Mobile Antenna Systems for 4G and 5G Applications with User Body Interaction

Kun Zhao

Doctoral Thesis School of Electrical Engineering

KTH Royal Institute of Technology Stockholm, Sweden 2017

| II

TRITA-EE 2017:137 KTH School of Electrical Engineering ISSN 1653-5146 SE - 100 44 Stockholm ISBN 978-91-7729-550-1 SWEDEN Akademisk avhandling som med tillstånd av Kungliga Tekniska högskolan framlägges till offentlig granskning för avläggande av teknologie doktorsexamen fredagen den 27 oktober kl. 10:00 i Kollegiesalen, Brinellvägen 8, Kungliga Tekniska högskolan, Stockholm. © Kun Zhao, October 2017 Tryck: Universitetsservice US AB

| III

Abstract

The cellular communication system has evolved from first generation (1G) analog systems that are only for voice communications, to 4th generation (4G) systems that can support high-throughput data transmission. Recently, the evolution of cellular systems is on the way to 5th generation (5G), which plans to utilize the spectrum over 6 GHz and into millimeter wave (mmWave) frequency range to deploy a wideband mobile network. To cope with the evolutions, various antenna systems have also been developed into mobile terminals in the past decades. The cellular antenna system in a mobile phone has been upgraded from a single antenna to a Multiple-Input Multiple-Output (MIMO) an-tenna system nowadays, which will be developed into an adaptive array system in future 5G commu-nications.

In the thesis, the user body effect on the performance of mobile antennas in a 4G/ Long-Term Evolution (LTE) mobile terminal is discussed first. In order to overcome the degradation of MIMO performance due to the user body effect, a distributed quad-elements MIMO antenna array for mobile terminals is introduced, which mitigates the body effect through an adaptive antenna switching method. Various novel bezel MIMO antennas for the mobile terminal application have also been pre-sented in the thesis. The proposed antennas operate in loop modes, and are robust to the impedance mismatching caused by the user effect. A double-ring shaped bezel structure is also introduced to extend the bandwidth of such kind of designs.

The study is further extended to frequency bands at 15 GHz and 28 GHz to investigate the user body effect on mobile antennas and wireless channels for future 5G communications. The body effect on antennas is obtained through 3D far-field measurements of the antenna radiation pattern with a real user. The corresponding impact on characteristics of the wireless channel is investigated with ray-tracing simulations. The results reveal that the user body will cause a strong shadowing loss at 15 GHz and 28 GHz, which will be critical to 5G cellular systems.

The electromagnetic field (EMF) exposure of a mobile terminal to the user’s body is also consid-ered in this thesis. For mobile terminals operating below 6 GHz, the specific absorption rate (SAR) is the core metric for EMF exposure compliance tests. The assessment of the simultaneous transmission SAR for MIMO antennas is the major focus of the thesis due to its complicated and time-consuming procedures. The SAR to peak location spacing ratio (SPLSR) is used by Federal Communications Commission (FCC) to determine the exclusion condition of simultaneous transmission SAR assess-ment, which is an important parameter for the MIMO antenna design in commercial phones. The SPLSR of various 2×2 MIMO antennas are compared in the thesis, to provide a benchmark for MIMO antenna designs in 4G/LTE terminals.

For future 5G terminals that operate above 6 GHz, the SAR is not valid anymore as a metric for EMF exposure compliance tests. Instead, the free space power density (PD) is adopted globally as the basic parameter of exposure limits. However, existing regulations on PD are not suitable for mobile terminal applications. Preliminary studies have been presented in this thesis to address the major challenges of PD assessment for the mobile terminal application.

Keywords Antenna, MIMO, mmWave, Mobile communication, SAR, Power density, User body effect, Channel modeling, Multiplex efficiency

| IV

Sammanfattning

De mobila kommunikationssystemen har utvecklats från första generationens (1G) analoga sy-stem, avsedda för enbart röstkommunikation, till fjärde generationens (4G) system som kan stödja högkompatibel datakommunikation. Den pågående utvecklingen går mot femte generationens (5G) system, som planeras för att utnyttja frekvensspektrumet från 15 GHz upp till millimetervågor (mmWave), för att kunna tillhandahålla ett bredbandigt mobilt nätverk. För att klara av utvecklingen av mobila terminaler har olika antennsystem utvecklats under de senaste decennierna. Antennsyste-met i en mobiltelefon har nu uppgraderats från att innehålla en enda antenn till ett MIMO-antennsy-stem (Multiple Input Multiple Output) innehållande flera antenner, vilket kan leda till adaptiva grup-pantenner i framtida 5G-system.

I avhandlingens första del diskuteras hur användarens kropp påverkar prestanda hos antennerna i en mobil terminal för 4G och för den långsiktiga utvecklingen efter 3G (LTE). Även den bakomlig-gande teorin diskuteras kortfattat. För att motverka försämringen av MIMO-prestanda i närvaron av användarens kropp introduceras för mobila terminaler en distribuerad fyrvägs MIMO-antennmatris, vilken kan minska påverkan från kroppen genom adaptiv växling mellan antennerna. Dessutom har nya chassi-baserade MIMO-antenner till 4G/LTE-mobilterminaler tagits fram för att möta efterfrå-gan på industriell produktion. De föreslagna chassi-antennerna arbetar i ett loop/spår-läge, vilket ger en stabil impedansanpassning även när användaren rör vid terminalen. En dubbel ringformad ram-konstruktion införs också för att utvidga bandbredden för denna typ av konstruktion.

Studien utvidgas sedan till frekvensbanden runt 15 GHz och 28 GHz för att undersöka hur använ-darens kropp påverkar mobila antenner och trådlösa kanaler i framtida 5G-kommunikation. Krop-pens inverkan på antennerna karaktäriseras genom 3D-fältmätningar av antennernas strålningdia-gram med en användare på plats, och motsvarande inverkan på de trådlösa kanalernas egenskaper studeras med hjälp av strålgångssimuleringar. Resultaten påvisar att människokroppen medför stora förluster på grund av att den blockerar och sprider effekt vid 15 GHz och 28 GHz, vilket är avgörande vid modelleringen av vägförlusterna utmed den trådlösa kanalen.

Det elektromagnetiska fältet (EMF) som absorberas av användaren kommer också att orsaka en temperaturökning i den mänskliga vävnaden, där de tillåtna nivåerna av EMF-exponering är strängt regulerade. För 4G-mobilterminaler som arbetar vid 6 GHz används den specifika absorptionsnivån (SAR) för att avgöra hur EMF-exponeringen överensstämmer med gränsvärdena.

Utvärderingen av SAR vid simultan sändning från MIMO-antenner är huvudmotivet för avhand-lingen, på grund av det kräver komplicerade och tidskrävande procedurer. Förhållandet mellan SAR och avståndet mellan dess utpräglade maxima (SPLSR) används av Federal Communications Com-mission (FCC) för att bestämma uteslutningsvillkoren vid simultan sändning. SPLSR är en viktig pa-rameter vid design av MIMO-antenner för kommersiella telefoner. Olika 2 × 2 MIMO antennkon-struktioner föreslås, och deras SPLSRs jämförs i avhandlingen för att tillhandahålla riktlinjer vid kon-struktionen av MIMO-antenner i 4G/LTE-terminaler.

För EMF-exponeringen från 5G-terminaler, som kommer att arbeta vid frekvenser över 15 GHz, är SAR-måttet inte relevant. Istället används globalt den elektromagnetiska vågens effekttäthet (PD) i fri rymd som det grundläggande måttet av exponeringen. De befintliga reglerna för PD är dock inte lämpliga för mobilterminaltillämpningar. Preliminära studier har presenterats i denna avhandling för att behandla de viktigaste utmaningarna vid användandet av PD-måttet för mobilterminaler.

Nyckelord Antenn, MIMO, Millimetervågor, Mobil kommunikation, SAR, Effekttäthet, Användarens kroppsef-fekt, Kanalmodellering, Multiplexeffektivitet

| V

Acknowledgement

First, I would like to express my sincerest gratitude to my supervisor Prof. Sailing He for taking a chance with me and offering the opportunity to start my Ph.D. education at KTH. I am grateful for his abundant support on my Ph.D. study. He helped me to develop critical thinking and motivated me to enter the world of research, which benefit me to grow up to be an independent researcher. I wish to thank Dr. Zhinong Ying from Sony Mobile for his excellent guidance on my research. His deep knowledge in the field of mobile antennas and original ideas have been a constant source of inspiration. In addition, his comments and suggestions have enlightened me on many different levels. I would also like to thank Dr. Shuai Zhang from Aalborg University who guided me in the world of antenna hand by hand, which greatly increases my knowledge and interest of antennas. I gratefully acknowledge to Mr. Thomas Bolin, Mr. Erik Bengtsson, Mr. Olof Zander and Dr. Peter Karlsson from Sony Mobile. I appreciate the fruitful discussions I had with them and their unlimited support during my time at Sony Mobile. I am grateful to all my research partners, including, but not limited to: Prof. Daniel Sjöberg, Prof. Gert Pedersen, Dr. Carl Gustafson, Jakob Helander, and Igor Syrytsi. Special thanks to Prof. Martin Norgren for his careful review of this thesis and the comments pro-vided. I would like to thank Dr. Nathaniel Taylor for the help of proof reading the thesis. I would also like to thank Dr. Oscar Quevedo Teruel and Prof. Lars Jonsson for providing many helps during my Ph.D. study. With the full administrative support, Carin Norberg and Ulrika Petterson, as well as system adminis-trator Peter Lönn are also the people that I have to acknowledge. I would like to thank former and current researchers and Ph.D. students in our department for the time we have been working together and the fun we had in the past five years. They are: Prof. Baoyou Zhu, Dr. Andres Alayon Glazunov, Dr. Pu Zhang, Dr. Hui Li, Dr. Helin Zhou, Dr. Fei Sun Dr. Kexin Liu, Dr. Roya Nikjoo, Dr. Mariana Dalarsson, Dr. Mengni Long, Dr. Lipeng Liu, Mahsa Ebrahimpouri, Christos Kolitsidas, Elena Kubyshkina, Fatemeh Ghasemifard, Andrei Osipov, Mauricio Aljure Rey, Jan Henning Jürgensen, Kateryna Morozovska, Sanja Duvnjak Zarkovic, Sajeesh Babu, Per Westerlund, Peyman Mazidi, Shuai Shi, Yue Cui, Bing Li, Qingbi Liao, Lei Wang, and Bo Xu among many others. It is a great fortune for me to spend time together with all of you. I would like also to thank my friends in Stockholm for the time we enjoy together, your help and accompany, they are: Victor Nygren, Rita Chedid, José M. Escudero, Raquel Alcalá Borao, Iman Lash-gari, Jasmine Elsayed, Daniel Luca, Nader Al-Ghazu, Martha Anthidou, and Björn Löfroth. Last but not least, I want to send all my thankfulness to my parents, and my wife Wen Jin for their endless love and support. Kun Zhao

| VI

| VII

List of papers

The candidate performed the main part of the work in the following five papers that are included in this thesis:

I. K. Zhao, S. Zhang, K. Ishimiya, Z. Ying, and S. He, “Body-insensitive Multi-Mode MIMO Terminal Antenna of Double-Ring Structure,” IEEE Trans. Antennas Propag., vol.63, no.5, pp.1925-1936, May 2015.

II. K. Zhao, J. Helander, D. Sjöberg, S. He, T. Bolin and Z. Ying, “User Body Effect on Phased

Array in User Equipment for the 5G mmWave Communication System,” IEEE Antennas Wireless Propag. Lett., vol. 16, pp. 864-867, 2017.

III. K. Zhao, C. Gustafson, Q. Liao, S. Zhang, T. Bolin, Z. Ying, and S. He, “Channel Characteris-tics and User Body Effects in an Outdoor Urban Scenario at 15 and 28 GHz ,” accepted in IEEE Trans. Antennas Propag., Special Issue on Antennas and Propagation Aspects of 5G Communications.

IV. K. Zhao, S. Zhang, Z. Ying, T. Bolin and S. He, “SAR Study of Different MIMO Antenna De-signs for LTE Application in Smart Mobile Handsets,” IEEE Trans. Antennas Propag., vol.61, no.6, pp.3270,3279, Jun. 2013.

V. K. Zhao, Z. Ying and S. He, “EMF Exposure Study Concerning mmWave Phased Array in Mobile Devices for 5G Communication,” IEEE Antennas Wireless Propag. Lett., vol. 15, pp. 1132-1135, 2016.

During the graduate study, the candidate has also participated in other closely related research activities. The resulting publications and patents, though not included in this thesis, are listed in the following: Journal Papers:

VI. K. Zhao, Z. Ying, and S. He, “Intrabody Communications Between Mobile Device and Wear-able Device at 26 MHz,” IEEE Antennas Wireless Propag. Lett, vol. 16, pp. 549-552, 2017.

VII. I. Syrytsin, S. Zhang, G. Pedersen, K. Zhao, T. Bolin and Z. Ying, “Statistical Investigation of the User Effects on Mobile Terminal Antennas for 5G Applications,” accepted in IEEE Trans. Antennas Propag., Special Issue on Antennas and Propagation Aspects of 5G Communica-tions

VIII. B. Xu, K. Zhao, B. Thors, D. Colombi, O. Lundberg, Z. Ying and S. He, “Power Density Meas-urements at 15 GHz for RF EMF Compliance Assessments of 5G User Equipment,” accepted in IEEE Trans. Antennas Propag., Special Issue on Antennas and Propagation Aspects of 5G Communications.

IX. J. Helander, K. Zhao, Z. Ying and D. Sjöberg, “Performance Analysis of Millimeter-Wave Phased Array Antennas in Cellular Handsets,” IEEE Antennas Wireless Propag. Lett., vol. 15, pp. 504-507, 2016.

X. S. Zhang, K. Zhao, Z. Ying and S. He, “Investigation of Diagonal Antenna-Chassis Mode in Mobile Terminal LTE MIMO Antennas for Bandwidth Enhancement,” IEEE Antennas Propag. Mag., vol. 57, no. 2, pp. 217-228, Apr. 2015.

XI. S. Zhang, K. Zhao, Z. Ying and S. He, “Adaptive Quad-Element Multi-Wideband Antenna Array for User-Effective LTE MIMO Mobile Terminals,” IEEE Trans. Antennas Propag., vol. 61, no. 8, pp. 4275-4283, Aug. 2013.

| VIII

Conference Papers:

XII. K. Zhao, Z. Ying and S. He, “Evaluation of combined TIS for high order MIMO system in

mobile terminal,” 11th European Conference on Antennas and Propagation (EUCAP), Paris, France, 2017, pp. 3684-3687.

XIII. K. Zhao, E. Bengtsson, Z. Ying and S. He, “Multiplexing efficiency of high order MIMO in mobile terminal in different propagation scenarios,” 10th European Conference on Anten-nas and Propagation (EuCAP), Davos, 2016, pp. 1-4.

XIV. K. Zhao, Z. Ying and S. He, “Double Ring Antenna Design for MIMO Application in Mobile Terminals,” 9th The European Conference on Antennas and Propagation (EuCAP), Lisbon, 2015, pp. 1-5.

XV. K. Zhao, Z. Ying and S. He “Antenna Designs of Smart Watch for Cellular Communications by using Metal Belt,” 9th The European Conference on Antennas and Propagation (EuCAP), Lisbon, 2015, pp. 1-5.

XVI. K. Zhao, Z. Ying and S. He, “SAR Study on MIMO Wi-Fi Antennas in LTE Mobile Terminals,” Progress In Electromagnetics Research Symposium (PIERS) , Best Poster Award, Aug 2014, Guangzhou, China

XVII. K. Zhao, S. Zhang, C. Chiu, Z. Ying and Sailing He, “SAR study for smart watch applications,”

IEEE Antennas and Propagation Society International Symposium (APSURSI), Memphis, TN, 2014, pp. 1198-1199.

XVIII. K. Zhao, S. Zhang, Z. Ying, and S. He, “MIMO performance study of different antennas for

LTE mobile phones in CTIA test mode,” 8th The European Conference on Antennas and Propagation (EuCAP), Gothenburg, 2013, pp. 727-731.

US Patents:

XIX. K. Zhao, S. Zhang, Z. Ying, E. Bengtsson, R. Ljung and S. He, “Multi-band wireless terminals with multiple antennas along an end portion, and related multi-band antenna systems,” US Patent, Grant, US8804560 B2, Aug. 2014.

XX. Z. Ying, K. Zhao, S. Zhang and S. He, “Double ring antenna with integrated non-cellular

antennas,” US Patent, Grant, US 9197270 B2, Nov. 2015.

XXI. S. Zhang, Z. Ying, K. Zhao and S. He, “Multi-band wireless terminals with multiple antennas along an end portion, and related multi-band antenna systems,” US Patent, Grant, US9673520 B2, Jun. 2017.

| IX

Acronyms

1G First Generation 3GPP 3rd Generation Partnership Project 4G 4th Generation 5G 5th Generation AIM Adaptive Impedance Network AoA AoD

Angle of Arrival Angle of Departure

APS Angular Power Spectrum AWGN Additive White Gaussian Noise CDF CFCM

Cumulative Distribution Function Components Field Combining Method

CSI Channel State Information CTIA DKED DRA

Cellular Telecommunications and Internet Association Double Knife-Edge Diffraction Dielectric Resonant Antenna

ECC Envelope Correlation Coefficient EMF Electromagnetic Fields FCC Federal Communications Commission FSPL GTD

Free Space Path Loss Geometrical Theory of Diffraction,

ICNIRP International Commission on Non-Ionizing Radiation IEEE Institute of Electrical and Electronics Engineers IFA Inverted-F Antenna

LOS Line-of-Sight LTE MEMS MFCM

Long-Term Evolution Microelectromechanical Systems System Magnitude Field Combining Method

MIMO Multiple-Input Multiple-Output mmWave Millimeter Wave MPAC Multi-Probe Anechoic Chamber MPC Multi-Path Component MPTP Maxim Permissible Transmitting Power MRC Maximum Ratio Combining MUX Multiplex Efficiency OTA Over-The-Air PD Power Density PDF PIFA

Probability Density Function Planar Inverted-F antenna

RF Radio Frequency Rx Receiver SAM Specific Anthropomorphic Mannequin SAR Specific Absorption Rate SD Standard Deviation SISO SIW

Single-Input Single-Output Substrate Integrated Waveguide

SNR Signal to Noise Ratio SPLSR SAR to Peak Location Separation Ratio SVD TCM

Singular Value Decomposition Theory of Characteristic Mode

| X

TIS Total Isotropic Sensitivity TRP Total Radiated Power Tx Transmitter UE UTD

User Equipment Uniform Theory of Diffraction

XPR Cross Pole Ratio

| XI

Contents

Chapter 1 Introduction ...................................................................................................................................... 1 1.1. Antennas in Modern Mobile Phones .................................................................................................... 1 1.2. MIMO Antenna Systems in 4G Communications ............................................................................... 3 1.3. Antenna Systems for 5G Communications ......................................................................................... 7 1.4. User Body Effects on Mobile Antenna Systems ............................................................................... 10 1.5. EMF Exposure of Mobile Terminals ................................................................................................... 11 1.6. Outline of The Thesis .......................................................................................................................... 11

Chapter 2 Evaluation of Mobile Antenna Systems and Modeling of Wireless Channels .......................... 13 2.1. Evaluation of MIMO Antenna System in Mobile Terminals .............................................................. 13 2.2. Evaluation of Antenna Array in Mobile Terminals for 5G Communications .................................. 15 2.3. Wireless Channel Modeling ................................................................................................................ 16

Chapter 3 MIMO Antenna Systems for 4G Mobile Terminals with Consideration of User Body Effect ... 19 3.1. User Body Effect on Mobile Antenna Systems ................................................................................. 19 3.2. User Body Effect on MIMO Antenna Systems in Mobile Terminals ................................................ 20 3.3. Body Insensitive MIMO Antenna Designs ......................................................................................... 22

3.3.1 Adaptive Quad-Element Antenna Array for User-Effective MIMO Mobile Terminals ............. 22 3.3.2. Impedance Mismatching Robust Metal Bezel Antenna Designs ............................................. 24

Chapter 4 User Body Effect on 5G Mobile Antenna Systems and Channel Characteristics .................... 31 4.1. User Body Effect on Antenna Systems at 15 GHz and 28 GHz ....................................................... 31

4.1.1. User Body Effect on Embedded Elements ................................................................................ 32 4.1.2. User Body Effect on Coverage Efficiency and Total Scan Pattern ......................................... 34 4.1.3. Statistic Investigation on User Body Effect .............................................................................. 36

4.2. User Body Effect on Channel Characteristics at 15 GHz and 28 GHz ............................................ 38 4.2.1. The Received Signal Strength without User Obstruction ........................................................ 39 4.2.2. The Received Signal Strength in Data Mode and Talk Mode ................................................... 40 4.2.3. The Received Signal Strength with Different User’s Orientations .......................................... 42

Chapter 5 EMF Exposure of Mobile Terminals to User Body ...................................................................... 45 5.1. SAR Assessment for Mobile Antennas ............................................................................................. 46

5.1.1. SAR Assessment for Multiple Antenna Systems ..................................................................... 47 5.1.2. Evaluation of SPLSR for MIMO Antennas ................................................................................. 47

5.2. Power Density Assessment for 5G Mobile Antennas ...................................................................... 52 5.2.1. Power Density of Linear Arrays ................................................................................................. 53 5.2.2. Power Density Assessment and Compliance Test Methods ................................................... 54

Conclusion and Future Work ......................................................................................................................... 57 Summary of Appended Papers ...................................................................................................................... 59 Bibliography .................................................................................................................................................... 61

CHAPTER 1 INTRODUCTION | 1

Chapter 1 Introduction Since the end of the 1970s, the cellular mobile communication system has been evolved from first generation (1G), which is only capable of voice communications, to today’s 4th generation (4G) sys-tem that can support a high-throughput data transmission. Meanwhile, to cope with the deployment of advanced cellular networks, a variety of mobile terminal antenna systems have been developed. Challenges of designing a state-of-the-art mobile terminal antenna include minimizing the antenna dimensions, fulfilling industrial design requirements, operation at multiple frequency bands, inte-grating the multiple-input multiple-output (MIMO) antenna systems, and so forth.

1.1. Antennas in Modern Mobile Phones External stubby antennas were used in mobile phones back in the 1990s, which is simple and easy to reuse. The external stubby antenna design is commonly realized by non-uniform helix [1], which func-tioning as a quarter wave monopole at a lower frequency band, and has a non-uniform diameter to control the higher frequency band (Fig. 1.1).

Figure 1.1. The external stubby antenna (source: Ericsson) with a non-uniform helix design, and the internal antenna with a PIFA type design (adapted from [1]).

From the end of the 1990s, the internal antenna has started to emerge and won its popularity within a few years due to its compact structure, mechanically robustness and cheap cost. Monopole type, Inverted-F antenna (IFA) type, Planar Inverted-F antenna (PIFA) type and loop type antennas are commonly used for internal mobile antenna designs, which are illustrated in Fig. 1.2.

Figure 1.2. The monopole type, IFA type, PIFA type and loop type antennas in mobile terminals.

CHAPTER 1 INTRODUCTION | 2

The dimensions of modern commercial mobile phones tend to be slimmer than ever; the thickness of a commercial phone is typically between 7 mm to 10 mm. Also, the increased display size shrinks the clearance space for antennas. In recent years, integrating the antennas onto the metal bezel of mobile terminals has become a popular trend in commercial productions (Fig. 1.3). The metal bezel antenna can provide a strong mechanical robustness and is also aesthetically pleasing, which is highly attrac-tive to industries and consumers. The most common way of designing a bezel antenna is to cut several slits on the bezel, which breaks it into isolated metal pieces. Then, the antenna types mentioned above (monopole, IFA and loop) can be integrated into the phone bezel.

Figure 1.3. The metal bezel antenna design on the iPhone 4, and the illustration of a bezel antenna structure.

In opposition to more compact dimensions, the cellular frequency bands that mobile antennas need to cover keep increasing [1]-[2] (see Fig. 1.4). The primary cellular bands are distributed between 700 MHz to 2.6 GHz at present, which is expected to be further spread with the evolution of Long-Term Evolution (LTE) system and also the next generation cellular network.

Figure 1.4. The number of 3GPP cellular RF bands growing from 2002 to 2014 (adapted from [1]).

Various technics have been developed for broadening the bandwidth of mobile antennas. As shown in Fig. 1.5, PIFA antenna with multiple cutting slots can generate multiple resonates, which is widely used in internal antenna designs. Parasitic elements can also be added close to the main radiator to provide extra resonates [3]. Feeding the antenna through capacitive coupling was found can enhanced antenna bandwidth effectively [4], and it has also become a popular technique for mobile antenna designs. Furthermore, exciting the chassis mode of the mobile handset is also a common method to enhance the antenna bandwidth [5].

CHAPTER 1 INTRODUCTION | 3

Figure 1.5. A PIFA with multiple cutting slots, a PIFA with a parasitic element, and a capacitive coupling fed PIFA (adapted from [1]).

1.2. MIMO Antenna Systems in 4G Communications Conventional communication systems with a single antenna in Transmitter (Tx) and Receiver (Rx) are called Single-input Single-output (SISO) systems. The SISO channel can be illustrated as in Fig. 1.6, and its narrow band channel model in a static environment is described as:

𝑦 = ℎ𝑥 + 𝑛 (1.1) where 𝑦 is the received signal, 𝑥 is the transmitted signal, ℎ is the channel impulse response, and n is the Additive White Gaussian Noise (AWGN). The channel capacity of the SISO channel is:

𝐶 = log 1 + |ℎ | (1.2)

where 𝑃 is the transmitted power and 𝜎 is the power spectrum of the AWGN. Therefore, the capac-ity of the SISO channel increases logarithmically with the transmitted power.

Figure 1.6. Channel models for SISO and MIMO communication systems.

CHAPTER 1 INTRODUCTION | 4

In 4G/LTE communication networks, MIMO antenna systems have been widely deployed, which can multiply the capacity of a radio link by exploiting multipath propagation. The MIMO channel with N antennas at Tx and M antennas at Rx is illustrated in Fig. 1.6, and the channel response is described by a vector matrix H:

𝐇 =ℎ ⋯ ℎ

⋮ ⋱ ⋮ℎ ⋯ ℎ

(1.3)

The MIMO system can be modeled as:

𝐲 = 𝐇𝐱 + 𝐧 (1.4)

where x, y, and n are the transmitted signal vector, the received signal vector and the AWGN vector, respectively. Without channel state information (CSI) at the Tx, the transmitted power will be distrib-uted evenly to all transmitting antennas, and the MIMO channel capacity in this situation can be written as:

𝐶 = log det 𝐈 + 𝐇𝐇 = ∑ 1 + 𝜆 (1.5)

The second equation in Eq. (1.5) is obtained through the singular value decomposition (SVD) of the channel matrix H = U𝚲𝐕 , where U and V are rotation unitary matrices. 𝚲 is a rectangular matrix whose diagonal elements 𝜆 are singular values of the channel matrix H, and K is the rank the of the channel matrix H. Eq. (1.5) converts a MIMO channel into multiple parallel channels through SVD, as shown in Fig. 1.7. It reveals that the MIMO technology enhances the channel capacity by supporting multiple parallel streams in the channel, which is a process known as spatial multiplexing.

Figure 1.7. Singular value decomposition of the MIMO channel. In addition to achieving a higher channel capacity through the spatial multiplexing, the MIMO system can also be used to facilitate spatially-independent fading to increase the received signal to noise ratio (SNR), which is known as the diversity scheme. In diversity mode, a MIMO system will transmit mul-tiple copies of the signal at Tx, which will go through independent channels to combat the deep fading. In a mobile terminal, the antenna diversity can usually be realized in three different ways: 1, the spa-tial diversity, which separates multiple antennas from one another physically; 2, the pattern diversity, which makes the multiple antenna elements radiate in different directions; 3, the polarization diver-sity, which combines two antennas with orthogonal polarizations. The diversity scheme needs techniques to combine the received signals from each antenna element, to recover the desired information. Among these techniques are switch combining, selection combin-ing, equal-gain combining and maximal-ratio combining. Switch combining will only receive the sig-nal from one antenna until the received SNR drops below a threshold level, while the selection com-bining will always choose the antenna with the highest SNR. On the other hand, equal-gain combining and maximal-ratio combining will coherently add received signals from all antennas in the MIMO

CHAPTER 1 INTRODUCTION | 5

array (Fig. 1.8). The maximal-ratio combining will also weigh all the signals, which is the optimal diversity technology if the CSI is perfectly known at the Rx [6]. The improvement of SNR due to the diversity scheme of a MIMO antenna relative to the SNR from a single antenna is described by the diversity gain, which is conditioned by the probability that the SNR is above a reference level.

Figure 1.8. The coherent combining of diversity scheme. Though a MIMO system can increase the channel capacity and the reliability of the communication system, its effectiveness is limited by the propagation environment as well as the antenna perfor-mance. Integrating the MIMO antenna system in a mobile terminal is challenging due to the limited space and complex electromagnetic environment [7]. A good MIMO antenna system needs not only the optimization of individual antenna but also requires reducing the interaction between multiple antenna elements, e.g., lowering the mutual coupling and the spatial correlation coefficient. Tech-niques regard the de-coupling and the de-correlation of MIMO antennas have been developed in the past decade, and several methods have been proposed and implemented in mobile terminal applica-tions. The most effective way is to design the antennas operate in orthogonal modes, which can be realized by combining an electrical dipole antenna and a magnetic dipole antenna [8]. Adding a scat-ter element between two antenna elements can also reduce the mutual coupling and the spatial cor-relation coefficient, where the scattering element can be realized by parasitic stubs, current chokes, or slots on the ground plane [9]-[10]. In [11], the technique using neutralization line to reduce the antenna mutual coupling is introduced. The neutralization line changes the current interaction be-tween the two antenna elements and leads to a reduced mutual coupling. In addition, the hybrid cou-pler and lumped element decoupling networks can also be added on antenna ports, which can effec-tively lower the mutual coupling between MIMO antennas [12].

Figure 1.9. The decoupling through the orthogonal mode MIMO antenna, the parasitic scattering element, the neutralization line and the decoupling network methods.

CHAPTER 1 INTRODUCTION | 6

Closely spaced multiple antennas have a strong mutual scattering effect, which can reduce the corre-lation coefficient of a MIMO antenna. The mutual scattering effect can be well controlled through the quality factor (Q) of the MIMO antenna. A multiband MIMO antenna has been proposed for mobile terminals based on this concept in [13]. In cellular bands below 1 GHz, the phone chassis usually plays a major role in radiation. Therefore, exciting the orthogonal current modes on the chassis can also reduce mutual coupling and correlation coefficient of the MIMO antenna. It has been shown in [14] that by optimizing antenna locations on the chassis, the diagonal chassis mode can be excited to re-duce the spatial correlation. In recently, the theory of characteristic mode (TCM) has become a pow-erful tool to design MIMO antenna in mobile terminal; it gives insights into physical mechanisms that how the MIMO performance of multiple antennas relate to their location on a mobile chassis [15]. The characteristic modes are intrinsically orthogonal to each other which makes it fit for MIMO an-tenna designs (see Fig. 1.10). Optimized MIMO antenna in mobile phones can be obtained by manip-ulating the phone chassis with the assistance of TCM analysis [16].

Figure 1.10. The characteristic currents of mode 1 and mode 2 of a phone chassis (150 mm x 70 mm) at 900 MHz.

Recently, the order of MIMO system in commercial phones has evolved from the 2×2 system to the 4×4 system (see Fig. 1.11), and it may even be developed into 8×8 in the future [17]-[18]. The high-order MIMO will further raise the difficulty of designing a mobile antenna system.

Figure 1.11. The topology of a 4×4 MIMO antenna system in a commercial mobile phone.

CHAPTER 1 INTRODUCTION | 7

1.3. Antenna Systems for 5G Communications MIMO technology increases the channel capacity by exploiting spatial multiplexing. In addition to this, a more straightforward way to increase the channel capacity linearly is to use a larger bandwidth, which has been recognized as one of the features of the next generation cellular system. Since 2012, the evolution of cellular mobile systems has started to move to the 5th generation (5G). The 5G communication system aims to provide innovated services driven by dramatically increased data traffic, e.g., high-definition video, virtual reality and so on. Consequently, more spectrum re-sources will be required to support such a high data throughput. As a significant amount of spectrum is available, frequency bands above 6GHz and up, into the millimeter wave (mmWave) range, have been tested recently as potential carrier frequencies for the 5G communication [19]-[20]. Currently, only 780 MHz spectrum bandwidth is allocated for cellular technologies globally between 700 MHz to 2.6 GHz [21], but there is over 1 GHz bandwidth available just at 28 GHz, as shown in Fig. 1.12.

Figure 1.12. The globe spectrum for 2G, 3G, 4G and the bands of interest for 5G communication systems.

Figure 1.13. The free space path loss at 900 MHz, 3 GHz, and 28 GHz.

CHAPTER 1 INTRODUCTION | 8

However, the propagation environment at higher frequencies will become less favorable for mobile communications. The free space path loss (FSPL) will be enlarged as the square of the frequencies if the antenna gain is fixed, as shown in Eq. (1.6). The FSPL at 28 GHz is about 30 dB higher than at 900 MHz (Fig. 1.13). Another important trait at such high frequencies is that the blocking effect from objects in the propagation environment will be more severe. The mmWave suffers a higher penetra-tion loss in conventional materials, as well as a larger diffraction loss due to a shorter wavelength [22].

FSPL = = (1.6)

One possible solution to overcome those issues is to use antenna arrays with beamforming in mobile terminals to compensate the higher losses. Moreover, to meet the demands of the mobile application, the beam steering function is also necessary for those arrays. Therefore, an adaptive array system is expected to be implemented in future 5G mobile terminals (Fig. 1.14).

Figure 1.14. The evolution of mobile antenna technologies from 4G to 5G. Beamforming is realized by assigning a weight coefficient to each antenna element, and it can be done in the digital domain, the analog domain or by a hybrid way [23]. Architectures of those beamforming methods are shown in Fig. 1.15. In digital beamforming, the coefficients are introduced on baseband signals for each radio frequency (RF) chain. The digital beamforming offers a better performance but with increased complexity and cost. The analog beamforming is done by applying coefficients to ana-log signals in the time domain, which is a simpler method but offers less flexibility. Hybrid beam-forming tries to combine advantages from both approaches: it can reduce the number of RF chains but still offer certain flexibilities in the system, which is a promising solution for 5G mobile commu-nication systems.

Figure 1.15. The digital, analog and hybrid beamforming architectures.

CHAPTER 1 INTRODUCTION | 9

Array designs in a mobile terminal must take into account practical restrictions of mobile terminals and requirements of mobile communications [24]. A cellular handset contains multiple dielectrics and electronic components, which will form anisotropic mediums around the array and distort the array’s radiation. For example, the phone case can cause a dome effect on the radiation pattern of an antenna element at 28 GHz, which is illustrated in Fig. 1.16. Moreover, a planar phased array will perform a sub-hemisphere coverage inherently. Therefore, it will be necessary to combine multiple arrays to achieve an omnidirectional coverage (See Fig. 1.17(a)).

Figure 1.16. The radiation pattern of an edge-mounted antenna element with and without a phone casing at 28 GHz. Antenna elements for 5G mobile terminals need to cover a wide bandwidth with dual-polarization. Meanwhile, they should be feasible to be integrated into a mobile terminal. Performances of five typical element designs are summarized in Table 1.1 which includes the patch antenna, the dipole antenna, the slot antenna, the substrate integrated waveguide (SIW) antenna and the dielectric reso-nant antenna (DRA) (Fig. 1. 17(b)). The performances listed in Table 1.1 are based on a fundamental design with a thin substrate.

(a) (b) Figure 1.17. (a) The spatial coverage of multiple arrays in mmWave 5G mobile terminal. (b) Some typical element designs for mmWave antenna array in mobile terminals. Phase shifters in antenna arrays usually introduce high insertion losses at mmWave frequency bands. Alternative antenna solutions have also been proposed for 5G mobile terminals to reduce the number or even avoid using phase shifters. A beam switch array with a passive beamforming circuit, e.g., a Bulter matrix, is an approach that can generate beams at predefined directions and can be fully inte-grated with an antenna array. A lens antenna or a reflector antenna is also a possible replacement for an adaptive array in mobile terminals at a higher frequency in mmWave band. By implementing multi feedings at different positions, it is possible to generate multiple prefixed high gain beams with those two techniques [25]. A lens antenna in mobile terminals can be formed by shaped dielectric or

CHAPTER 1 INTRODUCTION | 10

metasurface. A pattern reconfigurable element can also be used to enlarge the scanning angle of a phased array, where the antenna element pattern can be reconfigured by loading the antenna with Microelectromechanical systems (MEMS) switches to change the current path on it [26]-[28]. Table 1.1. Summarized comparison of mmWave element performance.

Antenna Type Bandwidth Ease of Dual-pol Ease of Integration Main Radiation

Direction Patch small feasible feasible Rear Dipole large difficult moderate Side Slot moderate moderate feasible Rear & Front SIW small difficult feasible Side DRA moderate feasible difficult Rear

1.4. User Body Effects on Mobile Antenna Systems In addition to the requirements above, the user body effect on the performance of mobile antenna systems is also critical since a mobile terminal is often used in the immediate vicinity of a human body. To ensure a reliable communication in real life, 3GPP, CTIA, and mobile network operators introduce requirements on the over-the-air (OTA) performance of wireless devices including the im-pact of the user's head and hand [29]. The OTA test of mobile phones needs to be carried out in free space, in the talk mode and the data mode, as shown in Fig. 1.18.

Figure 1.18. A mobile terminal in real life (source: Flickr user "Kārlis Dambrāns"), and in OTA test with the CTIA talk mode and the CTIA data mode. The user body can change the impedance matching, the radiated power, and the radiation patterns of mobile antennas, which will generally degrade the performance of communication systems. For ex-ample, the antenna in the iPhone 4 suffers as much as 6.7 dB mismatching loss when the user’s finger touches the bottom slit on the metal bezel [30], which results in dramatic signal drops of the phone in real life. For a MIMO system, the user body effect will also change the interaction between anten-nas, which needs to be evaluated carefully under different propagation environments. For the future 5G communication system that operates above 6 GHz, the user body effect on mobile antennas will behave differently from the sub-6 GHz frequency bands. A more pronounced shadowing effect is expected due to the shorter wavelength. The shadowing loss of a human body at frequencies above 15 GHz and into mmWave is typically around 20-40 dB, which is a sever type of loss, as it can change rapidly within a small timescale due to the movement of users. Therefore, it is necessary to have a good understanding of the user body effect on characteristics of mobile antenna systems, as well as wireless channels of the 5G communication system.

CHAPTER 1 INTRODUCTION | 11

1.5. EMF Exposure of Mobile Terminals In addition to the body effect on antennas, the user will also be exposed to the electromagnetic field (EMF) when the phone is transmitting signals (Fig. 1. 19). A commercial mobile device must be tested to comply with relevant regulatory limits on human exposure to EMF. Basic restrictions on EMF ex-posure are defined in terms of the specific absorption rate (SAR) at frequencies below 3 GHz, 6 GHz and 10 GHz from the Institute of Electrical and Electronics Engineers (IEEE) [31], the International Commission on Non-Ionizing Radiation (ICNIRP) [32], and the Federal Communications Commis-sion (FCC) [33], respectively. For the future 5G communication system that operates above those transition frequencies, the free space power density (PD) is adopted as the basic restriction. Re-strictions on EMF will limit the level of the maximum permissible transmitted power (MPTP) from mobile terminals; it is, therefore, critical for cellular network designs.

Figure 1.19. EMF exposure to a user from a mobile terminal. In a mobile terminal, the SAR when multiple antennas are simultaneously transmitting is required to be evaluated for compliance tests. The simultaneous transmission SAR shows different properties to the stand-alone SAR and is more complicated to be assessed. Hence, corresponding conditions for test exclusion of simultaneous transmission SAR have also been established by regulators, which is critical for commercial mobile antenna designs. For the future 5G applications operating at frequen-cies above transition frequencies, current regulations from IEEE, ICNIRP, and the FCC are still im-mature for cellular mobile devices, and therefore fundamental studies are needed on the PD proper-ties of mobile antenna systems and the compliance test methodology.

1.6. Outline of The Thesis The thesis focuses on antenna system designs for mobile terminal applications with consideration of user body interactions. Chapters 2-4 mainly focus on the antenna system designs for 4G and 5G mobile terminals and the corresponding user body effect on antenna systems and wireless channels.

Chapter 2 discusses the general figure of merit used for antenna performance evaluations for MIMO and adaptive arrays.

Chapter 3 focuses on MIMO antenna designs for mobile terminals, which are robust to the user body effect.

Chapter 4 investigates the user body effect on antenna array systems for the 5G communica-tion, as well as the impacts on wireless channel characteristics at 15 GHz and 28 GHz.

CHAPTER 1 INTRODUCTION | 12

In Chapter 5, the attention is shifted to the EMF exposure of mobile antennas.

In sections 5.1, features of the stand-alone SAR and the simultaneous transmission SAR of multiple antenna systems are discussed. The SAR to peak location separation ratio (SPLSR) of various MIMO antenna designs for the mobile terminal application is investigated.

In section 5.2, properties regarding PD of adaptive arrays in a mobile terminal are revealed at 15 GHz, and methods for the future EMF compliance test are discussed.

Finally, contributions of the thesis and topics for the future work in relevant fields are summarized in conclusion.

CHAPTER 2 EVALUATION OF MOBILE ANTENNA SYSTEMS AND MODELING OF WIRELESS CHANNELS | 13

Chapter 2 Evaluation of Mobile Antenna Systems and Modeling of Wireless Channels The requirements on SISO mobile antennas focus on efficiency and omnidirectional coverage. There-fore, the standard way to evaluate the SISO antenna in OTA testing is to measure power-based met-rics: the total radiated power (TRP) and the total isotropic sensitivity (TIS) [34]. The power-based metrics can be used to estimate the link budget of wireless channels under a line-of-sight (LOS) envi-ronment. However, due to the deployment of MIMO technology that increases the channel capacity by exploiting multipath propagation, those parameters are no longer sufficient for OTA testing of mo-bile terminals. The MIMO OTA performance is evaluated by measuring the data throughput in a multi-probe anechoic chamber (MPAC) under emulated propagation channels [35].

2.1. Evaluation of MIMO Antenna System in Mobile Terminals The relation between the MIMO channel capacity and MIMO antenna parameters is not as intuitive as in a SISO system because many factors can affect the capacity of a MIMO system simultaneously. Extensive studies have been carried out in order to fill this gap [36]-[39]. In [36], the multiplex effi-ciency (MUX) is introduced, which is defined as the loss of power efficiency to achieve the same chan-nel capacity when using a proposed MIMO antenna comparing to a reference MIMO array under the same propagation environment. Considering a M×M MIMO channel, the instantaneous channel capacity with no channel information at the transmitter can be written:

𝐶 = log det 𝐈 + 𝐇𝐇 = log det 𝐈 + 𝐑𝟏/𝟐𝐇𝒘𝐇𝒘 𝐑 /𝟐 ≈ 𝐶 + log det(𝐑) (2.1)

where 𝐶 = log det 𝐇𝒘𝐇𝒘 denotes the capacity of the ideal independent and identically

distributed (i.i.d.) Rayleigh channel, and the approximate equation can be obtained under a high SNR assumption [40]; 𝐑 is the receiving matrix which describes the impact of the proposed MIMO antenna on the channel capacity. Under a three-dimensional (3-D) isotropic environment, the receiving matrix 𝐑 characterizes the ef-ficiency, efficiency imbalance, and correlation among the multiple receive antennas. By using the re-lation det(𝐑) = det det(𝐑) / I , the impact from the MIMO antenna on channel capacity is trans-lated to a power measure, which is defined as MUX:

𝜂 = det(𝐑) / = (∏ 𝜂 ) det(𝐑) (2.2)

𝜂 stands for the total efficiency of the ith antenna in the MIMO array, which takes into account mis-match, dielectric, conductive and mutual coupling losses. The antenna total efficiency in a MIMO array can be expressed as:

𝜂 = 𝜂 , (1 − |𝑆 | − ∑ 𝑆 ) (2.3) where 𝜂 , stands for the radiation efficiency of the ith antenna, which is the ratio of the radi-ated power to the accepted power of the MIMO antenna. 𝐑 in Eq. (2.2) denotes the normalized antenna correlation matrix, its entry 𝑟 is the complex corre-lation coefficient between the ith and the jth antenna in the MIMO array. The correlation coefficient of a MIMO antenna system describes how independent different antennas are, and it can be calcu-lated through 3-D far field patterns of any two antennas:

CHAPTER 2 EVALUATION OF MOBILE ANTENNA SYSTEMS AND MODELING OF WIRELESS CHANNELS | 14

𝑟 = ∫( , ,∗

, ,∗)

∫( , , ) ∫( , , ) (2.4)

where 𝐸 , / and 𝐸 , / are the 𝜃 and 𝜙 polarized complex electric field (E-field) patterns of the ith and the jth antennas. 𝐺 , / and 𝐺 , / are the 𝜃 and 𝜙 polarized power gain patterns of the ith and the jth antennas. Ω stands for the solid angle (𝜃, 𝜙) and 𝑑Ω = sin𝜃𝑑𝜃𝑑𝜙. In mobile communication systems, the envelope correlation coefficient (ECC) is more commonly used, which is defined as ECC = 𝑟 . In a lossless MIMO array, the ECC can also be calculated through the S-parameters [41]:

ECC = ∗ ∗

( | | )( ) (2.5)

However, the antenna efficiency loss in a mobile terminal is ineligible due to the compact size of an-tennas, complex surroundings, and non-ideal materials. Efficiencies below 50% are not uncommon for mobile antennas. Therefore, the ECC from Eq. (2.5) can give an inaccurate value in this case. In [42]-[44], improved models for calculating the ECC through S-parameters have been introduced, which include the impact from losses in MIMO antenna arrays. For a 2×2 MIMO antenna in a 3-D isotropic propagation environment, the MUX can be written as:

𝜂 = 𝜂 𝜂 (1 − |𝑟 | ) (2.6) From Eq. (2.6), it can be seen that maximizing the geometric mean of MIMO antenna total efficiencies and minimizing the correlation coefficient in between are the key factors to achieve a high MIMO capacity. Maximizing the geometric mean of MIMO antenna total efficiencies will require an optimi-zation on the total efficiency of each antenna element as well as a miniature of the branch power imbalance. In addition to the 3D isotropic propagation environment, Gaussian and Laplacian distributed incident field have also been used in standard channel models, e.g., in a typical urban propagation environment. Under an arbitrary propagation environment, the MUX can be obtained by replacing the antenna total efficiency 𝜂 in Eq. (2.2) with the modified mean effective gain (MEG) [38]. The MEG combines effects from both the antenna radiation pattern and the incident field at the antenna system, which is the ratio between the mean received power at an antenna and the total mean incident power [45]. The modified MEG of the ith antenna in the MIMO array is defined as:

MEG = 2𝜂 ∫ ∫ [ 𝐷 , (Ω)𝑃 , (Ω) + 𝐷 , (Ω)𝑃 , (Ω)]𝑑Ω (2.7)

where 𝐷 /𝑃 and 𝐷 / 𝑃 are the antenna directivity pattern/the angular power spectrum (APS) of in-cident field in 𝜃 and 𝜙 polarizations. 𝜒 is the cross pole ratio (XPR), which is ratio of 𝜃 and 𝜙 polar-ized components in the channel. The modified MEG can account for the realized antenna gain and ensure that MEG = 𝜂 when 𝑃 and 𝑃 are 3D uniformly distributed. The complex correlation coefficient 𝑟 also needs to be modified according to the APS of the incident field, where it is defined as:

𝑟 = ∫ , ,∗

, ,∗

∫ , , ∫ , ,

(2.8)

By replacing the antenna total efficiency 𝜂 and 𝑟 in Eq. (2.4) with modified MEG and 𝑟 in (2.8), The MUX can be generalized to fit for arbitrary propagation environments. The impact of propagation environment on MUX of some 2×2 and 4×4 MIMO arrays has been investigated in [39], [46].

CHAPTER 2 EVALUATION OF MOBILE ANTENNA SYSTEMS AND MODELING OF WIRELESS CHANNELS | 15

2.2. Evaluation of Antenna Array in Mobile Terminals for 5G Communications Due to the randomness of mobile terminal’s orientation and propagation environment, omnidirec-tional coverage is preferred for a mobile terminal or a user equipment (UE) antenna system. This issue is usually taken less care in mobile antenna designs in sub-6 GHz cellular bands, as the antenna directivity is relatively low and the number of multipath components (MPCs) is also relatively large. For the future 5G system operating above 6 GHz, the issue of spatial coverage, however, is going to be more critical, since highly directional array systems will be used, and the number of MPCs is expected to be less due to the higher diffraction loss as well (see Fig. 2.1). Therefore, to ensure a stable perfor-mance of 5G cellular networks, the array system in mobile terminals should have a wide beam steering angle.

Figure 2.1. The propagation environment of the cellular system in sub-6 GHz bands with a single antenna, and above 6 GHz with a antenna array. In [47], the total scan pattern and the coverage efficiency are introduced to assess the spatial coverage of an antenna array system in mobile terminal. The total scan pattern is obtained by drawing out the highest gain at every angular point with all possible beam steering patterns. (Fig. 2.2).

𝐺 (Ω) = max[𝐺 (Ω), 𝐺 (Ω), … , 𝐺 (Ω)] (2.9)

The coverage efficiency is a quantitative description of the phased array’s spatial coverage; it is retrieved from its total scan pattern with respect to a threshold antenna gain that can ensure the link budget of the communication:

𝜇 =

(2.10)

The covered solid angle is calculated as the summation of solid angles in the total scan pattern that have gain higher than a threshold level 𝐺 (Ω) > 𝐺 . The maximum solid angle in (2.10) can be chosen as the surrounding sphere 4π or can be defined by the requirement of communication sys-tems. The coverage efficiency and total scan pattern give an intuitive insight of how large solid angle that an adaptive array can cover, which can be used to evaluate and optimize the array topology in a mobile phone [47]. Moreover, it can also be related to some system-level parameters, e.g., outage probability, under certain assumptions. For example, if the propagation environment is assumed to be the random LOS with only one incoming ray [48] (a pure LOS environment with consideration of the random location and the random orientation of a mobile terminal), a loss in the coverage efficiency can be interpreted as an increment in the outage probability of the communication system [paper II].

CHAPTER 2 EVALUATION OF MOBILE ANTENNA SYSTEMS AND MODELING OF WIRELESS CHANNELS | 16

Figure 2.2. The illustration of the total scan pattern and the coverage efficiency for a phased array in the mobile terminal.

2.3. Wireless Channel Modeling A wireless channel model is vital for the development of a communication system. A good wireless channel model should be able to reproduce the typical behavior of the channel efficiently, and also give insights into the relevant physical mechanisms of the radio channel. The wireless channel modeling approaches can be categorized into stochastic and deterministic modeling approaches. Stochastic approaches characterize the statistical behavior of the channel, which are crucial tools for designing and optimization of communication systems. On the other hand, deterministic modeling approaches are inherently site-specific, which are commonly realized by ray-tracing simulations, full wave simulations or measurements. When using computer simulations to obtain a site-specific determinist channel model, it usually requires a complex model of the environment to obtain accurate results, which is often computationally expensive. However, it is easier to reveal the physical propagation mechanisms behind channel parameters with deterministic approaches. Moreover, those methods include directional information of MPCs in the channel (angle of arrival (AoA) and angle of departure (AoD)). Therefore, the impact of mobile terminal antennas can be embedded into the channel separately, which makes it possible to extend the user body effect on the mobile antenna to the mobile wireless channels. A wireless channel can be described as a number of MPCs between a Tx antenna and an Rx antenna (see Fig. 2.3), and such a channel model is called the double directional channel model [49]. This model includes the directions of each MPC, which is an important parameter for antenna array sys-tems. Therefore, the double directional channel model can fit the demands of the channel modeling with adaptive arrays and the study of the user body effect on wireless channels. The generic expression of the double directional channel impulse response is:

ℎ (𝜏, Ω , Ω ) = ∑ 𝑎 𝛿(𝜏 − 𝜏 )𝛿 Ω − Ω 𝛿 Ω − Ω (2.11)

where n is the MPC index, and 𝑎 is the complex amplitude of an MPC. Ω and Ω are the AoA and AoD, respectively; and 𝛿 is the Dirac delta function.

CHAPTER 2 EVALUATION OF MOBILE ANTENNA SYSTEMS AND MODELING OF WIRELESS CHANNELS | 17

Figure 2.3. The illustration of the double-directional channel model. Based on Eq. (2.11), the channel impulse response between a Tx antenna and an Rx antenna can be obtained by including the complex antenna gain patterns:

ℎ (𝜏, Ω , Ω ) = ∑ 𝑎 𝛿(𝜏 − 𝜏 )𝛿 Ω − Ω 𝛿 Ω − Ω 𝛿(𝜏 − 𝜏 ) 𝐺 (Ω )𝐺 (Ω ) (2.12)

where 𝐺 and 𝐺 are the complex antenna patterns of the Tx and the Rx.

| 18

CHAPTER 3 MIMO ANTENNA SYSTEMS FOR 4G MOBILE TERMINALS WITH CONSIDERATION OF USER BODY EFFECT | 19

Chapter 3 MIMO Antenna Systems for 4G Mobile Terminals with Consideration of User Body Effect The user body effect on a mobile antenna can be categorized into three aspects: the antenna imped-ance mismatching, the loss of radiated power, and the change of radiation pattern (see Fig. 3.1) [50]-[56]. The first two effects are dominant when users are in the reactive near field of the antenna, and they are also the primary concerns in mobile antenna designs in sub-6 GHz frequency bands. The corresponding degradation of antenna performance can be evaluated by the loss of the antenna’s total efficiency.

Figure 3.1 The user body effect on the impedance matching, the radiated power and the radiation pattern of a mobile antenna. The user body effect on a MIMO antenna system is more complicated than on a SISO antenna because it will not only lead to a power loss of each antenna but can also change the interactions among mul-tiple antenna elements [57]-[59]. Moreover, the performance of a MIMO system also depends on the channel characteristics, and thus the corresponding user body effect will also vary under different propagation scenarios as well.

3.1. User Body Effect on Mobile Antenna Systems Human tissue is composed of permittivity lossy materials ( 𝜖 = 𝜖 − 𝑗 𝜖 ′′ ), and it will cause imped-ance mismatching and absorb radiated power of mobile antennas. Antenna impedance mismatching will not only reduce the antenna efficiency but also lead to a degradation on the output of power am-plifier. The shift of the resonant frequency of the mobile antenna that caused by a user body can be approximated by the perturbation theory of a filled cavity [60], as a mobile antenna operates at reso-nant wave modes. The human body is a non-magnetic material, and an approximation for the reso-nant frequency shift can be obtained when the perturbation is small:

= − ∫ [( ) ∙∗

]

∫ ( ∙∗

) (3.1)

where 𝜔 , 𝜖 and 𝐸 stands for angular frequency, relative permittivity and E-field, respectively; the subscripts 1 and 2 refer to the free space case and the case with a user body, respectively. The human tissue is composed of high permittivity material, which is mainly due to bones. For example, the CTIA equivalent liquid of human head tissue has the real part of the relative permittivity equals to 41.5 at 900 MHz. It will therefore decrease the resonant frequency of a mobile antenna. This conclusion agrees well with general observations. However, as Eq. (3.1) is just an approximation through a closed homogenous cavity with a single wavemode, it cannot be applied to give an accurate prediction of the resonant frequency shift for a mobile antenna. Occasionally, the user body may also increase the res-onant frequency, which has been observed in [61]. It was explained by the existence of multiple wave-modes closely coupled around the resonant frequency.

CHAPTER 3 MIMO ANTENNA SYSTEMS FOR 4G MOBILE TERMINALS WITH CONSIDERATION OF USER BODY EFFECT | 20

The human tissues will also absorb the radiated power of mobile antennas, due to dielectric damping and conductivity loss [62]. The power loss in a dielectric material can be expressed:

𝑃 = ∫ |𝐸| 𝑑𝑉 (3.2)

Therefore, the power loss depends on the internal E-field in human tissue. Due to the loss, the Q value of the antenna is reduced when the antenna is in the vicinity of a user, which will broaden the imped-ance bandwidth of the antenna [63]. It was shown in [64] that if the antenna resonant mode does not change dramatically due to the presence of a user, the absorption loss can be estimated through the changing of the Q value.

3.2. User Body Effect on MIMO Antenna Systems in Mobile Terminals In a 3D isotropic propagation environment, the performance of a MIMO antenna system depends on the total efficiency of each antenna element and the correlation coefficients among them. The pres-ence of the user body will cause a mismatch loss and a radiation efficiency loss on each antenna ele-ment. Moreover, the efficiency imbalance between multiple antennas might be enlarged in this case, which will also lead to an additional loss of MUX. On the other hand, the mutual coupling between antenna elements can usually be reduced, which can mitigate the mismatching loss slightly (see Eq. (2.3)) [65]. The user body will also affect the amplitude and phase of the antenna radiation patterns of a MIMO array, which will typically lead to a lower ECC to counteract the drop in MUX due that to the antenna efficiency loss. Overall, the loss in MUX of a MIMO antenna mainly comes from the antenna total efficiency loss when the user is nearby. The performances of a two-element MIMO terminal antenna in free space and in the CTIA data mode are compared in Fig. 3.2, which can be observed that the value of MUX is highly related to the average total efficiency of the MIMO antenna.

Figure 3.2 Performance characteristics of a two-element MIMO terminal antenna in free space and in the CTIA data mode.

CHAPTER 3 MIMO ANTENNA SYSTEMS FOR 4G MOBILE TERMINALS WITH CONSIDERATION OF USER BODY EFFECT | 21

In a propagation environment that the angular spread of the incident field is narrow, it is known that correlation coefficients among MIMO antennas will be increased [38],[46]. In addition, a user body will also introduce a shadowing loss on the incident field, which will change the MEG of each antenna. An example of a 4×4 MIMO system in the CTIA data mode under a Gaussian distributed incident field is presented below to illustrate the user body effect on high order MIMO system with a narrow angular spread incident field:

(a)

(b)

Figure 3.3 (a) The Gaussian distributed incident field with incident angle ( 𝜃 , 𝜙 ), and (b) The MUX of the 4×4 MIMO antenna in free space and in the CTIA data mode with different incident angles.

Figure 3.4 CDF of MUX with the Gaussian distributed incident field.

CHAPTER 3 MIMO ANTENNA SYSTEMS FOR 4G MOBILE TERMINALS WITH CONSIDERATION OF USER BODY EFFECT | 22

The incident field is assumed to be modeled as Gaussian distribution in both elevation angle 𝜃 and azimuth angle 𝜙 as shown in Eq. (3.3) [66],

𝑃 (𝜃, 𝜙) = 𝑃 (𝜃, 𝜙) ∝ exp − ( ) + ( ) (3.3)

where ( 𝜃 , 𝜙 ) are the center incident angles; the 𝜎 and 𝜎 are set to be 30˚. The resultant incident field is illustrated in Fig. 3.3 (a). In Fig. 3.3 (b), the spatial distribution of MUX according to different incident angle (𝜃 , 𝜙 ) in free space and in the CTIA data mode are plotted. The MUX of the MIMO antenna fluctuates with different incident angles due to the changing of MEG and correlation coefficient. It can be found that the hand shows a blockage effect on the incident field: the MUX within the hemisphere where the hand phan-tom is located (0° < 𝜙 < 180°) is about 10 dB lower than in the free space case. The cumulative distribution function (CDF) of the MUX is plotted in Fig. 3.4. Comparing the MUX in the isotropic environment and in the Gaussian distributed incident field when CDF = 5% (a typical value used in communication systems), the narrow APS results in a 7 dB loss in MUX. Additionally, the hand phantom will cause another 5.5 dB loss in MUX at this CDF level, which is about 0.8 dB higher than in the isotropic environment. Therefore, the MUX of MIMO antenna will be more sensi-tive to the user body effect in a narrow angular spread propagation environment. Since many factors that contribute the MUX will change simultaneously in an arbitrary propagation environment, it becomes more difficult to predict the user body effects on the MIMO performance than in a 3D isotropic propagation environment. Therefore, it is important to include the user body in the evaluation of MIMO performance in mobile terminals.

3.3. Body Insensitive MIMO Antenna Designs The effect of user body on mobile antennas varies with antenna types and also with the way that users hold their phone. Hence, it is difficult to obtain a design that can be robust in all situations. The im-pedance tuning network is an attractive technology for mobile phone antennas, which can re-match the antenna impedance due to the proximity of a user body [67]-[70]. In commercial productions, open-loop tuning networks have been commonly used, where the impedance on the antenna port is switched between pre-defined parameter sets. On the other hand, the closed-loop tuning technology has also grown rapidly in recent years, and it adjusts the parameters of the tuning network adaptively to match the antenna in all cases. The closed-loop tuning technology has also been used to mitigate the user effect based on the MIMO throughput performance [71]-[73]. A method of introducing a switchable antenna shielding is proposed in [74] to avoid the interaction between a user’s head and mobile antennas. Consequently, the absorption loss from the user’s head is reduced. Antenna switching is another method that can mitigate the degradation of antenna per-formance when users are nearby. In [75], a dynamic selection between two antennas was used to re-duce the loss caused by the user’s index finger. It is worthy to mention that the receiving diversity technology has been popularly used in commercial phones, but only a few can support transmitting diversity due to the high SAR of the antenna on top of the phone. For example, in the iPhone 4s and the iPhone 5 [76], the switch diversity scheme is implemented for transmitting to overcome the body effect on the antennas.

3.3.1 Adaptive Quad-Element Antenna Array for User-Effective MIMO Mobile Terminals In order to simultaneously support both the antenna switch diversity for mitigating the user body effects and the spatial multiplexing for a high data rate communication, an adaptive quad-element antenna system is introduced in [77]. The antenna array is shown in Fig. 3.5 (a), and it is designed to support a 2×2 MIMO transmission with choosing the pair of antennas which has the best MUX in

CHAPTER 3 MIMO ANTENNA SYSTEMS FOR 4G MOBILE TERMINALS WITH CONSIDERATION OF USER BODY EFFECT | 23

different use cases. The MIMO antenna is designed to operate in cellular bands of 750 MHz to 960 MHz and 1700 MHz to 2700 MHz. The four elements are separately distributed on the phone chassis, to ensure that they will not be covered by the user’s body at once. An example of the proposed antenna operates in the CTIA talk mode is shown in Fig. 3.5(b). The MUX varies with different antenna combinations, and the difference between the best and the worst combinations can be more than 3 dB in the frequency band below 1GHz. Among all combinations, the MIMO array with antenna 3 and 4 shows the highest MUX. The average total efficiency and the total efficiency balance of antenna 3 and 4 are better than any other combinations, as they suffer fewer absorption losses from head phantom than antenna 1 and 2. The ECC between antenna 3 and 4 also shows lower value than other combinations in the cellular band below 1 GHz, which is due to the radiation patterns distortion caused by the hand phantom,

(a)

(b)

Figure 3.5 (a) The quad element MIMO antenna for mobile terminals. (b) The quad element MIMO antenna performance in the CTIA talk mode (adapted from [77]).

CHAPTER 3 MIMO ANTENNA SYSTEMS FOR 4G MOBILE TERMINALS WITH CONSIDERATION OF USER BODY EFFECT | 24

3.3.2. Impedance Mismatching Robust Metal Bezel Antenna Designs Antenna designs that use phones’ metal bezel have become a popular choice. However, it also presents new challenges for antenna design, especially regarding the user body effect since the antenna can be touched directly by a user’s hand or head. The metal bezel antenna in the iPhone 4 suffered a 6.7 dB mismatching loss at the 850 MHz band when the bottom open slit on the bezel was touched by a finger [30]. The reason for such a high mismatching loss is mainly because the intensity of E-field is strong within the slit, where the open end of the bezel antenna is capacitively coupled to the grounded por-tion of the bezel. A simplified simulation model is presented in Fig. 3.6 to illustrate this phenomenon: an IFA type antenna is designed on the metal bezel of a 6-inch phone, which has a single resonance at 900 MHz. The E-field at the resonant frequency concentrates around the open slit on the left. A 1 cm3 cubic block filled with CTIA hand tissue (𝜖 = 30.0-12.5j at 900MHz) is then placed beside the slit to approximate a fingertip, as shown in Fig. 3.6(c). Consequently, the resonant frequency of this antenna is shifted from 900MHz to 400 MHz, as shown Fig. 3.6 (d). In recent commercial phones, those slits are usually placed in the positions which are not to be touched frequently in real life.

(a) (b)

(c) (d) Figure 3.6. (a) The IFA type bezel antenna design and (b) its E-field distribution at 900 MHz. (c) The simulation mode with a tissue block and (d) the shift of resonant frequency.

Basic Mechanism and SISO Antenna Design Avoiding strong capacitive coupling between the antenna portion and grounded portion of the bezel can reduce the mismatching loss that induced by users. One possible solution is to use a loop/slot type antenna on the bezel. The two ends of the antenna are shorted, which does not require any open slit on the metal bezel. This configuration can avoid strong perturbation on the E-field caused by the user’s hand, as the strongest E-field is within the middle part of the aperture between the bezel and the phone chassis, and its intensity can also be controlled by the changing clearance of the antenna. A single resonant model of a loop antenna on the phone bezel is proposed in Fig. 3.7, which has the same dimension and feeding as the IFA type antenna in Fig. 3.6. The same tissue block is also added and moved along the bezel antenna in 10 positions to verify the robustness of this antenna to the user

CHAPTER 3 MIMO ANTENNA SYSTEMS FOR 4G MOBILE TERMINALS WITH CONSIDERATION OF USER BODY EFFECT | 25

effect on antenna impedance matching. The results in Fig. 3.7 (d) show that the resonant frequency remains very stable when the tissue block touches on different parts of the antenna; the shift of the center resonant frequency is less than 30 MHz.

(a) (b)

(c) (d) Figure 3.7. (a) The metal bezel antenna design in loop model, and (b) its current distribution at 900 MHz. (c) The simulation mode with a tissue block and (d) the shift of resonant frequency.

Figure 3.8. The multi-band metal bezel antenna design and its reflection coefficient.

CHAPTER 3 MIMO ANTENNA SYSTEMS FOR 4G MOBILE TERMINALS WITH CONSIDERATION OF USER BODY EFFECT | 26

By optimizing the positions of the grounding connections for the metal bezel, a multiband antenna is realized on the metal bezel based on the loop mode, as shown in Fig. 3.8. The proposed design can cover 850–940 MHz and 1750–2200 MHz. A similar design has also been reported in [78], which demonstrates the feasibility of the loop/slot mode on the bezel for the SISO mobile antenna design.

Double-Ring MIMO Antenna Design A 2×2 MIMO antenna on the metal bezel of the phone is then proposed to meet demands for the 4G communication. To ensure a sufficient bandwidth for both antennas, a double-ring structure is intro-duced to support multiple modes [79] (which will be referred to as “double-ring antenna” from here and after). In Fig. 3.9, the 2×2 MIMO antenna with a double-ring structure for the mobile terminal application is presented. The single bezel structure is split into two thinner bezel rings, and an extra loop mode can then be excited between them, which is also orthogonal to the loop mode between the bezel and the phone chassis. PC/ABS plastic is filled in between the two rings to support the structure. By exciting the two modes simultaneously, the bandwidth of the MIMO antenna in the cellular band below 1 GHz can be broadened.

(a)

(b)

Figure 3.9. (a) The double-ring MIMO antenna design on phone bezel, and (b) the current distribution of port 1 at 830 MHz and 890 MHz (adapted from [79]).

CHAPTER 3 MIMO ANTENNA SYSTEMS FOR 4G MOBILE TERMINALS WITH CONSIDERATION OF USER BODY EFFECT | 27

In this design, port 1 and port 2 are located at the top and the bottom of the ground plane, respectively. The impedance bandwidth (S11 and S22 <-6dB) of the proposed MIMO antenna can cover cellular bands from 800 MHz to 930 MHz and from 1700 MHz to 2200 MHz. A dual resonance can be ob-served between 800MHz to 930 MHz due to the double-ring structure. For MIMO antenna designs, the optimization of the ECC and the total efficiency of each port are key factors, as shown in Eq. (2.6). The rule of thumb for the design target is that the total efficiency of each antenna should be above -4 dB and the ECC should below 0.5. The proposed MIMO antenna can satisfy the requirement within the operation bands, which is shown in Fig. 3.10. In particular, the ECC is close to zero between 850 MHz to 890 MHz even though the mutual coupling (S21) is very high, which is attributed to the mu-tual scattering effect between the two antennas [13].

(a) (b)

Figure 3.10. (a) S-parameters of the double-ring dMIMO antenna. (b) ECC and total efficiencies of the double-ring MIMO antenna (adapted from [79]). The user body effect on the impedance matching of bottom port (port 1) is investigated in the CTIA data mode, which is shown in Fig. 3.11. It can be observed that the resonant frequency below 1 GHz is shifted down about only 80 MHz, but the impedance bandwidth is broader due to the lossy material of the hand phantom. Therefore, the mismatching loss caused by the user’s hand effect is less than 0.6 dB. The antenna on top (port 2) of the phone suffers much less efficiency loss in the CTIA data mode due to the large dimension of the MIMO antenna, and the results are omitted here.

Figure 3.11. Reflection coefficients of port 1 in free space, and in the CTIA data mode with left and right hand.

CHAPTER 3 MIMO ANTENNA SYSTEMS FOR 4G MOBILE TERMINALS WITH CONSIDERATION OF USER BODY EFFECT | 28

Double-Ring MIMO Antenna Design with Integration of Non-cellular Antennas In commercial phones, the metal bezel is usually not only occupied by cellular antennas but also need to integrate non-cellular antennas, e.g., Wi-Fi or Bluetooth antennas. With the double-ring structure, the integration of Wi-Fi antennas and cellular antennas simultaneously on the seamless bezel can also be realized. One example is presented in [paper I]: two Wi-Fi antennas are integrated on the slot between the two bezel rings for the MIMO Wi-Fi application (Fig. 3.12), which operate in loop/slot modes as well.

(a)

(b)

Figure. 3.12. (a) The simulation model and antenna mockup of the double-ring antenna with MIMO cellular and MIMO Wi-Fi. (b) S-parameters of the MIMO cellular antenna and the MIMO Wi-Fi antenna (adapted from paper I). Both Wi-Fi antennas can cover 2.4 GHz to 2.5 GHz and 5.2 GHz to 5.8 GHz, two Wi-Fi bands. The impedance bandwidth (S11 and S22 <-6 dB) of the MIMO cellular antennas can cover the cellular bands of 830 MHz to 940 MHz, 1700 MHz to 2100 MHZ, and 2500 MHz to 2700 MHz. The total efficiencies are above -4 dB and the ECCs are below 0.5 within the operation band for the MIMO cellular and the MIMO Wi-Fi antennas. The user body effect on the impedance matching of the cellular antennas has been analyzed above. Therefore, we only focus on the total efficiency loss due to the user body effect for this design. The total efficiency loss of the MIMO cellular antenna is evaluated through simulations and measure-ments in the CTIA data mode. Additional to a relatively lower mismatching loss, a loop/slot mode has a half wavelength electrical length at the resonant frequency, it is larger than a monopole or an IFA type antenna which resonates at a quarter wavelength. Therefore, the probability of the loop/slot mode antenna being fully covered by the user’s hand is lower than for monopole and IFA type an-tenna, which also helps to improve the radiation performance of the antenna.

CHAPTER 3 MIMO ANTENNA SYSTEMS FOR 4G MOBILE TERMINALS WITH CONSIDERATION OF USER BODY EFFECT | 29

As shown in Fig. 3.13 (a), the average total efficiency loss is 5.3 dB of the bottom antenna (port 1) in the CTIA data mode from 830 MHz to 940 MHz with both on the left and the right hand. Meanwhile, the total efficiency loss of four conventional MIMO antennas (CA) designs on the bottom of the phone are also presented. The dimensions of those four MIMO antennas are similar to the double-ring an-tenna. It can be observed that the total efficiency loss from conventional antennas is usually between 5 dB to 12 dB in the CTIA data mode (see 3.13 (b)). Therefore, it can be concluded that the double-ring antenna design based on the loop/slot mode shows a relatively good radiation performance with the user body effect.

(a)

(b)

Figure 3.13. (a) The simulated and measured total efficiency loss in the CTIA data mode of the port 1 (bottom cellular antenna) of the double-ring (DR) antenna. (b) The comparison of total efficiency losses in the CTIA data mode of the double-ring (DR) antenna with four conventional MIMO antenna design (adapted from paper I).

CHAPTER 3 MIMO ANTENNA SYSTEMS FOR 4G MOBILE TERMINALS WITH CONSIDERATION OF USER BODY EFFECT | 30

In conclusion, the loop/slot mode can be a good candidate for the metal bezel antenna design. The antenna design can avoid using the open slits on the bezel which are usually has a strong intensity of E-field. Therefore, the antenna can maintain a relatively stable impedance matching when the user's hand is touching it. Moreover, the antenna needs no open slit on the bezel structure, which can pro-vide a more elegant appearance than conventional bezel antenna design. Based on this method, de-signs of both SISO and MIMO mobile antenna have been presented. In addition, the double-ring shaped bezel antenna designs have also been introduced, to further increase the bandwidth and inte-grated more antenna elements on the bezel structure. The cons of the proposed bezel antennas are mainly the large physical dimension, which increases the difficulty of integrating them in a small terminal. In addition, the impedance mismatching robustness of the proposed antenna may decrease if the bezel is too thin or the antenna clearance is too small. The mutual coupling in the proposed MIMO antenna design is also relatively high which limits the antenna total efficiency in free space. However, it is possible to reduce the mutual coupling by us-ing the TCM and exciting orthogonal modes of the two ports [80].

CHAPTER 4 USER BODY EFFECT ON 5G MOBILE ANTENNA SYSTEMS AND CHANNEL CHARACTERISTICS | 31

Chapter 4 User Body Effect on 5G Mobile Antenna Systems and Channel Characteristics It is widely accepted that the user body in the sub-6 GHz band mainly affects the antenna impedance matching and the radiated power. However, for the future 5G communication operates above 6 GHz and up, into the mmWave frequency range, the user body is not likely to disturb the reactive near-field of the mobile antenna strongly. Hence, those two effects tend to be weaker. It has been shown in [81] that the impedance matching and efficiency of a patch antenna array were virtually unaffected by the presence of the body at 60 GHz. On the other hand, a more pronounced shadowing zone are expected from a human body due to the shorter wavelength. A human body can be simplified as perfectly electrically conducting (PEC) cylinder or plate in mmWave frequency range when evaluating the blockage effect. In [82], the excess loss through a PEC cylinder at 2.45 GHz, 5.7 GHz, and 62 GHz are assessed through the geometrical theory of diffraction (GTD), which is 12.3 dB, 16.0 dB and 26.5 dB, respectively. In [83], the human body is modeled as a PEC plate at 73 GHz, where the shadowing loss is calculated based on the double knife-edge diffraction (DKED) method. Both GTD and DKED methods are illustrated in Fig. 4.1, and they have shown good agreement with measurement results. A water-filled anthropomorphic phantom was used in [84], and the shadowing loss through this phantom is measured with a vector network analyzer. Studies have also been carried out to charac-terize the effect of human body blockage on mmWawe communication system [85]-[86], which shows that the human blockage has a unneglectable effect on the cellular network.

GTD DEKD Figure 4.1. Illustration of GTD method through a PEC cylinder and DEKD method through screen blocker. However, previous studies are mainly based on simplified models of a human body, and without con-sidering the radiation from mobile terminal antennas. In addition, the concerns in previous studies are mostly the effect of human blockage on the channel parameters but give few information on the aspect of mobile antenna designs, which is one of the targets in this thesis.

4.1. User Body Effect on Antenna Systems at 15 GHz and 28 GHz In order to investigate the user body effects on mobile antenna systems for the future 5G communi-cations in a more realistic environment, two groups of antenna far-field measurement are carried out: one is at 15 GHz in Micro Vision Group, Paris, France; the other one is at 28 GHz in Aalborg Univer-sity, Aalborg, Denmark. Three antenna arrays with different elements are designed and fabricated at those two frequencies. The array unit is integrated on the top of a mobile phone prototype as shown in Fig. 4.2(a). The element designs and the corresponding polarizations are presented in Fig. 4.2(b); they will here and after be referred to as “notch antenna,” “slot antenna” and “edge-patch antenna,” where the edge path antenna is composed of mesh grid metalized vias on the side of a phone chassis [87]. Due to the absence of a standard body phantom at those frequency bands, the user body effect is evaluated with real users. The embedded far-field radiation patterns and the total efficiency of the proposed antenna elements are measured with and without the presence of a user to obtain the user body effect. The user stays in two poses which are similar to the CTIA data mode and the CTIA talk mode, and are shown in Fig. 4.2(c). Hence, they will be referred to as “data mode” and “talk mode” in this chapter.

CHAPTER 4 USER BODY EFFECT ON 5G MOBILE ANTENNA SYSTEMS AND CHANNEL CHARACTERISTICS | 32

(a)

(b)

(c)

Figure 4.2. (a) Antenna array unit and its integration in a phone mockup at 15 GHz and 28 GHz. (b) The three element designs with their polarizations. (b) The measurement setup in free space and with a user.

4.1.1. User Body Effect on Embedded Elements

Total Efficiency Loss The measured total efficiencies of the three embedded elements are presented in Table 4.1. The total efficiency takes into account mismatch, dielectric, conductive and mutual coupling loss. The total ef-ficiency loss due to the user body effect of the notch antenna element is only about 0.3 dB in data mode and 1.5 dB in talk mode at 15 GHz. The slot antenna element shows slightly higher losses with the user, which is about 1.5 dB and 2.5 dB in data mode and talk mode at 15 GHz, respectively. Table 4.1. The total efficiency of embedded elements.

Notch Slot Edge Patch Frequency (GHz) (GHz)

15 28 15 28 15 28

Without User (dB) -0.2 -0.4 -1 -1.4 -0.5 -0.2

Data mode (dB) -0.5 -3.4 -2.5 -4.8 -2 -2.8

Talk mode (dB) -2 -3.5 -3.5 -6.9 -2.2 -5

CHAPTER 4 USER BODY EFFECT ON 5G MOBILE ANTENNA SYSTEMS AND CHANNEL CHARACTERISTICS | 33

The higher loss in the slot antenna can be explained by the simulation model in Fig. 4.3, the vertically polarized slot antenna excites a strong surface current on the phone chassis, which leads to a higher absorption by the hand.

Figure 4.3. The surface current of the notch element and slot element in data mode at 15 GHz. The values of the total efficiency loss at 28 GHz are greater than at 15 GHz. However, as the measure-ments are carried out in a different system and with different users, it is difficult to compare those values directly. Overall, the values in total efficiency loss at 15 GHz and 28 GHz due to the presence of user body are considerably lower than their counterpart values in the sub 6-GHz bands.

Shadowing Loss The measured embedded radiation patterns with and without the user at 15 GHz are presented in Fig. 4.4(a)-(c). The polarization at each solid angle is presented by an elliptic line (dashed pink line: left hand, solid black line: right hand), and the peak gain is about 5-6 dBi for a single element pattern. In data mode, a human body shaped shadowing region can be observed. On the horizontal plane (eleva-tion angle θ = 90˚), the deepest shadowing loss is about 25 dB and the shadowing region is within the region 230˚< 𝜙 <320˚. In talk mode, since the antenna is closer to the user body, the shadowing effect from the user body becomes even stronger than in data mode, and the maximum shadowing loss is 30 dB. The measurement results at 28 GHz show similar shadowing regions as at 15 GHz, but the shadowing loss observed is on average 5 dB stronger than at 15 GHz. In 3GPP channel model TR 38.901 [88], a stochastic self-blockage region is introduced for portrait mode (see Fig. 4.4 (e)), and the attenuation inside this region is 30 dB. The values of shadowing loss obtained from our measurements are comparable to the 3GPP model self-blockage model. However, the shadowing region from the measurements in Fig. 4.4(d) shows a different shape than the 3GPP self-blockage model:

1. In the 3GPP model, the shadowing region only exists within the solid angles (60˚ <θ<140˚, 200˚< 𝜙 <320 ˚). In the real user measurements, the shadowing region has a larger spread in the elevation angle θ, which is mainly caused by the sitting pose of the real user.

2. In the real user measurements, the loss increases gradually within the shadowing region.

However, the loss in the 3GPP model is defined as 30 dB, which is constant within the shad-owing region.

3. The gain in the non-shadowing region is enhanced about 1-2 dB due to the reflection of the

user body, which is not considered in the 3GPP model. From the aspect of mobile antenna system designs: even though the three proposed elements can radiate to different directions in free space, their radiation patterns become visually similar in data mode and talk mode. This phenomenon indicates that any spatial diversity antenna system may not operate efficiently due to the scattering and the blockage of a user body. However, the polarization of each element remains mostly unchanged even with user body interactions, which inspires that some polarization diversity systems could be still valid.

CHAPTER 4 USER BODY EFFECT ON 5G MOBILE ANTENNA SYSTEMS AND CHANNEL CHARACTERISTICS | 34

(a)

(b)

(c)

(d) (e)

Figure 4.4. The measured radiation pattern of embedded elements (a) without the user, (b) in data mode, and (c) in talk mode (adapted from paper II), the polarization at each solid angle is presented by an elliptic line (dashed pink line: left hand, solid black line: right hand). (d) Shadowing gain patterns based on measurement of the notch element in data mode at 15 GHz, and (e) the 3GPP self-blocking model in portrait mode. The user shadowing gain pattern is obtained by taking the difference be-tween free space measurement and in data mode measurement with a user (adapted from paper III).

4.1.2. User Body Effect on Coverage Efficiency and Total Scan Pattern The total scan pattern and coverage efficiency of each array with and without the user are also evalu-ated at 15 GHz. The array system is more sensitive to the blockage in the channel because they collect energy from a very narrow angular spread, which gives them a higher probability of blocking than an omnidirectional antenna which can receive the signal from all possible multi paths [89]. The array pattern used in this study is synthesized from the single embedded element pattern to avoid the phase errors introduced by multiple measurements with different antenna ports. The array factor (AF) is calculated as [90]:

AF = ∑ 𝑎 𝑒 ( )( ) (4.1) and

𝑐𝑜𝑠𝛾 = sin 𝜃 cos 𝜑 𝑑𝑥 + sin 𝜃 cos 𝜑 𝑑𝑦 + 𝑐𝑜𝑠𝜃 𝑑�̂� (4.2)

CHAPTER 4 USER BODY EFFECT ON 5G MOBILE ANTENNA SYSTEMS AND CHANNEL CHARACTERISTICS | 35

where 𝑎 and 𝛽 is the amplitude and phase of the excitation signal, respectively; 𝑑 is the distance be-tween elements and k is the free space wave number. Since it is a popular choice during 3GPP stand-ardization discussions at present [91]-[92], we assume that only four elements linearly placed in one array with half-wavelength separation distance (4×1 linear array), The notch and slot array is placed along the top short edge of the phone chassis, and the edge-patch array is positioned along the long-edge of the phone. The progressive phase shift scheme with equal amplitude is adopted (𝑎 = 1) in this study, and the phase excitation 𝛽 progressively changes from -180˚ to 180˚. The total scan pattern and coverage efficiency of the proposed arrays are presented in Fig. 4.5 and Fig. 4.6. It can be observed that when the threshold gain level is low, e.g., -10 dBi, the coverage effi-ciency is dramatically reduced about 15 % and 30 % due to the user body blockage in data mode and talk mode, respectively. With increased threshold gain levels (e.g., the UE is moving further to the BS), the coverage efficiencies in free space and with the user tend to merge. The peak gain value is slightly higher in data mode and talk mode compared to the free space case, due to the scattering effect from the user body.

(a)

(b)

(c) Figure 4.5. The total scanned pattern of a 4×1 linear array (a) without the user, (b) in data mode and (c) in talk mode (adapted from paper II).

Figure 4.6. The coverage efficiency of a 4×1 linear array with and without a user (adapted from paper II).

CHAPTER 4 USER BODY EFFECT ON 5G MOBILE ANTENNA SYSTEMS AND CHANNEL CHARACTERISTICS | 36

4.1.3. Statistic Investigation on User Body Effect In [93], a statistic study of the user effect on mobile antennas at 28 GHz is carried out. The far-field radiation pattern of the notch antenna is measured with 12 different users (Fig. 4.7 (a)), and the an-tenna is placed both on the top and the bottom of the phone. It can be observed in Fig. 4.7 (b) that the total efficiency loss between different users can be up to 3 dB when the antenna is placed on the top of the phone, even though the users are guided to hold the phone in the same position.

(a)

(b) (c) Figure 4.7. (a) The measurements setup with 12 different users in data and talk mode. The variation of total efficiency loss of the notch antenna on top of the phone in (b) data mode and (c) talk mode with 12 users (adapted from [93]). The case in which the notch antenna is placed on the bottom of the phone, and radiates towards the user body is also investigated in [93]. It can be seen in Fig. 4.8 that the user body will scatter the electromagnetic wave and the shadowing region remains in this case. Therefore, the antenna array in the future 5G mobile phone should always be able to steer the beam away from the user body. The coverage efficiency analysis is limited to a single element in this study, which is shown in Fig. 4. 9. Depending on the dimensions of a user body, 20% absolute variation in coverage efficiency has been observed. In real life, since both the way that users hold their phone and the shape of users can be very different, the spatial coverage of a mobile antenna array can change rapidly, which will be another challenge of extending the 5G cellular network coverage.

CHAPTER 4 USER BODY EFFECT ON 5G MOBILE ANTENNA SYSTEMS AND CHANNEL CHARACTERISTICS | 37

Figure 4.8. The radiation pattern of the notch antenna on top and bottom of the phone in data mode.

(a) (b)

Figure 4.9. The variation in coverage efficiency of a notch element on (a) top and (b) bottom of the phone in data mode with 12 users (adapted from [93]). In conclusion, it is verified by the measurements that user body effect on mobile antennas will mainly present as the shadowing loss at 15 GHz and 28 GHz, which will profoundly reduce the spatial cover-age of an array system in mobile terminals. The level of shadowing loss is between 25 dB to 35 dB at 15GHz and 28 GHz bands, which can cause about a 100˚ shadowing region on the horizontal plane. The shadowing loss and the obstruction region will change according to the user body dimension and the relative position of the user and antenna. Therefore, it is expected to change rapidly in real life and can not be ignored in future 5G system design and channel modeling. Radiation patterns of proposed elements become similar under the same use case, which will restrain the effect of using antenna pattern diversity technology in real life. In addition, it will also be difficult to increase the spatial coverage by implementing multiple sub-arrays due to the same reason. How-ever, it is still necessary to have multiple arrays distributed separately in the phone to meet the de-mand of different use cases (see Fig. 4.10). The polarization diversity can be employed since the ele-ment polarization remains almost unchanged with and without the user presence.

Figure 4.10. Multiple sub-arrays in a mobile terminal to meet the demand of different use case.

CHAPTER 4 USER BODY EFFECT ON 5G MOBILE ANTENNA SYSTEMS AND CHANNEL CHARACTERISTICS | 38

4.2. User Body Effect on Channel Characteristics at 15 GHz and 28 GHz The value of shadowing loss and coverage efficiency cannot be translated directly to the loss of the received signal strength in a cellular system, as it will also depend on the wireless channels. The chan-nel characteristics can be obtained either through measurements with a channel sounding system or simulations of propagation models. Ray-tracing has been popularly used to obtain a deterministic channel model through computer simulations; it is based on Maxwell's equation following geomet-rical optics (GO) high-frequency criteria [94]. Ray-tracing determines that rays travel from a Tx loca-tion to an Rx location, and interactions between the rays and environments are calculated with GO and the uniform theory of diffraction (UTD). In this study, the user body effect on channel character-izes is determined by ray-tracing simulations, a flowchart in Fig. 4.11 describes how the user body effect on the channel characteristics is obtained in our study with ray-tracing simulations.

(a)

(b)

Figure 4.11. (a) Summary of all the possible interaction mechanisms undergone by rays with 2D building blocks, and a 3D ray-tracing simulation example with Wireless Insite 2.8. (b) The flowchart of the ray-tracing simulation that includes the user body effect on channel characteristics. In [paper III], the user body effect on the channel characterizes investigated in an urban environment in Kista, Stockholm, Sweden. The total study area is 550 m× 550 m, and it includes several buildings between 15-20 m, which is shown in Fig. 4.12. The physical environment of the study area is modeled by a detailed 3D map that includes, among other things, buildings, windows, and parked cars. A base station (BS) is hanging on the external wall of the red brick building, 8 m above the ground. More details are added to this red brick building and also the twin buildings opposite the BS whose facades are composed of metallic glass. The rest of the buildings are modeled as concrete walls with 30 cm thickness, the ground is modeled as wet terrain, and the cars are simplified as PEC. All the permittiv-ities of the material are adopted from International Telecommunication Union (ITU) database [95]-[96]. The simulation model has been verified with a measurement carried out by Ericson with their testbed [97] at 15 GHz in [paper III]. The measured radiation patterns are included in the ray-tracing simulations to obtain the impact of the user body on the downlink channel characterization. The transmitted power is assumed to be 30dBm. Since the focus is on the impact of user body, the antenna of the BS is simplified as a single omnidirectional antenna with dual polarizations, to exclude any potential effects from BS antenna systems.

CHAPTER 4 USER BODY EFFECT ON 5G MOBILE ANTENNA SYSTEMS AND CHANNEL CHARACTERISTICS | 39

Figure 4.12. The 3D ray-tracing simulation model of the urban environment in Kista, Stockholm, Sweden (adapted from paper III).

4.2.1. The Received Signal Strength without User Obstruction The results that without the user’s presence are shown first. The received signal distributions of the notch antenna element in a mobile phone at 15 GHz and 28 GHz are shown in Fig. 4.13. Because of the high diffraction loss, there is a sharp difference in the received signal strength between the LOS regions and the NLOS regions. In general, the received signal strength is above -60 dBm in LOS re-gions but lower than -80 dBm in NLOS regions at 15 GHz. Moreover, the received signal strength at 28 GHz is about 5 dB lower than at 15 GHz due to a higher FSPL.

Figure 4.13. The received signal distribution within the study area of the notch antenna element in the mobile terminal at 15 GHz and 28 GHz (adapted from paper III).

CHAPTER 4 USER BODY EFFECT ON 5G MOBILE ANTENNA SYSTEMS AND CHANNEL CHARACTERISTICS | 40

4.2.2. The Received Signal Strength in Data Mode and Talk Mode The received signal strength in data mode and talk mode are obtained by including the measured radiation patterns of antenna elements with the user (see Fig. 4.4 (a)). The received signal strengths of three proposed antenna element designs are calculated in data mode and talk mode, with two ori-entations of each use case. The positions and the relative orientations of the user are illustrated in Fig. 4.14. The NLOS regions are excluded in this part to highlight the effect from the user’s body. The received signals’ strength versus distance at 15 GHz are plotted in Fig. 4.15. The path losses of the three proposed element designs are visually similar in the same use case when orientation is the same. This result agrees with the observation in Fig. 4.4 (a), where the effective radiation patterns of differ-ent antenna elements are also similar when they are in the same use case. Therefore, the path loss model of the channel will mainly be determined by the relative positions of the BS, the UE and the user, instead of antenna element designs in UE. The results at 28 GHz presented in [paper III] suggest the same conclusion.

Figure 4.14. The position and the relative orientations of the user in each use case (adapted from paper III).

CHAPTER 4 USER BODY EFFECT ON 5G MOBILE ANTENNA SYSTEMS AND CHANNEL CHARACTERISTICS | 41

(a)

(b)

(c)

(d)

(e)

Figure 4.15. The received signal strength at 15 GHz versus distance (a) without the user, (b) in data mode 1, (c) in data mode 2, (d) in talk mode 1, and (e) in talk mode 2.

CHAPTER 4 USER BODY EFFECT ON 5G MOBILE ANTENNA SYSTEMS AND CHANNEL CHARACTERISTICS | 42

4.2.3. The Received Signal Strength with Different User’s Orientations The user’s orientation is random in real life. This impact must be included in the study of channel characteristics since we have observed strong effects from the user’s orientation. The received signal strengths of the notch element are simulated with the user’s orientation is rotated from 0˚ to 350˚ in the horizontal plane at 15 GHz. The probability distribution functions (PDFs) of the received signal strength are calculated in three typically urban environments, which are the street canyon, the corner, and the parking lot. The three routes and the corresponding AoAs are shown in Fig. 4.16. The AoAs in the street canyon scenario and the corner scenario are relatively narrow due to the waveguide effect and the poor scattering environment. The corner scenario also includes an LOS to NLOS transition. On the other hand, the parking lot scenario is a rich scattering environment which has a more uniform distribution of AoA in the horizontal plane.

Figure 4.16. The Rx positions and AoA in the urban canyon, the corner, and the parking lot scenarios (adapted from paper III).

(a) (b)

(c)

Figure 4.17. The PDF of received signal strength in (a) the street canyon scenario, (b) the corner scenario, and (c) the park-ing lot scenario with different user orientation at 15 GHz (adapted from paper III).

CHAPTER 4 USER BODY EFFECT ON 5G MOBILE ANTENNA SYSTEMS AND CHANNEL CHARACTERISTICS | 43

The probability density function (PDF) of the received signal strengths at 15 GHz is shown in Fig. 4.17, and the mean and the standard deviation (SD) at 15 GHz and 28 GHz are listed in Table 4.2. In the urban canyon and the corner scenarios, the variation of the received signal is enlarged in data mode and in talk mode comparing to the case that without a user. Moreover, The PDF in talk mode also exhibits a more pronounced tail probability than in the data mode, as the UE is closer to the user body and the shadowing effect is increased. In contrast, the received signal strength is less dispersed in the parking lot scenario, which is attributed by the more uniform distribution of AoA. Table 4.2. The mean and standard deviation of the signal strength sampling distributions.

Scenario User case

15 GHz Mean [dBm]/SD [dB]

28 GHz Mean [dBm]/SD [dB]

Urban Canyon Corner

Without user

-52.2 /5.8 -62.5/5.5

Data mode

-52.7/7.3 -62.7/7.3

Talk mode

-55.3/10.2 -64.1/10.9

Corner Parking

Without user

-54.5/9.7 -65.4/9.9

Data mode

-55.2/10.7 -66.5/11.9

Talk mode

-58.2/13.3 -66.7/14.0

Parking

Without user

-48.4/3.3 -59.3/3.3

Data mode

-48.3/3.2 -59/5.0

Talk mode

-49.2/5.4 -56.5/5.5

In conclusion, channel models for the future 5G communication system will highly depend on the LOS/NLOS condition as well as the relative orientation of the UE antenna, the BS antenna, and the user. The user body will introduce a shadowing region in the effective radiation pattern of the UE antenna, and the effect on the UE received signal strength is expected to change rapidly in a poor scattering environment. Therefore, it is necessary to introduce a parameter in the channel model for the future 5G communication system to describe the self-blockage effect. This parameter should also be adjustable based on different propagation environments and different use cases. Antenna arrays in mobile terminals can be used to mitigate the user body effect. However, considering the limited space in a mobile terminal, the number of antenna elements will likely be restricted to 4-8 in the frequency range of 15 GHz to 40 GHz, and the corresponding array gain will be up to 6-9 dB. The effect of using an 8-element linear array on received signal strength of UE has been investigated in [Paper III], its effect on overcoming the shadowing loss induced by the user’s body is limited.

CHAPTER 4 USER BODY EFFECT ON 5G MOBILE ANTENNA SYSTEMS AND CHANNEL CHARACTERISTICS | 44

Figure 4.18. The illustration of a macro cell vs. a small cell concept. The optimization on the BS locations and the co-operated network should be considered as necessary solutions for overcoming the blockage effect from the user’s body. In order to ensure a stable commu-nication link, at least two BSs may be needed within the LOS region of the UE, which can illuminate the UE from different angles. Therefore, a much denser deployment of BS will be required to ensure a good outdoor coverage for the 5G mmWave communication. The small cell concept (Fig. 4.18) with miniature BSs which can be placed a couple of hundred meters apart in the city, will be a promising solution for the 5G network operating in the mmWave band to overcome high path losses and block-ages from objects in the channel [98].

CHAPTER 5 EMF EXPOSURE OF MOBILE TERMINALS TO USER BODY | 45

Chapter 5 EMF Exposure of Mobile Terminals to User Body When the user body is in the vicinity of a mobile phone, it will not only degrade the performance of the antennas but also be exposed to the radio frequency (RF) electromagnetic field (EMF). The expo-sure of the human body to EMF will lead to an increased thermal effect in human tissues due to the absorbed energy, and the level of the exposure has been strictly limited by governments and regula-tory agencies. The FCC Guide offers the following "bottom line" editorial: "ALL cell phones must meet the FCC’s RF exposure standard, which is set at a level well below that at which laboratory testing indicates, and medical and biological experts generally agree, adverse health effects could occur." [99]. The specific absorption rate (SAR) has been adopted as the controlling criterion for human exposure compliance tests of mobile phones by IEEE, ICNIRP, and FCC in the frequency range above 100 kHz but below 3 GHz, 6 GHz and 10 GHz [31]-[33], respectively. The unit of SAR is W/kg, and it indicates the rate of energy absorbed by user body per unit mass ( ( )). The SAR for a mobile phone can be calculated by the E-field within the human tissue:

SAR = (5.1)

where 𝜎 is the tissue’s conductity and 𝜌 is the tissue’s density; E is the root mean square (RMS) E-field magnitude in human tissues, which is primarily interesting because it directly relates to the value of SAR. The antenna is the primary radiation source in a mobile phone, and the human exposure level mainly depends on emissions from the antenna. As we have discussed, the user body will mostly appear in the reactive near field of the antenna in sub-6 GHz band, where the E-field and the H-field are decou-pled. However, the E-field inside human tissue is usually weakly related to the E-field distribution in antenna near field (see Fig. 5.1), as the high permittivity of human tissue will lead to an attenuation of the normal component of E-field [100]-[101]. Practically, it is always found that the SAR distribu-tion is highly correlated with the current distribution of the mobile antennas.

Figure 5.1. The E-field distribution of a dipole antenna in front of a flat phantom filled with body tissue-equivalent liquid at 750 MHz. The maximum E-field of the dipole antenna is located on the two ends of the antenna, but the maximum E-field in the tissue is located near the antenna center feeding point.

CHAPTER 5 EMF EXPOSURE OF MOBILE TERMINALS TO USER BODY | 46

The SAR averaged over the whole body is usually taken as the fundamental parameter to establish the restriction of the exposure, and the value is limited to 4 W/kg which is accepted worldwide. However, in the world of mobile phone, the local SAR when the power absorption takes place in a confined body region is more important, since it can be much higher than the whole-body averaged SAR. For exam-ple, a mobile phone transmits signal next to a user’s head. Limits on local SAR have been introduced by FCC and ICNIRP for mobile phones: the limitations are 1.6 W/kg averaged over 1 g tissue mass for the FCC and 2W/kg averaged over 10 g tissue mass for the ICNIRP. The limitation from the FCC is stricter than the ICNIRP limit, due to its lower limited value and the smaller average mass. The IC-NIRP limit is recognized by Europe Union, China, and many other countries. Meanwhile, the major markets that adopted the FCC regulation are mainly U.S and Australia. India switched from the EU limits to the US limits for mobile handsets in 2012.

5.1. SAR Assessment for Mobile Antennas The SAR assessment for compliance tests is a complicated procedure. An accurate SAR measurement requires a high-resolution scanning of the E-field inside the equivalent human liquid which is filled in a human-shaped mannequin. The DASYS system from SPEAG is one of the most commonly used SAR measurement setups, which is exhibited in Fig. 5.2 (a). Recent years, immediate SAR measure-ment systems with 2D grid sensors are also popularly used, e.g., SPEAG iSAR system. This kind of systems can perform a much faster measurement of SAR but with less accuracy. Three positions for the local SAR evaluations are usually required for mobile terminals [102]: the cheek touch position and the tilt position beside the head phantom for the talking scenario (see Fig. 5.2 (b)), and the position in front of a flat phantom for the body worn scenario. The specific anthro-pomorphic mannequin (SAM) head phantom filled with head tissue equivalent liquid and the flat phantom filled with body tissue equivalent liquid are used. The permittivity and conductivity of the equivalent liquid are frequency dependent, and to measure the SAR for a mobile phone through all the operational bands, various liquids are usually needed. For a complete compliance test of a multi-mode antenna in a mobile phone, the whole procedure can take up to weeks.

(a)

(b)

Figure 5.2. (a) The DASY SAR measurement system and the iSAR SAR measurement system from SPEAG. (b) The check touch and tilt positions on the left side of the head phantom (adapted from [102]).

CHAPTER 5 EMF EXPOSURE OF MOBILE TERMINALS TO USER BODY | 47

Studies regarding SAR of mobile antennas have been carried out for years, and several methods have been proposed to reduce the SAR values. One of the conventional methods is to place the antenna upon the ground plane, where the ground plane can serve as a shield and decouple the antenna near field from the human body [103]. Moreover, as only the peak SAR value matters for compliance test, increasing the physical dimension of the radiator and reducing the surface current density is another way to lower the SAR value. In practice, this idea can be realized by exciting the characteristic model of the phone chassis [104] or adding parasitic elements [105]-[106]. Some other methods such as using high permittivity material to control the near field distribution of the antenna, controlling the current distribution with the defected ground plane and using metamaterials or ferrite sheets to de-couple the antenna can also reduce the SAR value effectively. However, considering the practical lim-itations of a mobile phone, those methods are usually more difficult to be implemented in a commer-cial phone.

5.1.1. SAR Assessment for Multiple Antenna Systems When multiple antennas in a mobile terminal that are transmitting simultaneously, not only the stand-alone SAR of each antenna but also the simultaneous transmission SAR must be in compliance with standard regulations. In Fig. 5.3, the stand-alone SAR and simultaneous transmission SAR of a monopole type cellular antenna and a PIFA type Wi-Fi in a mobile terminal beside a head phantom are presented. The accepted power of cellular antennas and Wi-Fi antennas are set to be 23 dBm and 19 dBm, respectively, where the power loss due to the impedance mismatching and the mutual cou-pling are excluded. The stand-alone SAR is calculated when one port is transmitting, and the other ports are terminated with 50 ohm. It can be observed that the value of the simultaneous transmission SAR is higher than the stand-alone SAR values, and the distribution of the simultaneous transmission SAR is more spread over the surface of head phantom.

(a) (b) (c)

Figure 5.3. Stand-alone SAR of (a) the cellular antenna, and (b) the Wi-Fi antenna. (c) Simultaneous transmission SAR of the cellular antenna and the Wi-Fi antenna. The evaluation procedure of simultaneous transmission SAR is time-consuming; It requires large, high-resolution scans for each transmitter in order to overlay and combine the SAR distribution from multiple antennas. Research on the simultaneous transmission SAR has also been carried out in past years. In [107]–[109], the simultaneous transmission SAR values of various MIMO antenna topolo-gies are studied, which provides a benchmark for MIMO antenna designs in mobile phones. In [110]-[111], estimations on the simultaneous transmission SAR with scalar E-field probes are investigated, which are intended to reduce the measurement time for SAR compliance tests.

5.1.2. Evaluation of SPLSR for MIMO Antennas For simplifying the assessment procedure of SAR for a mobile terminal with multi-antenna systems, regulators proposed several conditions of the test exclusion for the simultaneous transmission SAR. The FCC adopts the following conditions for the exception of simultaneous transmission SAR testing

CHAPTER 5 EMF EXPOSURE OF MOBILE TERMINALS TO USER BODY | 48

[112]: 1. the sum of the stand-alone SAR of all simultaneous transmission antennas is below 1.6 W/kg; 2. if the SAR to peak location separation ratio (SPLSR) of simultaneous transmission antenna pair is below 0.3. The SPLSR defined as the ratio between the summation of two stand-alone SAR values and the distance between the SAR hot spot of each transmitter, which can be expressed as:

SPLSR = (SAR + SAR )/𝐷 (5.2) where SAR1 and SAR2 are the 1g stand-alone SAR values of transmitters 1 and 2, respectively. D is SAR peak-locations spacing in centimeters (cm). The SPLSR describes how closely those pair of two SAR values are related. The threshold level of 0.3 is rounded off from 0.32, which equals to the FCC SAR limitation 1.6 W/kg divided by 5 cm. The SPLSR based on the observation that a SAR hot spot is usually within a region with a radius of 2–3 cm, and the SAR regions of the two transmitters are assumed to overlap significantly if the separation distance is below 5 cm [113]. An example of SPLSR in a compliance test is presented in Fig. 5.4.

Figure 5.4. An illustration of SPLSR in a compliance measurement. The SPLSR cannot be translated directly to the value of simultaneous transmission SAR. However, it is critical for SAR compliance tests, and mobile phone vendors usually put great efforts on reducing this value. Therefore, it should be considered in priority to multiple antenna system designs with sim-ultaneous transmission functions, and a thorough understanding of SPLSR is necessary. In [paper IV], the SPLSRs of various cellular MIMO antenna designs in mobile terminals are studied, which provide a benchmark for the mobile MIMO antenna design. In the thesis, the SPLSR of three symmetrical MIMO antenna designs for the mobile terminal appli-cation are presented, which are shown in Fig. 5.5. The proposed models are so called: the semi on-ground antenna, the ground free co-located antenna, and the on-ground antenna. The effects of different frequency (0.75 GHz and 1.9 GHz) and different ground plane lengths (90 mm, 110 mm, 130 mm and 150 mm) on SPLSR of MIMO antennas are also shown. The accepted power level is set to be 23 dBm for each port, and the loss due to the impedance mismatching and the mutual coupling vari-ations are excluded. The SAR beside head phantom is more complicated than on the flat phantom, as it is affected both by the antenna near field distribution and the distance to the curved surface of head phantom. Therefore, the SAR and SPLSR of MIMO antenna beside the head phantom in the cheek touch position is focused in this thesis. Results for the flat phantom can be found in [paper IV].

CHAPTER 5 EMF EXPOSURE OF MOBILE TERMINALS TO USER BODY | 49

Figure 5.5. Diagrams of the proposed MIMO antennas, and the simulation models of SAR in cheek touch position on the right side of the head phantom (adapted from Paper IV). The 1g stand-alone SAR values of the port 1 of the three MIMO antennas are shown in Fig. 5.6. Based on the measurement setup of SAR in the cheek touch position, the antenna on the bottom of the phone will be moved further away to the cheek with an increased ground plane length.

1. At 1.9 GHz, the SAR value decreases rapidly with an increased ground plane length, as the antenna form is the major radiator at this frequency. The ground free antenna shows the highest SAR value, which is 3-4 times greater than the semi on-ground and the on-ground antenna.

2. At 0.75 GHz, the variation of SAR values with different ground plane length and different grounding states (on-ground, semi on-ground, ground free) are considerably small; the rea-son is that the ground plane itself is excited as a part of the radiator at frequency band below 1 GHz, and it is attached directly to the cheek of the head phantom. Therefore, the ground plane itself contributes mostly to the stand-alone SAR value at 0.75 GHz. A comparison of the current distribution at 0.75 GHz and 1.9 GHz of the on-ground PIFA antenna is shown in Fig. 5.7.

The 1g stand-alone SAR of the port 2 of proposed MIMO antennas are also presented in Fig. 5.5. In compliance tests, the acoustic output unit of the phone needs to be placed against the ear reference point. Therefore, the relative position of the antenna on the top of the phone and the head phantom is fixed. Consequently, the SAR values remain almost unchanged with an increased ground plane length for the semi on-ground antenna and the on-ground antenna. The shielding effect of the ground plane becomes the decisive factor of the SAR value, where the SAR of the semi on-ground antenna is almost double compared to the on-ground antenna.

CHAPTER 5 EMF EXPOSURE OF MOBILE TERMINALS TO USER BODY | 50

(a) (b)

Figure 5.6. 1g stand-alone SAR values on the right side of the head phantom of (a) the port 1 and (b) the port 2.

Figure 5.7. Current distribution of the on-ground antenna at 0.75 GHz and 1.9 GHz. The separation distance between SAR hot spots is also critical for SPLSR. Antenna engineers usually maximize the distance by placing two antennas to be as far as possible. However, due to the curved surface of the SAM head phantom and the near field distribution of the antenna, the distance between SAR hot spots usually does not equal the physical distance between the antenna elements. The SAR hot spot separation distance is investigated in four situations: the first two cases are when both antennas transmit at 0.75 GHz or at 1.9 GHz as a MIMO transmission. The other two cases are when one antenna transmits at 0.75 GHz and the other antenna transmits at 1.9 GHz. The latter two cases can be seen as one for voice communications and the other one for data communications. The SAR hot spot separation distance is plotted against the ground plane length in Fig. 5.8. With an increased ground plane length, the physical distance between the two antenna elements is enlarged for the semi on-ground MIMO antenna and the on-ground MIMO antenna. However, the distance between the SAR hot spots can be even smaller. For example, when the ground plane length of the on-ground MIMO antenna is increased from 90 mm to 130 mm, both hot spots of top and bottom antenna elements shift towards to the middle of the ground plane (see Fig. 5.9). Consequently, the distance between the SAR hot spots is reduced from 5.5 cm to 3.5 cm. For the co-located antenna, the physical distance between two antennas remains unchanged with an increased ground plane length. However, the hot spots of the SAR can also tend to be closer. In the case of MIMO transmission at 0.75 GHz, the separation distance between SAR hot spots is only 0.5 cm when the ground plane length equals to 150 mm.

CHAPTER 5 EMF EXPOSURE OF MOBILE TERMINALS TO USER BODY | 51

(a) (b) (c)

Figure 5.8. Distances between SAR hot spots of (a) the semi on-ground antenna, (b) the co-located ground free antenna and (c) the on-ground antenna.

(a)

(b)

Figure 5.9. SAR hot spots of the parallel placed on-ground MIMO antenna at 0.75 GHz with ground plane length of (a) 90mm and (b) 130 mm.

(a) (b) (c)

Figure 5.10. SPLSR of (a) the semi on-ground antenna, (b) the ground free co-located antenna and (c) the on-ground antenna.

CHAPTER 5 EMF EXPOSURE OF MOBILE TERMINALS TO USER BODY | 52

The corresponding values of SPLSR for the three cases are plotted in Fig. 5.10. The accepted power is reduced to 20 dBm for each port in the calculation of SPLSR to maintain the total accepted power to be 23 dBm. The results above and in [paper IV] can provide a benchmark for MIMO antenna designs in mobile phones, and it can be concluded that for a MIMO antenna system with low SPLSR:

1. The stand-alone SAR can be effectively reduced by the on-ground antenna structure, which is particularly important for a small terminal or the antenna on the top of the phone.

2. The stand-alone SAR of the bottom antenna in the phone can be reduced with a larger ground plane at frequency bands above 1 GHz. However, the stand-alone SAR values of the bottom antenna below 1 GHz are usually less sensitive to the ground plane dimension, as the ground plane will become a major role in radiation.

3. Increasing the physical distance between the antenna elements with a larger ground plane does

not necessarily lead to a smaller separation distance between SAR hot spots. The radiation from the ground plane will also tend to be enhanced, especially at frequency bands below 1 GHz.

4. Though the co-located MIMO antenna design provides an attractive choice for compact MIMO array designs, the distance between their SAR hot spots can be very close when the ground plane is strongly excited as a radiator. Therefore, the stand-alone SAR must be very well con-trolled for such kind of design.

It is worth mentioning that a new formula for calculating the SPLSR has been introduced by FCC after the study had been done:

SPLSR = ( ) .< 0.04 (5.3)

where 𝐷 is the peak-locations spacing in millimeters (mm). The SAR values were given a higher weight in the new equations, which suggests that more attention should be paid to minimizing the stand-alone SAR.

5.2. Power Density Assessment for 5G Mobile Antennas For the 5G communication operating above 6 GHz and up into mmWave frequency range, the power deposition becomes more superficial, and SAR inside the user’s body becomes less meaningful for the exposure restrictions. Hence, the external (free space) power density (PD) has been taken as the basic figure of merit by standardization commissions and regulators. The incident PD (S) is calculated as:

𝑆 = 𝐸 × 𝐻∗ (5.4) The PD limitation for the general public is set to be 10 W/m2 by the FCC, which is currently to be considered as a spatial peak value. However, as the spatial peak is not a well-defined quantity, where the results will depend on the exposure assessment method, e.g., the probe dimensions, the FCC also proposed the PD limits for an averaging area of 1 cm2 recently (though it has not been adopted into FCC regulations) [114]. The ICNIRP exposure guidelines specified a maximum power density of 10W/m2 for the general public but taken as a spatial average over a 20 cm2 of area. In the frequency range between 3 GHz to 30 GHz, the IEEE standard averages the power density over any contiguous area corresponding to 100λ 2 where λ is the free space wavelength. Above 30 GHz, the averaging is to be conducted over any contiguous area of 100 cm2. The study on the power density of mobile antenna system will be particularly important for the 5G communication as it will directly impact on the maximum permissible transmitted power (MPTP) of mobile terminals, which will further restrict the coverage of the cellular communication system. In [115], the MPTP of a half-wave dipole antenna to be compliant with the PD limits in Table 5.1 has been calculated in a frequency range from 1 GHz to 70 GHz at a distance of 2 cm in front of the antenna. Due to the change in the basic restriction quantity, there is a 5.5 dB drop and a 6.5 dB drop of the

CHAPTER 5 EMF EXPOSURE OF MOBILE TERMINALS TO USER BODY | 53

MPTP over the transition frequency in the FCC and ICNIRP regulations, respectively. For the IEEE regulation, the discontinuity of MPTP is a considerably smaller, but it has not been adopted by any authority. Another investigation in [116] indicates that the maximum increased temperature at the transition from SAR to PD is reduced about three to four times for the FCC and the ICNIRP. In order to provide an insight into the future 5G communication system, a systematic study of the EMF expo-sure for phased arrays transmitting above 10 GHz is carried out in [117]. In addition to the effect of frequency, the impacts from the numbers of array elements, the evaluation distance to the array, the beam steering range, and the array topology have been investigated.

Table 5.1. The basic restrictions on power density for general public. FCC ICNIRP IEEE

Frequency (GHz) 6-100 10-300 3-30 30-100

PD limitation (W/m2) 10 (spatial peak / 1 cm2)

10 (20 cm2)

10 (100λ2)

10 (100 cm2)

5.2.1. Power Density of Linear Arrays In [paper V], the power density properties of an 8×1 linear patch array is investigated at 15 GHz and 28 GHz (Fig. 5.11.). The spatial peak value of the maximum instantaneous PD is calculated at different distances above the patch array, to give an estimation of the highest exposure level from a mobile terminal. The results with 23 dBm accepted power are plotted in Fig. 5.12. The evaluation distance of the PD compliance test has not been defined yet, but since the mobile terminal will always operate close to a user body, it is more likely to evaluate the PD in the near field region of the antenna array. In the near field region of the antenna, the spatial peak PD is smaller with a larger antenna aperture. Therefore, from the array design point of view, an array with a larger aperture will be beneficial for getting a lower PD. In addition, a larger antenna array will also lead to a higher array gain in the far field, which can lower output power level from the UE and further reduce the PD in near field region.

Figure 5.11. Simulation models of patch arrays and the maximum instantaneous power density distribution on the surface of the array at 15 GHz and 28 GHz (adapted from paper V).

Figure 5.12. Spatial peak value of the maximum instantaneous PD of the linear patch array in the mobile terminal at 15 GHz and 28 GHz (adapted from paper V).

CHAPTER 5 EMF EXPOSURE OF MOBILE TERMINALS TO USER BODY | 54

Figure 5.13. MPTP to comply with the FCC exposure metrics of the linear patch array in a mobile terminal at 15 GHz (adapted from paper V). However, the level of maximum PD in the near field region is not comparable with the current limi-tation 10 W/m2. In Fig. 5.13, the MPTP to comply with the 10W/m2 is calculated as in Eq. (5.5).

MPTP(𝑑) = ( )

(5.5)

where 𝑃 is the total radiated power, 𝑆 is the PD limit, 𝑆 is the spatial peak PD of the proposed array, and d is the evaluation distance. The MPTP of the patch array is only about 7.5 dBm at 15 GHz, when the evaluation distance is 5 mm above the antenna array. This value is significantly lower than the maximum output power of mobile phones in cellular systems today (20-30 dBm).

5.2.2. Power Density Assessment and Compliance Test Methods Additional to the sharp discontinuity in MPTP, the assessment of PD in the near field region of the antenna array is also challenging. The E-field and the H-field are decoupled in the reactive near field, which requires measurements on both E-field and H-field with the phase information to determine the “true” power density. Moreover, the reactive near field can change rapidly with a small position shift, which also requires a high-resolution scanning. In [118], the PD of the patch array at 15 GHz is measured with a scalar E-field probe SPEAG EF3DV3, where the measurement setup is shown in Fig. 5.14. It indicates that the spatial peak value of the time-averaged power density 𝑆 = Re(𝐸 × 𝐻∗) can be estimated through a plane-wave equivalent value 𝑆 =

|𝐸| even in the near field of the antenna (see Fig. 5.15), though their position may not overlap. However, with an increased average area (e.g. 1 cm2 in the proposed FCC regulation and 20 cm2 in the ICNIRP regulation), the discrepancy between those two values are enlarged due to the estimation errors that are accumulated.

Figure 5.14. Near-field measurement system with E-field probe SPEAG EF3DV3 (adapted from [118]).

CHAPTER 5 EMF EXPOSURE OF MOBILE TERMINALS TO USER BODY | 55

Figure 5.15. True PD compared with plane wave equivalent PD (adapted from [118]). The methodology of compliance test with a scalar E-field probe is also investigated in [118], in which the PD of the patch array is evaluated with two plane-wave approximation methods. The proposed two approaches are called: components field combining method (CFCM) and magnitude field com-bining method (MFCM):

SPW,CFCMav (z=d)=max

x,y

∬ ∑ ∑ Eτi(x,y,d)N

i=12

τ=x,y,zAavdxdy

2η0Aav (5.6)

SPW,MFCMav (z=d)=max

x,y

∬ ∑ Ei(x,y,d)Ni=1

2Aav

dxdy

2η0Aav (5.7)

where Eτ

i is the x, y, or z component of the electric field produced by the ith antenna element. The CFCM assumes that the fields are temporally “in phase,” and MFCM assumes that the fields are spa-tially and temporally “in phase.” Both methods provide a conservative estimation of the PD of an antenna array. However, the results of the compliance assessment need not be the “true PD”: they should just be a reference value for manufacturers, operators, and regulators to determine whether the limits are exceeded. Furthermore, the assessment method needs to fit for the antenna systems in the 5G mobile terminals and the mainstream test equipment. The PD of a phased array system will vary with different input phase combinations. Therefore, it may require a significant amount of measurement to determine the worst-case scenario. By taking the conservative estimations in (5.6) and (5.7), this procedure can be simplified. The MPTP obtained with “true” PD (time-average Poynting vector), and the conservative estimation in (5.6) and (5.7) are com-pared in Fig. 5.16 for the 8×1 linear patch array at 15 GHz with a progressive phase shifting scheme. The conservativeness of the two methods is in the range of 5-10 dB, where the largest deviation is found for the ICNIRP exposure metric due to its larger averaging area.

Figure 5.16. MPTP to comply with the FCC and ICNIRP exposure metrics at a distance of 5 mm from UE as function of progressive phase shift φi (adapted from [118]).

CHAPTER 5 EMF EXPOSURE OF MOBILE TERMINALS TO USER BODY | 56

In addition, a theoretical boundary of PD based on convex optimization [119] can be derived. This method is particularly attractive to a full digital controlled array for which both amplitude and phase of the input signal can be arbitrary values. In conclusion, less attention has been paid to EMF exposure at frequencies above 6 GHz. However, it has started to change because there is a clear demand for RF EMF exposure assessment methods for the upcoming 5G communication system. Currently, the major exposure limits will result in a transmission power of the mobile device significantly below the value of existing cellular communi-cation systems. It may have a large negative impact on the performance of the 5G mobile communi-cation system if this issue cannot be properly resolved. In addition, compliance assessment methods are also needed. The compliance assessment does not require a true value of EMF exposure, but it should be able to indicate a reasonable reference level to determine whether the limits have been exceeded or not. The corresponding assessment methods may also need to handle different use cases to allow a higher output power in certain scenarios. For example, the mobile terminal may direct the transmitted energy away from the user body, which should be allowed to transmit with higher output power.

CONCLUSION AND FUTURE WORK | 57

Conclusion and Future Work Mobile antenna systems for 4G and the 5G applications with considering the user body interactions are discussed in this thesis. The evolution of communication networks requires an advanced antenna system in mobile terminals. The MIMO antenna system is a key component for 4G mobile terminals. Comparing to the SISO antenna system, the user body effect on a MIMO antenna system is more complicated: it will not only degrade the single antenna performance but also change the interactions among multiple antennas. Moreover, the MIMO performance is also limited by the propagation en-vironment, and it will become more sensitive to the user body effect with a narrow angular spread incident field. A quad-elements MIMO antenna array is presented in this thesis, which can operate adaptively in a 2×2 MIMO mode to minimize the user body effect on the MUX of the MIMO antenna. The MIMO antenna designs in mobile terminals also need to handle the requirements of modern industrial designs. Two metal bezel MIMO antennas based on loop/slot modes have also been pre-sented in the thesis. The proposed MIMO antennas can be realized in a seamless bezel structure and are robust to the mismatching loss caused by the vicinity of a user. By splitting a single bezel into two parallel ones, a double-ring metal bezel structure is obtained, which can enhance the bandwidth of the proposed MIMO antennas. In addition, it has also been demonstrated that a 2×2 MIMO cellular antenna and a 2×2 MIMO Wi-Fi antenna can be integrated simultaneously on the double-ring metal bezel structure, which proves the practicality of this design. The future mobile MIMO antenna designs at sub-6 GHz band need to be focused on high order MIMO, e.g., 4×4 or 8×8 MIMO antenna systems. In addition, the TV white spaces (600 MHz) are looking forward to be opened for cellular communications, which will be challenging to be covered by compact mobile terminal antennas. Moving to 5G communications which aim to use the spectrum above 6 GHz as carrier frequencies, it brings new challenges for mobile antenna system designs. The user body shows a more pronounced shadowing effect, which can significantly decrease the spatial coverage of antenna arrays in a mobile terminal. Consequently, the path loss model of the wireless channel will be highly depended on relative orientations of users, especially in a poor scattering propagation environment. Hence, the received signal strength of future 5G UEs would be intermittent if this issue cannot be addressed properly. Due to the limited space inside a mobile terminal, the shadowing loss caused by the user cannot be solved solely by utilizing an array system in mobile terminals; an optimization on cellular network planning is also necessary. The array elements in future 5G terminals need to cover a broad bandwidth with dual polarization, and should also be able to remain a stable radiation pattern cross the operation bands. In addition, they need to co-exist with cellular antennas which operate below 6 GHz. Antenna packaging related issues need to be taken care as well, since the active components may need to be integrated together with antenna elements to lower the loss in RF chain. Furthermore, multiple arrays will be required in a mobile terminal to realize an omnidirectional coverage, and to fit for different use cases. Analysis on channel characteristics should also be extended to a more realistic situation. For example, a down-link channel model with multiple users, which can include the effect of interference and scattering from nearby users. Advanced antenna systems in 4G and 5G mobile terminals do not only suffer from a more complicated user effect but also cause a more intricate EMF exposure assessment. The SAR evaluation for multiple antenna systems is a complex and time-consuming procedure. However, a good understanding of the test exclusion conditions can be beneficial for simplifying the assessment. The SPLSR of various MIMO antenna designs has been discussed this thesis. The separation distance between SAR hot spots is critical for SPLSR: it is not related to the physical distance between antenna elements but will be minimal when the antenna ground plane is strongly excited as the radiator. This information should be considered in priority to the MIMO antenna design in a mobile phone. The EMF exposure assessment for mobile terminals that operate over 6 GHz has also begun to get attention due to the demand for 5G mobile communication systems. The EMF properties of linear patch arrays have been shown in this thesis: despite the discontinuity in MPTP due to the change of

CONCLUSION AND FUTURE WORK | 58

basic metric from SAR to PD, a larger antenna aperture can be a benefit for obtaining a lower PD in the antenna near field. The PD assessment with a scalar E-field probe has also been experimented in antenna near-field region, which shows that it is promising to use a scalar probe to estimate the value of PD for the compliance test. Future work will be focused on providing improved exposure limits and practical assessment methods for 5G mobile terminals.

SUMMARY OF APPENDED PAPERS | 59

Summary of Appended Papers Paper I: In this paper, a novel multimode MIMO antenna system is proposed on the metal bezel of a mobile phone. The proposed metal bezel antenna system is composed of a double-ring structure, and two cellular antennas and two Wi-Fi antennas are integrated into this metal bezel of the phone. With the multi-mode excitation, the MIMO cellular antenna can operate at 830–900 MHz, 1700–2200 MHz, and 2400–2700 MHz, for 2G, 3G, and LTE bands, respectively. The MIMO Wi-Fi antenna can cover two Wi-Fi bands from 2.4 to 2.5 GHz and from 5.2 to 5.8 GHz. The proposed MIMO cellular antenna operates mainly in loop modes, which does not require any open slit on the bezel structure. The proposed MIMO antenna is also robust to the mismatching loss due to the user body effect. It has been shown in this paper that the proposed structure has a relatively lower efficiency loss than con-ventional mobile antennas in both the CTIA talk and data modes. Measurements are also carried out to verify the accuracy of the simulation results Paper II: The user body effect on the antenna and phased array system in a mobile terminal for the 5G communication system operating at 15 GHz is investigated in this paper. The user body effects are determined through 3D measurement of antenna far-field radiation patterns with three antenna element designs and a real user. It has been found that at 15 GHz, the user body effects are mainly presented in the way as shadowing loss, which is about 25 dB to 30 dB. The shadowing effect from the user body will generate a degradation of the coverage efficiency of the phased array system in the mobile terminal and can cause an increase in the outage probability of the system if the randomness of the mobile terminal orientation is considered. Paper III: The user body effects on channel characteristics an urban scenario at 15 GHz and 28 GHz are investigated in this paper. The propagation channel is obtained with a ray-tracing sim-ulator, and the accuracy is verified with measurement results at 15 GHz. The user body effects are brought in by including the measured 3D radiation pattern of antenna elements in a mobile terminal with a user. The results show that the user’s body will cause a strong shadowing loss and can determine the path loss model at 15 GHz and 28 GHz. Moreover, it also tends to generate a significant fluctuation on the received signal strength of the UE when the orientation of the user is changing, which is crucial to channel modeling studies.

Paper IV: This paper focuses on the SAR and SPLSR of the dual-element LTE MIMO antenna in mobile phones. Four designs of dual-element MIMO antenna are studied under four typical LTE frequency points (0.75 GHz, 0.85 GHz, 1.9 GHz, and 2.1/2.6 GHz). In addition, the influence of the ground plane length (90 mm to 150 mm) is also included since it will impact on the separa-tion distance of SAR hot spots. Both SAR and SPLSR are studied beside head phantom and above a flat phantom, and the simulation results are verified with measurement results from iSAR and DASY 4 systems. The results reveal that the chassis model shows a huge impact on the SAR and SPLSR values in the frequency bands below 1 GHz. The antenna positions on the ground plane (on-ground, semi on-ground and ground free) need to a tradeoff between antenna dimension, required bandwidth, and the SAR values. The MIMO antenna designs on the same end of the phone need more careful treatment to lower their SAR and SPLSR values. Paper V: The PD property of linear phased arrays in mobile terminals at 15 and 28 GHz are investigated for the future 5G applications. Uniform linear patch arrays are used, and different array configurations and phase of the input signal are compared. The spatial peak values of max-imum instant PD are calculated for each case to give a preserved estimation on the EMF expo-sure level of the 5G mobile antennas. Suggestions for the PD evaluation are also discussed.

SUMMARY OF APPENDED PAPERS | 60

BIBLIOGRAPHY | 61

Bibliography [1] Z. Ying, “Antennas in cellular phones for mobile communications,” Proc. IEEE, vol. 100, no. 7, pp. 2286–2296, July 2012.

[2] Spectrummonitoring.com - Frequencies ITU Region 1. spectrummonitoring.com. 2014-11-06. Retrieved 2014-12-29.

[3] K. Fujimoto and J. R. James, Mobile Antenna Systems Handbook, 3rd ed. Reading, MA: Artech House, 2008, ch. 5.

[4] P. Vainikainen, J. Ollikainen, O. Kivekas and K. Kelander, “Resonator-based analysis of the combination of mobile handset antenna and chassis,” IEEE Trans. Antennas Propag., vol. 50, no. 10, pp. 1433-1444, Oct 2002.

[5] C. R. Rowell, “Capacitive loaded PIFA for compact mobile telephone handsets,” IEEE Trans. Antennas Propag., vol. 45, no. 5, pp. 837–842, May 1997

[6] 3GPP TS 21.101 “Technical Specifications and Technical Reports for a UTRAN-based 3GPP system”. Mar. 2013.

[7] A. J. Paulraj, D. Gesbert, and C. Papadias, “Encyclopedia for Electrical Engineering,” New York, Wiley, 2000, ch. Antenna arrays for wireless communications, pp. 531–563.

[8] H. Li, B. K. Lau, Z. Ying, and S. He, “Decoupling of multiple antennas in terminals with chassis excitation using polarization diversity, angle diversity and current control,” IEEE Trans. Antennas Propag., Nov. 2011.

[9] B. K. Lau and J. B. Andersen, “Simple andefficient decoupling of compact arrays with parasitic scatterers,” IEEE Trans. Antennas Propag., vol. 60, no. 2, pt. 1, pp. 464–472, Feb. 2012.

[10] S. Zhang, B. K. Lau, Y. Tan, Z. Ying and S. He, “Mutual Coupling Reduction of Two PIFAs With a T-Shape Slot Impedance Transformer for MIMO Mobile Terminals,” IEEE Trans. Antennas Propag., vol. 60, no. 3, pp. 1521-1531, March 2012.

[11] A. Diallo, C. Luxey, P. L. Thuc, R. Staraj, and G. Kossiavas, “Study and reduction of the mutual coupling between two mobile phone PIFAs operating in the DCS1800 and UMTS bands,” IEEE Trans. Antennas Propag., vol. 54, no. 11, pt. 1, pp. 3063–3074, Nov. 2006.

[12] K. Zhao, S. Zhang, C. Y. Chiu, Z. Ying and S. He, “Impact of size and decoupling element on some fundamental compact MIMO antennas,” 2014 International Workshop on Antenna Technology: Small Antennas, Novel EM Structures and Ma-terials, and Applications (iWAT), Sydney, NSW, 2014, pp. 136-139.

[13] S. Zhang, A. A. Glazunov, Z. Ying and S. He, “Reduction of the Envelope Correlation Coefficient with Improved Total Effi-ciency for Mobile LTE MIMO Antenna Arrays: Mutual Scattering Mode,” IEEE Trans. Antennas Propag., vol. 61, no. 6, pp. 3280-3291, June 2013.

[14] S. Zhang, K. Zhao, Z. Ying and S. He, “Investigation of Diagonal Antenna-Chassis Mode in Mobile Terminal LTE MIMO Antennas for Bandwidth Enhancement,” IEEE Antennas Propag. Mag., vol. 57, no. 2, pp. 217-228, April 2015.

[15] H. Li, Y. Tan, B. K. Lau, Z. Ying and S. He, “Characteristic Mode Based Tradeoff Analysis of Antenna-Chassis Interactions for Multiple Antenna Terminals,” IEEE Trans. Antennas Propag., vol. 60, no. 2, pp. 490-502, Feb. 2012.

[16] H. Li, Z. T. Miers and B. K. Lau, “Design of Orthogonal MIMO Handset Antennas Based on Characteristic Mode Manipu-lation at Frequency Bands Below 1 GHz,” IEEE Trans. Antennas Propag., vol. 62, no. 5, pp. 2756-2766, May 2014.

[17] “Samsung Galaxy S7, the first 4×4 MIMO smartphone”. Oct, 2016. [Online]. Available: http://cellularinsights.com/sam-sung-galaxy-s7-the-first-4x4-mimo-smartphone/,

[18] “Qualcomm Announces Showcase of Connectivity Leadership with Debut of Several Live Demonstrations”. Feb 2016. [Online]. Available:https://www.qualcomm.com/news/releases/2016/02/17/qualcomm-announces-showcase-connectiv-ity-leadership-debut-several-live,

[19] Z. Pi and F. Khan, “An introduction to millimeter-wave mobile broadband systems,” IEEE Commun. Mag., vol. 49, no. 6, pp. 101–107, Jun. 2011.

[20] T. S. Rappaport et al., “Millimeter wave mobile communications for 5G cellular: It will work!” IEEE Access, vol. 1, pp. 335–349, May 2013.

[21] T. S. Rappaport et al., “Millimeter Wave Wireless Communications” New Jersey, Prentice Hall, Sep, 2014

[22] I. Rodriguez et al., “Analysis of 38 GHz mmWave Propagation Characteristics of Urban Scenarios,” 21th European Wireless Conference, Budapest, Hungary, 2015, pp. 1-8.

[23] 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.

[24] W. Hong, K. h. Baek and S. Ko, “Millimeter-wave 5G Antennas for Smartphones: Overview and Experimental Demonstra-tion,” accepted in IEEE Trans. Antennas Propag.

[25] W. Hong et al., “Multi-Beam Antenna Technologies for 5G Wireless Communications,” accepted in IEEE Trans. Antennas Propag.

BIBLIOGRAPHY | 62

[26] X. Ding, B. Z. Wang and G. Q. He, “Research on a Millimeter-Wave Phased Array With Wide-Angle Scanning Performance,” IEEE Trans. Antennas Propag., vol. 61, no. 10, pp. 5319-5324, Oct. 2013.

[27] D. Sanchez-Escuderos, M. Ferrando-Bataller, M. Baquero-Escudero and J. I. Herranz, “Reconfigurable Slot-Array Antenna With RF-MEMS,” IEEE Antennas and Wireless Propag. Lett., vol. 10, pp. 721-725, 2011.

[28] Y. Yang, Y. Cai, K. Y. Chan, R. Ramer and Y. J. Guo, “MEMS-loaded millimeter wave frequency reconfigurable quasi-Yagi dipole antenna,” Asia-Pacific Microwave Conference 2011, Melbourne, VIC, 2011, pp. 1318-1321.

[29] CTIA Certification Department Program, Test Plan for Wireless Device Over the Air Performance Method of Measurement for Radiated RF Power and Receiver Performance. [Online]. Available: www.ctia.org.

[30] W. N. Allen and D. Peroulis, “Bandwidth-optimal single-tunable-element matching network for antenna tuning in mobile handsets,” in IEEE Microwave Symposium Digest, (MTT 2011), Baltimore, USA, 5–10 June 2011.

[31] IEEE Standard for Safety Levels with Respect to Human Exposure to Radio Frequency Electromagnetic Fields, 3 kHz to 300 GHz, IEEE Standard C95.1-2005, 2005.

[32] ICNIRP, “Guidelines for limiting exposure to time-varying electric, magnetic, and electromagnetic fields (up to 300 GHz),” Health Phys., vol. 74, no. 4, pp. 494_522, Apr. 1998.

[33] R. F. Cleveland, Jr., D. M. Sylvar, and J. L. Ulcek, “Evaluating compliance with FCC guidelines for human exposure to radiofrequency electromagnetic fields,” Washington, DC, USA: FCC, Aug. 1997.

[34] 3GPP TR 25.914, “Technical Specification Group Radio Access Network: Measurements of radio performances for UMTS terminals in speech mode,” V7.0.0, Tech. Rep., Jun. 2006.

[35] Test Plan for 2x2 Downlink MIMO and Transmit Diversity Over-the-Air Performance, CTIA, Apr. 2017.

[36] R. Tian, B. K. Lau, and Z. Ying, "Multiplexing efficiency of MIMO antennas," IEEE Antennas and Wireless Propag. Lett., vol. 10, pp. 183 -186, 2011.

[37] X. Chen, "Throughput Multiplexing Efficiency for MIMO Antenna Characterization," IEEE Antennas Wireless Propagat. Lett., vol. 12, pp. 1208-1211, 2013.

[38] R. Tian, B. K. Lau and Z. Ying, “Multiplexing efficiency of MIMO antennas in arbitrary propagation scenarios,” 6th Euro-pean Conference on Antennas and Propagation (EUCAP), Prague, 2012, pp. 373-377.

[39] R. Tian, B. K. Lau and Z. Ying, “Multiplexing efficiency of MIMO antennas with user effects,” Proceedings of the IEEE International Symposium on Antennas and Propagation, Chicago, IL, 2012, pp. 1-2.

[40] A. Paulraj, R. Nabar, and D. Gore, Introduction to Space-TimeWireless Communications. Cambridge, U.K.: Cambridge Univ. Press, 2003.

[41] S. Blanch, J. Romeu and I. Corbella, ‘‘Exact representation of antenna system diversity performance from input parameter description,’’ Electron. Lett., vol. 39, no. 9, May 2003.

[42] P. Hallbjörner, ‘‘The significance of radiation efficiencies when using S-parameters to calculate the received signal correla-tion from two antennas,’’ IEEE Antennas Wireless Propagat. Lett., vol. 4, pp. 97-99, 2005.

[43] I. Salonen, C. Icheln, and P. Vainikainen, ‘‘The dependency of pattern correlation on mutual coupling and losses in antenna arrays,’’ Microwave Opt. Technol. Lett., vol. 47, no.2, pp. 145-147, 2005.

[44] H. Li, X. Lin, B. K. Lau and S. He, “Equivalent Circuit Based Calculation of Signal Correlation in Lossy MIMO Antennas,” IEEE Trans. Antennas Propag., vol. 61, no. 10, pp. 5214-5222, Oct. 2013.

[45] T. Taga, “Analysis for mean effective gain of mobile antennas in land mobile radio environments, ” IEEE Trans. Veh. Tech-nol., vol. 39, pp. 117-131, May 1990.

[46] K. Zhao, E. Bengtsson, Z. Ying and S. He, “Multiplexing efficiency of high order MIMO in mobile terminal in different propagation scenarios,” 10th European Conference on Antennas and Propagation (EuCAP), Davos, 2016, pp. 1-4.

[47] J. Helander, K. Zhao, Z. Ying and D. Sjöberg, "Performance Analysis of Millimeter-Wave Phased Array Antennas in Cellular Handsets," IEEE Antennas Wireless Propagat. Lett., vol. 15, pp. 504-507, 2016.

[48] P.S. Kildal, X. Chen, M. Gustafsson, Z. Shen, " MIMO Characterization on System Level of 5G Microbase Stations Subject to Randomness in LOS", IEEE Access, vol. 2, pp. 1062-1075, 2014.

[49] M. Steinbauer, A. F. Molisch and E. Bonek, "The double-directional radio channel," IEEE Antennas and Propagation Mag-azine, vol. 43, no. 4, pp. 51-63, Aug 2001.

[50] J. Toftgard, S. N. Hornsleth, and J. B. Andersen, “Effects on portable antennas of the presence of a person,” IEEE Trans. Antennas Propag., vol. 41, no. 6, pp. 739–746, June 1993.

[51] K. R. Boyle, Y. Yuan, and L. P. Ligthart, “Analysis of mobile phone antenna impedance variations with user proximity,” IEEE Trans. Antennas Propag., vol. 55, no. 2, pp. 364–372, February 2007.

BIBLIOGRAPHY | 63

[52] R. Valkonen, S. Myllymi, A. Huttunen, J. Holopainen, J. Ilvonen, P. Vainikainen, and H. Jantunen, “Compensation of finger effect on a mobile terminal antenna by antenna selection,” in Proceedings of the 12th International Conference on Electro-magnetics in Advanced Applications (ICEAA10), Sydney, Australia, 20–24 September 2010, pp. 364–367.

[53] M. Pelosi, O. Franek, M. B. Knudsen, and G. F. Pedersen, “Influence of dielectric loading on PIFA antennas in close prox-imity to user’s body,” Electron. Lett., vol. 45, no. 5, pp. 246–248, February 2009.

[54] J. Anguera, A. Andújar, A. Camps, C. Puente, and C. Picher, “Mitigation of the finger loading effect in handset antennas,” 4th European Conference on Antennas and Propagation, (EuCAP 2010), Barcelona, Spain, 12–16 April 2010

[55] L. Chung-Huan, E. Ofli, N. Chavannes, E. Cherubini, H. U. Gerber, and N. Kuster, “Effects of hand phantom on mobile phone antenna performance,” IEEE Trans. Antennas Propag., vol. 57, no. 9, pp. 2763–2770, July 2009.

[56] J. Poutanen, J. Villanen, C. Icheln, and P. Vainikainen, “Behavior of mobile terminal antennas near human tissue at a wide frequency range,” in Proceedings of the International Workshop on Antenna Technology: Small Antennas and Novel Met-amaterials (iWAT 2008), Keyaki-kaikan, Japan, 4–6 March 2008, pp. 219–222.

[57] A. A. H. Azremi, K. Haneda, and P. Vainikainen, “Site-specific evaluation of a MIMO channel capacity for multi-antenna mobile terminals in proximity to a human hand,” 5th European Conference on Antennas and Propagation (EUCAP 2011), Rome, Italy, 11–15 April 2011, pp. 538–542.

[58] V. Plicanic Samuelsson, B. K. Lau, A. Derneryd, and Z. Ying, “Channel capacity performance of multi-band dual antenna in proximity of a user,” IEEE Int. Workshop Antenna Technol. (IWAT’2009), Santa Monica, CA, USA, Mar. 2-4, 2009, pp. 5-8.

[59] V. Plicanic, B. K. Lau, A. Derneryd, and Z. Ying, “Actual diversity performance of a multiband diversity antenna with hand and head effects,” IEEE Trans. Antennas Propag., vol. 57, no. 5, pp. 1547-1556, May2009.

[60] R. F. Harrington, Time-Harmonic Electromagnetic Fields, 2nd ed. New York: John Wiley & Sons, 2001.

[61] J. Holopainen, R. Valkonen, O. Kivekäs, J. Ilvonen, L. Martínez, P. Vainikainen, J. R. Kelly, and P. S. Hall, “Equivalent circuit model-based approach on the user body effect of a mobile terminal antenna,” in Proceedings of the Loughborough Antennas and Propagation Conference, (LAPC 2010), Loughborough, U.K., 8–9 November 2010, pp. 217–220.

[62] D. M. Pozar, Microwave Engineering, 4th ed. John Wiley & Sons, 2012.

[63] A. D. Yaghjian and S. R. Best, “Impedance, bandwidth, and Q of antennas,” IEEE Trans. Antennas Propag., vol. 53, no. 4, pp. 1298–1324, April 2005.

[64] J. Ilvonen, O. Kivekäs, J. Holopainen, R. Valkonen, K. Rasilainen, and P. Vainikainen, “Mobile Terminal Antenna Perfor-mance with the User’s Hand: Effect of Antenna Dimensioning and Location,” IEEE Antennas Wireless Propagat. Lett., vol. 10, pp. 772–775, 2011.

[65] P. S. Kildal and K. Rosengren, “Electromagnetic analysis of effective and apparent diversity gain of two parallel dipoles,” IEEE Antennas Wireless Propagat. Lett., vol. 2, no. 1, pp. 9–13, 2003.

[66] R. Vaughan and J. B. Andersen, Channels, Propagation and Antennas for Mobile Communications. London: The IEE, 2003

[67] J. Ilvonen, R. Valkonen, J. Holopainen, and V. Viikari, “Multiband Frequency Reconfigurable 4G Handset Antenna with MIMO Capability,” Progress In Electromagnetics Research, vol. 148, pp. 233–243, 2014.

[68] S.C. Del Barrio, M. Pelosi, O. Franek, and G.F. Pedersen, “The Effect of the User’s Body on High-Q and Low-Q Planar Inverted F Antennas for LTE Frequencies,” 75th Vehicular Technology Conference (VTC Spring), 2012 IEEE 75th, pp. 1–4, May, 2012.

[69] A. Cihangir, F. Ferrero, C. Luxey, G. Jacquemod, E. Larique, R. Robin, P. Brachat, “Tunable Antennas Using MEMS Switches for LTE Mobile Terminals,” Loughborough Antennas and Propagation Conference (LAPC 2013).

[70] A. Cihangir, F. Ferrero, C. Luxey, G. Jacquemod, A. Reinhardt, L. Dussopt, N. Lorphelin, N. Obeid, P. Brachat, “Investi-gation of Tunable Matching Circuits for Multiband 4G Handsets,” 8th European Conference on Antennas and Propaga-tion (EuCAP), 2014

[71] I. Vasilev, V. Plicanic, R. Tian, and B. K. Lau, “Measured adaptive matching performance of a MIMO terminal with user effects,” IEEE Antennas Wireless Propagat. Lett., pp. 1720-1723, 2013.

[72] I. Vasilev, V. Plicanic and B. K. Lau, “Impact of Antenna Design on MIMO Performance for Compact Terminals With Adap-tive Impedance Matching,” IEEE Trans. Antennas Propag., vol. 64, no. 4, pp. 1454-1465, April 2016.

[73] I. Vasilev, J. Lindstrand, V. Plicanic, H. Sjöland and B. K. Lau, “Experimental Investigation of Adaptive Impedance Match-ing for a MIMO Terminal With CMOS-SOI Tuners,” IEEE Trans. Micro. Theory Techn., vol. 64, no. 5, pp. 1622-1633, May 2016.

[74] J. Ilvonen, R. Valkonen, O. Kivekäs, P. Li, and P. Vainikainen, “Antenna Shielding Method Reducing Interaction Between User and Mobile Terminal Antenna,” Electron. Lett., vol. 47, no. 16, pp. 896–897, August 2011.

[75] M. Berg, M. Sonkki and E. T. Salonen, “Absorption Loss Reduction in a Mobile Terminal with Switchable Monopole Anten-nas,” IEEE Trans. Antennas Propag., vol. 59, no. 11, pp. 4379-4383, Nov. 2011.

BIBLIOGRAPHY | 64

[76] “The iPhone 5 review” http://www.anandtech.com/show/6330/the-iphone-5-review/18

[77] S. Zhang, K. Zhao, Z. Ying and S. He, “Adaptive Quad-Element Multi-Wideband Antenna Array for User-Effective LTE MIMO Mobile Terminals,” IEEE Trans. Antennas Propag., vol. 61, no. 8, pp. 4275-4283, Aug. 2013.

[78] Y. L. Ban, Y. F. Qiang, Z. Chen, K. Kang and J. H. Guo, “A Dual-Loop Antenna Design for Hepta-Band WWAN/LTE Metal-Rimmed Smartphone Applications,” IEEE Trans. Antennas Propag., vol. 63, no. 1, pp. 48-58, Jan. 2015.

[79] K. Zhao, Z. Ying and S. He, “Double ring Antenna Design for MIMO Application in Mobile Terminals, ” The 9th European Conference on Antennas and Propagation (EuCAP), Lisbon, 2015, pp. 1-5.

[80] Z. Miers, H. Li and B. K. Lau, "Design of bezel antennas for multiband MIMO terminals using Characteristic Modes," The 8th European Conference on Antennas and Propagation (EuCAP), Hague, 2014, pp. 2556-2560.

[81] N. Chahat, M. Zhadobov, L. Le Coq, S. I. Alekseev, and R. Sauleau, “Characterization of the interactions between a 60-GHz antenna and the human body in an off-body scenario,” IEEE Trans. Antennas Propag., vol. 60, no. 12, pp. 5958–5965, 2012.

[82] F. Villanese, N. E. Evans and W. G. Scanlon, “Pedestrian-induced fading for indoor channels at 2.45, 5.7 and 62 GHz,” in Proc. IEEE 52nd Vehicular Technology Conference (VTC2000-Fall), Sept. 2000, pp. 43-48.

[83] G. R. MacCartney, S. Deng, S. Sun, and T.S. Rappaport, “Millimeterwave human blockage at 73 GHz with a simple double knife-edge diffraction model and extension for directional antennas,” IEEE 84th Veh. Technol. Conf. (VTC-Fall), Sep. 2016, pp.1-6.

[84] C. Gustafson and F. Tufvesson, “Characterization of 60 GHz shadowing by human bodies and simple phantoms, ” Radi-oengineering, vol. 21, no. 4, pp. 979-984, 2012.

[85] T. Bai and R. W. Heath, “Analysis of self-body blocking effects in millimeter wave cellular networks,” 48th Asilomar Con-ference on Signals, Systems and Computers, Pacific Grove, CA, 2014, pp. 1921-1925.

[86] M. Abouelseoud and G. Charlton, “The effect of human blockage on the performance of millimeter-wave access link for outdoor coverage,” IEEE 77th Vehicular Technology Conference (VTC2013), Jun. 2013, pp. 1-5.

[87] W. Hong, K. Baek, Y. Lee and Y. Kim, “Design and analysis of a low-profile 28 GHz beam steering antenna solution for Future 5G cellular applications,” IEEE MTT-S Int. Microw. Symp. Dig., Jun. 2014, pp. 1–4. June 2014

[88] 3GPP. Technical specification group radio access network; channel model for frequency spectrum above 6 GHz. TR 38.900, 3rd Generation Partnership Project (3GPP), June. 2016.

[89] T. Bai and R. W. Heath, Jr., “Analysis of self-body blocking effects in millimeter wave cellular networks,” in Proc. 48th Asilomar Conf. Signals, Syst. Comput., Nov. 2014, pp. 1921–1925.

[90] C. A. Balanis, Antenna Theory: Analysis and Design, 3rd Ed. John Wiley and Sons, 2009.

[91] Intel Corporation on UE power class for mmWave, R4-1706508

[92] Qualcomm Incorporated Output Power Requirement for mmW, R4-1706624

[93] I. Syrytsin, S. Zhang, G. Pedersen, K. Zhao, T. Bolin and Z. Ying, “Statistical Investigation of the User Effects on Mobile Terminal Antennas for 5G Applications,” accepted in IEEE Trans. Antennas Propag.

[94] F. Mani, “Improved ray-tracing for advanced radio propagation channel modeling,” in PHD dissertation, June 2012. [95] Rec. ITU-R P.530-16 “Propagation data and prediction methods required for the design of terrestrial line-of-sight sys-

tems,” Jul. 2015 [96] Rec. ITU-R P.527-3 “Electrical characteristics of the surface of the earth”, 1992. [97] P. Okvist, H. Asplund, A. Simonsson, B. Halvarsson, J. Medbo, and N. Seifi, “15 GHz propagation properties assessed with

5G radio access prototype,” in Proc. 26th Annu. IEEE Int. Symp. Pers., Indoor, Mobile Radio Commun (PIMRC), Sep. 2015, pp. 2220–2224.

[98] "Small Cells". Ericsson.com. [Online]. Available: https://www.ericsson.com/ourportfolio/networks-products/small-cells?nav=productcatagory006%257Cfgb_101_0516

[99] "Specific Absorption Rate (SAR) For Cell Phones: What It Means For You". Federal Communications Commission. Dec. 2012

[100] J. C. Lin, “Electromagnetic Fields in Biological Systems,” CRC Press, Oct. 2011.

[101] O. Kivekas, T. Lehtiniemi, P. Vainikainen, “On the general energy-absorption mechanism in the human tissue,” Micro-wave Opt. Technol. Lett., vol. 43, pp. 195-201, Nov. 2004.

[102] IEEE 1528-2003, ‘‘IEEE Recommended Practice for Determining the Peak Spatial-Average Specific Absorption Rate (SAR) in the Human Head from Wireless Communications Devices: Measurement Techniques.’’

[103] J. Trajkovikj and A. K. Skrivervik, “Comparison of SAR of UHF wearable antennas,” 10th European Conference on An-tennas and Propagation (EuCAP), Davos, 2016, pp. 1-4.

BIBLIOGRAPHY | 65

[104] S. M. Ali and H. Gu, “Effects of Chassis Currents on Hearing Aids Compatibility in the Handset,” IEEE Trans. Electro-magn. Compat., vol. 52, no. 4, pp. 837-842, Nov. 2010.

[105] M. R. Islam and M. Ali, “Ground Current Modification of Mobile Terminal Antennas and Its Effects,” IEEE Antennas Wireless Propagat. Lett., vol. 10, pp. 438-441, 2011.

[106] Y. Koga, M. Kai, K. Fujieda and H. Egawa, “Low-SAR multiband antenna with ground slit and parasitic elements for tablet computers,” 2016 IEEE International Symposium on Antennas and Propagation (APSURSI), Fajardo, 2016, pp. 449-450.

[107] H. Li, A. Tsiaras and B. K. Lau, “Analysis and Estimation of MIMO-SAR for Multi-antenna Mobile Handsets,” IEEE Trans. Antennas Propag., vol. 65, no. 3, pp. 1522-1527, March 2017.

[108] H. Li, A. Tsiaras, B. Derat, and B. K. Lau, “Analysis of SAR on flat phantom for different multi-antenna mobile terminals,” 8th Eur. Conf. Antennas Propag. (EuCAP), The Hague, The Netherlands, Apr. 2014, pp. 1989–1993.

[109] M. S. Wang et al., “Evaluation and optimization of the specific absorption rate for multiantenna systems,” IEEE Trans. Electromagn. Compat., vol. 53, no. 3, pp. 628–637, Aug. 2011.

[110] N. Perentos, S. Iskra, A. Faraone, R. J. McKenzie, G. Bit-Babik, and V. Anderson, “Exposure compliance methodologies for multiple input multiple output (MIMO) enabled networks and terminals,” IEEE Trans. Antennas Propag.,vol. 60, no. 2, pp. 644–653, Feb. 2012.

[111] D. T. Le, L. Hamada, S. Watanabe and T. Onishi, “A Fast Estimation Technique for Evaluating the Specific Absorption Rate of Multiple-Antenna Transmitting Devices,” IEEE Trans. Antennas Propag.,vol. 65, no. 4, pp. 1947-1957, Apr. 2017.

[112] “SAR Evaluation Consideration for Handsets with Multiple Transmitters and Antennas,” FCC, Feb. 2008

[113] B. M. Hochwald and D. J. Love, "Minimizing exposure to electromagnetic radiation in portable devices," 2012 Infor-mation Theory and Applications Workshop, San Diego, CA, 2012, pp. 255-261.

[114] “Notice of proposed rulemaking,” FCC, Commission Documents 15- 138, Oct. 2015.

[115] D. Colombi, B. Thors and C. Törnevik, “Implications of EMF Exposure Limits on Output Power Levels for 5G Devices Above 6 GHz,” IEEE Antennas Wireless Propagat. Lett., vol. 14, pp. 1247-1249, 2015.

[116] K. Foster and D. Colombi, “Thermal response of tissue to RF exposure from canonical dipoles at frequencies for future mobile communication systems,” Electron. Lett., vol. 53, no. 5, pp. 360-362, 2017.

[117] B. Thors, D. Colombi, Z. Ying, T. Bolin and C. Törnevik, “Exposure to RF EMF From Array Antennas in 5G Mobile Com-munication Equipment,” IEEE Access, vol. 4, pp. 7469-7478, 2016.

[118] B. Xu, K. Zhao, B. Thors, D. Colombi, O. Lundberg, Z. Ying and S. He, “Power Density Measurements at 15 GHz for RF EMF Compliance Assessments of 5G User Equipment,” accepted in IEEE Trans. Antennas Propag..

[119] B. Xu, M. Gustafsson, S. Shi, K. Zhao, Z. Ying, and S. He, “RF EMF Exposure Compliance of mmWave Array Antennas for 5G User Equipment Application,” submitted to IEEE Trans. Antennas Propag..

BIBLIOGRAPHY | 66