Conformal Body-Worn Smart Antenna System for Wideband UHF Operation
Dissertation
Presented in Partial Fulfillment of the Requirements for the Degree Doctor of Philosophy in the Graduate School of The Ohio State University
By
Gil Young Lee, M.S., B.S.
Graduate Program in Electrical and Computer Engineering
The Ohio State University
2012
Dissertation Committee:
John L. Volakis, Advisor Chi-Chih Chen, Co-Advisor Fernando L. Teixeira Dimitris Psychoudakis c Copyright by
Gil Young Lee
2012 Abstract
There is much interest in developing body-centric wireless communication systems
(BWCS) for mobile health care systems. However, the realization of a BWCS is chal- lenging due to the body’s interference with the antenna’s operation. More specifically, body-worn antennas suffer from impedance detuning, pattern deformation, and gain reduction caused by the body. Therefore, it is important to consider these effects in evaluating body-worn antennas. In this regard, a diversity technique is proposed to improve body-worn antenna performance.
More specifically, a channel decomposition method (CDM) is proposed and used to evaluate body-worn antenna systems. The CDM significantly reduces computa- tion time when evaluate body-worn antennas and is applicable to various surrounding environments without recalculation of the more complex interaction. A second con- tribution of this dissertation is design of a diversity systems which automatically determines the minimum number of antennas while maximizing performance. This approach is employed to design body-worn antenna diversity systems for given com- munication scenarios. The results obtained via this process demonstrated that this simple method can substantially reduced computation time in designing body-worn antenna diversity system. As a demonstration of the proposed methodology, a vest- mounted UHF body-worn antenna diversity system (BWADS) is developed using 4 light-weight antennas. The proposed BWADS is transparent and unobtrusive to the
ii users but provides performance superior to commercial antennas. A variety of tests were performed to validate the proposed BWADS. It was found that the proposed
BWADS provided 7 dB (outdoor) to 16.5 dB (indoor) of higher gain as compared to commercial antennas. The dissertation concludes by proposing other applications of the developed body-worn antennas and design methods.
iii To my wife, daughters, and son
iv Acknowledgments
First and the foremost, I would like to thank Almighty God for bestowing His blessings upon me and giving me the strength to carry out and complete this work.
I would especially like to thank my advisor Professor John L. Volakis for all of the support and invaluable guidance that he provided in completing this work and in pursuing my Ph.D. degree. I cannot help admiring him for his integrity and diligence.
It is an honor for me to be his student.
I am also extremely grateful to my co-advisor Professor Chi-Chih Chen for all his support and advice in the technical details of my dorctoral work. He encouraged me to overcome many issues I have faced during my research and I really enjoyed academic discussion with him. He is like a brother to me and I look forward to working with him in future.
I would also like to express my deep gratitude to my good friend and colleague,
Dr. Dimitris. Psychoudakis, who has been very supportive and generous in sharing his knowledge. My gratitude is extended to Professor Fernando L. Teixeira and
Professor Philip Schniter for serving my committee and providing insightful ideas and suggestions. Their lectures also were always great and I learned a lot of valuable knowledges in electromagentics and signal processing from them.
v I offer my blessings to all the staff and fellow students of the ElectroScience Lab- oratory. I really want to thank for their valuable discussion and friendship. Spe- cial thanks to officemates - Jae-Young Chung, Yijun Zhou, Jing Zhao, Ming Chen,
Mustafa Kuloglu, Ugur Olgun, Erdinc Irci, Ioannis Tzanidis, and Tao Peng. My gratitude is also extended for other friends at ESL - Chun-Sik Chae, James Park, Jun
Seok Lee, Keum-Su Song, Kyung-young Jung, Haksu Moon, Pil Sung Park, Youngseo
Ko, Justin Kasemodel, Praphun Naenna, and William Moulder.
Finally, I would like to thank my family. The unconditional love and support of my parents and parents-in-law have been the greatest motivation during the graduate study. And most importantly, I would like to express the deepest gratitude to my wife, In Kyung, who always loved, encouraged, and supported me. I also thank to my two daughters and a son - Ye Eun, Ye Won, and Jum beom. You are all of my pleasures and hopes. I love you!
vi Vita
January 11, 1975 ...... Born - Kwangchon, Chungnam-Do, Korea March, 1997 ...... B.E., Electronics Eng., Korea Air-Force Academy, Cheongju, Chungbuk-Do, Korea February, 2001 ...... B.S., Electrical Eng. & Computer Sci., Seoul National University, Seoul, Korea February, 2004 ...... M.S., Electrical Eng. & Computer Sci., Seoul National University, Seoul, Korea June, 2011 ...... M.S., Electrical & Computer Eng., The Ohio State University, Columbus, Ohio, USA
Publications
G.-Y. Lee, C.-C Chen, D. Psychoudakis, and J. L. Volakis, “MIMO for Body-Worn Antennas: Approach and Measurements,” IEEE International Symposium on An- tennas and Propagation & USNC/URSI National Radio Science Meeting, San Diego, CA, USA, Jul. 5–11, 2008.
D. Psychoudakis, G.-Y. Lee, C.-C Chen, and J. L. Volakis, “Body-worn diversity antennas for Squad Area Networks (SAN),” XXIXth General Assembly of Union Radio Science International (URSI), Chicago, IL, USA, Aug. 07-16, 2008.
G.-Y. Lee, C.-C Chen, D. Psychoudakis, and J. L. Volakis, “Diversity Evaluation for Multiple Body-Worn Antennas,” URSI-National Radio Science Meeting, Boulder, CO, USA, Jan. 5–8, 2009.
vii J. L. Volakis, G.-Y. Lee, D. Psychoudakis, and C.-C Chen, “Multiple Body-Worn Antenna Diversity,” IEEE International Workshop on Antenna Technology, Santa Monica, CA, USA, Mar. 2–4, 2009.
D. Psychoudakis, G.-Y. Lee, C.-C Chen, and J. L. Volakis, “Diversity Evaluation for Multiple Body-Worn Antennas,” 3rd European Conference on Antennas and Propa- gation, Berlin, Germany, Mar. 23–27, 2009.
G.-Y. Lee, C.-C Chen, D. Psychoudakis, and J. L. Volakis, “A Novel Evaluation Method for Body-Worn Radio Systems,” IEEE International Symposium on Anten- nas and Propagation & USNC/URSI National Radio Science Meeting, Charleston, SC, USA, Jun. 1–5, 2009.
G.-Y. Lee, C.-C Chen, D. Psychoudakis, and J. L. Volakis, “Systematic Design Approach for Diversity Antenna Systems,” 31st Antenna Measurement Techniques Association (AMTA) Symposium, Salt Lake City, UT, USA, Nov. 1–6, 2009.
D. Psychoudakis, G.-Y. Lee, C.-C Chen, and J. L. Volakis, “Military UHF Body- Worn Antennas for Armored Vests,” 4th European Conference on Antennas and Propagation, Barcelona, Spain, Apr. 12–16, 2010.
G.-Y. Lee, C.-C Chen, D. Psychoudakis, and J. L. Volakis, “A Computationally Efficient Method for Body-Worn Antenna Diversity Design,” 26th Applied Computa- tional Electromagnetics Society (ACES) conference, Tampere, Finland, Apr. 26–29, 2010.
G.-Y. Lee, C.-C Chen, D. Psychoudakis, and J. L. Volakis, “Body-Worn Antenna Diversity Design using FEKO,” 26th Applied Computational Electromagnetics Society (ACES) conference, Tampere, Finland, Apr. 26–29, 2010.
G.-Y. Lee, C.-C Chen, D. Psychoudakis, and J. L. Volakis, “Multiple Antenna Design Method for Mobile Platform Diversity Systems,” IEEE International Symposium on Antennas and Propagation & USNC/URSI National Radio Science Meeting, Toronto, Canada, Jul. 11–17, 2010.
G.-Y. Lee, C.-C Chen, D. Psychoudakis, and J. L. Volakis, “Channel Decomposition Method for Designing Body-Worn Antenna Diversity Systems,” IEEE Transactions on Antennas and Propagation, vol. 59, no. 1, pp. 254–262, Jan. 2011.
viii G.-Y. Lee, C.-C Chen, D. Psychoudakis, and J. L. Volakis, “Omnidirectional Vest- Mounted Body-Worn Antenna System for UHF Operation,” IEEE Antennas and Wireless Propagation Letters, vol. 10, pp. 581–583, Jun. 2011.
Fields of Study
Major Field: Electrical and Computer Engineering
Studies in: Electromagnetics and Antenna Design Communications and Signal Processing
ix Table of Contents
Page
Abstract...... ii
Dedication...... iv
Acknowledgments...... v
Vita ...... vii
ListofTables...... xiii
ListofFigures ...... xiv
1. Introduction...... 1
2. Human Body Model for Body-Centric Wireless Communication . .... 9
2.1 Introduction ...... 9 2.2 Review of Electromagnetic Properties of the Human Body . . . . . 10 2.2.1 Numerical Expression of the Dielectric Properties for Human Tissues ...... 10 2.2.2 Dielectric Properties of Human Tissue ...... 11 2.3 Propagation Modes for Body-Centric Wireless Communications . . 15 2.3.1 Penetrating and Reflecting Wave Analysis ...... 17 2.3.2 Creeping Wave Analysis ...... 20 2.3.3 Space Wave Analysis for Off-Body Communication . . . . . 22 2.4 EM Human Torso Model for Body-Centric Wireless Communication 25 2.4.1 Human Torso Model for In-Body Communication ...... 25 2.4.2 Human Torso Model for On-Body Communication . . . . . 28 2.4.3 Human Torso Model for Off-Body Communication . . . . . 30 2.5 Summary ...... 31
x 3. Body-WornAntennaEvaluationMethod ...... 33
3.1 Introduction ...... 33 3.2 Body-WornAntennaElement ...... 34 3.3 Evaluation Method for Diversity System ...... 35 3.3.1 Channel Decomposition Method for Diversity Evaluation . . 39 3.3.2 CDM Validation ...... 42 3.4 Multiple Body-Worn Antenna Diversity System ...... 47 3.4.1 SingleBody-WornAntennaStudy ...... 48 3.4.2 DualBody-WornAntennaStudy ...... 51 3.4.3 Multiple Body-Worn Antenna Diversity ...... 52 3.5 Diversity Module Realization and Measurements ...... 54 3.5.1 Diversity Module Realization ...... 54 3.5.2 IndoorMeasurement ...... 56 3.6 Conclusion ...... 59
4. Systematic Design Method for Body-Worn Antenna Diversity Systems . 60
4.1 Introduction ...... 60 4.2 Systematic Design Method for Body-Worn Diversity Systems . . . 62 4.2.1 Definition of the Design Parameters ...... 62 4.2.2 Systematic Design Process for Antenna Diversity ...... 64 4.3 Diversity Design Examples for Body-Worn Antenna ...... 66 4.3.1 Communication Scenario I (Ground to Ground Communica- tion)...... 68 4.3.2 Communication Scenario II (All-Purpose Communication) . 71 4.4 Measurement and Validation ...... 74 4.5 Conclusions...... 77
5. Vest-mounted Body-Worn Antenna Diversity System for Wideband UHF Operation ...... 78
5.1 Introduction ...... 78 5.2 Conformal Wideband Body-Worn Antenna Design for UHF Operation 79 5.2.1 Antenna Elements and Diversity Configuration ...... 80 5.2.2 Antenna Performance on Body ...... 83 5.3 Compact Diversity Module Realization ...... 88 5.3.1 Operation of the Diversity Module ...... 88 5.3.2 Diversity Module Fabrication using COTS Components . . 90 5.4 Field Test for Body-Worn Antenna Diversity System ...... 93 5.4.1 Summary of the Reference Antenna ...... 93
xi 5.4.2 TestSetup ...... 94 5.4.3 OutdoorTests ...... 96 5.4.4 IndoorTests ...... 103 5.4.5 FieldTestSummary ...... 105 5.5 Conclusions...... 105
6. FutureWorks ...... 108
6.1 Introduction ...... 108 6.2 Vehicular Antenna Diversity System ...... 109 6.3 Vital Sign Monitoring System using Body-Worn Antennas . . . . . 112 6.4 Body-Worn Antenna System for RF Power Harvest ...... 115
7. Conclusions ...... 121
Bibliography ...... 125
xii List of Tables
Table Page
2.1 Parameters required to find human tissue dielectric constant and con- ductivity[23]...... 12
2.2 Electromagnetic properties of skin (dry), muscle, and fat at selected frequencies (30 MHz, 433 MHz, 915 MHz, 2.45 GHz, and 12 GHz). . . 16
2.3 Calculated SWR, reflection coefficient, and transmitted E-field at each frequency...... 18
4.1 Comparison of selected antenna positions, pattern coverage, and corre- lation coefficients for n-antenna diversity between the proposed method and full evaluation method for scenario I (Abbreviations for antenna positions; 1st letter- B: back, F: front, S: shoulder, T: Thigh, 2nd letter- C: center, H: Horizontal, L: left, R: right)...... 70
4.2 Comparison of selected antenna positions, pattern coverage, and corre- lation coefficients for n-antenna diversity between the proposed method and full evaluation method for scenario II...... 73
5.1 Summary of calculated SAR values and limitations...... 88
5.2 Summary of the technical specifications of the diversity module. . . . 92
5.3 Summary of the test results in outdoor and indoor environments. . . 107
6.1 Electromagnetic properties of deflated lung and inflated lung at se- lected frequencies (30 MHz, 433 MHz, 915 MHz, 2.45 GHz, and 12 GHz)...... 113
xiii List of Figures
Figure Page
1.1 Brief illustration of the works in the dissertation: (a) concept of chan- nel decomposition method (CDM), (b) best body-worn antenna diver- sity for omni-directional pattern, (c) vest-mounted body-worn antenna diversity system for wideband UHF operation ...... 5
2.1 Electromagnetic properties of skin, fat, muscle, and lung (deflated); (a) relative permittivity (real value), (b) conductivity, and (c) penetration depthinmm...... 13
2.2 Electromagnetic properties of heart, liver, lung (deflated), and lung (inflated); (a) relative permittivity (real value), (b) conductivity, and (c)penetrationdepthinmm...... 14
2.3 Infinite human tissue layers (skin - fat - muscle - fat - skin) and radiat- ing sources near the body surface; (a) plane wave source, (b) vertical dipole antenna, and (c) horizontal dipole antenna...... 17
2.4 Penetrating and reflecting wave on the boundary of human tissues: (a) at whole observed region, (b) zoom-in view focused on human tissue layers...... 19
2.5 Near field plot (E-field) around the radiating antenna and human body layer: (a) vertical dipole antenna near the body surface, (b) horizontal dipole antenna near the body surface, (c) vertical dipole antenna in the free space, and (d) E-field magnitude along y-axis at z=0.4 λ (center oftheantenna)...... 21
2.6 Plane wave incident obliquely on a plane dielectric boundary: (a) par- allel polarized wave, (b) perpendicular polarized wave...... 22
xiv 2.7 Electric far-field pattern along the elevation angle for different distances betweenantennaandhumanbody...... 23
2.8 The reference human torso model: (a) thorax cut image from visible human project, (b) top view, (c) front view of torso model...... 26
2.9 (a) Top view and (b) front view of the approximated concentric 4-layer model, and (c) comparison of normalized E-field magnitude inside the humanbodyfortwomodels...... 27
2.10 Top view of (a) 4-layer model, (b) 3-layer model, (c) homogeneous model, (d) transmission coefficients between two antennas for each model. 29
2.11 (a) Homogeneous human model, (b) 3-layer model (c) azimuth radia- tion pattern (x-y plane), (d) elevation radiation pattern (x-z plane). . 31
3.1 AMF antenna structure and dimensions for operation at 560 MHz. . . 35
3.2 Performance of the AMF antenna element when placed near the body’s torso: (a) the simulated reflection coefficients S11 for 3 distances from the body, (b) measured and simulated radiation patterns in the hor- izontal (x-y) plane at 550 MHz (antenna is 12 mm from the body’s surface)...... 36
3.3 Communication channel model for the CDM...... 38
3.4 Flowchart of the diversity evaluation process using the CDM...... 40
3.5 Simple channel model for validating CDM...... 43
3.6 (a) Amplitude distribution and (b) polarization angle distribution of the incident wave in the environment depicted in Fig 3.5...... 45
3.7 Comparison of the channel capacity CCDF calculated using the Chan- nel Decomposition Method with direct simulations (no decomposition). 46
3.8 Investigated mounting positions for the AMF antenna: (a) front view, (b) back view, (c) left side view, and (d) right side view...... 49
xv 3.9 Projected upper-hemisphere radiation pattern onto the x-y plane for selected single antenna positions: (a) front torso, (b) left arm, and (c) leftshoulder...... 50
3.10 Channel capacity CCDF curves for different antenna pairs...... 51
3.11 (a) Channel capacity per configuration, and (b) list of antenna config- urationspernumberinginFig. 3.8...... 53
3.12 Schematic diagram of the implemented 4-channel diversity module. . 55
3.13 The 4-antenna diversity indoor measurement set-up...... 56
3.14 OSU-ESL hallway environment and the motion path used for the in- doormeasurement...... 57
3.15 Comparison of the normalized received power between single antennas (mounted at different location on the body) and the diversity system (measurementsweredoneinthehallway)...... 58
4.1 ConceptoftheCDM...... 61
4.2 Flow chart of the systematic antenna diversity design approach. . . . 64
4.3 (a) Investigated mounting positions for the antenna diversity configu- rations and antenna elements operating at (b) 550 MHz and (c) 350 MHz,respectively...... 67
4.4 Illustration of the communication scenario I for ground to ground com- munication...... 68
4.5 Illustration of the communication scenario II for all-purpose commu- nication...... 71
4.6 Outdoor measurement setup with human mannequin and diversity module...... 75
4.7 Normalized received power from the 4-antenna diversity module as compared to each antenna’s individual response...... 75
xvi 4.8 Comparison of the effective antenna gain CCDF curves between simu- lations and measurements for the best single antenna, 2-antenna, and 4-antennadiversitysystems...... 76
5.1 A typical vest configuration [52] with selected antenna mounting loca- tions and the proposed antenna geometries with dimensions; (a) front view, (b) rear view of the vest, (c) gorget antenna on the front, (d) rear antennas on the left back and the right back, and (e) horizontal antennaonthebackwaist...... 81
5.2 Several antenna candidates for gorget antenna mounted on the front vest; (a) antenna 1, (b) antenna 2, (c) antenna 3, and comparisons of (d)S11and(e)peakgain...... 83
5.3 Several antenna candidates for rear antennas mounted on the back vest; (a) antenna 1, (b) antenna 2, (c) antenna 3, and comparisons of (d)antennaS11and(e)peakgain...... 84
5.4 Peak gain comparison for each antenna in free space and on the body (antenna is 25 mm from the body surface); (a) gorget antenna, (b) rear antenna,(c)waistantenna...... 85
5.5 Azimuth radiation pattern comparisons between simulation and mea- surement; (a) simulated resuts at 250 MHz, (b) measured resuts at 250 MHz, (c) simulated resuts at 350 MHz, (d) measured resuts at 350 MHz. 86
5.6 (a) Hugo upper torso model and (b) simulation setup for SAR calculation. 87
5.7 (a) Schematic diagram and (b) operation algorithm of the diversity module...... 89
5.8 Fabricated diversity module; (a) Bottom view (RF switch), (b) Top view (micro controller and voltage regulators), (c) packaged diversity module...... 91
5.9 (a) Test setup for the diversity module and (b) test results...... 92
5.10 (a) Measurement and (b) simulation model for monopole whip antenna, and (c) peak gain comparisons between measurement and simulation. 94
5.11 Test setup for BWADS and monopole whip antenna...... 95
xvii 5.12 Test fields in OSU ESL facilities; (a) outdoor test field with several obstacles,(b)indoortestfield...... 96
5.13 Illustration of the outdoor test Scenario I (top figure), and test results for vertical transmitter (2nd row figures) and for horizontal transmitter (3rdrowfigures)...... 98
5.14 Illustration of the outdoor test Scenario II (top figure), and test results for vertical transmitter (2nd row figures) and for horizontal transmitter (3rdrowfigures)...... 99
5.15 Illustration of the outdoor test Scenario III (top figure), and test results for vertical transmitter (2nd row figures) and for horizontal transmitter (3rdrowfigures)...... 101
5.16 Illustration of the outdoor test Scenario IV (top figure), and test results for vertical transmitter (2nd row figures) and for horizontal transmitter (3rdrowfigures)...... 102
5.17 Illustration of the indoor test Scenario I (top figure), and test results forverticaltransmitter(bottomfigures)...... 103
5.18 Illustration of the indoor test Scenario II (top figure), and test results forverticaltransmitter(bottomfigures)...... 104
5.19 CCDF comparison of received power including all scenarios for (a) outdoorand(b)indoortest...... 106
6.1 Simulation model for vehicular antenna diversity system and commu- nicationscenarioconsidered...... 110
6.2 Best antenna diversity configuration and antenna gain CCDF curves for single, 2-, 3-, and 4-antenna diversity system...... 111
6.3 Diversity performance comparison between the worst and the best 4- antennadiversitysystem...... 112
6.4 Simulation model of vital sign monitoring system and results; (a) side view, (b) top view of the model, and (c) path gain comparison. . . . . 114
xviii 6.5 (a) Measurement setup and (b) ambient RF power measured by 0 dBi receiver at the backyard of OSU-ESL building (Wi-Fi signal is mea- suredatthehallwayofOSU-ESLbuilding)...... 116
6.6 (a) Front view and (b) rear view of body-worn antennas, (c) antenna element structure, (d) 3-D radiation pattern and (e) peak gain of the arrayantenna...... 117
6.7 Calculated RF power received from array antenna and each single an- tennainhorizontaldirection...... 118
6.8 Schematics of the modified Greinacher rectifier [61]...... 119
6.9 CCDF curves of the generated DC power from the proposed body-worn rectennasystems...... 120
xix Chapter 1: Introduction
The rapid growth in wearable sensors and wireless communications has enabled the development of a new generation of wireless sensor networks, specifically wireless body area networks (WBANs). WBANs are special sensor networks designed to autonomously connect various sensors and appliances, located inside and outside of a human body. The wireless nature of the network offers numerous new innovative applications to improve health care and quality of life. Specifically, a mobile health care system is one of the substantial applications of the WBANs [1, 2]. Various sensors are attached on the body or even implanted under the skin to measure the heartbeat, the body temperature, etc. Then, the collected information is transferred to an external medical database or a doctor through a personal communication device for remote health monitoring. Since all sensors and appliances are wirelessly connected, the patient experiences a greater physical mobility and is no longer compelled to stay in hospital. The benefit of WBANs is also applicable to communication systems for specialized occupations such as firefighters, police officers, etc. However, there is a challenge in implementing the system to provide reliable wireless links between on- body sensors, personal communication devices, and external communication nodes. In this regard, body-centric wireless communication systems (BWCS) become important in connecting all wireless devices on the body or its proximity.
1 BWCS can be classified into three communication domains as follows [3]:
1. In-body communication: This refers to communication between implanted sen-
sors and devices on the body surface.
2. On-body communication: The main goal of on-body communication is to pro-
vide a reliable wireless connectivity among wearable sensors, and mobile wireless
devices mounted on the body.
3. Off-body communication: This link provides connectivity from a personal area
network to other existing wireless communication domains (e.g. base station
for cellular communication).
Each of the above is significantly affected by the electrical properties of the human body. Therefore, device performance must be considered in presence of the human body. Chapter 2 discusses the properties of the human body and its effects on each communication devices using electromagnetic (EM) human body models. My focus throughout the dissertation is, however, on off-body communications of the wearable devices.
Electromagnetic field interactions between the human body and antennas have been documented in the literature [4–10]. Based on these studies one may conclude the following:
1. Antenna impedance strongly depends on placement position and distance from
the body. The resonant frequency of the antenna decreases as distance de-
creases, and impedance bandwidth increases as distance decreases.
2. Antenna efficiency and gain are significantly reduced at distances less than 0.1
λ from the body due to the body’s loss and negative reflection from the body.
2 3. Antenna pattern is also deformed by the presence of the body. One can consider
the body as a reflector of the antenna. Thus, it causes directional pattern instead
of a usually desired omni-directional pattern. The pattern shape and peak gain
also depend on the distance from the body.
To overcome human body effects on body-worn antenna performance addressed above, some authors have proposed metallic or electromagnetic bandgap (EBG) sur- faces to reduce back radiation and increase efficiency [11–13]. However, EBG surfaces are narrowband and difficult to adapt to lower frequencies, and may also cause un- desirable radiation patterns. Therefore, they are not suited for broadband UHF wearable communication.
As an alternative, diversity techniques can offer increased performance for body- worn antennas. Among previous researches in this area, Serra et al. [14] considered dual antenna diversity for the 2.45 GHz ISM band. However, they focused on the connectivity between antennas on the same body (on-body communication), and lim- ited their study to two antennas. Other papers [15, 16] considered antenna diversity for off-body communication, but their conclusions were restricted to specific indoor environments. Moreover, all of the above cases refer to narrowband applications. For the first time, in this dissertation, the diversity techniques are adapted for wideband
UHF body-worn communications.
As body-centric communications are associated with continuously changing en- vironments, a generalized antenna diversity evaluation method is desired when de- signing body-worn antennas. It is desired to include the surrounding environments as well as the effect of the human body for evaluation of the body-worn antennas.
However, critical issues regarding computational resources for EM calculations arise:
3 1. The human body itself consists of various tissues with different EM proper-
ties, and substantial computational resources are required for EM simulations
without simplifying approximations.
2. It is very difficult to consider the surrounding environments with accurate nu-
merical methods like method of moments (MoM) and the finite element method
(FEM). Modeling the environment also requires significant computational re-
sources.
3. Both the human body and the mobile environment change continuously over
time. Body movement causes rapid variations in antenna parameters (impedance,
gain, pattern, etc.) affecting performance. The mobile environment can also
vary over time, however its characteristics do not change as abruptly. These
effects should be considered in any evaluation method.
4. A large number of configuration options should be investigated when developing
multiple body-worn antenna diversity. The total computation time required will
be multiplied by the number of diversity configurations.
With the above in mind, in this dissertation we develop an efficient and versa- tile evaluation method for designing body-worn antennas with diversity. A channel decomposition and systematic design methods are proposed to accomplish this. The methods are subsequently employed to design a vest-mounted body-worn diversity system.
Illustrations of the research carried out in this dissertation are shown in Fig. 1.1.
Some of the key contributions of the dissertation are given below:
4 Communication channel
Human body channel Propagation channel
h1 Diversity h module 2
Rx Tx h3 SNR Re c SNR Rx SNR Tx (a)
Selected antenna Ant 1 Ant 3 Ant 2 Ant 4 Ant 1 000 gg
---1-1110000
---2-2220000
---3-3330000 Ant1- Front center (FC) Ant2- Back center (BC) ---4-4440000 Ant3- Left shoulder (LS) Ant4- Right shoulder (RS)
Normalized received [dB] power ---5-5550000 Diversity module output
000 33303000 66606000 99909000 111212220000 111515550000 111818880000 222121110000 222424440000 222727770000 3330300000 3333330000 333636660000 Azimuth angle [deg] (b)
Front of vest Back of vest Diversity module
Antennas
(c)
Figure 1.1: Brief illustration of the works in the dissertation: (a) concept of chan- nel decomposition method (CDM), (b) best body-worn antenna diversity for omni- directional pattern, (c) vest-mounted body-worn antenna diversity system for wide- band UHF operation .
5 1. Developed a channel decomposition method (CDM): The proposed CDM was
used to evaluate body-worn antenna performance in various environments. The
CDM significantly reduced computation time (more than 10 times) without loss
of accuracy for body-worn antenna evaluations.
2. Developed a systematic body-worn antenna diversity design method: The pro-
posed method, incorporated with CDM, automatically finds the best antenna
mounting positions to achieve acceptable performance. The approach is sub-
sequently applied to design a body-worn antenna diversity system for specific
communication environments. Importantly, the proposed approach reduced the
number of diversity configurations (from 495 cases to 12 cases for 4-antenna
diversity from 12 available antenna positions) and computation time.
3. Designed and fabricated a vest-mounted UHF body-worn antenna system with
diversity: Using the above methods, a practical body-worn antenna diversity
system was developed to provide a nearly omni-directional pattern over 225-450
MHz. The light-weight and conformal antennas were hidden in a standard vest
and in a manner not to impede the wearers’ activities. A compact diversity
module (79 × 41 × 28 mm) was also constructed to automatically select the
best channel among 4-antennas in real time. The proposed system was tested
in various environments. It was demonstrated that the range of communication
was increased by 2.2 and 6.7 times as compared to a commercial whip antenna
for outdoor and indoor environments, respectively.
The organization of the dissertation is as follows:
6 Chapter 2 reviews the human body properties from VHF to X-band. Antenna
propagation on the surface of the human body is considered for in-body, on-body, and
off-body communication. The results determine which antenna type is proper for each
body-centric communication. Based on these propagation studies, a simplified human
torso model is suggested. It is shown that the simplified (homogeneous) model is
adequate for designing antenna as relates to off-body communications. This simplified
model is then employed for body-worn evaluation in the subsequent chapters.
Chapter 3 proposes and develops the channel decomposition method (CDM) for
body-worn antenna evaluation. The details of the CDM are introduced, and valida-
tions are provided by comparison to full wave simulations. The CDM is subsequently
used to find the optimal antenna locations and design a body-worn diversity system
for isotropic pattern coverage. To realize the diversity system, a prototype 4-channel
diversity module was fabricated. Indoor measurement results are also provided to
demonstrate the effectiveness of the proposed body-worn diversity system in rich
multi-path environments.
In chapter 3, the antenna locations to realize diversity are selected via an em- pirical method. Although this selection process provides some insights into efficient antenna location, it requires substantial simulation time to reach an optimal solution.
With the goal of reducing computation time, chapter 4 introduces a systematic design method for body-worn diversity antennas. First, the important parameters are iden- tified. This is followed by the design method ignoring the mutual coupling between antennas. Several design examples are also given to demonstrate the effectiveness of the approach. As part of the chapter, fundamental assumption of the method is also
7 justified based on full wave simulations. Measurements are also given to validate the
overall method.
Chapter 5 introduces the vest-mounted UHF body-worn antenna diversity system
(BWADS). This system is intended to provide a reliable communication system for law-enforcement officers with the goal of making the antennas unobtrusive. First, the antenna elements are designed and a compact high speed diversity module is devel- oped to implement the diversity system. The module automatically selects the best antenna among multiple ones in real time. After integrating the antennas with the diversity module, numerous field tests are performed in outdoor and indoor environ- ments. The benefits of the system are highlighted at the end of Chapter 5 with a comparison to a commercial whip antenna.
Chapter 6 suggests future research topics relating to the body-worn antenna sys- tem. First, a diversity system is proposed for a vehicular platform using the method proposed in chapter 4. Second, a vital sign monitoring system using body-worn an- tennas is suggested for monitoring the respiration and vital signs. Third, Body-worn antennas are proposed to harvest ambient RF power. These can generate dc power from ambient RF signals and provide electrical power to mobile devices.
8 Chapter 2: Human Body Model for Body-Centric Wireless Communication
2.1 Introduction
Because body-centric wireless communications always involve the interaction of electromagnetic waves with the human body, it is important to understand the body’s electromagnetic properties when we design body-worn antennas or implanted wireless sensors. As these properties also depend on signal frequency and on tissue types, it becomes necessary to consider different human models and properties for each frequency band and application.
In this regard, the relevant properties of the human body from VHF to X-band are studied and dominant propagation modes around the body are discussed. Based on these data, approximate human torso models are built for each propagation mode us- ing multiple concentric cylinders emulating the skin, muscle, organ, etc. The simplest model for each propagation mode is proposed for efficient calculation without loss of accuracy. Furthermore, the proposed human model will be adopted for designing the body-worn antenna system.
9 2.2 Review of Electromagnetic Properties of the Human Body
The electromagnetic properties of human body tissues affect the propagation,
reflection, attenuation, and other behaviors of electromagnetic fields around the body.
These properties depend strongly on the types of tissue and frequency. Temperature
and blood or fluid perfusion also affect these properties but these secondary effects
are normally not considered. We shall only consider relative permittivity (ǫr) and conductivity (σ). Because the body is so weakly magnetic, the relative magnetic permeability (µr) is assumed to be 1, except for special applications such as magnetic resonance imaging (MRI).
2.2.1 Numerical Expression of the Dielectric Properties for Human Tissues
The dielectric properties of body tissues have been studied by several authors
[17–22]. The most comprehensive and referenced study is [23]. Gabriel measured over 25 tissue types in the frequency range from 1 MHz to 20 GHz using an open- ended coaxial probe technique [24]. Most of these measurements were carried out with excised animal tissue from freshly killed sheep at 20◦C and 37◦C. In addition, accessible parts of the human body such as palm, sole and forearm skin were measured in vivo. To describe the frequency dependence of the tissue dielectric properties, a model based on the summation of 4-Cole-Cole expressions is used [23] and the relative dielectric constant is expressed as
4 ∆ǫm σi ǫr (ω)= ǫ∞ + + , (2.1) (1−αm) jωǫ m=1 1+(jωτm) ! 0 X
10 where ǫ∞ is the material’s relative permittivity at terahertz frequency; ǫ0 is the free
space permittivity; σi is the ionic conductivity; ǫm, τm, and αm are material parame-
ters for each dispersion region. Since the dielectric constant is generally complex, it
consists of real and imaginary components:
′ ′′ ǫr (ω)= ǫr (ω) − jǫr (ω) . (2.2)
Also, conductivity and penetration depth (or skin depth) of the tissue can be expressed as
′′ σ (ω)= ǫr (ω) · ǫ0 · ω, (2.3)
− 1 1 2 ′ ′ 2 2 1 µ ǫ0ǫr(ω) σ(ω) δ(ω)= 1+ ′ − 1 . (2.4) ω 2 ωǫ0ǫr(ω) ! The penetration depth is defined as the depth at which the inte nsity of the radi- ation inside the material falls to 1/e of the original value at the surface.
The parameters required to determine the dielectric properties of selected human
tissues are tabulated in Table. 2.1. Using these parameters along with equation (2.1)-
(2.4), we can find the important dielectric properties of human tissues of interest.
2.2.2 Dielectric Properties of Human Tissue
Fig. 2.1 shows the relative permittivity, conductivity, and penetration depth of
selected tissues of the human torso in the frequency range from 10 MHz to 20 GHz.
Fig. 2.2 also shows the same properties of relatively large organs inside the human
torso (heart, liver and lungs).
We can observe several important properties of the human body:
11 Table 2.1: Parameters required to find human tissue dielectric constant and conduc- tivity [23].
ε∞ σi ∆ε1 τ1 (ps) α1 ∆ε2 τ2 (ns) α2 ∆ε3 τ3 (µs) α3 ∆ε4 τ4 (ms) α4 Bone - cancellous 2.5 0.07 18 13.263 0.22 300 79.577 0.25 2.0e+4 159.155 0.2 2.0e+7 15.915 0 Bone - cortical 2.5 0.02 10 13.263 0.2 180 79.577 0.2 5.0e+3 159.155 0.2 1.0e+5 15.915 0 Fat 2.5 0.01 3 7.958 0.2 15 15.915 0.1 3.3e+4 159.155 0.05 1.0e+7 7.958 0.01 Heart 4 0.05 50 7.958 0.1 1200 159.155 0.05 4.5e+5 72.343 0.22 2.5e+7 4.547 0 Liver 4 0.02 39 8.842 0.1 6000 530.516 0.2 5.0e+4 22.736 0.2 3.0e+7 15.915 0.05 Lung - deflated 4 0.2 45 7.958 0.1 1000 159.155 0.1 5.0e+5 159.155 0.2 1.0e+7 15.915 0 Lung - inflated 2.5 0.03 18 7.958 0.1 500 63.662 0.1 2.5e+5 159.155 0.2 4.0e+7 7.958 0 Muscle 4 0.2 50 7.234 0.1 7000 353.678 0.1 1.2e+6 318.31 0.1 2.5e+7 2.274 0 Skin - dry 4 0 32 7.234 0 1100 32.481 0.2 0.0e+0 159.155 0.2 0.0e+0 15.915 0.2 Skin - wet 4 0 39 7.958 0.1 280 79.577 0 3.0e+4 1.592 0.16 3.0e+4 1.592 0.2
1. Human body tissues can be classified into two types: the first type has low
water content (i.e. fat, bone, etc.), low dielectric constant, and low loss; the
second type has high water content (i.e. skin, muscle, heart, etc.), high dielectric
constant, and high loss (see Fig. 2.1).
2. Almost all internal organs with high water content tissue have very similar EM
properties. Only an inflated lung has lower dielectric constant and conductivity
due to increased air content (see Fig. 2.2).
3. The conductivity (or penetration depth) of all tissues increase (or decrease)
as signal frequency increases. In other words, human body becomes lossy and
incident waves are rapidly attenuated at high frequencies (see Fig. 2.1 and 2.2).
These observations about human tissues led to some insights for body-worn an- tenna applications. Here are some examples:
12 3 10 Skin-dry Fat ') r Muscle ε 2 Lung-deflated 10 Bone-cortical
1 10 Relative permittivity ( permittivity Relative
0 10 1 2 3 4 10 10 10 10 Frequency [MHz] (a)
2 10
1 10
0 10
-1 10 Conductivity [S/m] Conductivity
-2 10 1 2 3 4 10 10 10 10 Frequency [MHz] (b)
3 10
2 10
1 10 Penetration depth [mm] depth Penetration
0 10 1 2 3 4 10 10 10 10 Frequency [MHz] (c)
Figure 2.1: Electromagnetic properties of skin, fat, muscle, and lung (deflated); (a) relative permittivity (real value), (b) conductivity, and (c) penetration depth in mm.
13 3 10 Heart Liver ') r
ε Lung-deflated
2 Lung-inflated 10
1 10 Relative permittivity ( permittivity Relative
0 10 1 2 3 4 10 10 10 10 Frequency [MHz] (a)
2 10
1 10
0 10
-1 10 Conductivity [S/m] Conductivity
-2 10 1 2 3 4 10 10 10 10 Frequency [MHz] (b)
3 10
2 10
1 10 Penetration depth [mm] depth Penetration
0 10 1 2 3 4 10 10 10 10 Frequency [MHz] (c)
Figure 2.2: Electromagnetic properties of heart, liver, lung (deflated), and lung (in- flated); (a) relative permittivity (real value), (b) conductivity, and (c) penetration depth in mm.
14 1. The human torso can be modeled by multi-layer structures using skin, fat, mus-
cle, organs, and bone in the order they appear in human body layers. Since the
organs have EM properties similar to muscle, they can be omitted in approxi-
mate human models, with the exception of inflated lungs.
2. The different EM properties of lungs in different respiratory state (exhalation
and inhalation) make it possible to sense the human’s respiratory rate and vital
sign.
3. In-body communication (between an outer antenna and implanted sensor) should
be performed at lower frequency due to loss characteristics of the human body.
Previously, in-body communication has been performed at 13.56 MHz via in-
ductive coupling and at 402-405 MHz, allocated for medical implanted commu-
nication systems.
4. Penetration depth in real human bodies will be much less than the minimum
penetration depth, because the incident waves experience a multiple reflections
at skin - fat - muscle interfaces (having relative permittivity 46.1 - 5.6 - 56.9 at
433 MHz).
The EM properties of the skin, muscle and fat (representing high water content and low water content) are also tabulated in Table 2.2 at selected frequencies.
2.3 Propagation Modes for Body-Centric Wireless Commu- nications
Body-centric wireless communications can be divided into three domains [3]:
1. Off-body communication (human-to-human or human-to-base station).
15 Table 2.2: Electromagnetic properties of skin (dry), muscle, and fat at selected fre- quencies (30 MHz, 433 MHz, 915 MHz, 2.45 GHz, and 12 GHz). Tissues EM properties f=30 MHz f=433 MHz f=915 MHz f=2.45 GHz f=12 GHz K Relative permittivity ( r’) 152.9 46.08 41.33 38.01 29.33 Skin Conductivity (σ) [S/m] 0.3415 0.7017 0.8715 1.464 10.33 - dry Penetration depth (δ) [mm] 222 54 40 23 3 K Relative permittivity ( r’) 8.11 5.57 5.46 5.28 4.46 Fat Conductivity (σ) [S/m] 0.033 0.042 0.051 0.105 0.728 Penetration depth (δ) [mm] 614 304 242 117 16 K Relative permittivity ( r’) 91.81 56.87 55.0 52.73 40.1 Muscle Conductivity (σ) [S/m] 0.658 0.805 0.948 1.739 13.540 Penetration depth (δ) [mm] 127 52 42 22 3
2. on-body communication (multiple antennas placed on the same body surface).
3. in-body communication (implanted sensor and antenna placed on the body
surface).
Since each communication domain utilizes different propagation modes around the body, it is important to characterize each propagation mode. We can categorize the propagation modes for each body-centric wireless communication as follows:
1. Space wave that propagates away from the body (for off-body communication).
2. Creeping wave that propagates along the body surface (for on-body communi-
cation).
3. Penetrating wave that propagates into the body (for in-body communication).
To analyze each propagation mode, model of infinite planar human tissue layers
(emulating human torso) with three types of radiating sources were considered as
16 shown in Fig. 2.3. Three types of tissue are used. To calculate the penetrating wave
and reflected wave, an ideal plane wave source is assumed as shown in Fig. 2.3(a). For
other propagating waves, vertical and horizontal dipole antennas are used as shown in
Fig. 2.3(b)-(c). Each layer is assumed to be infinite in size to avoid other scattering
and diffraction effects. The thickness of each tissue layer is t1 = 2.4 mm, t2 = 13.5 mm, and t3 = 168 mm [25]. The total thickness of the body layer is 200 mm. The
EM properties of the tissues are specified in Table 2.2.
Plane wave Q/2 dipole antenna r ) + = jk0z r E xE0e k
d d zˆ 1 2
t1 Skin xˆ Skin Skin Fat Fat Fat t2
t3 Muscle Muscle Muscle
t2 Fat Fat Fat t1 Skin Skin Skin
(a) (b) (c)
Figure 2.3: Infinite human tissue layers (skin - fat - muscle - fat - skin) and radiating sources near the body surface; (a) plane wave source, (b) vertical dipole antenna, and (c) horizontal dipole antenna.
2.3.1 Penetrating and Reflecting Wave Analysis
To study the penetrating wave and reflected wave behavior of the human body,
+jk0z the plane wave represented by E~ =xE ˆ 0e (E0 = 1 V/m) is directed along the
-z axis onto an infinite planar tissue layers located from z=0 mm to z=-200 mm as
17 shown in Fig. 2.3(a). The magnitude of the E field is calculated using FEKO and
is shown in Fig. 2.4 from z=300 mm to z=-300 mm. One can readily observe large
reflections of incoming waves at the boundary (z=0 mm) between air and human
tissue (skin). Standing waves are formed in the air region. The magnitude of the
reflection coefficient is calculated using the standing wave ratio (SWR) and expressed
by
SWR − 1 |Γ| = . (2.5) SWR + 1
Table 2.3 summarizes the calculated SWR, reflection coefficient, and transmitted
E-field at each frequency. We note that most of the incoming waves are reflected at the boundary and incident waves cannot be efficiently transmitted into the body for in-body communication. The reflected wave also cancels the off-body radiation
Table 2.3: Calculated SWR, reflection coefficient, and transmitted E-field at each frequency. f=30 MHz f=433 MHz f=915 MHz f=2.45 GHz f=12 GHz SWR 29.29 5.84 2.45 10.89 4.94 Γ 0.934 0.707 0.421 0.832 0.663
Et [V/m] 0.0458 0.016 0.001 0.00005 0
field when the radiation source is very close to the body (relative to its wavelength)
because the phase of the reflection coefficient is close to 180◦ (see E-field at z=0+ mm in Fig. 2.4(a)) thereby reducing the efficiency of the body mounted antenna. Overall, only a small amount of the E-field can enter the body and only small ratio of the field can transfer throught the body at higher frequency as described in Table 2.2. From
18 twoeosre ein b omi iwfcsdo ua tissu human on focused o view boundary zoom-in the (b) on region, wave observed reflecting whole and at Penetrating 2.4: Figure
IExI [V/m] skin (2.4mm) IExI [V/m] 2.45 GHz 2.45 915 MHz 915 MHz 433 2.45 GHz 915 MHz 433 MHz 30MHz 12GHz 30 MHz 12 fat (13.5 fat mm) GHz muscle(168 mm) z z [mm] 19 (b) (a) z [mm] fat (13.5 fat mm) ua ise:(a) tissues: human f skin (2.4 skin mm) layers. e Planewave ( E 0 =1 V/m) =1 these two observations, we can infer the design tips for in-body communication and
off-body communication antennas. The antenna for in-body communication must
be mounted directly on the body (without any gap) to reduce the reflection at the
boundary between air and skin. For off-body communication the gap between antenna
and body should be carefully chosen to avoid the severe gain drop caused by reflection
cancellation. This will be shown later.
2.3.2 Creeping Wave Analysis
Radiations from a vertical and a horizontal (relative to the body surface) dipole
antennas are compared to study the creeping wave mode at 2.45 GHz as depicted in
Fig. 2.3(b)-(c). Also, the vertical dipole antenna in free space is compared with one
on the body. All the antennas are located at z=0.4 λ and the body layers are placed between z=0 mm and z= -200 mm. Fig. 2.5(a)-(c) compares the near fields for (1) vertical dipole, (2) horizontal dipole near body, and (3) vertical dipole in free space.
Fields from both vertical antennas propagate well along the body surface, while fields from the horizontal antenna do not. Fig. 2.5(d) plots field amplitude along the body surface as a function of lateral distance from antenna. At y=5 λ, the calculated normalized E-field magnitudes are -40, -52, -42 dBV/m for the case Fig. 2.5(a), (b),
(c), respectively. It shows that the vertical antenna near the body produces stronger
field transmission than that of the horizontal antenna. It is also slightly stronger than the same antenna in free space. These observations can be explained from reflection when waves are obliquely incident as shown in Fig. 2.6 and the reflection coefficients for parallel and perpendicular polarized waves are expressed by (2.6) and (2.7) [26], respectively.
20 1 1 ENorm [dBV/m] 0
0 −10 0
− 20 − 1 − 30 −1 z in wavelength z − 40 in wavelength z
− 2 − 50 − 2 − 5.0 0 1 2 3 4 5 − 5.0 0 1 2 3 4 5 y in wavelength y in wavelength (a) (b)
1 0 ver. dipole on the body -10 hor. dipole on the body ver. dipole in free space 0 -20
-30 [dBV/m]
−1 Norm -40 E
z in wavelength z -50
− 2 -60 0 1 2 3 4 5 − 5.0 0 1 2 3 4 5 y in wavelength y in wavelength (c) (d)
Figure 2.5: Near field plot (E-field) around the radiating antenna and human body layer: (a) vertical dipole antenna near the body surface, (b) horizontal dipole antenna near the body surface, (c) vertical dipole antenna in the free space, and (d) E-field magnitude along y-axis at z=0.4 λ (center of the antenna).
η2cosθt − η1cosθi Γ// = . (2.6) η2cosθt + η1cosθi
η2cosθi − η1cosθt Γ⊥ = . (2.7) η2cosθi + η1cosθt In the case of parallel polarization (Fig. 2.5(a) and 2.6(a)), the reflection coefficient
◦ converges to +1 when the incident angle increases to θi ≈ 90 and the incident and reflected waves add. As a result the body boundary helps to transfer EM power along the surface. In the case of perpendicular polarization, the reflection coefficient
21 r r r E// r ˆr ˆr H ⊥ k k⊥ // r r ⊗ Et ˆt E t ˆt r // k// r ⊥ k⊥ r r H // r E⊥ r θ θ t θ θ t r t H // r t H ⊥ r θ θ i i i E// ˆi r ˆi k// i k⊥ r E⊥ i H // r i H ⊥ Medium 1 Medium 2 Medium 1 Medium 2 η η η η 1 2 1 2 (a) (b)
Figure 2.6: Plane wave incident obliquely on a plane dielectric boundary: (a) parallel polarized wave, (b) perpendicular polarized wave.
◦ converges to -1 when the incident angle increases to θi ≈ 90 (Fig. 2.5(b) and 2.6(b)), causing the field cancellation along the boundary. Thus, a vertically (relative to the body surface) polarized antenna should be selected for on-body communication.
Higher frequency (i.e. 2.45 GHz) is also preferred for this application because of the physical size limitations on the antenna.
2.3.3 Space Wave Analysis for Off-Body Communication
Space waves emitted by body-worn antennas radiating away from the body consist of a direct radiating field from the antenna and a reflected wave from the body surface.
Thus, it is affected by the gap between the antenna and human body (reflecting surface). To investigate the effects of the gap, I considered a horizontal dipole antenna for off-body communication as shown in Fig. 2.3(c) placed at different distances (d2) from the body surface. Fig. 2.7 shows the electric far-field patterns at 433 MHz for
22 various antenna distances from the body surface. The free space antenna is used as a reference.
θ = 0 20zˆ in free space xˆ d =0.05λ 18 2 d =0.1λ 2 16 d2 λ
¢ d =0.2 θ = − ¡ θ = 90 2 90 Skin Fat d =0.3λ 14 2 Muscle d =0.4λ Fat 2 12 Skin d =0.5λ 2 10
8
6
Electric Electric farfield magnitude [V] 4
2
0 -90 -60 -30 0 30 60 90 Elevation angle, N [deg]
Figure 2.7: Electric far-field pattern along the elevation angle for different distances between antenna and human body.
In this case, one can observe three important phenomena:
1. There are similar radiation patterns for distances up to d2 = 0.2 λ. When
the distance is greater than 0.3 λ, the main beam is broadened or split in the
horizontal direction. This can be explained by an array model consisting of
the original antenna and an image antenna at z = -d2. Since the magnitude of
23 the reflection coefficient is less than one and its polarity is negative, the image
antenna has a smaller excitation voltage with reversed polarity.
2. The strongest radiation field is observed at d2 = 0.2 λ in the main beam di-
rection. It is also stronger than the reference antenna (in free space) because
the reflected wave is added to the direct radiation field at this distance and the
phase of the reflection coefficient is approximately π (negative polarity). This is
consistent with Fig. 2.4(a) in which the first maximum of the total field occurs
at approximately 150 mm (≈ 0.22 λ at 433 MHz) as shown in Fig. 2.4(a).
3. The radiation field is weaker when the antenna is close to the body, because the
reflected wave and the direct radiation field cancel.
From the above, one can determine the best antenna location for efficient off- body communication. The antenna should be placed at a certain electrical distance depending on the effective reflection coefficient of the human body. Since the human body can be treated as a metallic surface (large reflection with negative polarity), a distance of λ/4 is usually selected. However, this cannot be applied to wideband antennas and low-frequency antennas because the electrical distance associated with the frequency and the physical distance corresponding to λ/4 is too large to be prac- tical in low frequency applications. In this case, it is necessary to reduce the reflection from the body using wideband electromagnetic band gap (EBG) surfaces to reduce the back radiation [13].
24 2.4 EM Human Torso Model for Body-Centric Wireless Com- munication
Many numerical human models have been developed for computational simula- tions of body-centric wireless communication. For accurate calculation of body effects, an anatomically realistic human model composed of a set of fine voxels is desired. Hu- man voxel models for the entire body have been developed by Dimbylow [27]. This model was based on magnetic resonance images (MRIs) of an adult male and was segmented into 37 different tissues. Subsequently, higher-resolution human models have been developed by other researchers [28–30]. Anatomical human models require large computational resources to generate accurate solutions to EM problems.
In this light, several approximate human torso models are proposed for in-body, on-body, and off-body communication. A concentric elliptical cylinder composed of multiple tissue layers is considered, and the simplest model having reasonably small error in EM analysis is selected for each application.
2.4.1 Human Torso Model for In-Body Communication
To establish a human torso model, the communication between on-body antenna and a sensor implanted on the heart is considered. The calculation is performed at
433 MHz because the antenna size is realizable and the penetration depth at this frequency is also suitable for implanted sensor applications.
First, a reference human model is constructed based on the thorax cut image from the visible human project [31] as shown in Fig. 2.8. The reference model consists of skin, fat, muscle, lung, heart, and spinal cord. The outer diameter of the model is 200
25 Muscle
Spinal cord Lung Lung
Heart Fat
Skin
(a) y
x Spinal cord Heart Lung Lung Lung Heart
z Muscle Fat Spinal cord
Skin y (b) (c)
Figure 2.8: The reference human torso model: (a) thorax cut image from visible human project, (b) top view, (c) front view of torso model.
mm (along the short axis) and 320 mm (along the long axis). To evaluate communi- cation performance, the penetrating field is calculated inside the human model when the plane wave is incident from the +x direction. The calculated E-fields from the reference model and the approximate concentric 4-layer model (skin-fat-muscle-lung) are compared in Fig. 2.9. Overall, both results are in a good agreement except in the region between x = 0 mm to x = 48.4 mm. The difference (less than 1 dB) is caused by
EM property differences between heart and lung (deflated). This result shows that
E-fields inside different organs vary slightly, but this error is small. Thus, 4-layer
26 Skin Fat Skin
Muscle Muscle Lung Lung 0 48 4. Fat 84 1. 97 6. 100 mm
(a) (b) 0
-5 y
[dBV/m] -10 x Norm E
-15
Reference model 4-layer model -20 0 10 20 30 40 50 60 70 80 90 100 x [mm]
(c)
Figure 2.9: (a) Top view and (b) front view of the approximated concentric 4-layer model, and (c) comparison of normalized E-field magnitude inside the human body for two models.
27 model (skin-fat-muscle-lung) is sufficient for simulation of in-body communication antennas.
2.4.2 Human Torso Model for On-Body Communication
The main propagation mode for on-body communication is the creeping waves which are not significantly affected by the core medium (organ in this case). It is noted that the creeping wave cannot penetrate into the core medium at higher fre- quency as shown in Fig 2.5(a). Three human torso models (homogeneous, 3-layer,
4-layer concentric cylinder) are constructed and compared using transmission coeffi- cients between two antennas mounted on the same body as shown in Fig. 2.10. The homogeneous model is a phantom whose relative permittivity and conductivity are equivalent to 2/3 times that of muscle. The 4-layer model is the same as one shown in Fig. 2.9 and the 3-layer model is similar to the 4-layer model except that the lung is replaced with muscle. Fig. 2.10(d) shows the transmission coefficients from
2.4 to 2.5 GHz between two antennas, it is noted that the 4 and 3-layer model show excellent agreement but the homogeneous model has about 2 dB higher values. In the 4 and 3-layer models, fat has high contrastive EM properties to skin and muscle.
Thus penetrating waves into the body layer do not propagate significantly beyond the fat layer. This results in a loss of transmission power via creeping wave in the 4 and 3-layer models. Therefore, it is necessary to include skin, fat, and muscle layers in an approximate human model for on-body communication, but organ tissue is not important, especially at higher frequency. Finally, the 3-layer model is selected as the simplest accurate human model for on-body communication.
28 Fat Fat Skin
Lung Muscle
Muscle
0.25λ 0.48λ Skin dipole
(a) (b)
Homogeneous body
(c) -30
-32
-34
[dBV/m] -36 21 S
-38
-40 2.4 2.45 2.5 Frequency [GHz]
(d)
Figure 2.10: Top view of (a) 4-layer model, (b) 3-layer model, (c) homogeneous model, (d) transmission coefficients between two antennas for each model.
29 2.4.3 Human Torso Model for Off-Body Communication
Space waves are the main propagation mode for off-body communication. The space wave radiating in the off-body direction is composed of a direct radiated wave from the source and a reflected wave from the body surface. Thus the reflection coefficient of the body is important in constructing a human model for off-body communication.
To evaluate the human model for off-body communication, radiation patterns for the homogeneous model and the 3 layer model at 433 MHz are calculated and compared as shown in Fig. 2.11. Both models are the same as the model presented in
Fig. 2.10(b) and (c). The λ/2 dipole antennas are located in the front of the human model at 0.1 λ distance and parallel to the body surface. The azimuth and elevation
radiation patterns are shown in Fig. 2.11(c) and (d). Overall, they are in good
agreement except for back radiation. They also have a small difference in gain (less
than 0.5 dB) in the main beam direction, but this can be compensated by modifying
the EM properties of the homogeneous body. In conclusion, the homogeneous model
with proper EM properties is sufficient to represent the human body model for off-
body communication.
30 z
x x
y 0.1λ
(a) (b)
φ Gain [dB] θ Gain [dB] Homogeneous model 3-layer model
(c) (d)
Figure 2.11: (a) Homogeneous human model, (b) 3-layer model (c) azimuth radiation pattern (x-y plane), (d) elevation radiation pattern (x-z plane).
2.5 Summary
In this chapter, the human body model was investigated for body-centric wireless communication. First, electromagnetic (EM) properties of the human body were re- viewed from VHF to X-band. EM properties of the human body depend on frequency and tissue types. Human body tissues were categorized into high water content tis- sue (high dielectric constant and loss) and low water content tissue (low dielectric
31 constant and loss) according to their EM properties. Fat, bone and inflated lung are grouped with low water content tissue while skin, muscle and other main organs are grouped with high water content tissue.
Second, wave propagation modes around the body were studied for in-body, on- body, and off-body communication, respectively. Also, proper antenna type and mounting configuration on the body were suggested for each body-centric communi- cation.
Finally, approximate numerical human torso models using concentric elliptical cylinders with multi-tissue layer were suggested for each body-centric communica- tion. A 4-layer (skin-fat-muscle-organ) model, and a 3-layer (skin-fat-muscle), homo- geneous model were suggested for in-body, on-body, off-body communication, respec- tively. All models were simplified for calculation efficiency without loss of accuracy.
In the following, horizontally placed dipole antennas along the human body and the homogenous human model will be used in body-worn antenna design for off-body communication.
32 Chapter 3: Body-Worn Antenna Evaluation Method
3.1 Introduction
Electromagnetic field interactions between the human body and antennas have been documented in the literature [4–10]. These studies show that the antenna’s impedance depends strongly on its position and its distance from the body and con- clude that the body effects cause efficiency losses. These effects are intensified for conformal wearable body-worn antenna systems, making it essential to explore viable solutions to overcome pattern deformation and gain reduction. To this end, some au- thors have proposed metallic or electromagnetic bandgap (EBG) surfaces to reduce back radiation and increase efficiency [11–13], reporting 63∼80 % efficiency at 2.45
GHz and 2.8 GHz. However, EBG surfaces are narrowband and difficult to adapt to lower frequencies, and may also cause undesirable radiation patterns. Therefore, they are not suited for broadband UHF/VHF wearable communications.
As an alternative, diversity techniques can offer increased data throughput for body-worn antennas. Among previous works in this area, Serra et al. [14] considered dual antenna diversity for the 2.45 GHz ISM band. However, these authors focused on the connectivity between antennas on the same body, and limited their study to two antennas. Other papers [15, 16] considered antenna diversity for off-body communication, but their conclusions were restricted to specific indoor environments.
33 As body-centric communications are associated with continuously changing environ-
ments, a generalized antenna diversity evaluation method is being sought for designing
body-worn antennas.
In this respect, this chapter introduces an efficient and versatile methodology for designing multiple antenna diversity systems for any environment. The multi- antenna system is integrated into a unique module that employs a diversity decision process to maximize channel capacity. The underlying design utilizes a new channel decomposition technique to account for interaction with the nearby environment.
The structure of the chapter is as follows. First, we introduce the antenna element used throughout this chapter. Next, we describe the Channel Decomposition Method
(CDM) for capacity evaluation and demonstrate its validity via simulations. Subse- quently, CDM is used to design a body-worn system for omni-directional coverage.
As such, optimum mounting positions for the body-worn antennas are determined to provide isotropic pattern coverage using the minimum number of antennas. In the last section, we provide the details of implementing a diversity module to retrofit ex- isting single channel radios with diversity capability. Measurements are presented and compared with simulations to evaluate the body-worn diversity system in a multipath environment.
3.2 Body-Worn Antenna Element
Before presenting the channel decomposition method, it is important to introduce the body-worn antenna used throughout this chapter. The subject antenna is a scaled version of the Asymmetric Meandered Flare (AMF) body-worn element presented in
[32]. It is modified to operate at 560 MHz instead of 300 MHz, the center frequency
34 Figure 3.1: AMF antenna structure and dimensions for operation at 560 MHz.
of the original antenna (see Fig. 3.1 for dimensions). Fig. 3.2 depicts the antenna performance for on-body operation. Specifically, Fig. 3.2(a) presents the simulated reflection coefficients at various distances from the body and Fig. 3.2(b) provides a comparison between measurements and calculations for the pattern at 550 MHz.
These results show that the body presence affects three antenna characteristics: gain, input impedance, and radiation pattern. Diversity techniques, presented in the next section, are intended to alleviate antenna performance compromises due to the body presence.
3.3 Evaluation Method for Diversity System
As mentioned in the previous section, the negative effects of the body’s presence can be reduced by employing multiple antennas. An effective way of evaluating which antenna configuration offers the highest advantages, is to study the increase in channel capacity for a given communication channel.
35 0
-5
-10
-15
-20 S11 S11 [dB] -25 2mm gap 12mm gap -30 free standing ant -35 400 450 500 550 600 650 700 Frequency [MHz]
(a) 0 zˆ -5 . yˆ xˆ -10 -15 gain ¤ -20
gain £ -25 -30
-35
Realized Realized Gain [dB] -40
-45 Solid line : Simulation Marked line : Measurement -50 0 30 60 90 120 150 180 210 240 270 300 330 360 Azimuth angle [deg]
(b)
Figure 3.2: Performance of the AMF antenna element when placed near the body’s torso: (a) the simulated reflection coefficients S11 for 3 distances from the body, (b) measured and simulated radiation patterns in the horizontal (x-y) plane at 550 MHz (antenna is 12 mm from the body’s surface).
36 For a multi-antenna diversity system, a generalization of the channel capacity
[33, 34] is given by
C SNR H∗H bits/sec = log det 1+ Rx , (3.1) B 2 Norm2 Hz
where B denotes bandwidth, SNRRx refers to the average signal-to-noise ratio (SNR)
at the receiving antenna (see Fig. 3.3), and H is the channel transfer function matrix.
In the case of a single antenna transmitter, like the one considered in this chapter,
the channel transfer function matrix becomes a vector given by
h11 . H = . , (3.2) hNR1 th where hi1 represents the ratio of the voltage at the i receiver over that at the transmitter. Since this ratio includes the channel propagation effects, the H matrix is typically normalized by the Frobenius norm to exclude path losses. However, as path losses don’t depend solely on the environment but also on antenna location, it is important not to exclude the antenna characteristics from capacity evaluations.
Therefore, instead of using the Frobenius norm, we considered the average SNR over all different antenna configurations to fairly compare channel capacities. Thus, the subject norm used in (3.1) is of the form:
K N (NR)k 1 2 1 Norm = v |hi1| . (3.3) uK · N (NR) u k=1 n=1 i=1 k u X X X t In this, K denotes the number of different configurations, N is the number of
samples per configuration, and (NR)k refers to the number of receive antennas of the kth configuration.
37 Communication channel
Human body channel Propagation channel
h1 Diversity h module 2
Rx Tx h3 SNR Re c SNR Rx SNR Tx
Figure 3.3: Communication channel model for the CDM.
To determine the channel transfer matrix, we introduce a Channel Decomposition
Method (CDM). Specifically, the communication channel is decomposed into two regions as shown in Fig. 3.3:
1. the human body, including the antennas (denoted as the human body channel
in Fig. 3.3) and
2. the surrounding environment (denoted as the propagation channel in Fig. 3.3).
Below, we introduce and validate the CDM to evaluate the channel transfer func- tion. As such, we can effectively characterize the body-worn antenna diversity system for any environment using channel statistics. As can be understood, the proposed
CDM can be readily applied to diversity systems for platforms other than the human body.
38 3.3.1 Channel Decomposition Method for Diversity Evalua- tion
Conventionally, the entire channel environment is used to evaluate the channel capacity [15, 16]. However, since the human body is a rapidly changing platform as opposed to a building or even a vehicle, this approach is not attractive for general channel evaluation. Instead, the communication channel is divided into the human body and propagation channel components, shown in Fig. 3.3. As expected, the body’s movement causes rapid variations in the antenna parameters (impedance, gain, pattern, etc.) affecting performance. On the other hand, the propagation channel can vary over time, but its characteristics do not change as abruptly. Therefore, one can evaluate the propagation channel statistics separately. To do so, a ray tracing ap- proach for outdoor channel environments can be adopted. However, for body effects, a more accurate method such as the Method of Moments (MoM) or Finite Element
Method (FEM) must be used. Alternatively, measurements can be taken in place of simulated data for validation or for improved accuracy. That is, different analysis methods can be used for evaluating the channel components, providing flexibility in analyzing communication channel characteristics.
Fig. 3.4 outlines the proposed channel decomposition method. As inferred above, the human body channel consists of the body-mounted antennas in absence of other environmental obstacles or scatterers. The human body channel contribution is eval-
th uated by collecting the far-field pattern (E~i) for the i antenna on the body with all other antennas terminated at matched loads. Concurrently, the propagation channel or environmental contributions can be estimated by means of field measurements, large scale simulations or even tabulated statistical models assembled in the form of
39 Human body channel Propagation channel
Channel statistics Simulation / Measurement Polarization angle distribution (ϕ ) Amplitude distribution ( A)
Data building (far field pattern) Monte-Carlo method r Regenerate the A & ϕ E = θˆ()E + φˆ()E rand rand i θ i φ i from channel statistics
Data modification r = ⋅ θˆ ϕ +φˆ ϕ hi Ei ( cos rand sin rand )Arand
Selection diversity H = max{ h , ⋅⋅⋅ ,h } 1 N R
Channel capacity calculation C HH H bits / s = log det I + SNR 2 Rx 2 B Norm Hz
Statistical analysis Find distribution of capacity Find outage capacity at 90% reliability
Figure 3.4: Flowchart of the diversity evaluation process using the CDM.
40 polarization angle (ϕ) and amplitude (A) distributions. Once the body and propa-
gation channel are characterized, they can be combined to form the channel matrix
components (hi). Generally, the environment (propagation channel) contributions
modulate the human body pattern data via a Monte-Carlo process and can be ex-
pressed as
hi = E~i · θˆcos ϕrand + φˆsin ϕrand Arand, (3.4) where ϕrand is the polarization angle and Arand is the field amplitude random variables
generated by a Monte-Carlo process (the latter draws random values ϕrand and Arand
based on the propagation channel statistics). These random variables reflect the
impact of polarization mismatch and incoming attenuation due to the multipath
environment, respectively. Also, as usual,
~ ˆ ˆ Ei = θ (Eθ)i + φ (Eφ)i , (3.5)
where Eθ and Eφ are the θ and φ components of the antenna E-field pattern on the
body, respectively. The channel capacity is subsequently calculated using (3.1) and
employed to estimate diversity performance and communication reliability.
In this section, the human body contributions are evaluated via the simulation
package FEKO (MoM) [35]. Specifically, field patterns (500 MHz – 600 MHz at 10
MHz increments) were calculated from φ=0◦ to 360◦ at 2◦ increments in the horizon-
tal plane. The human body used in these simulations was modeled as a homogeneous
body, having a relative permittivity of ǫr=56.7 and conductivity, σ=0.94. It was placed on a ground which was emulated by ǫr=9 and a loss tangent of tan δ=0.01. To accurately apply the CDM, the propagation channel should not affect the body-worn antenna impedance. This typically occurs when nearby objects are more than 0.4λ
41 away from the body at the lowest frequency of operation. This distance sufficiently re-
duced multi-bouncing effects between the human body and the adjacent objects. We
also note that use of a homogeneous human model (instead of a more detailed layered
model) should not affect accuracy because our interest lies in off-body performance.
At the employed frequencies (λ ∼60 cm) the outer layer tissue losses dominate, block- ing any interaction effects from the interior human organs as discussed in Chapter
2.
Next, we proceed to validate the CDM by simulating the scattering environment and evaluating two body-worn antenna configurations.
3.3.2 CDM Validation
To validate the proposed CDM, 2-antenna and 4-antenna diversity configurations were chosen and placed in a simple channel environment (emulating a multipath channel) as illustrated in Fig. 3.5. Specifically, the channel environment consists of two lossy brick walls (ǫr=4, σ=2) and three lossless dielectric spheres (ǫr=36)
with vertically polarized transmitters, evenly distributed from 0◦ to 360◦ in 2◦ step over the horizontal plane (in far field region). For CDM, the human body (chan- nel 1) and propagation channel (channel 2) were simulated separately and statistics
(namely incident wave polarization and incident field amplitude) were obtained from simulations.
To model the effects of channel 2 only, the human body and AMF antenna ele- ments were replaced by a set of λ/2 vertical dipoles. Their far field patterns were calculated at all azimuth angles in the horizontal plane, and polarization/amplitude data were extracted. Subsequently, appropriate distribution functions were fitted to
42 CHANNEL 1 dielectric sphere
lossy brick zˆ wall yˆ × xˆ
infinite ground
(a) side view
infinite ground
Human
CHANNEL 1 yˆ ˆ zˆ . x
(b) top view
Figure 3.5: Simple channel model for validating CDM.
43 the simulated data based on a Chi-Squared test. The polarization angles (ϕrand) and
amplitudes (Arand) were then pseudo-randomly generated using a Monte-Carlo pro-
cess. The generated statistical data, shown in Fig. 3.6, allow us to conclude that the
best fit for the amplitude distribution is a Rice distribution given by
r r2 + A2 Ar p (r)= exp − I , (3.6) σ2 2σ2 0 σ2
where A is the mean, σ is the standard deviation and I0 is the zeroth order modified
Bessel function of the first kind. This result was expected since the Rice distribution
is best at describing sparsely populated scattering environments with dominant line-
of-sight contributions (as the one illustrated in Fig. 3.5). For our specific calculations
we found that A = 1.245 V/m and σ = 0.541 V/m.
The polarization angle was instead fitted to an exponential distribution given by
1 r p (r)= exp − . (3.7) Γ Γ a a ◦ ◦ Here, the mean polarization angle (Γa) is 3.08 with 0 denoting vertical polarization.
From Fig. 3.6(b), it can be inferred that the given channel does not significantly
affect the incident wave’s polarization angle.
In addition to implementing the CDM, we also simulated the human model and
the environment as a single channel for validation. To adequately represent human
activity, the human model was rotated about its axis in 10◦ increments. The collected
far field data were then used to evaluate the channel capacity to be compared to the
CDM results.
Fig. 3.7 shows the complementary cumulative distribution function (CCDF)
curves of the channel capacity obtained via the CDM and full geometry simulation.
It is evident that the CCDF curves from the two approaches are in good agreement.
44 0.05 Bar : simulation data Line: fitted Rice distribution 0.04 error χ 2 = 0.0098
0.03 Probability 0.02
0.01
0 0 1 2 3 4 E-field amplitude [V/m]
(a)
Bar : simulation data 0.25 Line: fitted exponential distribution
0.2 error χ 2 = 0.0146
0.15
Probability 0.1
0.05
0 0 10 20 30 40 50 60 70 80 90 Polarization angle [deg]
(b)
Figure 3.6: (a) Amplitude distribution and (b) polarization angle distribution of the incident wave in the environment depicted in Fig 3.5.
45 1.1 1 4-antenna 0.9 2-antenna 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 Solid line : no decomposition Probability (capacity abscissa) Probability> (capacity Dashed line: CDM 0 0 1 2 3 4 5 6 7 8 9 10 Channel capacity (bps/Hz)
Figure 3.7: Comparison of the channel capacity CCDF calculated using the Channel Decomposition Method with direct simulations (no decomposition).
This holds for the 2-antenna and 4-antenna diversity configurations. Of course, the
CDM was much faster by more than a factor of 10 (5.7 hours vs. 66.7 hours for a
2.21 GHz dual core CPU).
In the next section, we proceed to employ the CDM to evaluate the optimum antenna mounting configuration aimed at maximizing channel capacity in a non- scattering environment. The obtained results can be easily used to evaluate the systems in a more complex environment (by adding the contribution from other sta- tistical models).
46 3.4 Multiple Body-Worn Antenna Diversity System
Having established the CDM validity, we next proceed to evaluate specific body- worn diversity configurations. To begin with, we note that the high dielectric constant and losses due to the human body tissue [23] affect off-body radio communication in three ways: 1) the loss degrades antenna efficiency and gain, 2) the body proximity detunes the operational frequency, and 3) the body presence causes pattern defor- mation and polarization detuning. In spite of these issues, to ensure communication reliability, reception is needed from all incoming directions, regardless of polarization, channel environment and body posture. Ideally, the radiation pattern of the receiv- ing antenna should fully cover the upper hemisphere for all polarizations and body orientations. Therefore, it is prudent to design around a more general propagation channel instead of a specific complex environment. These observations prompted us to adopt the following assumptions in our subsequent studies:
1. The reference channel is isotropic [36] having a uniform Angle of Arrival (AoA)
distribution over the entire upper hemisphere (0◦ to 360◦ in azimuth, 0◦ to 90◦
in elevation) and a uniform polarization distribution (between 0◦∼ 90◦) for all
incidence angles.
2. The human body and antennas are residing on an infinite ground plane.
3. Three human postures (standing, kneeling, and prone) with a full 360◦ rotation
are used to evaluate capacity.
4. The SNR at the receiving antenna (SNRRx) is fixed at 20 dB for all cases.
47 It is understandable that the channel assumption mentioned in 1) is not realistic.
It is nevertheless quite representative, and therefore useful for evaluating the capacity independent of the propagation environment. Furthermore, adding an alternate envi- ronment using statistical data is a straightforward process via the CDM (see previous section). In the subsequent sections optimum antenna diversity will be compared to that of isotropic reception.
3.4.1 Single Body-Worn Antenna Study
Before considering multiple antennas, it is important to determine the best body location for a single antenna. This study was done using 12 mm spacing between the antenna and the body to account for clothing. The antenna used is the AMF described in section 3.2 [32].
The considered mounting positions are illustrated in Fig. 3.8 and example perfor- mance evaluations are given in Fig. 3.9. The circular plots shown in Fig. 3.9 are two dimensional projections of the upper hemispherical gain pattern onto the x-y plane.
The center of the circle refers to zenith and the bottom edge of the circular periph- ery refers to the forward looking direction (θ=90◦, φ=0◦). As dark red represents stronger radiation, it is apparent that the best (single) antenna location is on the shoulder. This is due to the reduced obstruction by the human body and the equally good radiation of the horizontal and vertical polarizations. We also note that for this specific location, half of the antenna is on the shoulder and the other half extends along the arm.
48 (12) (5) (5) (12) (5) (12) (1) (2) (3) (8) (9) (10) (6) (13)
(4) (11)
(7) (14)
(a) (b) (c) (d)
Figure 3.8: Investigated mounting positions for the AMF antenna: (a) front view, (b) back view, (c) left side view, and (d) right side view.
49 zˆ gθ gφ [dB] . yˆ xˆ .zˆ yˆ (a)
φ xˆ
(b)
(c)
Figure 3.9: Projected upper-hemisphere radiation pattern onto the x-y plane for selected single antenna positions: (a) front torso, (b) left arm, and (c) left shoulder.
50 1.1 1 0.9 0.8 0.7 0.6 0.5 0.4 0.3 Both torsos 0.2 Both arms 0.1 Both shoulders Probability (capacity>abscissa) 0 0 1 2 3 4 5 6 7 8 9 10 Channel capacity [bps/Hz]
Figure 3.10: Channel capacity CCDF curves for different antenna pairs.
3.4.2 Dual Body-Worn Antenna Study
The conclusions drawn from the single antenna study are further confirmed by channel capacity computations. These calculations were carried out for symmetri- cally located antennas. Specifically, in Fig. 3.10 we compare the CCDF for various antenna pairs: front/back torso, left/right shoulders, and left/right arms. As noted before, three postures (standing, kneeling, and prone) were considered in evaluating the performance. Clearly, in agreement with Fig. 3.9, the shoulder positions offered the best channel capacity (6.0 bps/Hz). By contrast, the antenna pair on the torso delivered only 4.6 bps/Hz capacity. Therefore, the shoulder antenna positions were
51 considered as the primary body-worn antenna locations in selecting the best diversity configuration.
The upper hemisphere radiation patterns in Fig. 3.9 provide further insight on choosing additional antenna locations (to achieve better coverage). For instance, we observe that additional antennas should be considered to improve horizontal polariza- tion reception (φ-polarization) and thus enhance channel capacity. Next, we proceed to add more antennas and evaluate their impact on capacity.
3.4.3 Multiple Body-Worn Antenna Diversity
Overall, four diversity configurations were investigated using 2, 4, 6, and 8 an- tennas. The channel capacities of all these configurations are depicted in Fig. 3.11.
Not surprisingly, the best configuration (solid circle) from each set includes antennas at both shoulders (position 5, 12 in Fig. 3.8). With the best single antenna (left shoulder positioned) used as benchmark, it is clear that all diversity configurations show channel capacity improvements of 1.4, 2.2, 2.4, and 2.7 bps/Hz for the 2, 4, 6, and 8 antenna diversity configuration, respectively. Each of these provides 4.3, 6.8,
7.4, and 8.3 dB of diversity gain. However, diminishing returns are observed as the number of antennas exceeds 4. Specifically, only 7 % increase in channel capacity was observed between 4 and 8 antennas (from 6.8 bps/Hz to 7.3 bps/Hz), concluding that the 4-antenna configuration (front/back torso and both shoulders) is the most reasonable and cost effective diversity setup.
In the next section, we describe a practical realization of a diversity module and apply the above findings to produce a wearable antenna prototype.
52 8
7
6
5
4
3
2
1
Channel capacity Channel at 90% reliability [bps/Hz] 2-antennas 4-antennas 6-antennas 8-ants. 0 2-2 2-4 2-5 4-1 4-2 4-3 4-4 4-5 4-6 6-1 6-2 8-2 2-1 2-3 4-7 6-3 6-4 8-1
Single Configuration
(a)
Config- uration Antenna positions Single 12 2-1 2, 9 2-2 3.10 2-3 5, 12 2-4 6, 13 2-5 7, 14 4-1 1, 3, 8, 10 4-2 2, 5, 9, 12 4-3 2, 6, 9, 13 4-4 3, 4, 10, 11 4-5 4, 5, 11, 12 4-6 4, 6, 11, 13 4-7 5, 7, 12, 14 6-1 1, 2, 3, 8, 9, 10 6-2 1, 3, 6, 8, 10, 13 6-3 3, 4, 6, 10, 11, 13 6-4 2, 5, 7, 9, 12 , 14 8-1 1, 2, 3, 6, 8, 9, 10, 13 8-2 1, 3, 5, 7, 8, 10, 12, 14
(b)
Figure 3.11: (a) Channel capacity per configuration, and (b) list of antenna configu- rations per numbering in Fig. 3.8. 53 3.5 Diversity Module Realization and Measurements
To validate the findings of our previous section, a 4-channel diversity module was designed and fabricated using the selection diversity scheme operating at 100 MHz
– 1000 MHz. Using the fabricated module, the performance of the proposed body- worn antennas was measured and evaluated in an indoor setting (the Ohio State
University ElectroScience Laboratory hallway). In place of an actual human body, a full-size plastic mannequin was used, filled with body-tissue emulating liquid (water:
52 %, sugar: 47 %, salt: 1 %).
3.5.1 Diversity Module Realization
To implement a receiving diversity module, there are three choices for combining multiple signals: Maximum Ratio Combining (MRC), Equal Gain Combining (EGC), and selection diversity. In general, MRC and EGC achieve better communication per- formance than selection diversity, but the complexities of realizing a coherent phase circuit (having a precise and stable phase tracking) in a rapidly changing multipath fading environment are paramount. In contrast, the proposed selection diversity scheme easily achieves stable operation even in a fast multipath environment and is also simple to implement [37, 38]. Moreover, selection diversity gave better per- formance than other schemes in some communication environments (in which high correlated noise exists) [39]. Therefore, selection diversity was chosen for diversity module realization.
A schematic diagram of the proposed 4-channel diversity module is depicted in
Fig. 3.12. A portion of the RF signal from each antenna is routed through a di- rectional coupler to an RF power detector. Each RF power detector produces an
54 Coupler RF_out 1
RF_in 1 RF power RSSI_1 RF_cpl 1 detector
RF_out 2 RF_in 2 RSSI_2 RF switch RF_out RF_cpl 2
RF_in 3 RF_out 3 Switch RSSI_3 control RF_cpl 3
RF_in 4 RF_out 4 RSSI_4 RF_cpl 4 Analog_in
Digital data Data Acquisition Unit Control signal Digital_out LabVIEW
Figure 3.12: Schematic diagram of the implemented 4-channel diversity module.
analog Received Signal Strength Indication (RSSI) voltage proportional to the input
RF power. This RSSI voltage is then digitized and recorded via a data acquisition
unit (DAQ) connected to a LabVIEW [40] equipped computer. LabVIEW is then
used to process the four RSSI signals and control the 4-to-1 RF switch. The RF
switch is controlled via the DAQ to route the strongest signal selected by LabVIEW.
To reduce RSSI noise, a digital smoothing filter is implemented in LabVIEW and the
signals are weighted for calibrating the power detectors. The fabricated module can
measure the RF signal and control the RF switch at a maximum rate of 106 times per second. It is, thus, suitable for real time operation. Using this module we carried out the measurements described in the next section.
55 3.5.2 Indoor Measurement
Indoor measurements were carried out to demonstrate the effectiveness of the proposed diversity system in a rich multi-path fading environment. The body as- sembly with selected 4 antennas (12 mm thickness of styrofoam is used to emulate a gap between clothes and body) was mounted on a cart and tested in a hallway of the ElectroScience Laboratory building as shown in Fig. 3.13. The motion path
Figure 3.13: The 4-antenna diversity indoor measurement set-up.
56 33.6 m 1.2 m 2.4 m 180 °°° 180 °°°
6 m 6 m 6 m 6 m 6 m Tx 1.5 m 180 °°° 180 °°°
Figure 3.14: OSU-ESL hallway environment and the motion path used for the indoor measurement.
for data collection is illustrated in Fig. 3.14. As seen, a combination of translations and rotations were performed to deliver realistic scenarios. Specifically, the assembly
(diversity module + body-worn antenna system) was initially moved away from the transmitting antenna by 6 m for 60 seconds and then rotated 180◦ clockwise for 60 more seconds. This process was repeated four times until the assembly reached the center of the hallway. Concurrently, the individual channel data and the diversity module output were measured in real-time. The indoor measurement results are shown in Fig. 3.15. In this figure, the normalized received power from each antenna channel is plotted together with the module output. It is evident that the diversity module successfully operates to track the strongest signal throughout the entire path.
In contrast, the signals from the individual antennas suffer from severe drop-outs.
57 Selected antenna Ant 1 Ant 4 Ant 2 Ant 4 Ant 1 Ant4 Ant 3 Ant 2 Ant 3 Ant1 Ant1- front torso 0 Ant2- back torso Ant3- L. shoulder -10 Ant4- R. shoulder Diversity module output
-20
-30
-40
-50
Normalizedreceived power [dB] 0 60 120 180 240 300 360 420 480 540 Time [sec]
move rotate move rotate move rotate move rotate move 6 m 180 o 6 m 180 o 6 m 180 o 6 m 180 o 6 m
Figure 3.15: Comparison of the normalized received power between single antennas (mounted at different location on the body) and the diversity system (measurements were done in the hallway).
58 Overall, the diversity module provided up to 40 dB improvement over any single
antenna configuration.
3.6 Conclusion
A Channel Decomposition Method (CDM) was introduced for generalized on- platform antenna diversity evaluations and used to design a multi-antenna body- worn diversity system. The CDM separates the entire communication channel into the platform (human body) and the propagation sub-channel. It evaluates the full wave response of the platform (by means of measurements or simulations) and em- ploys statistical data to provide the environment effects. Both polarization and am- plitude characteristics were considered and results showed a 10-fold computational speed increase. CDM was then used to design a body-worn antenna diversity system and found that the 4-antenna configuration provided a good compromise in terms of capacity vs. number of antennas.
The designed 4-antenna diversity system (using selection diversity) was also fab- ricated incorporating the optimized antenna mounting configuration. The indoor measurement showed the significant improvement in communication performance.
Specifically, the proposed diversity system provided increased reliability (upto 40 dB improvement), while each single antenna suffered from deep nulls by the multi-path fading indoor environment.
59 Chapter 4: Systematic Design Method for Body-Worn Antenna Diversity Systems
4.1 Introduction
Antenna diversity is one of the most widely used and simplest methods for mit- igating shadowing and fading effects associated with rich multipath mobile radio environments. Antenna diversity becomes even more attractive for wearable appli- cations due to human body losses and shadowing [4, 6, 15, 16]. When designing a body-worn antenna diversity system, one is likely to face the following challenges: 1) large computational burden for analyzing the antenna, body, and surrounding envi- ronment together; 2) complex optimization process that searches for the minimum number of antennas to achieve maximum performance.
Diversity performance is usually assessed by channel capacity or diversity gain, while antenna correlation data are used for diversity design [34, 36, 41]. Channel ca- pacity provides a practical measure of communication performance (maximum data throughput), and diversity gain shows system enhancement via multiple antennas compared to a single antenna. Correlation between antenna elements presents a de- sign guide for antenna location and orientation. To my knowledge, there is no system- atic design process to select antenna mounting locations and the number of antennas needed for optimal performance in the presence of different platform environments
60 (for example, the human body for body-worn antenna). The need to provide reliable communication viz. maximum angle and polarization coverage complicates antenna choice even further.
In this chapter, I propose a systematic design method for a body-worn antenna di- versity system. The method incorporates the Channel Decomposition Method (CDM) presented in the previous chapter to reduce computation time by dividing the design domain into the human body and propagation channels (see Fig. 4.1). More sig- nificantly, the proposed design process automatically finds the antenna mounting positions for acceptable performance using the minimum number of antennas. This approach is then employed to design a body-worn antenna diversity system for specific communication environments.
The structure of this chapter is as follows. First, the systematic design method for body-worn antenna systems is introduced. Important parameters of the method are defined and the fundamental assumptions of the method are described and justi-
fied. Next, several body-worn diversity system design examples are presented, with the results verifying the effectiveness of the method and justifying the fundamental assumptions. To validate the method measurement results are provided in a simple
True environment Human Propagation channel +
Figure 4.1: Concept of the CDM.
61 communication environment.
4.2 Systematic Design Method for Body-Worn Diversity Sys- tems
It is difficult to find optimum diversity configurations for a body-worn diversity system that accounts for both the human body and the surrounding environment (ve- hicle, diversity, etc.). More specifically, if there are 12 possible antenna locations for a
4-antenna diversity system, we have to consider 495 (= 12C4) diversity configurations to find the optimal diversity system. Although we applied the CDM presented in the previous chapter, the optimization process requires a great deal of computation time and data processing. Thus, a simple and systematic design method is needed to
find an optimal diversity configuration from the multitude of possibilities. The goal is to achieve a specified antenna gain over the desired physical communication sector
(pattern coverage) using the minimum number of antennas.
4.2.1 Definition of the Design Parameters
Three parameters are used to define the optimal diversity system: 1) required antenna gain (Gmin), 2) pattern coverage (C) to gauge diversity performance, and 3)
antenna correlation coefficient (ρ) to gauge design efficiency.
Typically, the required antenna gain is determined by the communication link
budget, and pattern coverage is set to 90 % for voice communication or 99 % for data
communication. The pattern coverage of the ith single antenna is defined by
N {Gi(θ,φ) > Gmin} Ci = , (4.1) Nt
where Nt refers to the total number of pattern samples (or the total area of the
communication sector to be covered), and N represents the number of pattern samples
62 exceeding the minimum antenna gain (or the area of communication sector satisfying
the given goal). For multiple antennas, the pattern coverage becomes
N {max (Gi(θ,φ),Gj(θ,φ), ··· ) > Gmin} Cij··· = . (4.2) Nt In this expression, selection diversity is applied for the diversity module by the
max(·) operation in the numerator. The pattern coverage represents the performance
of the diversity system.
On the other hand, the antenna correlation coefficient (ρ) represents the similarity
in voltage fluctuations received by the two antennas. This quantity is related to the
antenna radiation patterns, their relative positions, and incoming waves’ polariza-
tions. The correlation coefficient is defined [41, 42] as:
∗ 2 E ViVj ρij = ∗ ∗ , (4.3) E [V iVi ] E V jVj
th where Vi is the induced voltage at the i antenna and E[·] indicates the expected value. It is also noted that the voltage Vi and Vj are phasor quantities representing amplitude and phase, thus their values are affected by not only the antenna patterns but also by the relative positions of the antennas. By using the field patterns of the antennas, this equation can be modified to 2 ~ ~ ~ ∗ ~ Ei(θ,φ) · h(θ,φ) · Ej (θ,φ) · h(θ,φ) θ,φ ρ = h i , (4.4) ij P 2 2 ~ ~ ~ ~ Ei(θ,φ) · h(θ,φ) · Ej(θ,φ) · h(θ,φ) θ,φ θ,φ P P where ~h(θ,φ) is the incoming waves’ polarization vector which can be expressed as
~h(θ,φ)= θˆcosψ + φˆsinψ. (4.5)
In the diversity system, the correlation coefficient is used as a metric for diversity
design efficiency because it represents the similarity of the antenna patterns. For
63 example, highly correlated antennas have similar patterns, and they are not efficient for the diversity system.
4.2.2 Systematic Design Process for Antenna Diversity
Fig. 4.2 describes the proposed design procedure. The design goals established at the outset are antenna gain (Gmin), pattern coverage (Cgoal) and allowable correlation coefficient (ρmax). Typically, if the correlation coefficient is less than 0.3 the diversity system consisting of antennas with low correlation is considered an efficient system
[42].
Establishment of the design goal ¦ - Required antenna gain (Gmin), pattern coverage(Cgoal), and allowed correlation ( max) CDM calculation of antenna pattern on platform Propagation channel characterization - Calculate vertical and horizontal component of gain th pattern for i antenna - Polarization angle (¥) / amplitude (A) distribution θ φ θ φ of the incident field in a given propagation channel gθ ,i ( , ), g φ ,i ( , )
Data modification using propagation channel statistics
- Regenerate the polarization angle ( ¥rand) and amplitude (Arand) via Monte-Carlo process θ φ = ()θ φ 2 ϕ + θ φ 2 ϕ ⋅ 2 Gi ( , ) gθ ,i ( , )cos rand gφ,i ( , )sin rand Arand
Correlation for pair of antennas Pattern coverage for n-antenna diversity 2 * - Define all n-antenna diversity configurations (start with n=2) 2 E[]VV ρ = ρ = i j - increase the number of antenna (n) in every repeated step ij s,ij E[][]VV * E V V * N{max()G (θ,φ),G (θ,φ),G (θ,φ)L > G } i i j j = i j k min where, V is induced voltage at ith antenna and Cijk... i Nt E[•] is expected value operator n = n +1 Finding configuration satisfying the goal > ρ ρ ρ L < ρ Cijk... Cgoal and ij , ik , jk , max No Yes Optimum diversity configuration with minimum number of antennas
Figure 4.2: Flow chart of the systematic antenna diversity design approach.
64 After initialization, CDM is employed to evaluate a single antenna in a specific
communication environment. This involves calculating the radiation patterns (gain)
for all antenna locations using a single antenna and adjusting the data to include the
th propagation channel. From this we can obtain the effective gain (Gi) for the i single antenna in the given communication environment defined by
2 cos sin 2 Gi(θ,φ)= gθ,i(θ,φ) ψrand + gφ,i(θ,φ) ψrand · Arand. (4.6) q q The details of this procedure were explained in the previous chapter.
Next, the pattern coverage (C) and correlation (ρ) for all 2-antenna diversity
configurations are calculated. For our case, selection diversity was used. However,
other diversity schemes can be applied by altering equation (4.2) for pattern coverage.
The decision making process involves a comparison of the calculated pattern coverage
(C) and correlation (ρ) subject to the goals, namely Cgoal and ρmax, respectively. If
no configuration satisfies the goals, the antenna number is increased and the steps
are repeated until the goal is satisfied. Eventually, we can find the optimum diversity
configuration with the minimum number of antennas and the best pattern coverage
satisfying the given goal.
The proposed design procedure is simple and straightforward. Furthermore it
provides the best diversity configuration with a minimum number of antennas. To
do so, the method assumes that mutual coupling between antennas is negligible.
This is a valid assumption in well-designed diversity configurations (e.g. having low
correlations between antennas). This will be demonstrated in the following section
with design examples.
65 4.3 Diversity Design Examples for Body-Worn Antenna
In this section, several design examples are shown for validation of the proposed
design method. Fig. 4.3 depicts the 12 available antenna positions for body-worn
antenna diversity systems and antenna elements operating at 350 and 550 MHz,
respectively. To validate the basic assumption in the proposed method, specifically
that mutual coupling between selected antennas in the well-designed diversity system
is negligible, two types of simulation are performed and compared.
1. Proposed method: The radiation pattern of each antenna at all locations is
calculated without any antennas at the other locations as suggested in the
proposed method. Thus, the pattern data does not include any mutual coupling
effects with other antennas in the given diversity configurations, but only 12
simulations are required to perform the proposed design method.
2. Full evaluation method without any assumption: The radiation pattern of
each antenna at each location is calculated with other antennas terminated
by matched load (50 Ω) in the given diversity configurations. Therefore, all
mutual coupling effects are included in the pattern data for the given diversity
configurations. In this case, we need to simulate 132 (= 12C2 ×2) configurations
for 2-antenna diversity, 660 (= 12C3 × 3) configurations for 3-antenna diversity,
and 1980 (= 12C4 × 4) configurations for 4-antenna diversity, respectively.
By comparing the above two cases, we can show that the assumption made in the proposed method is valid.
The proposed design method is next applied to two communication scenarios: the
first is ground to ground communication in a rural area; the second is all purpose
66 SR SL
SL SR
FR FC FL BL BC BR
FH BH
TL TR
(a) 186 mm
23 mm
(b) 259 mm
32 mm
(c)
Figure 4.3: (a) Investigated mounting positions for the antenna diversity configura- tions and antenna elements operating at (b) 550 MHz and (c) 350 MHz, respectively.
67 communication in an urban area. The details of each communication scenario are
described in the following sub-sections.
4.3.1 Communication Scenario I (Ground to Ground Com- munication)
Fig. 4.4 illustrates the ground to ground communication scenario in a rural area.
The details of the communication scenario and design goals are described as follows:
1. Required antenna gain (Gmin), pattern coverage (C), and allowable correlation
(ρmax) are set to -8.5 dBi, 90 %, and 0.3, respectively for the antenna operating
at 550 MHz (antenna gain is set to -14 dBi for the antenna operating at 350
MHz).
2. Incoming waves are uniformly distributed from θ = 80◦ to θ = 90◦ and all
horizontal directions.
3. All incident fields are assumed to be vertically polarized.
FEKO was used to model the human body for these simulations. The human body was modeled as a lossy homogeneous body of relative permittivity ǫr = 56.7
Vertical polarization
o 80 < § < 90
Figure 4.4: Illustration of the communication scenario I for ground to ground com- munication.
68 and conductivity, σ = 0.94. An infinite ground plane (ǫr = 9 and tanδ = 0.01) is used in all simulation cases. Antennas are located at each location shown in Fig. 4.3 with 12 mm separation from the body.
After applying the proposed design method and full evaluation method, the best diversity configurations and their pattern coverage and correlation coefficients are summarized and compared in Table 4.1(a) for 550 MHz operation. Note that the best antenna locations for each n-diversity system are same in both methods. Fur- thermore, the pattern coverage and correlation coefficients given by the two methods are within 0.2 % error for the pattern coverage and 0.0001 difference for the correla- tion coefficient. Finally, a 4-antenna diversity system (with antennas located on BC,
FL, TL and TR) satisfies all design goals and is the optimum. Table 4.1(b) summa- rizes the evaluation results for 350 MHz. The best diversity configurations found by both evaluation methods are also same for all n-antenna diversity systems, and other values are very close to each other. In all cases, the correlation coefficients are very small (the maximum correlation is only 0.153), thus all selected configurations are very efficient.
Based on the above results, we can conclude that the assumption made in the pro- posed method, that mutual coupling between antennas selected as the best diversity configuration is negligible, is a valid assumption. From this, we can conclude that the proposed systematic design method for body-worn antenna diversity systems is very accurate. Also, the method reduces the number of configurations requiring EM simulations compared to the basic evaluation in the design process. For example,
1980 EM simulations would be required to evaluate all 4-antenna diversity configura- tions. This number is reduced to only 12 EM simulations by the proposed procedure.
69 Table 4.1: Comparison of selected antenna positions, pattern coverage, and correla- tion coefficients for n-antenna diversity between the proposed method and full evalu- ation method for scenario I (Abbreviations for antenna positions; 1st letter- B: back, F: front, S: shoulder, T: Thigh, 2nd letter- C: center, H: Horizontal, L: left, R: right). (a) At 550 MHz,
Proposed Method Full evaluation method (without mutual coupling effects) (with mutual coupling)
Number (n) Pattern Pattern Best antenna Correlation Best antenna Correlation of antennas coverage coverage positions coefficient positions coefficient for diversity (%) (%) 2 TL, TR 78.6 0.0001 (TL, TR) TL, TR 78.4 0.00002 (TL, TR) 0.153 (FL, TL) 0.153 (FL, TL) 3 FL, TL, TR 84.9 0.001 (FL, TR) FL, TL, TR 84.9 0.001 (FL, TR) 0.0001 (TL, TR) 0.00002 (TL, TR) 0.0001 (BC, FL) 0.0001 (BC, FL) 0.004 (BC, TL) 0.004 (BC, TL)
BC, FL, 0.004 (BC, TR) BC, FL, 0.004 (BC, TR) 4 91.0 91.1 TL, TR 0.153 (FL, TL) TL, TR 0.153 (FL, TL) 0.001 (FL, TR) 0.001 (FL, TR) 0.0001 (TL, TR) 0.00002 (TL, TR)
(b) At 350 MHz.
Proposed Method Full evaluation method (without mutual coupling effects) (with mutual coupling)
Number (n) Pattern Pattern Best antenna Correlation Best antenna Correlation of antennas coverage coverage positions coefficient positions coefficient for diversity (%) (%) 2 BC, FC 84.1 0.022 (BC, FC) BC, FC 84.4 0.015 (BC, FC) 0.014 (BC, FL) 0.009 (BC, FL) 3 BC, FL, FR 89.6 0.013 (BC, FR) BC, FL, FR 89.6 0.008 (BC, FR) 0.039 (FL, FR) 0.039 (FL, FR) 0.128 (BL, BR) 0.127 (BL, BR) 0.004 (BL, FL) 0.005 (BL, FL)
BL, BR, 0.004 (BL, FR) BL, BR, 0.006 (BL, FR) 4 91.6 91.7 FL, FR 0.004 (BR, FL) FL, FR 0.006 (BR, FL) 0.004 (BR, FR) 0.005 (BR, FR) 0.039 (FL, FR) 0.040 (FL, FR)
70 Random linear
polarization