Angle (dBi) Bandwidth
[21] 18° -6 10.7-11.7 2.75×2.75×0.59 3.6%
[23] 10° -10 8.7-9.7 1.8×1.8×0.51 3.6%
[24] 22° -5 9.6-10.4 2.75×2.75×0.45 4%
[26] 15° -8 12-15.5 3.1×3.1×0.55 4.4%
This
30° -19 9.5-10.4 2.9×2.9×0.37 4%
work
discrepancies between the simulated and measured results are primarily due to the
tolerance in the soldering of the SMA connectors.
The radiation patterns in x-z plane of the PRS antenna at 5.4, 5.5 and 5.6GHz are
shown in Fig. 17. The simulated and measured radiation patterns are seen to be in
good agreement with each other. At 5.4, 5.5 and 5.6GHz, the beam deflection angles
are ±25º, ±30º and ±35º, respectively. The increase of deflection angle with
operating frequency is mainly because of that the deflection angle of the feeding
source increases with the frequency, as mentioned in Section 4.2. At the center
frequency of 5.5GHz, the PRS antenna can radiate three pencil-shape beams
towards 0º and ±30º with SLLs of less than -19dB. The simulated and measured
peak realized gains of the PRS antenna in the three states are shown in Fig. 18. At
5.5GHz, the predicted simulation peak realized gains of the PRS antenna in States
108 PAPER II
0, 1 and 2 are 10.4, 9.5 and 10dBi, respectively. The minor differences (< 1dB)
between the simulated and measured results are mainly attributed to the tolerance
of the measurement system and slight pattern shape discrepancy, the latter of which
was caused by inaccuracies in the PIN diode’s equivalent circuits. The simulated
gain variation between the broadside and the deflected beams at 5.4, 5.5 and 5.6GHz
are less than 1.4, 0.9, and 0.7dBi, respectively.
A comparison of recent reconfigurable PRS antennas using PIN diodes based on the
maximum beam deflection angle, SLL, realized gain, and volume is presented in
Table 4. The proposed PRS antenna achieves larger beam deflection angle and lower
SLL than other antennas, which is due to the use of inductive PRS in combination
with the beam-reconfigurable feeding source. The inductive PRS of the proposed
antenna leads to a lower antenna profile with moderate gains [30], and the gain
variation from the broadside beam to the deflected beam is relatively small.
Compared with the phased array feeding source in [26], the proposed feeding source
with parasitic elements offers a simpler structure to achieve the 4% overlapped
impedance bandwidth, without extra matching network and phase shifters, which
reduces the insertion loss and further enhances the beam steering range.
6 Conclusion
A reconfigurable PRS antenna with inductive PRS and parasitic elements in the
feeding source is proposed in this work for beam steering. First, the working
principle of beam deflected PRS antenna is analyzed using ray-tracing. It is
concluded that the inductive PRS facilitates larger beam deflection angle and lower
SLL than the capacitive one, given the same reflection phase difference. Based on
the ray-tracing analysis, it is shown that the beam deflected feeding source can
further enhance the beam deflection angle of the reconfigurable PRS antenna limited
by the gain variation between the different unit-cell states in the two-part PRS. A
prototype is then designed and fabricated for the center frequency of 5.5GHz using
these insights. The measured results show that the PRS antenna provides three
pencil-shape beams towards 0° and ±30º with SLLs and gain variation lower than -
19dB and 0.9dBi, respectively, at 5.5GHz. The measured overlapped impedance
bandwidth is 5.41-5.63GHz for all states.
References
[1] A. Gupta and R. K. Jha, “A survey of 5G network: architecture and emerging
technologies,” IEEE Access, vol. 3, pp. 1206-1232, 2015.
PAPER II 109
[2] J. Navarro Ortiz, P. Romero Diaz, S. Sendra, and P. Ameigeiras, J. J. Ramos-
Munoz, and J. M. Lopez-Soler, “A survey on 5G usage scenarios and traffic
models,” IEEE Commun. Surv. Tut., vol. 22, no. 2, pp. 905-929, 2020.
communication systems: applications, requirements, technolo-gies, challenges,
2020.
[4] W. Hong, et al., “Multibeam antenna technologies for 5G wireless
communications,” IEEE Trans. Antennas Propag., vol. 65, no. 12, pp. 6231-
6249, 2017.
[5] K. Tekkouk, M. Ettorre, L. Le Coq, and R. Sauleau, “Multibeam SIW slotted
waveguide antenna system fed by a compact dual-layer Rotman lens,” IEEE
Trans. Antennas Propag., vol. 64, no. 2, pp. 504-514, Feb. 2016.
[6] C. C. Chang, R. Lee, and T. Shih, “Design of a beam switching/steering butler
matrix for phased array system,” IEEE Trans. Antennas Propag., vol. 58, no.
2, pp. 367-374, Feb. 2010.
[7] C. G. Christodoulou, Y. Tawk, S. A. Lane, and S. R. Erwin, “Reconfigurable
antennas for wireless and space applications,” Proc. IEEE, vol. 100, no. 7, pp.
2250-2261, July 2012.
[8] W. Li, et al., “Programmable coding metasurface reflector for reconfigurable
multibeam antenna application,” IEEE Trans. Antennas Propag., vol. 69, no. 1,
pp. 296-301, Jan. 2021.
[9] S. V. Hum and J. Perruisseau-Carrier, “Reconfigurable reflectarrays and array
lenses for dynamic antenna beam control: a review,” IEEE Trans. Antennas
[11] L. Zhou, X. Chen, and X. Duan, “Fabry-Pérot resonator antenna with high
aperture efficiency using a double-layer nonuniform superstrate,” IEEE Trans.
Antennas Propag, vol. 66, no. 4, pp. 2061-2066, Apr. 2018.
[12] A. Goudarzi, M. M. Honari, and R. Mirzavand, “Resonant cavity antennas for
5G communication systems: a review,” Electron., vol. 9, no. 7, pp. 1080, July
2020.
[13] A. Ourir, S. N. Burokur, and A. de Lustrac, “Phase-varying metamaterial for
compact steerable directive antennas,” Electron. Lett., vol. 43, no. 9, pp. 493-
494, Apr. 2007.
110 PAPER II
[14] H. Nakano, S. Mitsui, and J. Yamauchi, “Tilted-beam high gain antenna system
composed of a patch antenna and periodically arrayed loops,” IEEE Trans.
[15] B. Ratni, W. A. Merzouk, A. de Lustrac, S. Villers, G. P. Piau, and S. N.
Burokur, “Design of phase-modulated metasurfaces for beam steering in
Fabry-Pérot cavity antennas,” IEEE Antennas Wireless Propag. Lett., vol. 16,
pp. 1401-1404, 2017.
[16] M. U. Afzal, and K. P. Esselle, “Steering the beam of medium-to-high gain
antennas using near-field phase transformation,” IEEE Trans. Antennas.
Propag., vol. 65, no. 4, pp. 1680-1690, Apr. 2017.
[17] X. Yang, Y. Liu, H. Lei, Y. Jia, P. Zhu, and Z. Zhou, “A radiation pattern
reconfigurable Fabry-Pérot antenna based on liquid metal,” IEEE Trans.
Antennas Propag., vol. 68, no. 11, pp. 7658-7663, Nov. 2020.
[18] R. Guzman-Quiros, J. L. Gomez-Tornero, A. R. Weily, and Y. J. Guo,
“Electronically steerable 1-D Fabry-Pérot leaky-wave antenna employing a
tunable high impedance surface,” IEEE Trans. Antennas Propag., vol. 60, no.
11, pp. 5046-5055, Nov. 2012.
“Electronic full-space scanning with 1-D Fabry-Pérot LWA using
electromagnetic band-gap,” IEEE Antennas Wireless Propag. Lett., vol. 11, pp.
1426-1429, Nov. 2012.
[20] A. Ourir, S. N. Burokur, and A. de Lustrac, “Electronic beam steering of an
active metamaterial-based directive subwavelength cavity,” in Proc. 2nd Eur.
Conf. Antennas Propag. (EuCAP), Edinburgh, U.K., Nov. 2007, pp. 1-4.
[21] L. Ji, P. Qin, J. Li, and L. Zhang, “1-D electronic beam-steering partially
reflective surface antenna,” IEEE Access, vol. 7, pp. 115959-115965, 2019.
[22] R. Guzman Quiros, A. R. Weily, J. L. Gomez Tornero, and Y. J. Guo, “A
Fabry-Pérot antenna with two-dimensional electronic beam scanning,” IEEE
Trans. Antennas Propag., vol. 64, no. 4, pp. 1536-1541, Apr. 2016.
[23] P. Xie, G. Wang, H. Li, and J. Liang, “A dual-polarized two-dimensional beam-
steering Fabry-Pérot cavity antenna with a reconfigurable partially reflecting
surface,” IEEE Antennas Wireless Propag. Lett., vol. 16, pp. 2370-2374, 2017.
[24] L. Ji, Z. Y. Zhang, and N. W. Liu, “A two-dimensional beam steering partially
reflective surface (PRS) antenna using a reconfigurable FSS structure,” IEEE
Antennas Wireless Propag. Lett., vol. 18, no. 6, pp. 1076-1080, Jun. 2019.
[25] T. Debogovic and J. Perruisseau-Carrier, “Array-fed partially reflective surface
antenna with independent scanning and beamwidth dynamic control,” IEEE
Trans. Antennas Propag., vol. 62, no. 1, pp. 446-449, Jan. 2014.
PAPER II 111
[26] L. Ji, Y. J. Guo, P. Y. Qin, S. X. Gong, and R. Mittra, “A reconfigurable
partially reflective surface (PRS) antenna for beam steering,” IEEE Trans.
[27] M. D. Hougs, O. S. Kim, and O. Breinbjerg, “A ray-tracing method to
analyzing modulated planar Fabry-Perot antennas,” in Proc. 9th Europ. Conf.
Antennas Propag. (EuCAP), Lisbon, Portugal, Apr. 12-17, 2015, pp. 1-4.
[29] Y. Luo, Y. He, S. Xu, and G. Yang, “Programmable zeroth-order resonance
with uniform manipulation using the nonlinearity of PIN diodes,” IEEE
Antennas Wireless Propag. Lett., vol. 18, no. 11, pp. 2419-2423, Nov. 2019.
[30] Q. Liang and B. K. Lau, “Comparison of capacitive and inductive partially
reflective surface antenna using ray-tracing,” in Proc. 16th Europ. Conf.
Antennas Propag. (EuCAP’2022), Madrid, Spain, Mar. 27- Apr. 1, 2022.
[31] Q. Liang and B. K. Lau, “Analysis of partially reflective surface antenna with
different reflection magnitudes using ray-tracing,” in Proc. 2022 2022 IEEE
Int. Symp. Antennas Propag. USNC/URSI Nat. Radio Sci. Meeting (APS’2023),
Denver, USA, July 10-15, 2022.
12, pp. 1834-1838, May 2018.
[33] A. Foroozesh and L. Shafai, “Investigation into the effects of the reflection
2010.
[34] A. H. Abdelrahman, F. Yang, A. Z. Elsherbeni, and P. Nayeri, Analysis and
Design of Transmitarray Antennas. Morgan & Claypool, 2017.
[35] D. M. Pozar. Microwave Engineering, 3rd ed., John Wiley & Sons, Inc., 2005.
[36] Skyworks. Products: SMP1340-079LF. [Online]. Available: https://
www.skyworksinc.com/en/Products/Diodes/SMP1340-Series
[37] Q. Liang and B. K. Lau, “Co-designed millimeter-wave and sub-6GHz antenna
for 5G smartphones,” IEEE Antennas Wireless Propag. Lett., vol. 21, no. 10,
pp. 1995-1999, Oct. 2022.
Paper III
Dual-Band Shared-Aperture Antenna
with Single-Layer Partially Reflecting
Surface
Dual-band shared-aperture Fabry-Pérot cavity (DS-FPC) antennas with

single-layer partially reflective surface (PRS) are inherently limited in the
achievable frequency ratio. This paper analyzes such antennas to derive the
feasible range of frequency ratio and thereafter proposes a dual-band shared-
aperture antenna that can fill the frequency ratio gap. The proposed antenna
integrates a short backfire antenna (SBA) for the low band (LB) and a Fabry-
Pérot cavity (FPC) antenna for the high band (HB), utilizing a shared single-
layer PRS and a parasitic element. In LB, using a parasitic element that is
transparent for HB, the SBA works well despite its low profile. In HB, the FPC
antenna works with the first-order FPC mode. A prototype working at 5.5GHz
and 9GHz was fabricated and measured. The overall height of the antenna is
0.36 and 0.59 wavelength in LB and HB, respectively. The prototype achieves
7.3% and 6.7% of measured overlapping 10dB impedance bandwidth and 3dB
realized gain bandwidth, as well as peak realized gains of 10.3 and 14.6dBi, in
LB and HB, respectively.

Qiuyan Liang, Buon Kiong Lau and Gaonan Zhou,
“Dual-Band Shared-Aperture Antenna with Single-Layer Partially Reflecting Surface,”

IEEE Trans. Antenna Propag., 2023, Submitted.
PAPER III 117
1 Introduction
The demand for high speed wireless connectivity has grown tremendously over the
past decade [1], [2]. To meet this demand, a diverse range of standards with different
operating frequencies and protocols have been put forward and adopted for mobile
communications [3]. This trend has led to the increasing need for multi-band
antennas. A shared-aperture antenna is an attractive multi-band solution since it
integrates antennas of different frequency bands in a shared space and provides high
space utilization [4]-[7]. In addition, due to the adoption of higher frequency bands
and hence smaller cells, there is also a growing need for low-cost antennas that can
offer higher gains than classical antennas.
The Fabry-Pérot cavity (FPC) antenna can achieve medium-to-high gains with low
cost and low complexity [8]-[9]. In recent years, dual-band shared-aperture FPC
(DS-FPC) antennas have been proposed for a variety of applications, such as
synthetic aperture radars, satellite communications, dual-band wireless local area
network (WLAN) and mobile communications [10]-[23]. Based on the working
principle of FPC antenna, the waves from the feeding source are reflected multiple
times in the cavity formed by a partial reflective surface (PRS) and a fully reflecting
ground, with each incident wave at the PRS also producing a transmitted wave [8].
An FPC antenna achieves resonance when the multiple transmitted waves are in-
phase, which requires that the cavity length l (or PRS height) satisfies the classical
equation
φ λ λ
l   r  0.5   N , (1)
 2π  2 2
where λ is the wavelength in free space, φr is the reflection phase of the PRS and N
(N = 0, 1, 2, ….) is the order of the FPC mode [8], [10]. For a DS-FPC antenna, the
FPC antenna’s resonance condition (1) needs to be satisfied in both bands.
DS-FPC antennas could be implemented with a single-layer PRS [10]-[14]. In [10]-
[12], the DS-FPC antennas are realized with a shared PRS height l but with different
N’s for the low band (LB) and high band (HB). The achieved HB-to-LB frequency
ratios are 1.88, 1.9 and 1.79, respectively. In [13], with a shared PRS height l and
the same N for the two bands, the achieved frequency ratio is 1.1. It can be observed
that using a shared single-layer PRS, the frequency ratio tends to be close to the
ratio of N’s, the orders of the utilized FPC modes, which can be explained from (1),
as follows: A unit cell of a single-layer PRS can be considered as a two-port system,
and the reflection phase of the PRS is related to its reflection magnitude [24]. To
achieve high gain for the antenna, a sufficiently high reflection magnitude is
required [16], which results in the reflection phase φr being constrained within a
specific range. Therefore, with the same l and the narrow range of feasible φr values,
the ratio of the two wavelengths (and the two corresponding operating frequencies)
is mainly determined by the ratio of N’s for the two bands as obtained from (1).
118 PAPER III
In [14], the frequency ratio of 1.49 is obtained with a shared PRS height l and the
same N by employing a PRS substrate with thickness of 0.16 in HB (30GHz).
However, this method may not be practical for some low frequency applications due
to the relatively thick PRS substrate needed. Therefore, existing DS-FPC antennas
implemented with single-layer PRS do not offer flexible frequency ratio in general,
which limit their usefulness in dual-band applications that involve some commonly
used frequency bands, such as S-band (2-4GHz), C-band (4-8GHz), and X-band (8-
12GHz).
To realize a more flexible frequency ratio, artificial-magnetic-conductor (AMC)
ground planes have been adopted in DS-FPC antennas to compensate for the
propagation phases in LB [15] or HB [16]. However, these antennas exhibit
relatively narrow overlapping bandwidths between the 10dB impedance bandwidths
and the 3dB realized gain bandwidths in the corresponding bands. Another approach
to realize a flexible frequency ratio is to use two separate PRS layers for the two
FPCs working in the two bands [17]-[22]. With different cavity heights l’s but with
the same N, the resonance condition can be satisfied individually in both bands.
However, the overall profiles of these antennas are relatively high, especially for
HB due to the presence of LB resonant cavity. Moreover, the two required PRS
layers also increase the complexity of the antenna structure.
In addition to designing dual-band FPCs, dual-band shared-aperture antennas can
also be implemented using two types of antennas with different working principles.
A dual-band shared-aperture antenna combining a folded transmitarray antenna
working in HB and a FPC antenna working in LB with a large frequency ratio of
2.8 has been proposed [23]. However, since phase-shifting surface and electrically
larger volume are required for the folded transmitarray antenna to work normally,
four layers of PRS are employed in the antenna, which leads to a bulky overall
structure. Therefore, it is important to integrate a suitable type of antenna with a
FPC antenna to realize a dual-band shared-aperture antenna with a flexible
frequency ratio, a low-complexity structure, and a low antenna profile.
In this context, this paper proposes a new dual-band shared-aperture antenna design
integrating a short backfire antenna (SBA) working in LB and a FPC antenna
working in HB by using a shared single-layer PRS and a parasitic element. The
contributions are:
 The range of feasible frequency ratio in a traditional DS-FPC antenna with
single-layer PRS is calculated for the first time. This range is obtained from the
ranges of the PRS’s reflection phase in the two bands, given a specific
reflection magnitude range for gain enhancement.
 A PRS with different unit cells for the inner and outer parts is utilized to realize
different reflection coefficient distributions in LB and HB. In LB, the operation
of a low-profile SBA is achieved by utilizing a PRS with nonuniform
PAPER III 119
distribution in the reflection coefficient and a parasitic element that is
transparent for HB. In HB, the FPC antenna is resonant at the first-order FPC
mode. The proposed antenna provides more flexible frequency ratio with a
simple structure and it fills the frequency ratio gap of traditional DS-FPC
antennas.
 To demonstrate the significance of the proposed design philosophy, a prototype
working at 5.5GHz and 9GHz with HB-to-LB frequency ratio of 1.64 is
designed, fabricated and measured. The antenna has an overall profile height
of 0.36 and 0.59 in LB and HB, respectively. The measured overlapping
10dB impedance bandwidths and 3dB realized gain bandwidths in LB and HB
are 7.3% and 6.7%, respectively. The measured peak realized gains in LB and
HB are 10.3 and 14.6dBi, respectively. The proposed antenna not only provides
a flexible frequency ratio, but it also compares favorably with recent dual-band
shared aperture designs in impedance-gain bandwidth, height, and overall size.
The rest of the paper is organized as follows: Section 2 analyzes the range of feasible
frequency ratio for traditional DS-FPC antennas with single-layer PRS. The design
of the proposed dual-band shared-aperture antenna is described in detail in Section
3. Section 4 presents the measurement results of the fabricated prototype. Finally,
Section 5 provides the conclusions.
2 Frequency Ratio Analysis For DS-FPC Antennas

As mentioned in the introduction, traditional DS-FPC antennas with single-layer
PRS exhibit inflexible frequency ratio since the reflection phases of the PRS is
constrained within a specific range. This section analyzes the above limitation to
determine the feasible range of the frequency ratio, both without and with the
requirement for a low-profile antenna structure. In this paper, the studied range of
frequency ratio is between 1 and 2, for potential dual-band applications that combine
commonly used bands in the S-band, C-band, and X-band.
2.1 Analysis of Reflection Phase Range for Enhanced Gain

As mentioned, high gain PRS antennas require sufficiently large reflection
magnitude to confine the power between the PRS and the fully ground plane [16].
In this subsection, the PRS type and the reflection phase variation near the resonance
frequency with maximum reflection is first analyzed. Then, the relationship between
the reflection magnitude and reflection phase is derived, which shows that
capacitive and inductive PRSs have opposite trends in reflection phase as reflection
magnitude decreases. These relationships define the upper and lower boundaries of
120 PAPER III
Table 1: Variations of PRS properties near resonant frequency.
Frequency f f0  Δf f0 f0 + Δf
Reactance <0 0 >0
φr π + Δφr π π  Δφr
PRS Type Capacitive - Inductive
the reflection phase, given a specific range of reflection magnitude for gain
enhancement, as detailed in the following.
A unit cell of a single-layer PRS can be considered as a two-port system regardless
of the unit cell shape. It is equivalent to a series LC circuit and acts essentially as a
short circuit at the resonance frequency (i.e., f0), providing maximum reflection
[24]-[25]. In circuit terms, the PRS properties (i.e., equivalent reactance and
reflection phase) vary near the resonance frequency [26], as listed in Table 1. At the
resonance frequency f0, the equivalent reactance is 0Ω and the reflection phase of
the PRS is π. As the frequency increases from f0  Δf to f0 + Δf, the equivalent
reactance increases monotonically from a negative to a positive value, whereas the
reflection phase of the PRS decreases monotonically from π + Δφr to π – Δφr (Δf
and Δφr are small and positive values). The PRS is changed from being a capacitive
one to an inductive one with the frequency variation, according to the sign of the
equivalent reactance [27]. Therefore, the reflection phase of the capacitive PRS unit
cell is larger than π, whereas that of the inductive PRS unit cell is less than π near
the resonance frequency with maximum reflection.
Assuming that the PRS layer is lossless, symmetric and reciprocal, and neglecting
the higher order harmonics of the PRS layer, the relationships between the reflection
magnitude ρ and the reflection and transmission phases φr and φt have been derived
[24]. The expressions, which are independent of the unit cell structure, are given by
[24]
ρ   sin φt , (2)
π
φr  φt   . (3)
2
Inserting (3) into (2), the reflection magnitude ρ near the resonant frequency with
maximum reflection becomes
ρ   cos φr . (4)
From (4), and based on the range of reflection phase φr of two types of PRS provided
in Table 1, it can be deduced that when the reflection magnitude ρ decreases, the
reflection phase φr of the inductive PRS decreases, whereas that of the capacitive
PRS increases. Given the range of reflection magnitude 1 > ρ > 0.707 (-3dB) where
there is more power in the reflected waves than the transmitted ones to facilitate
PAPER III 121
gain enhancement, the corresponding reflection phase ranges are (3π/4, π) and (π,
5π/4) for inductive and capacitive PRSs, respectively.
2.2 Analysis of Frequency Ratio Range for Enhanced Gain

Traditional DS-FPC antennas with single-layer PRS use a shared resonant cavity for
both bands. By using a proper unit cell structure, the shared PRS exhibits desirable
reflection magnitudes and phases to satisfy the FPC resonance condition in both LB
and HB. From (1), with a shared l for both bands, the frequency ratio can be
expressed as
φrH
 0.5  N H
fH
 2π , (5)
fL φr L
 0.5  N L
2π
where fL and fH represent the resonant frequency of the FPC in LB and HB,
respectively. NL and NH are the orders of the FPC modes for LB and HB, respectively.
NL is chosen as 1 to minimize the cavity height (while being a physical solution).
Therefore, with a chosen NH, the frequency ratio is determined by the upper and
lower limits of the reflection phases φrH and φrL. Since the range of φr has been
obtained in Section 2.1, the range of frequency ratio for a specific NH can be
calculated.
When NH = 1, both the LB and HB FPCs operate at the first-order FPC mode. The
frequency ratio range of (1, 1.29) can be obtained with φrH > φrL, where 1.29
corresponds to the PRS having the reflection phases of φrH = 5π/4 (capacitive) and
φrL = 3π/4 (inductive) in HB and LB, respectively. The upper limit of 1.29
corresponds to a PRS height of l = 0.56λH, calculated from (1). The frequency ratio
could be increased further by increasing φrH or decreasing φrL, but the reflection
magnitudes of the PRS in HB and LB will further decrease according to (4), leading
to a lower gain due to the lack of reflection.
Furthermore, for a fixed frequency ratio, there is a tradeoff between the gain
performances in the two bands. For example, given a frequency ratio, when a higher
reflection magnitude is chosen for a higher gain in HB (corresponding to a smaller
φrH for capacitive PRS unit cell), the calculated φrL for LB will decrease according
to (5). The smaller φrL corresponds to a lower reflection magnitude, which leads to
a lower antenna gain in LB.
When NH = 2, the HB FPC operates at the second-order FPC mode with a larger
profile (or height) than the case of NH = 1. The frequency ratio range of (1.67, 2.43)
can be obtained, where 1.67 corresponds to the PRS with the reflection phases of
φrH = 3π/4 (inductive) and φrL = 5π/4 (capacitive), and 2.43 corresponds to the PRS
with the reflection phases of φrH = 5π/4 (capacitive) and φrL = 3π/4 (inductive),
122 PAPER III
Antenna gains
Frequency ratio gap
Decreasing Decreasing
gains gains NH = 2
NH = 1
1 1.29 1.67 2 fH/fL

Challenging for NH = 1
Figure 1: Illustration of the frequency ratio gap and predicted gain variation.
respectively. The lower limit of 1.67 corresponds to a PRS height of l = 0.94λH, as

calculated from (1). Similarly, the frequency ratio outside this range can only be
achieved by decreasing the reflection magnitudes of the PRS, which decreases the
antenna gains in both bands according to (4).
The achievable frequency ratio ranges given 1 > ρ > 0.707 (-3dB) and the predicted
trends in gain variation of traditional DS-FPC antennas with NH = 1 and NH = 2 are
illustrated in Fig. 1, within the range of interest (1, 2). Therefore, the frequency ratio
gap is given by [1.29, 1.67].
2.3 FPC Mode Order Consideration for Low-Profile Single-Layer PRS

Antenna
As discussed, the frequency ratio gap given the reflection magnitude requirement of
1 > ρ > 0.707 cannot be filled with either NH = 1 or NH = 2 for a traditional shared
dual-band FPC formed by a single-layer PRS, due to gain degradation. However,
achieving the frequency ratio of beyond 1.67 with NH = 2 comes at a cost. It can be
seen from (1) a higher FPC order leads to a larger PRS height given a specific
reflection phase of the PRS. In fact, the analysis in Section 2.2 shows that a
relatively large PRS height of 0.94λH is required with NH = 2 to achieve a design at
the upper boundary of the frequency ratio gap (i.e., 1.67). Therefore, if low-profile
design is the only requirement, then NH = 1 would be preferable to NH = 2. However,
as depicted in Fig. 1, utilizing a frequency ratio beyond 1.29 with NH = 1 will result
in smaller and smaller antenna gain. For example, consider the extreme case where
PAPER III 123
the frequency ratio is 1.67 (lower limit for NH = 2), which can be achieved with φrH
= 3π/2 (capacitive) and φrL = π/2 (inductive) in HB and LB, respectively, for the
case of NH = 1. The corresponding theoretical reflection magnitudes of the PRS, as
calculated from (4) with these two reflection phases are zero, implying no gain
enhancement to the feeding source in both bands. In contrast, this ratio is feasible
for NH = 2, as discussed in Section 2.2. Therefore, when a low-profile solution is
desired, it is even more challenging to design a DS-FPC antenna with single-layer
PRS for flexible frequency ratio.
3 Proposed Dual-Band Shared-Aperture Antenna

Given the aforesaid fundamental limitations in realizing a low-profile DS-FPC
antenna with single-layer PRS, a shared-aperture design approach is adopted in this
work, where the single-layer PRS is utilized for FPC resonance in HB and a different
function in LB. To demonstrate the frequency ratio flexibility of the approach, a
dual-band shared-aperture antenna with single-layer PRS is designed to achieve the
frequency ratio of 1.64 (LB at 5.5GHz, HB at 9GHz). The ratio is chosen to be
significantly above the upper boundary of frequency ratio for NH = 1 (i.e., 1.29). A
similar dual-band requirement in 5GHz and 9GHz was previously considered for a
dual-layer PRS design [20].
Unlike the traditional DS-FPC antenna with a shared dual-band FPC, the proposed
antenna integrates a FPC antenna and a short backfire antenna (SBA), which is a
type of high gain antenna with a typical profile of 0.5 free-space wavelength (0.5)
[28]. The single-layer PRS is designed to be capacitive, to achieve a higher gain for
the FPC antenna [27]. In addition, the PRS height calculated from (1) for the
capacitive PRS is larger than 0.5, so the FPC antenna is designed for HB, such that
the shared PRS height can still be close to the typical SBA profile of 0.5 in LB.
Moreover, as will be detailed in Section 3.2, a low-profile SBA working in LB is
designed with a PRS and a parasitic element, to facilitate a smooth integration with
the HB FPC antenna with a large frequency ratio of 1.64.
3.1 Antenna Structure and Shared-Aperture Principle

The simulation model of the proposed antenna is depicted in Fig. 2. It consists of a
PRS, a reflecting ground, feeding sources for the two bands and a parasitic element
placed above the LB feeding source. The supporting Nylon spacers in the simulated
model has the diameter of 2mm (with dielectric constant εr = 4.3 and loss tangent
tan δ = 0.004). The PRS consists of a single-layer substrate (RO4350 with εr = 3.66,
tan δ = 0.004, and thickness t1 = 0.508mm). The unit cells are printed on the bottom
side of the substrate.
124 PAPER III
PRS
Parasitic
element h
z
Nylon
y spacer
x Ground
(a)
l4 l5
y l6 y Parasitic element
x HB patch x
l3
array w5
wL
p1
Port 2 l1
w1 lf
p2 wp
Port 1
l2 LB patch antenna
w2 wH
w3 w4
(b) (c)
Figure 2: Geometry of the proposed antenna. (a) 3D view (l = 19mm, h = 10mm). (b)
Feeding sources (l1 = 9mm, l2 = 24.3mm, l3 = 14.9mm, l4 = 12.6mm, l5 = 3mm, l6 = 3.4mm,
w1 = 1.3mm, w2 = 2.6mm, w3 = 1.8mm, w4 = 0.4mm, wL = 13.8mm, lf = 11.1mm, wH = 8mm,
lH = 7.6mm). (c) Parasitic element (w5 = 1.3mm, p1 = 6.5mm, p2 = 12.5mm, wp = 21mm).
As shown in Fig. 3, the PRS of 8  8 unit cells uses a different unit cell structure for
the inner four elements as compared to the outer ones. Each inner unit cell consists
of two square slots, whereas each outer unit cell consists of only one square slot.
The substrate for the feeding source is also RO4350 but with the thickness of t2 =
1.016mm. The HB antenna employs a 1 × 2 patch array as the feeding source. The
feeding network for the patch array is printed on the top layer of the substrate and
excited with Port 2. The LB antenna employs a patch antenna with a parasitic
element above as the feeding source. The substrate for the parasitic element is the
same as that of the PRS. The parasitic element consists of an outer ring and an inner
patch printed on the bottom side of the substrate.
PAPER III 125
Outer unit cell
b p
go
g2
a1 a2 p
g1
Inner unit cell
Figure 3: Layout of PRS (a1 = 6.5mm, a2 = 4mm, b = 8.2mm, p = 10mm, go = 0.4mm, g1 =

0.5mm, g2 = 0.8mm).
(a) (b)
Figure 4: (a) Reflection magnitudes and (b) reflection phases of the unit cells.
The reflection coefficients of the unit cells are shown in Fig. 4. At 9GHz, both the
inner and outer unit cells are designed for the reflection magnitude and phase of ρ1H
= ρ2H = 0.82 and φ1H = φ2H = 214 (capacitive), respectively. With large reflection
magnitudes, the PRS enables the antenna to operate as FPC antenna in HB, as
illustrated in Fig. 5(a). The theoretical PRS height for the first order FPC mode (NH
= 1), as determined by φ1H and φ2H is l = 18.2mm. The optimized value for the final
structure is l = 19mm, to account for practical aspects (e.g., oblique wave incidence,
presence of substrate materials) that are neglected in the theoretical model as well
as to achieve a good tradeoff in the gain and bandwidth performance.
126 PAPER III
PRS
z
x
(a)
Subreflector Ray1 Ray2
0.5λ
Feeding
source
z
x
Main reflector
(b)
Inner PRS Ray1 Ray2

Outer PRS Parasitic Outer PRS
element
0.35λ
z
x
(c)
Figure 5: Working principle of (a) proposed HB antenna, (b) traditional SBA, and (c)
proposed LB antenna.
In contrast, the inner and outer unit cells of the PRS are designed to behave
differently at 5.5GHz, such that the LB antenna operates as an SBA. As shown in
Fig. 5(b), a traditional SBA consists of a main reflector, a small subreflector and a
feeding source between the two reflectors [28]. The waves from the feeding source
are reflected by the two reflectors and the wave components along different paths
satisfy an in-phase condition as shown in Fig. 5(b) [29]. The difference in the path
lengths of Ray1 and Ray2 are mainly attributed to the initial upward path of Ray1
from the feeding source to the subreflector, the reflection at the metallic surface of
the subreflector (giving an equivalent path length of 0.5), and part of the continuing
downward path of Ray1 from the subreflector until it reaches the height of the
feeding source. For Ray1 and Ray2 to be in phase, this additional path of Ray1
PAPER III 127
relative to Ray2 should be close to a positive integer multiple of 1. Therefore, the
proposed PRS is designed for high reflection in the inner parts and low reflection in
the outer parts, to mimic a subreflector in LB. Specifically, at 5.5GHz, the inner unit
cell has the reflection magnitude and phase of ρ1L = 0.92 and φ1L = 158 (inductive),
respectively, whereas the outer unit cell has the reflection magnitude and phase of
ρ2L = 0.41 and φ2L = 115 (inductive), respectively. As will be elaborated in the next
subsection, the relatively low reflection magnitude of the outer unit cell is also
designed to enable the SBA operation at a profile lower than the conventional
requirement of around 0.5.
As opposed to a traditional SBA, the “floating” LB feeding source is replaced with
a parasitic element and a feeding source (patch antenna) built onto the main reflector,
as shown in Fig. 5(c). The parasitic element, with energy coupled from the excited
patch antenna (fed with Port 1), acts as an equivalent feeding source between the
subreflector (inner PRS) and the main reflector (ground plane). Apart from
simplifying the feeding structure, the parasitic element also functions as a stacked
patch and introduces a second (higher) resonance, which can be used to enhance the
LB impedance bandwidth. As such, the frequency of the higher resonance can be
tuned by the distance between the feeding patch and the parasitic element, and a
tradeoff exists between the bandwidth and the depth of the dual-resonances.
Moreover, the patch’s feeding position lf (see Fig. 2(b)) is also optimized for
impedance matching. However, it was observed in the simulation that the slight
variations in the height of the parasitic element within this tradeoff has little effect
on the LB antenna gain. Furthermore, the parasitic element is designed as a
transparent unit cell in HB, with a low reflection magnitude of 0.02 at 9GHz. Given
its self-resonant structure and low coupling design (with large empty space around
the metallic parts), the standalone unit cell can be expected to show similar behavior
as a periodic structure. Hence, the parasitic element has minimal impact on the HB
FPC mode.
3.2 Working Principle of LB SBA with Reduced Profile Height

Having described the basic operation of the HB FPC antenna and the LB SBA
antenna, the focus in this subsection is on describing how the traditional SBA (in
Fig. 5(b)) was modified to work for the PRS height determined for normal FPC
antenna operation in HB.
As stated earlier, the height of the PRS (above the ground plane) is l = 19mmm, or
0.35λL at 5.5GHz (λL is the free-space wavelength at 5.5GHz), which is smaller than
the typical subreflector height of 0.5 used in a traditional SBA. It can be seen in
Fig. 5(c) that the vertical distance between the subreflector (inner PRS) and the
equivalent feeding source (parasitic element) is reduced (being ~0.14λL) in the
proposed antenna, as compared to the traditional SBA shown in Fig. 5(b). The
128 PAPER III
x
L2
y
Z0 L1 Z0
Normal
C incidence
z
Oblique
y incidence
(a) (b)
Figure 6: (a) Equivalent circuit of outer PRS unit cell, and (b) equivalent cross section for
normal incidence and oblique incidence.
reduced height shortens the propagation path of Ray1 and disturbs the in-phase
condition between Ray1 and Ray2, which can lead to a reduced antenna gain.
For the proposed LB antenna, the expected gain degradation due to the reduced PRS
height is mitigated by utilizing the outer unit cells of PRS. An outer unit cell of the
PRS with an etched slot has the equivalent circuit shown in Fig. 6(a) [30]. When the
incident angle of the wave increases, the effective cross section area of the unit cell
changes as shown in Fig. 6(b). As a result, the equivalent capacitance C between the
inner patch and the outer edge decreases and the equivalent inductance L2 of the
inner patch increases, which leads to a decrease in the equivalent reactance of the
inductive outer unit cell and an increase in the transmission phase [26]. As the
simulation results in Fig. 7 show, the outer unit cell has positive and increasing
transmission phase at 5.5GHz as the incident angle increases. This means that a
smaller phase delay is introduced to the wave with a larger incident angle. As can
be seen in Fig. 5(c), the incident angle of Ray2 is larger than that of Ray1, so the
phase delay introduced by the outer unit cells would be larger for Ray1 than for
Ray2, thus reducing the phase difference due to the reduced profile height.
In addition to Ray1 and Ray2 that are reflected by the two reflectors (i.e.,
subreflector and ground), there are also wave components from the parasitic element
and the feeding patch (not depicted in Fig. 5) that transmit directly through the outer
PRS. And since the transmitted power is more dominant than the reflected power
for the outer unit cells of the PRS (that has low reflection magnitude in LB), the
outer PRS cells have a function equivalent to a lens. The outer cells at the edges of
the PRS provide more phase compensation for the transmitted waves than those near
the middle, due to the larger angles of the incident waves. This leads to a flatter
phase front to the transmitted wave components across the PRS, which facilitates a
higher gain than that of a low-profile SBA that has no outer unit cell. Therefore, it
can be concluded the outer unit cells’ transmission phase variation with incident
angle enables the in-phase condition to be achieved by an SBA (for both direct and
indirect propagation paths), despite the PRS (specifically, its inner unit cells with
the role of a subreflector) having a reduced profile.
PAPER III 129
Figure 7: Transmission phase variation of outer unit cell with oblique incident angle at
5.5GHz.
z
y
x
Figure 8: SBA without outer PRS unit cell, used for comparison.
To better quantify the gain enhancement effect of the outer unit cells, a LB SBA
antenna without the outer PRS unit cells (see Fig. 8) is simulated at 5.5GHz with l
= 0.35λL, for comparison with the proposed antenna (with outer PRS unit cells). The
SBA antenna without the outer PRS unit cells and with l = 0.5λL is also simulated
for comparison. The resulting radiation patterns in the y-z plane are shown in Fig. 9.
It can be seen that, without the outer unit cells, the gain is decreased significantly
from 11.9dBi to 4.7dBi when the antenna profile is reduced from l = 0.5λL to 0.35λL.
However, by employing the outer unit cells, the gain of the SBA is almost fully
restored to that of the typical profile of 0.5λL (i.e., 11.1dBi), representing an increase
130 PAPER III
Figure 9: Radiation patterns of SBAs in y-z plane at 5.5GHz.
of 6.4dBi relative to that of the reduced profile SBA (with l = 0.35λL). In addition,
the sidelobe level of the proposed antenna is 16dB lower than that of the SBA with
0.5λL, which can be attributed to the PRS extending the effective antenna aperture
and providing suitable amplitude tapering for sidelobe reduction.
3.3 Comparison of LB SBA with Traditional DS-FPC Antenna

Based on the theoretical analysis in Section 2.2, a large frequency ratio can result in
low antenna gains for traditional DS-FPC antennas with single-layer PRS, when the
FPC resonance condition is satisfied in both bands with NL = NH = 1.
To further verify the gain enhancement effect of the proposed dual-band shared-
aperture antenna, a quasi-DS-FPC antenna consisting of a uniform PRS, a ground
plane, and two feeding sources for the two bands is simulated. As before, the
targeted frequency ratio is 1.64, to cover the 5.5GHz and 9GHz bands. Compared
to the proposed antenna, the quasi-DS-FPC antenna has no parasitic element and the
PRS consists of only the outer unit cells shown in Fig. 3.
Since the unit cells of the quasi-DS-FPC antenna have the same reflection
magnitude as both the outer and inner ones of the proposed antenna at 9GHz, the
quasi-DS-FPC antenna satisfies the FPC resonance condition in HB and hence
retains similar HB performance as the proposed antenna. However, its unit cells
provide φrL = φ2L = 115 and φrH = φ2H = 214 at fL = 5.5GHz and fH = 9GHz,
respectively, as obtained for the outer unit cells in Section 3.1. Therefore, the DS-
PAPER III 131
Figure 10: Radiation patterns of quasi-DS-FPC antenna and proposed SBA in y-z plane at
5.5GHz.
FPC resonance condition in (5) is not satisfied in LB, given NL = NH = 1, due to

insufficient PRS height in LB. This leads to gain degradation of the quasi-DS-FPC
antenna in LB. To quantify the benefit of replacing the quasi-FPC operation with
the low-profile SBA of the proposed antenna, the radiation patterns in the y-z plane
at 5.5GHz are shown in Fig. 10 for the quasi-DS-FPC antenna and the proposed
antenna. As can be seen, the gain of the proposed antenna is 3.7dBi higher than that
of the quasi-DS-FPC antenna at 5.5GHz. It is noted that the DS-FPC condition can
be satisfied for a shared PRS height with a larger φ2H and a smaller φ2L, but the
antenna gain in both bands will be low due to the small reflection magnitude, as
explained in Section 2.2.
3.4 Analysis of Frequency Ratio Flexibility

The proposed dual-band antenna realizes the frequency ratio of 1.64 with NH = 1.
This ratio is far above the frequency ratio limitation for NH = 1 of 1.29, and hence it
cannot be realized with good gain by a traditional DS-FPC antenna with NH = 1, as
analyzed in Section 3.3. On the other hand, the frequency ratio beyond 1.67 can be
achieved for NH = 2, albeit with a higher antenna profile.
To analyze the potential for the proposed SBA-FPC antenna to fill the entire
frequency ratio gap of [1.29, 1.67], the frequency ratio flexibility of the proposed
antenna is investigated for a fixed LB center frequency fL of 5.5GHz. For the
proposed antenna, the PRS height l as determined by the HB FPC condition
132 PAPER III
(a) (b)
Figure 11: (a) Radiation patterns in x-z and y-z planes at 5.5GHz and (b) reflection
coefficient of LB SBA with l = 0.44λL (24mm), h = 7mm and lf = 11.5mm.
guarantees good HB performance. A smaller frequency ratio than that of the

proposed antenna (i.e., fH / fL < 1.64) corresponds to a smaller fH (i.e., fH < 9GHz)
and a larger electrical length in LB. To ensure that the SBA part of the proposed
antenna can still work well for a smaller frequency ratio (i.e., 1.29 < fH / fL < 1.64),
the LB SBA with different PRS heights (electrical lengths) is simulated at 5.5GHz.
With the frequency ratio of fH / fL = 1.64 (LB at 5.5GHz, HB at 9GHz), the PRS
height l as determined by the HB FPC condition is 0.57λH, which corresponds to
0.35λL in LB. However, for the frequency ratio of fH / fL = 1.3 (LB at 5.5GHz, HB at
7.15GHz), the PRS height is l = 0.57λH (24mm) assuming unit cells with the same
ρ1H = ρ2H = 0.82 and φ1H = φ2H = 214 at 7.15GHz. This height (24mm) corresponds
to 0.44λL. Therefore, a LB SBA with l = 0.44λL is simulated at 5.5GHz. As
mentioned in Section 3.1, the impedance matching of the LB dual resonances is
optimized by adjusting the parasitic element height h and the patch’s feeding
position lf. With the vertical distance between the subreflector and the parasitic
element of 0.28λL, the waves reflected by the two reflectors (see Fig. 5(c)) can
roughly satisfy the in-phase condition. Moreover, the outer unit cells of the PRS as
an equivalent lens provides gain enhancement, as mentioned in Section 3.2. The LB
radiation pattern cuts and reflection coefficient magnitude are given in Fig. 11. It
can be seen that the LB SBA antenna can still achieve a similar high gain of 11.2dBi
as for the previous ratio 1.64 (simulated gain of 10.7dBi). Therefore, it can be
projected that the LB SBA antenna can be integrated with the HB FPC antenna with
flexible frequency ratio from 1.3 to 1.64 with different profile heights. It is noted
that the required reflection magnitudes and phases of the PRS in two frequency
bands with different frequency ratios can be realized by choosing proper structural
parameters for the unit cells.
PAPER III 133
To compare the antenna gain obtained by the proposed design method with that
obtained by the traditional DS-FPC antenna design method for the frequency ratio
of 1.3, a DS-FPC antenna with l = 0.44λL at 5.5GHz is simulated. This comparison
is interesting since the ratio of 1.3 is close to the upper boundary of 1.27 for NH = 1.
This means that a traditional DS-FPC antenna may provide competitive gain
performance and should therefore be investigated. The required reflection phase of
the PRS for the LB FPC antenna is 123, which is calculated from (5) with the PRS
height of l = 0.44λL, NL = NH = 1 and φ2H = 214 (for the same HB performance).
The simulated gain of the LB FPC antenna is 9.4dBi at 5.5GHz, which is almost
2dB lower than the proposed antenna with the ratio of 1.3. The lower gain of the LB
FPC antenna is due to the tradeoff of the reflection magnitudes of the PRS in two
frequency bands, as mentioned in Section 2.2. Here, the utilized PRS yields a low
reflection magnitude of 0.52 (corresponding to the reflection of 123) in LB, given
a high reflection magnitude of 0.82 (corresponding to the reflection of 214) in HB.
Therefore, the proposed antenna not only provides flexible frequency ratio while
retaining a low-complexity structure, it also facilitates competitive gain
performance relative to traditional DS-FPC antennas with single-layer PRS even for
the case where the frequency ratio is close to the upper bound of the frequency ratio
limit for NH = 1.
4 Measurement Verification
A prototype of the proposed dual-band shared-aperture antenna presented in Section
3.1 was fabricated (see Fig. 12). Nylon spacers are used to support the PRS substrate
and parasitic element above the ground plane. The size of the antenna is 80 × 80 ×
19.5mm3.
Figure 13 shows the simulated and measured S-parameters of the proposed antenna
in LB and HB. The measured 10dB impedance bandwidth of Port 1 for LB is
0.56GHz (5.29-5.85GHz). The measured bandwidth is larger than the simulated one,
which can be due to the effect of soldering and slight differences in the heights of
the PRS and parasitic element between the fabricated prototype and the simulation
model. As described in Section 3.1, the LB dual resonance is enabled by the parasitic
element. In the operating LB, the measured isolation |S21| between Port 1 and Port 2
is larger than 24dB. The measured 10dB impedance bandwidth of Port 2 for HB is
0.77GHz (8.66-9.43GHz). The dual resonance observed in HB is introduced by the
feeding network for the patch array, depicted in Fig. 2(b). In the operating HB, the
measured isolation |S21| between Port 1 and Port 2 is likewise larger than 24dB.
134 PAPER III
Figure 12: Prototype of the proposed dual-band shared-aperture antenna.
The simulated and measured radiation patterns of the proposed antenna in the E-
plane (x-z plane) and H-plane (y-z plane) are provided for 5.5GHz (LB) and 9GHz
(HB) in Fig. 14. At 5.5GHz, the antenna facilitates a pencil-shape beam with the
measured realized gain of 10.2dBi and side lobe level (SLL) of -17dB. The
measured realized gain and SLL are 13.9dBi and -12dB, respectively, at 9GHz. The
measured cross-polarization levels of the dual-band antenna are higher than 29.0 dB
and 24.0 dB at 5.5GHz and 9GHz, respectively. Figure 15 shows the realized gains
of the antenna in both bands. The simulated peak realized gains of the antenna in
LB and HB are 10.7dBi at 5.5GHz and 15dBi at 9.1GHz, respectively. The
measured results are in general consistent with the simulation ones. The differences
between the measured and simulated results are primarily due to fabrication and
measurement system tolerances. It can be seen from the measured results in Figs.
13 and 15 that the overlapping bandwidth between the 10dB impedance bandwidths
and 3dB realized gain bandwidths in LB and HB are 7.3% (5.3-5.7GHz) and 6.7%
(8.8-9.4GHz), respectively.
To facilitate comparison with state-of-the-art designs, Table 2 presents the key
parameters of 8 recent dual-band shared-aperture antennas that utilize PRS,
alongside those of the proposed antenna. These antennas are designed based on the
typical approaches mentioned in the introduction. The key parameters include the
PAPER III 135
(a)
(b)
Figure 13: Simulated (S) and measured (M) S-parameters of proposed antenna in (a) LB
and (b) HB.
frequency ratio, the number of PRS layers, the overlapping bandwidth between
10dB impedance bandwidth and 3dB realized gain bandwidth, the peak gain and
the volume. It can be observed that the achieved frequency ratios of the DS-FPC
antennas with single-layer PRS tend to be close to 1 or 2 (i.e., the ratios of FPC
modes), which is consistent with the frequency ratio range analysis in Section 2.2.
As verified in Section 3.4, the proposed antenna can fill the frequency ratio gap [1.3,
1.64] with proper structural design, without using traditional dual-band FPC
136 PAPER III
(a)
(b)
Figure 14: Simulated (S) and measured (M) radiation pattern cuts of proposed antenna in
(a) LB and (b) HB.
antennas. Moreover, the proposed antenna can achieve better overlapping

bandwidth in both bands than other single-layer designs. When compared with the
dual-band shared aperture antennas with multiple PRS layers that give flexible
frequency ratio, the proposed antenna with single-layer PRS facilitates a simpler
structure. The advantages of flexible frequency ratio, simple structure, relatively
low antenna profile and high space utilization efficiency make the antenna a good
candidate for dual-band applications.
PAPER III 137
(a)
(b)
Figure 15: Simulated (solid line) and measured (dashed line) realized gains of proposed
antenna in (a) LB and (b) HB.
5 Conclusion
In this paper, the frequency ratio range of the traditional DS-FPC antennas with
single-layer PRS is analyzed and a frequency ratio gap is derived. A dual-band
shared-aperture antenna integrating a LB SBA and a HB FPC antenna with a shared
138 PAPER III
PAPER III 139
single-layer PRS is proposed to fill the gap. The PRS exhibits different reflection
coefficient distributions in LB and HB. With a parasitic element which is transparent
in HB and a nonuniform PRS with relatively small reflection magnitude in the outer
unit cells, the LB SBA can be take advantage of the PRS to yield a reduced profile,
without affecting the HB FPC antenna. In HB, the FPC antenna works with the first-
order FPC mode. Without relying on a traditional dual-band resonance cavity, the
proposed antenna facilitates more flexible frequency ratio. A prototype working at
5.5GHz and 9GHz was designed and fabricated. The overall profile of the antenna
is 0.36 and 0.59 wavelengths in LB and HB, respectively. The antenna achieves 7.3%
and 6.7% overlapping bandwidths between the 10dB impedance bandwidths and
3dB realized gain bandwidths in LB and HB. It offers the simulated peak realized
gains of 10.7dBi and 15dBi, respectively.
References
[1] J. Navarro Ortiz, P. Romero Diaz, S. Sendra, P. Ameigeiras, J. J. Ramos-
Munoz, and J. M. Lopez-Soler, “A survey on 5G usage scenarios and traffic
models,” IEEE Commun. Surveys & Tutorials, vol. 22, no. 2, pp. 905-929, Feb.
2020.
communication systems: applications, requirements, technologies, challenges,
July 2020.
[3] A. Dogra, R. K. Jha, and S. Jain, “A survey on beyond 5G network with the
advent of 6G: architecture and emerging technologies,” IEEE Access, vol. 9,
pp. 67512-67547, Oct. 2021.
[4] T. Li and Z. N. Chen, “Metasurface-based shared-aperture 5G S-/K-band
antenna using characteristic mode analysis,” IEEE Trans. Antennas Propag.,
vol. 66, no. 12, pp. 6742-6750, Dec. 2018.
[5] G. Zhou, B. Sun, Q. Liang, S. Wu, Y. Yang, and Y. Cai, “Triband dual-
polarized shared-aperture antenna for 2G/3G/4G/5G base station applications,”
IEEE Trans. Antennas Propag., vol. 69, no. 1, pp. 97-108, Jan. 2021.
[6] F. Qin, and et al., “A triband low-profile high-gain planar antenna using Fabry-
Perot cavity,” IEEE Trans. Antennas Propag., vol. 65, no. 5, pp. 2683-2688,
May 2017.
[7] P. Mei, X. Q. Lin, G. F. Pedersen, and S. Zhang, “Design of a triple-band
shared-aperture antenna with high figures of merit,” IEEE Trans. Antennas
Propag., vol. 69, no. 12, pp. 8884-8889, Dec. 2021.
140 PAPER III
[9] A. Goudarzi, M. M. Honari, and R. Mirzavand, “Resonant cavity antennas for
5G communication systems: a review,” Electronics, vol. 9, no. 7, pp. 1080,
July 2020.
[10] F. Meng and S. K. Sharma, “A dual-band high-gain resonant cavity antenna
with a single layer superstrate,” IEEE Trans. Antennas Propag., vol. 63, no. 5,
pp. 2320-2325, May 2015.
[11] Y. Lv, X. Ding, and B. Wang, “Dual-wideband high-gain Fabry-Perot cavity
antenna,” IEEE Access, vol. 8, pp. 4754-4760, Dec. 2020.
[12] J. Chen, Y. Zhao, Y. Ge, and L. Xing, “Dual-band high-gain Fabry-Perot cavity
antenna with a shared-aperture FSS layer,” IET Microwaves, Antennas Propag.,
vol. 12, no. 13, pp. 2007-2011, Feb. 2018.
[13] C. Chen, Z. G. Liu, H. Wang, and Y. Guo, “Metamaterial-inspired self-
polarizing dual-band dual-orthogonal circularly polarized Fabry-Pérot
resonator antennas,” IEEE Trans. Antennas Propag., vol. 67, no. 2, pp. 1329-
1334, Feb. 2019.
[14] E. B. Lima, J. R. Costa, and C. A. Fernandes, “Multiple-beam focal-plane dual-
band Fabry-Pérot cavity antenna with reduced beam degradation,” IEEE Trans.
Antennas Propag., vol. 67, no. 7, pp. 4348-4356, Jul. 2019.
[15] A. Pirhadi, M. Hakkak, F. Keshmiri, and R. Karimzadeh Baee, “Design of
compact dual band high directive electromagnetic bandgap (EBG) resonator
antenna using atificial magnetic conductor,” IEEE Trans. Antennas Propag.,
vol. 55, no. 6, pp. 1682-1690, Jun. 2007.
[16] J. Qi, Q. Wang, F. Deng, Z. Zeng, and J. Qiu, “Low-profile uni-cavity high-
gain FPC antenna covering entire global 2.4 GHz and 5 GHz WIFI-bands using
uncorrelated dual-band PRS and phase compensation AMC,” IEEE Trans.
Antennas Propag., early access. July 2022.
[17] H. Moghadas, M. Daneshmand, and P. Mousavi, “A dual-band high-gain
resonant cavity antenna with orthogonal polarizations,” IEEE Antennas
Wireless Propag. Lett., vol. 10, pp. 1220-1223, 2011.
[18] B. A. Zeb, Y. Ge, K. P. Esselle, Z. Sun, and M. E. Tobar, “A simple dual-band
electromagnetic band gap resonator antenna based on inverted reflection phase
gradient,” IEEE Trans. Antennas Propag., vol. 60, no. 10, pp. 4522-4529, Oct.
2012.
[19] B. A. Zeb, N. Nikolic, and K. P. Esselle, “A high-gain dual-band EBG
resonator antenna with circular polarization,” IEEE Antennas Wireless Propag.
Lett., vol. 14, pp. 108-111, Sep. 2015.
PAPER III 141
[20] F. Qin, and et. al., “A simple low-cost shared-aperture dual-band dual-
polarized high-gain antenna for synthetic aperture radars,” IEEE Trans.
Antennas Propag., vol. 64, no. 7, pp. 2914-2922, Jul. 2016.
[21] M. L. Abdelghani, H. Attia, and T. A. Denidni, “Dual- and wideband Fabry-
Propag. Lett., vol. 16, pp. 473-476, Jan. 2017.
[22] Z. Liu, R. Yin, Z. Ying, W. Lu, and K. Tseng, “Dual-band and shared-aperture
Fabry-Perot cavity antenna,” IEEE Antennas Wireless Propag. Lett., vol. 20,
no. 9, pp. 1686–1690, Sep. 2021.
[23] P. Mei, S. Zhang, and G. F. Pedersen, “A dual-polarized and high-gain X-/Ka-
band shared-aperture antenna with high aperture reuse efficiency,” IEEE Trans.
Antennas Propag, vol. 69, no. 3, pp. 1334-1344, Mar. 2021.
[24] A. H. Abdelrahman, F. Yang, A. Z. Elsherbeni, and P. Nayeri, Analysis and
Design of Transmitarray Antennas, Morgan & Claypool, 2017.
[25] Ben A. Munk, Frequency Selective Surfaces: Theory and Design, John Wiley
& Sons, 2000.
[26] D. M. Pozar, Microwave engineering, 3rd ed., John Wiley & Sons, 2005.
[27] L. Zhou, X. Chen, Y. Cui, and X. Duan, “Comparative effects of capacitive
and inductive superstrates on the RCA’s gain,” IET Microw. Antennas Propag.,
vol. 12, pp. 1834-1838, May 2018.
Feb. 2005.
[29] G. Zhou, B. Sun, Q. Liang, Y. Yang, and J. Lan, “Beam-deflection short
backfire antenna using phase-modulated metasurface,” IEEE Trans. Antennas
[30] V. K. Kanth, and S. Raghavan, “Complementary frequency selective surface
array optimization using equivalent circuit model,” in Proc. IEEE MTT-S Int.
Microw. RF Conf. (IMaRC), Ahmedabad, India, Dec. 2017, pp. 1-4.
Paper IV
Co-Designed Millimeter-Wave and Sub-
6GHz Antenna for 5G Smartphones
This letter proposes a co-designed millimeter-wave (mm-wave) and sub-6GHz

antenna system. The antenna system consists of four 28GHz mm-wave arrays
with reconfigurable radiation patterns and two sub-6GHz antennas fed with
two corner capacitive coupling elements (CCEs). Each corner CCE is formed
by the connected ground planes of two mm-wave arrays in the shared-aperture
configuration. The two CCEs are separately matched to cover two sub-6GHz
bands. Each mm-wave array consists of an active patch element and two
parasitic patch elements loaded with PIN diodes, realizing 90-degree beam
scanning range with two states of the PIN diodes. The measured results of the
fabricated prototype show good agreement with the simulated ones. The
prototyped mm-wave arrays cover the band 27.5-28.35GHz, and each achieves
90º beam scanning at 28GHz, with measured peak realized gain of 7.9 dBi. The
CCE ports cover the two sub-6GHz bands of 0.79-0.96GHz and 1.7-5GHz, with
measured isolation of above 17dB and 20dB, respectively. The mm-wave band
isolation is above 26dB.

Qiuyan Liang, Hanieh Aliakbari and Buon Kiong Lau,
“Co-Designed Millimeter-Wave and Sub-6GHz Antenna for 5G Smartphones,”

IEEE Antennas Wireless Propag. Lett., vol. 21, no. 10, pp. 1995-1999, Oct. 2022.
PAPER IV 147
1 Introduction
With the widespread adoption of smartphones and the increasing use of bandwidth-
hungry apps, there is demand for ever higher data rates in wireless communications.
Millimeter wave (mm-wave) technology can facilitate higher data rates, due to more
bandwidth being available at higher frequencies [1], [2]. But to compensate for high
path loss in mm-wave bands to ensure sufficient coverage area, mm-wave antennas
need to form steerable beams with high gains.
One popular approach for mm-wave beam-steering is to use conventional phased
array antennas [3]-[7]. However, phase shifters can incur considerable insertion loss
and phased array elements can occupy a relatively large volume in a smartphone [8].
To facilitate a compact radiator and avoid the use of phase shifters, a 28GHz array
with parasitic elements is proposed [8]. The beam-steering is realized by shorting
the parasitic elements via four transmission lines of different lengths. However, no
real switch is used in the measurement and the transmission lines occupy
considerable printed circuit board (PCB) space.
To fit multiple antennas working in widely separated bands into the limited space
of a smartphone, co-design of the mm-wave antenna and sub-6GHz antenna has
been studied [9]-[17]. In [9] and [10], the mm-wave arrays with feeding networks
and the sub-6GHz antennas (chip antennas/monopoles) are designed in separate
spaces. In [11] and [12], the slot structure acting as a mm-wave connected array is
cleverly reused as a defected ground structure to improves the isolation between the
sub-6GHz antennas. However, the array and sub-6GHz antennas still occupy
separate spaces.
To improve aperture utilization, a frequency-reconfigurable slot antenna with a
varactor diode working in a 4G band is reused as a mm-wave antenna based on the
connected slot array concept [13]. However, its 4G band is limited to 2.05-2.7GHz.
As another share-aperture approach, the mm-wave array module is embedded into
the metal bezel present in some smartphones, with the bezel serving as the sub-
6GHz antenna [14]-[16]. This method helps to reduce the blockage of the mm-wave
antenna radiation due to the metallic frame. Similarly, the addition of grating strips
facilitates the reuse of the PCB space occupied by a low-band planar inverted-F
antenna (PIFA) for implementing a mm-wave antenna array [17]. However, the
mm-wave antennas in [14]-[17] are still phased arrays with lossy feeding networks,
and some sub-6GHz antennas (e.g., PIFA) occupy relatively large spaces.
In this work, a co-designed smartphone antenna system is proposed to accommodate
four mm-wave arrays and two sub-6GHz antennas in a compact space. The mm-
wave array employs parasitic elements loaded with PIN diodes to achieve beam-
steering, in the same manner as Yagi-Uda antenna [18]. Instead of using self-
resonant elements, the sub-6GHz antennas are excited by non-resonant capacitive
coupling elements (CCEs) [19], which are becoming popular to realize low cellular
148 PAPER IV
z
Ls CCE1 y
t x
Copper Chassis
Port1 Port3
Port4
CCE2
Substrate
Lg
Port2
Matching network 1 Matching network 2
t
dc
Wc L1 L3
Wg
L2
C1
Lc y
x
Figure 1: Overall structure of proposed co-designed antenna (Ls = 4mm, Lc = 17mm, Wc =

8mm, t = 0.508mm, Lg = 120mm, Wg = 60mm, dc = 3mm).
band antennas due to their compactness and simple structure (see [20] and
references therein). The metal ground planes of the mm-wave arrays are shared by
the corner CCEs that excite the sub-6GHz bands. This shared-aperture configuration
with compact mm-wave arrays on the corner CCEs facilitates sleek integration into
5G smartphones. The fabricated prototype confirms that each mm-wave array can
realize 90° scanning range with good impedance matching in the desired frequency
band of 27.5-28.35GHz. Therefore, the four mm-wave arrays on the two CCEs
enable full 360° coverage. The two sub-6GHz antennas are well matched in the
operating bands of 0.79-0.96GHz and 1.7-5GHz, respectively.
2 Co-Designed Antenna System

Figure 1 provides the three-dimensional (3D) view and top view of the proposed co-
designed antenna system. It consists of four mm-wave arrays mounted on two corner
CCEs and a 120mm × 60mm chassis excited by the two CCEs to cover two sub-
6GHz bands. The proposed antenna system is further described below.
PAPER IV 149
Active element x
Passive element z d
Sub1 e
b W
Lf
a
a La Lp
Shorting pin PIN diode
Ground plane
Sub2 g
y
x V
z
(a) (b)
x y
z ds Wd x
rs DC biasing pad
(c) (d)
Figure 2: Geometry of mm-wave array fed by Port3. (a) Three-dimensional exploded view.
(b) Top view (a = 1mm, b = 1.1mm, d = 0.5mm, e = 0.7mm, La = 4.6mm, Lf = 2.6mm, Lp =
3.2mm, W = 3.3mm, g = 0.3mm). (c) DC biasing lines (rs = 1.8mm, ds = 2.2mm, Wd =
0.6mm). (d) Side view.
2.1 Mm-Wave Antenna Array

As shown in Fig. 2, the antenna array uses two layers of substrate (Rogers 5880,
with thickness of 0.508mm, relative permittivity of 2.2 and loss tangent of 0.0009)
for the radiating layer (Sub1) and the direct current (DC) control layer (Sub2). The
array is composed of an active element fed with a 50Ω coaxial cable and two passive
elements symmetrically located on two sides of the active element. Each passive
element is loaded with two PIN diodes bonded over a square slot etched on the
ground. The DC biasing lines are printed on the bottom layer of Sub2. Two shorting
pins connect the DC biasing lines, the passive elements and the metal sheets within
the etched slots. The two sets of PIN diodes for the two passive elements are
installed with opposite bias directions and their DC biasing lines with isolation fan
stubs are connected in parallel to share a DC feeding pad. The opposite bias ensures
that, when a DC voltage is applied, the PIN diodes of one passive element will
always be in the opposite state to those of the other passive element (i.e., ON and
OFF states for passive elements 1 and 2, respectively, or vice-versa). The PIN diodes,
produced by MACOM (Model no. MA4GP907), allow for operation up to
millimeter frequencies [21]. Its equivalent circuits for the ON and OFF states in
mm-wave bands, as shown in Fig. 3, were used in the simulation model, where the
insertion loss of the PIN diode in the ON and OFF state is modelled with a 5.2Ω and
a 10kΩ resistor, respectively.
150 PAPER IV
CT =0.025pF
L =30nH L =30nH
Rf =5.2Ω
Rp =10kΩ
(a) (b)
Figure 3: Equivalent circuits of PIN diode in (a) ON state and (b) OFF state.
The parasitic element is connected to/disconnected from the ground plane when the
beneath PIN diodes are turned ON/OFF, which decreases/increases its effective
electrical size, such that it acts as a director/reflector [22]. The beam of this array is
steered to the director and away from the reflector based on the principle of Yagi-
Uda antenna. By applying positive or negative DC voltage between the DC biasing
pad and the ground, two symmetrical beams can be achieved.
The effects of the structural parameters were investigated using ANSYS HFSS 2021.
The simulation results show that the beam deflection angle and sidelobe level (SLL)
are mainly dependent on the size of the parasitic elements (controlled by a) and the
distance between the parasitic elements and the active one (d) (see Fig. 2(b)). For
example, decreasing a or increasing d will lead to increased beam deflection angle
and SLL, as well as narrower main beam. The appropriate a and d values were then
optimized to obtain 45° beam deflection (i.e., mirror symmetric beams for the two
possible states of the diodes) and low SLL. The impedance bandwidth of the antenna
becomes wider when d decreases, which is because of that a second (higher)
resonance is introduced by the parasitic element in the ON state. Considering the
radiation pattern performance and the fabrication tolerance requirements, the
distance d = 0.5mm was finally chosen.
2.2 Sub-6GHz Antennas

Since the CCE should be placed in the region of maximum electric field maximum
strength of the mode to be excited (i.e., the fundamental dipole mode) [23], two
CCEs are placed at two diagonally opposite corners of the chassis. In addition, these
corner locations enable the four mm-wave antennas on the two CCEs to cover the
entire field of view and mitigate blockage from the user’s hand(s). CCE1 and CCE2
excite the chassis through matching networks (see Figs. 1 and 4) designed in
Betamatch [24] to realize two sub-6GHz antennas, i.e., Port1 and Port2 cover the
low band (LB) of 0.79-0.96GHz and the high band (HB) of 1.71-5GHz, respectively.
A larger CCE size facilitates broader bandwidth of Port1, thus the CCE parameters
are chosen considering the trade-off between the bandwidth and size [25]. The
PAPER IV 151
L1 =10nH L3 =0.7nH
CCE1 CCE2
Port1 Port2
L2 =12nH
C1 =0.4pF
(a) (b)
Figure 4: Schematics of matching networks for (a) CCE1 and (b) CCE2.
d1
CCE1
d2
CCE2
Figure 5: Proposed antenna system with metal frame.
bandwidth of Port1 and the isolation between Port1 and Port2 in the LB mainly
depend on the inductance L1. With decreasing L1, the LB bandwidth increases
whereas the Port1-Port2 isolation decreases. To achieve a good trade-off between
bandwidth and isolation, the matching elements in matching network 1 were chosen
to be L1 = 10nH and L2 = 12nH. The bandwidth of Port2 in the HB mainly depends
on the capacitance value C1. The Port1-Port2 isolation in HB is not significantly
affected by the matching elements. The optimized matching elements in matching
network 2 are C1 = 0.4pF and L3 = 0.7nH. It is noted that the loading effect of the
mm-wave connectors has been included in the design of matching networks, to
facilitate experimental validation. In practice, these connectors are not needed and
the matching network can be updated by changing the matching circuit parameter
values (e.g., L1 = 18nH and L2 = 10nH for Port1). Moreover, more matching
elements can give a larger Port2 bandwidth (e.g., 1.37-6.71GHz using five elements)
[24].
In practical applications, smartphones are equipped with a touch screen and some
come with metal bezels (side frames). It is found that adding a metal plate (of the
same size as the chassis) 4mm above the chassis (and grounded through a shorting
pin at the chassis center) to model the screen does not affect the fundamental mode
of the chassis [26], and the impedance matching in the sub-6GHz bands can be
restored by updating the matching networks. To study the effect of metal bezels,
four separate vertical metal strips of 8mm width were located along (but not
152 PAPER IV
Figure 6: Prototype of proposed antenna system and measurement setups.
connected to) the four sides of the chassis, as depicted in Fig. 5. When the distance
d1 between the strips and the CCE1 is larger than 5mm (0.015 wavelength in free
space at 0.875GHz), Port1 retains over 90% of the original bandwidth (i.e., the case
with no strip). The distance d2 for the CCE2 to retain at least 90% of the original
bandwidth is 2mm (0.023 wavelength in free space at 3.5GHz).
3 Simulated and Measured Results

A prototype of the smartphone antenna system was fabricated (see Fig. 6). The mm-
wave CAB.058 coaxial cables are used to feed the mm-wave arrays on CCE1 in the
measurement, which are intended to verify the beam scanning range of the mm-
wave arrays on each CCE. In real implementation, the feeding of the mm-wave
antenna should be realized with more advanced integrated technology to minimize
any possible interference. The DC voltage is applied through a substrate with the
DC circuits, which is attached on the back side of the chassis. An FR4 frame is
designed to support the antennas in the measurement.
3.1 S-Parameters
Fig. 7 shows the S-parameter results of the two sub-6GHz antennas. The measured
6dB impedance bandwidth (VSWR of 3:1) of Port1 is 0.38GHz (0.75-1.13GHz),
PAPER IV 153
Figure 7: Simulated (S) and measured (M) S-parameters of sub-6GHz antennas.
Figure 8: Simulated (S) and measured (M) S-parameters of mm-wave array.
covering the LTE800/850/900 bands. In the operating band (i.e., LB), the measured
isolation of the sub-6GHz ports |S21| is larger than 17dB and that between Port1 and
mm-wave port Port3 |S31| is larger than 43dB. The measured 6dB impedance
bandwidth of Port2 is 3.6GHz (1.70-5.30GHz), covering the LTE1700-2600 and 5G
NR n77-79 bands. In this upper band (i.e., HB), the measured isolations with other
ports are over 20dB.
The S-parameter results of the mm-wave array are shown in Fig. 8. Port3 and Port4
have the same simulation results for reflection coefficient due to symmetry. The
simulated 10dB bandwidth of the mm-wave antenna is around 2GHz (26.56-
28.54GHz). It is noted that, if needed, the mm-wave antenna bandwidth can be
significantly enhanced (e.g., to 3GHz) by using a stacked patch as the active element.
154 PAPER IV
Measured Simulated
(a)
Measured Simulated
(b)
Figure 9: Simulated and measured normalized radiation patterns at sub-6GHz band in (a)
LB (0.875 GHz) and (b) HB (3.5GHz).
The measured resonances are slightly higher than those in simulation and the
measured bandwidths are narrower than the simulated ones, due to the tolerance in
the soldering of the mm-wave cables. Such discrepancy is common in mm-wave
bands due to the relatively small wavelengths [13]-[17]. The measured 10dB
bandwidths of Port3 and Port4 are 1.15GHz (27.25-28.4GHz) a 1.1GHz (27.4-
28.5GHz), respectively, covering the 5G NR n261 band. In the operating mm-wave
band, the measured isolations with other ports are over 26dB.
3.2 Radiation Patterns

The radiation pattern results of the fabricated prototype in the sub-6GHz bands were
obtained from a SATIMO multi-probe spherical near-field system [27]. Figure 9
shows the normalized simulated and measured 2-D pattern cuts at θ = 90° (azimuth
cut), φ = 0° (elevation cut), and φ = 90° (elevation cut) for Port1 and Port2 at 0.875
GHz in LB and 3.5 GHz in HB, respectively. The fundamental mode of the chassis
is excited by the CCEs in the two sub-6GHz antennas.
PAPER IV 155
(a) (b)
Figure 10: Simulated (S) and measured (M) realized gain and efficiency in sub-6GHz in (a)
LB and (b) HB.
Figure 11: Simulated (S) and measured (M) radiation patterns of mm-wave arrays with
positive (pos) and negative (neg) voltage at 28 GHz.
The discrepancies between the simulated and measured patterns are primarily due to
the presence of a feed cable in the near field of the structure. The simulated and
measured realized gain and efficiency in the sub-6GHz bands are shown in Fig. 10.
The measured efficiencies are higher than the simulated ones at some frequency
points due to the discrepancies between the measured and simulated S-parameters.
156 PAPER IV
Table 1: Comparison between the proposed and previous 4G/5G designs.
Frequency range Total antenna Phased array (mm- Diodes
(GHz) volume (mm3) wave)
[13] 2.05-2.7 and 23-29 35×12.28×0.381 yes yes
0.7-0.96, 1.71-2.69
[14] 75×10×7 yes no
and 25-30
0.76-0.98, 1.24-2.87, Active and dummy
[16] 21.7×7.75×0.64 no
and 22-28.4 elements
0.74-0.96, 1.7-2.2,
[17] 70×9×0.764 yes no
and 22-31
This 0.75-1.13, 1.7-5.3
34×8×1.016×2 no yes
work and 27.5-28.35
Figure 11 shows the radiation patterns of two mm-wave arrays on CCE1 at 28GHz,
which were measured with an in-house pattern measurement system utilizing the
Rohde and Schwarz vector network analyzer ZVA67. Applying the
positive/negative voltages on the DC biasing pads of the two mm-wave arrays, four
deflected beams were realized. The two mm-wave arrays on CCE1 can achieve 180°
coverage range with half power beamwidth. The peak measured realized gains are
7.4dBi or 7.9dBi (Port3 with positive or negative voltage) and 7.1dBi or 7.7dBi
(Port4 positive or negative voltage). The peak measured realized gains are less than
the simulated ones by 1.4dBi (Port3 positive), 1.2dBi (Port3 negative), 1.6dBi
(Port4 positive) and 1.3dBi (Port4 negative), respectively. The realized gain
difference is primarily due to the loss in the mm-wave cables and slight pattern
shape discrepancy, the latter of which was caused by inaccuracies in the PIN diode’s
equivalent circuits and the tolerance of the measurement system.
3.3 Comparison with Other Sub-6GHz/Mm-Wave Antennas

A comparison of recent co-designed antennas for sub-6GHz and mm-wave bands is
presented in Table 1. The sub-6GHz antennas in [14] and [17] employ self-
resonance structures, which occupy larger volumes. Phased arrays, which require
feeding networks, are used for the mm-wave antennas in [13], [14] and [17]. The
proposed antenna system can cover wider sub-6GHz bands despite the use of two
compact CCEs. With the parasitic elements, the proposed mm-wave array achieves
beam scanning without the need for complex feeding networks. Moreover, the
shared-aperture configuration of the corner CCEs and mm-wave arrays enables a
compact antenna volume.
PAPER IV 157
4 Conclusion
A co-designed mm-wave and sub-6GHz antenna system for 5G smartphone
application is proposed in this letter. Each mm-wave array antenna uses two
parasitic elements loaded with PIN diodes to realize beam scanning. The four mm-
wave arrays share the aperture of the CCEs, with the latter providing coverage of
two sub-6GHz bands, which facilitates a compact antenna structure. The measured
results show that the sub-6GHz antennas cover the bands of 0.79-0.96GHz and 1.71-
5GHz. The mm-wave array provides 90° scanning range with measured realized
gain of up to 7.9dBi at 28GHz. Possible future work includes adding ports in the
sub-6GHz bands for MIMO operation by means of creating and exciting more
resonant modes [20] as well as using more parasitic elements and reconfigurable
states for higher gain in mm-wave bands.
5 Acknowledgement
The authors would like to thank A. Johansson of Lund University and D. Pugachev
of Sigma Connectivity for their help in prototype fabrication and pattern
measurement.
References
[1] T. S. Rappaport, et al., “Millimeter wave mobile communications for 5G
cellular: It will work!” IEEE Access, vol. 1, pp. 335-349, 2013.
[2] W. Hong, K. H. Baek, and S. Ko, “Millimeter-wave 5G antennas for
smartphones: Overview and experimental demonstration,” IEEE Trans.
[3] N. Ojaroudiparchin, M. Shen, S. Zhang, and G. F. Pedersen, “A switchable 3-
D-coverage-phased array antenna package for 5G mobile terminals,” IEEE
Antennas Wireless Propag. Lett., vol. 15, pp. 1747-1750, Feb. 2016.
[4] S. Zhang, X. Chen, I. Syrytsin, and G. F. Pedersen, “A planar switchable 3-D-
coverage phased array antenna and its user effects for 28-GHz mobile terminal
applications,” IEEE Trans. Antennas Propag., vol. 65, no. 12, pp. 6413-6421,
Dec. 2017.
[5] I. Syrytsin, S. Zhang, G. F. Pedersen, and A. S. Morris, “Compact quad-mode
planar phased array with wideband for 5G mobile terminals,” IEEE Trans.
Antennas Propag., vol. 66, no. 9, pp. 4648-4657, Sep. 2018.
158 PAPER IV
[6] W. Hong, K. Baek, Youngju Lee, and Yoon Geon Kim, “Design and analysis
of a low-profile 28 GHz beam steering antenna solution for future 5G cellular
applications,” in Proc. IEEE MTT-S Int. Microw. Symp. (IMS), Tampa, USA,
Jun. 1-6, 2014, pp. 1-4.
[7] N. Ojaroudiparchin, M. Shen, and G. F. Pedersen, “A 28 GHz FR-4 compatible
phased array antenna for 5G mobile phone applications,” in Proc. Int. Symp.
Antennas Propag. (ISAP), Hobart, Australia, Nov. 9-12, 2015, pp. 1-4.
[8] S. Zhang, I. Syrytsin, and G. F. Pedersen, “Compact beam-steerable antenna
array with two passive parasitic elements for 5G mobile terminals at 28 GHz,”
IEEE Trans. Antennas Propag., vol. 66, no. 10, pp. 5193-5203, Oct. 2018.
[9] C. Lee, M. K. Khattak, and S. Kahng, “Wideband 5G beamforming printed
array clutched by LTE-A 4 × 4-multiple-input-multiple-output antennas with
high isolation,” IET Microw. Antennas Propag., vol. 12, no. 8, pp. 1407-1413,
Mar. 2018.
[10] R. Hussain, A. T. Alreshaid, S. K. Podilchak, and M. S. Sharawi, “Compact 4G
MIMO antenna integrated with a 5G array for current and future mobile
handsets,” IET Microw. Antennas Propag., vol. 11, no. 2, pp. 271-279, Jan.
2017.
[11] M. Ikram, R. Hussain, and M. S. Sharawi, “4G/5G antenna system with dual
function planar connected array,” IET Microw. Antennas Propag., vol. 11, no.
12, pp. 1760-1764, Sep. 2017.
[12] M. S. Sharawi, M. Ikram, and A. Shamim, “A two concentric slot loop based
connected array MIMO antenna system for 4G/5G terminals,” IEEE Trans.
[13] M. Ikram, E. A. Abbas, N. Nguyen-Trong, K. H. Sayidmarie, and A. Abbosh,
“Integrated frequency-reconfigurable slot antenna and connected slot antenna
array for 4G and 5G mobile handsets,” IEEE Trans. Antennas Propag., vol. 67,
no. 12, pp. 7225-7233, Dec. 2019.
[14] J. Kurvinen, H. Kähkönen, A. Lehtovuori, J. Ala-Laurinaho, and V. Viikari,
“Co-designed mm-Wave and LTE handset antennas,” IEEE Trans. Antennas
Propag., vol. 67, no. 3, pp. 1545-1553, Mar. 2019.
[15] R. M. Moreno, J. A. Laurinaho, A. Khripkov, J. Ilvonen, and V. Viikari, “Dual-
polarized mm-wave endfire antenna for mobile devices,” IEEE Trans.
Antennas Propag., vol. 68, no. 8, pp. 5924-5934, Aug. 2020.
[16] R. C. Rocio, S. Zhang, K. Zhao, and G. F. Pedersen, “Mm-wave beam-steerable
endfire array embedded in a slotted metal-frame LTE antenna,” IEEE Trans.
Antennas Propag., vol. 68, no. 5, pp. 3685-3694, May 2020.
PAPER IV 159
[17] M. M. Samadi Taheri, A. Abdipour, S. Zhang, and G. F. Pedersen, “Integrated
millimeter-wave wideband end-fire 5G beam steerable array and low-
frequency 4G LTE antenna in mobile terminals,” IEEE Trans. Veh. Technol.,
vol. 68, no. 4, pp. 4042-4046, Apr. 2019.
[18] J. Huang and A. C. Densmore, “Microstrip Yagi array antenna for mobile
satellite vehicle application,” IEEE Trans. Antennas Propag., vol. 39, no. 7, pp.
1024-1030, July 1991.
[19] J. Villanen, J. Ollikainen, O. Kivekas, and P. Vainikainen, “Coupling element
based mobile terminal antenna structures,” IEEE Trans. Antennas Propag., vol.
54, no. 7, pp. 2142-2153, July 2006.
[20] H. Aliakbari and B. K. Lau, “Low-profile two-port MIMO terminal antenna for
low LTE bands with wideband multimodal excitation,” IEEE Open J. Antennas
Propag., vol. 1, pp. 368-378, 2020.
[21] MACOM. (2016). Products: MA4GP907. [Online]. Available:
http://cdn.macom.com/datasheets/MA4GP907.pdf
[22] W. Deng, X. Yang, C. Shen, J. Zhao, and B. Wang, “A dual-polarized pattern
reconfigurable Yagi patch antenna for microbase stations,” IEEE Trans.
Antennas Propag., vol. 65, no. 10, pp. 5095-5102, Oct. 2017.
[23] R. Valkonen, M. Kaltiokallio, and C. Icheln, “Capacitive coupling element
antennas for multi-standard mobile handsets,” IEEE Trans. on Antennas
Propag., vol. 61, no. 5, pp. 2783-2791, May 2013.
[24] Betamatch. (2019). MNW Scan Pte Ltd, Version 3.7.6. Accessed: Nov. 11, 2021.
[Online]. Available: http://www.mnw-scan.com
[25] H. Aliakbari and B. K. Lau, “Impact of capacitive coupling element design on
antenna bandwidth,” in Proc. 12th Europ. Conf. Antennas Propag. (EuCAP),
London, UK, Apr. 9-13, 2018, pp. 1-4.
[26] H. Aliakbari, L. Y. Nie, and B. K. Lau, “Large screen enabled tri-port MIMO
handset antenna for low LTE bands,” IEEE Open Journal of Antennas and
Propag., vol. 2, pp. 911-920, Aug. 2021.
[27] Satimo Stargate. (2019). Measurement System. Accessed: Aug. 26, 2021.
[Online]. Available: http://www.satimo.com

Thesis Qiuyan Lund University

Uploaded by

Copyright:

Available Formats

Thesis Qiuyan Lund University

Uploaded by

Document Information

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

Thesis Qiuyan Lund University

Uploaded by

Copyright:

Available Formats

Low-Complexity Multiband and Beam-Reconfigurable Antennas for Beyond 5G

Citation for published version (APA):

Total number of authors:

Read more about Creative commons licenses: https://creativecommons.org/licenses/

Series of licentiate and doctoral theses

This thesis is produced by using the Microsoft Word

Department of Electrical and Information Technology

During my doctoral studies, I had also contributed to the following publications,

1.1 Overview of Antenna Development Trends

stations in the deployment of future wireless communication networks [4].

User devices User devices

1.2 PRS Antennas for Base Stations and Design

1.2.1. Beam-reconfigurable PRS antennas

A reconfigurable PRS is mainly implemented using three approaches: mechanical

1.2.2. Shared-aperture antennas with PRS

1.2.3. Ray-tracing analysis for PRS antenna

1.3 Co-Designed Antennas for Mobile Terminals and

Figure 1.6: Simplified antenna design environment of a mobile terminal.

1.4 Research Questions and Thesis Goals

 To evaluate the effects of several practical design considerations on the

1.5 Thesis Structure

2.1 Working Principle of Conventional PRS Antenna

2.1.1. Ray-tracing analysis considering normal incidence

Feeding source Ground plane

2.1.2. Ray-tracing analysis considering oblique incidence

Phase front of normal incidence

0 0' 1 1' n n' n+1 (n+1)'

Feeding source Ground plane

2.1.3. Proper reflection magnitude

2.2 PRS Properties

2.2.1. PRS unit cell types

Single layer of PRS

Figure 2.3: Illustration of waves on both sides of a single-layer PRS [47].

The resonant frequency is f r  ωr 2π  1 2π Leq Ceq , where the equivalent

The resonant frequency is ft  ωt 2π  1 2π Leq Ceq , where the equivalent

2.2.2. Reflection and transmission coefficients of PRS

Figure 2.8: Illustration of a simulated unit cell model.

2.2.3. Effect of PRS properties on broadside beam

2.3 Beam-Reconfigurable PRS Antenna Design

Feeding source Ground plane

PRS Part1, φr1, φt1 PRS Part2, φr2, φt2

Feeding source Ground plane

PRS Part1, φr1, φt1 PRS Part2, φr2, φt2

Feeding source Ground plane

2.3.1. Working principle of beam deflected PRS

2.3.2. Effect of PRS type on beam deflection capability

2.3.3. Performance metrics for beam-reconfigurable PRS antenna

2.3.4. Beam deflected feeding source

2.4 Shared-Aperture Antenna Design Using PRS

2.4.1. Frequency ratio range analysis for DS-FPC antennas

Subreflector Ray1 Ray2

Main reflector Main reflector

2.4.2. Dual-band shared-aperture antenna design using PRS

2.4.3. Performance metrics for shared-aperture antenna using PRS

3.1 Working Principle of Co-Designed Antennas

3.2 Investigation of Further Performance Improvements

3.2.1. Investigation of mm-wave array

impedance bandwidth of the mm-wave antenna is around 2GHz (26.56-28.54GHz).

3.2.2. Investigation of sub-6GHz antennas

3.3 Effects of Practical Design Considerations