Citation
5.8-Gigahertz Doppler Radar Design for Improved Coverage and System Miniaturization of Noncontact Vibration and Distance Detection

Material Information

Title:
5.8-Gigahertz Doppler Radar Design for Improved Coverage and System Miniaturization of Noncontact Vibration and Distance Detection
Creator:
Nieh, Chien-Ming
Place of Publication:
[Gainesville, Fla.]
Florida
Publisher:
University of Florida
Publication Date:
Language:
english
Physical Description:
1 online resource (117 p.)

Thesis/Dissertation Information

Degree:
Doctorate ( Ph.D.)
Degree Grantor:
University of Florida
Degree Disciplines:
Electrical and Computer Engineering
Committee Chair:
LIN,JENSHAN
Committee Co-Chair:
EISENSTADT,WILLIAM R
Committee Members:
YOON,YONG KYU
LONG,MAUREEN T
Graduation Date:
12/19/2014

Subjects

Subjects / Keywords:
Antenna arrays ( jstor )
Antenna design ( jstor )
Antennas ( jstor )
Patch antennas ( jstor )
Radar ( jstor )
Signals ( jstor )
Simulations ( jstor )
Steering ( jstor )
Vibration ( jstor )
Vital signs ( jstor )
Electrical and Computer Engineering -- Dissertations, Academic -- UF
antenna -- beam-steering -- doppler -- non-contact -- radar -- sensor -- wireless
Genre:
bibliography ( marcgt )
theses ( marcgt )
government publication (state, provincial, terriorial, dependent) ( marcgt )
born-digital ( sobekcm )
Electronic Thesis or Dissertation
Electrical and Computer Engineering thesis, Ph.D.

Notes

Abstract:
This dissertation begins with the basic theory of Doppler radar system and motivation of this dissertation in Chapter 1. Challenges on the previous version of 5.8 GHz Doppler radar system will be discussed. For commercial purpose, the antenna coverage would be a challenge with a fixed-beam antenna. It demonstrates the misalignment issue under the test. An adaptive beam-steering antenna normally occupies a larger area. Chapter 2 introduces new adaptive beam-steering antenna with the focused beam. Better signal quality is achieved without manually alignment of an antenna. The new antenna size is the same as the conventional patch-array antenna. Misalignment of either transmitter or receiver antenna would degrade the signal quality. The antenna at Tx has larger effects on the misalignment due to the short-range re-radiation effect. In Chapter 3, a new method to detect the distance to the target with continuous wave (CW) radar and an adaptive beam-steering antenna is introduced. The system can concurrently detect the vibration and the distance to the target. To further reduce the size, system architecture with one shared transmitter and receiver antenna is demonstrated in Chapter 4. Novel miniature three-branchline coupler is designed and implemented to save the area. Wider bandwidth and smaller size are the benefits of this coupler. Chapter 5 integrates the Doppler vital-sign radar system with this new coupler into one printed circuit board (PCB). Only one antenna is used in the system to save the area and eliminate the coupling between transmitter and receiver antennas. System package is another possible issue to degrade the signal and is designed for our system. In Chapter 6, two hand-held devices integrated with the Bluetooth are designed and implemented. Low weight and low power consumption radar provide higher flexibility for mobile detection. In Chapter 7, this chapter summarizes the whole work and describes the future work in this field. This work demonstrates the adaptive beam-steering antenna to get better signal quality and the new method to detect the distance to the target besides the vibration rate. The miniature branchline coupler offers more compact size for our system. ( en )
General Note:
In the series University of Florida Digital Collections.
General Note:
Includes vita.
Bibliography:
Includes bibliographical references.
Source of Description:
Description based on online resource; title from PDF title page.
Source of Description:
This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Thesis:
Thesis (Ph.D.)--University of Florida, 2014.
Local:
Adviser: LIN,JENSHAN.
Local:
Co-adviser: EISENSTADT,WILLIAM R.
Electronic Access:
RESTRICTED TO UF STUDENTS, STAFF, FACULTY, AND ON-CAMPUS USE UNTIL 2015-12-31
Statement of Responsibility:
by Chien-Ming Nieh.

Record Information

Source Institution:
UFRGP
Rights Management:
Applicable rights reserved.
Embargo Date:
12/31/2015
Resource Identifier:
974007399 ( OCLC )
Classification:
LD1780 2014 ( lcc )

Downloads

This item has the following downloads:


Full Text

PAGE 1

1 5.8 GIGAHERTZ DOPPLER RADAR DESIGN FOR I MPR OVED COVERAGE AND SYSTEM MINIATURIZATION OF NONCONTACT VIBRATION AND DISTANCE DETECTION By C HIEN M ING N IEH A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2014

PAGE 2

2 © 2014 Chien Ming (Jemmy) Nieh

PAGE 3

3 To my family

PAGE 4

4 ACKNOWLEDGMENTS I would like to express my sincere appreciation to my advisor Dr. Jenshan Lin for his mentoring, patience , and encouragement through my Ph.D. life. I really enjoy doing the research under his guidance. He gives me a great flexibility to study the topic in which I am interested. His patience and supportiveness help me finish my degree smoothly. I also feel tha nkful to my Ph.D. committee members, Dr. William Eisenstadt, Dr. Yong Kyu Yoon, and Dr. Muareen Therese Long for their valuable advice and precious time. I am also thankful to my colleagues in the lab of Radio Frequency Circuits and System Research (RFCSR ) at UF: Dr. Xiaogang Yu , Dr. Yan Yan, Dr. Raul Chinga, Dr. Taesong Huang, Dr. Te Yu ( Jason ) Kao, Jianxuan Tu, Changyu Wei, Meiyu Li, Ron Chi Kuo, Tien Yu Huang, and Jaime Garnica for the useful discussion and I am happy to co work with you . I also feel th ankful to our previous members: Prof . Changzhi Li at T exas Tech University and Dr. Austin Chen at Skyworks for the useful guidance and instruction . My special thanks to our visiting scholar: Dr. Tze Ming Shen, Chieh Lin Ho, Prof . Chien Hung Chen, Dr. XinZh i Shi, Wei Ting ( Terrence ) Chen , Shuhei Yoshida , Shichang Chen , Dr. Gi Bum Lee , Shichang Chen , Prof. Xinzhi Shi and Prof . Chao Hsiung Tseng for sharing their research experience . During my internship, I would love to thanks my colleagues at Intel : Dr. Jong bae Park, Dr. Myunghyun Ha, Dr. Jennifer Tsai, Dr. Mo Liu, Dr. Ala, Dr. Min Wang and Dr. Ruihua Ding for your kind instruction. I would like to thanks Prof. J ennifer A . Rice and S hanyue Guan of the Civil and Coastal Engineering Department in University of Florida for helping the seismic simulator experiment, and Dr. J iun P eng Chen of National Taiwan University for helping the antenna measurement. , Skyworks solutions, Rogers, Maxim, Hittite Microwave, and RFMD for their support .

PAGE 5

5 Last but not least, I want to thanks my family and my girlfriend for their unconditi onal and unlimited support.

PAGE 6

6 TABLE OF CONTENTS p a g e ACKNOWLEDGMENTS ................................ ................................ ................................ ............... 4 LIST OF TABLES ................................ ................................ ................................ ........................... 8 LIST OF FIGURES ................................ ................................ ................................ ......................... 9 LIST OF ABBREV IATIONS ................................ ................................ ................................ ........ 13 ABSTRACT ................................ ................................ ................................ ................................ ... 15 CHAPTER 1 INTRODUCTION AND MOTIVATION ................................ ................................ .............. 17 1.1 Doppler Noncontact Vibration Radar ................................ ................................ .......... 17 1.2 System Implementation ................................ ................................ ............................... 21 1.3 System Alignment ................................ ................................ ................................ ........ 23 2 ADAPTI VE BEAM STEERING ANTENNA DESIGN ................................ ....................... 26 2.1 Introduction ................................ ................................ ................................ .................. 26 2.2 2×2 Patch Array Antenna Design ................................ ................................ ................ 27 2.3 Antenna Radiation Pattern Effects ................................ ................................ ............... 31 2.4 Adaptive Beam Steering Antenna Design ................................ ................................ ... 38 2.5 Human Vital Sign Detection and Comparison ................................ ............................ 47 3 CONCURRENT DETECTION OF VIBRATION AND DISTANCE USING AN ADAPTIVE BEAM STEERING ANTENNA ................................ ................................ ....... 50 3.1 Introduction ................................ ................................ ................................ .................. 50 3.2 Vital Sign Detection with a Beam Steering Antenna in Real Time ............................ 51 3.3 Distance Detection Theory ................................ ................................ .......................... 55 3.3.1 DC Offset Analysis ................................ ................................ ........................... 55 3.3.2 Distance Detection with an Adaptive Beam Steering Antenna ........................ 58 3.4 Experiment and Analysis ................................ ................................ ............................. 60 4 MINIATURE DIRECTIONAL COUPLER DESIGN ................................ ........................... 69 4.1 Introduction ................................ ................................ ................................ .................. 69 4.2 Miniature Three Branchline Coupler Design ................................ .............................. 72 4.3 Measurement Results and Analysis ................................ ................................ ............. 78 5 RADAR SENSOR SYSTEM WITH MINIATURE COUPLER DESIGN ........................... 83 5.1 Introduction ................................ ................................ ................................ .................. 83

PAGE 7

7 5.2 Link Budget and System Design ................................ ................................ .................. 83 5.3 Tx Leakage Consideration ................................ ................................ ........................... 88 5.4 Measurement Results and Analysis ................................ ................................ ............. 91 5.5 System Package ................................ ................................ ................................ ........... 94 6 HAND HELD DOPPLER RADAR TRANSCEIVER ................................ .......................... 97 6.1 Introduction ................................ ................................ ................................ .................. 97 6.2 System Design ................................ ................................ ................................ ............. 98 6.3 Implementation and Experiment Results ................................ ................................ ... 104 SUMMARY ................................ ................................ ................................ ................................ . 110 LIST OF REFERENCES ................................ ................................ ................................ ............. 112 BIOGRAPHICAL SKETCH ................................ ................................ ................................ ....... 1 17

PAGE 8

8 LIST OF TABLES Table p age 2 1 Truth table of beam steering functions ................................ ................................ ................. 44 3 1 Distance calculation results from the measurement data ................................ ...................... 65 4 1 Parameter summary of miniature microstrip line ................................ ................................ . 73

PAGE 9

9 LIST OF FIGURES Figure page 1 1 Illustration of radar sensor system ................................ ................................ ..................... 18 1 2 Concept of the optimal and the null detection point ................................ .......................... 20 1 3 Photograph of the Doppler noncontact vital sign radar system in the front side and back side. ................................ ................................ ................................ ............................ 22 1 4 Measurement setups of the misalignment experiment. ................................ ...................... 23 1 5 Measurement results in time domain when either Tx antenna or Rx antenna is tilted 30° angle. ................................ ................................ ................................ ........................... 25 2 1 2×2 patch array antenna is design ed and simulated in HFSS v13. . ................................ ... 29 2 2 Photograph of the 2×2 patch array antenna and the re flection coefficient in the measurement result ................................ ................................ ................................ ............ 30 2 3 Homodyne architecture of the Doppler noncontact vibration detection radar sy stem ...... 32 2 4 Simulated radiation patterns of the 2×2 array antenna (left) and the single patch antenna (right) in ANSYS HFSS ................................ ................................ ....................... 33 2 5 Measured reflection coefficients of the single patch antenna (blue dotted line) and the 2×2 array antenna (solid red line). ................................ ................................ ............... 33 2 6 Measurement setup. ................................ ................................ ................................ ........... 34 2 7 Vibration signals measured at four different combinations of antennas on Tx and Rx. ................................ ................................ ................................ ................................ ............ 35 2 8 Impedance of the single patch antenna and the 2×2 fixed beam patch array antenna ...... 36 2 9 Measured res ults in frequency domain ................................ ................................ .............. 37 2 10 Schematic of the adaptive beam steering antenna. ................................ ............................ 39 2 11 Phase shifter is a switched line type design. ................................ ................................ ...... 40 2 12 Simulated Results between two microstrip lines ................................ ............................... 41 2 13 Entire structure of the adaptive beam steering antenna is cascaded and simulated in ADS. ................................ ................................ ................................ ................................ ... 42 2 14 Simulated radiation pattern of 2 2 patch array antenna when the beam steers in 22°, 0°, and 22°. ................................ ................................ ................................ ......................... 42

PAGE 10

10 2 15 Reflection coefficient (S11) in both simulation and measurement. ................................ ... 43 2 16 Photographs of (A) adaptive beam steering antenna and (B) fixed beam antenna ........... 44 2 17 Photograph of antenna chamber ................................ ................................ ......................... 45 2 18 Measured radiation patterns including both co polarization and cross polarization in the E plane at different beam steering angles . ................................ ................................ ... 46 2 19 Photograph of measurement setup ................................ ................................ ..................... 47 2 20 V ital sign measurement result s in time domain ................................ ................................ . 48 3 1 Photograph of the Doppler noncontact vibration radar transceiver ................................ ... 52 3 2 C ontrol signal for adaptive beam steering antenna in LABVIEW . ................................ ... 52 3 3 Control signals and s ystems b y the field programmable gate array (FPGA) .................... 53 3 4 Control signals of the adaptive beam steering antenna. ................................ .................... 54 3 5 Block diagram of the Doppler noncontact vibration radar system for mechanical vibration detection. ................................ ................................ ................................ ............ 55 3 6 Constellation plot showing that the dc offset of the measured data is off the origin and the radius, Ar, is not equal to unity. ................................ ................................ ............ 57 3 7 Doppler noncontact vibration radar system with an adaptive beam steering antenna can radiate the beam in two different directions fast and alternately. ............................... 58 3 8 Testing environment of the Doppler vibration radar system. ................................ ............ 60 3 9 Baseband I(t) and Q(t) signals of the measurement data. Three different dc offsets are observed for both I(t) and Q(t) channels. ................................ ................................ ..... 61 3 10 Frequency spectrum of the measurement data when an adaptive beam steering antenna steers in three directions. ................................ ................................ ...................... 62 3 11 (A) Measured an d (B) recovered results in the constellation plot. ................................ .... 63 3 12 Target is measured at three different distances with two beam steering angles. The error of this method is less than 8%. ................................ ................................ .................. 65 3 13 Photograph of the seismic simulator experiment ................................ ............................... 66 3 14 Measured results before and after the seismic event emulated by the shaker ................... 67 4 1 Photograph of the Dopp ler radar system architecture with one branchline coupler and one antenna ................................ ................................ ................................ ........................ 69

PAGE 11

11 4 2 Structures and their ideal simulation results of the conventional two branch and three branchline couplers. ................................ ................................ ................................ .. 70 4 3 Equivalent model of the microstrip line ................................ ................................ ......... 72 4 4 Simulation results of an ideal single microstrip line and an ideal T model microstrip line structure in ADS ................................ ................................ ................................ ......... 74 4 5 model microstrip line ....................... 74 4 6 Size comparison between the conventional and T model two branchline couplers in HFSS. ................................ ................................ ................................ ................................ . 75 4 7 Size comparison between the conventional and T model three branchline couplers in HFSS. ................................ ................................ ................................ ................................ . 76 4 8 Final layout with the size information of the T model three branchline coupler .............. 76 4 9 EM simulation of the T model three branchline coupler in HFSS. ................................ ... 77 4 10 EM simulation result when the solder mask thickness is 110 um and the input microstrip line varies 10%. ................................ ................................ ................................ 77 4 11 Photograph of the miniature three branchline couplers ................................ ..................... 78 4 12 Measurement results of both couplers with the dummy ground (left) and without the dummy ground (right). ................................ ................................ ................................ ....... 79 4 13 Basic structure of the two branchline coupler and the even odd analysis ......................... 79 4 14 Simulation results of the branchline coupler when the impedance termination ................................ ................................ ............................... 82 5 1 Schematic of the new architecture of Doppler vital sign radar sensor .............................. 84 5 2 Layout of the Dopple r vital sign radar system with single antenna in top, bottom, and side view ................................ ................................ ................................ ............................ 86 5 3 Flow diagram of automatic gain control in a Doppler radar system ................................ .. 87 5 4 Schematic and the layout of the filter to isolate the digital noise from the vital sign signal ................................ ................................ ................................ ................................ .. 88 5 5 Link budget of Tx leakage for low/high power modes ................................ ...................... 89 5 6 Isolation comparison between different SMA connectors and loads. ................................ 90 5 7 Photograph of the board in both the top view and the bottom view. The filter and bias board is to isolate the di gital noise from the vital sign signal ................................ ........... 91

PAGE 12

12 5 8 Final design comparison between the previous and current design. The size is greatly reduced. ................................ ................................ ................................ .............................. 92 5 9 Comparison of the previous design and current design on the vital sign signal contaminated by the digital noise ................................ ................................ ...................... 93 5 10 Function of the automatic gain control in the measurement result. ................................ ... 93 5 11 Antenna performance at different gap between the package box and the antenna in the simulation result ................................ ................................ ................................ ........... 95 5 12 Antenna measurement data at different values of the gaps ................................ ................ 95 5 13 Photograph of the final system within the package ................................ ........................... 96 6 1 Simulated return loss and Radiation pattern of the single patch antenna in HFSS ........... 99 6 2 Design layout and the simulated antenna gain of the IFA at 2.45 GHz. ............................ 99 6 3 Simulated reflection coefficient of IFA ................................ ................................ ........... 100 6 4 Layout of the miniature two branchline coupler with a single patch antenna ................. 101 6 5 S parameter of the miniature two branchline coupler with a single patch antenna ........ 101 6 6 Layout of the Wilkinson power divider with the Tx antenn a. ................................ ......... 102 6 7 S parameter of the Wilkinson power divider with the Tx antenna. ................................ . 103 6 8 Design layout of two Doppler vibration radar systems. ................................ .................. 103 6 9 Photograph of the IFA and the single patch antenna. ................................ ...................... 104 6 10 Measured reflection coefficient s . ................................ ................................ ..................... 105 6 11 Photograph of both systems and the weight of the system with two antennas. ............... 106 6 12 Photograph of radar systems after soldering all the components an d cutting the redundant space ................................ ................................ ................................ ................ 106 6 13 Testing environment of the Doppler vibration radar system. ................................ .......... 107 6 14 Measured results after the Fourier transform in the frequency domain. .......................... 108

PAGE 13

13 LIST OF ABBREVIATIONS ABS Acrylonitrile butadiene styrene ADC Analog to digital conversion ADS Advanced design system CSD Complex signal demodulation CW Continuous wave DAQ Data acquisition DIP D ual in line parallel DSP Digital signal processing EM Electromagnetic ENIG E lectro less nickel immersion gold FCC Federal communications commission FFT Fast Fourier transforms FMCW Frequency modulated continuous wave FPGA Field programmable gate array HFSS High frequency structural simulator HPBW half power beamwidth IC Integrated circuit IF A Inverted F antenna I/Q in phase/quadrature phase LNA Low noise amplifier LO Local oscillator MCU microcontroller unit P 1dB in Input of 1 dB gain compression point

PAGE 14

14 PA Power amplifier PCB Printed circuit board RCS R adar cross section RF Radio frequency Rx Receiver SMA SubMiniature version A SNR S ignal to noise ratio SPDT S ingle pole double through SP4T S ingle pole four through TDR T ime domain reflectometer T x Transmitter VCO Voltage control oscillator VGA Variable gain control amplifier VNA V ector network analyzer

PAGE 15

15 Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy 5.8 GIGAHERTZ DOPPLER RADAR DESIGN FOR I M P R OVED C OVERAGE AND SYSTEM MINIATURIZATION OF NONCONTACT VIBRATION AND DISTANCE D ETECTION By Chien Ming Nieh December 2014 Chair: Jenshan Lin Major: Electrical and Computer Engineeri ng This dissertation begins with the basic theory of Doppler radar system and motivation of this dissertation in Chapter 1. Challenges on the previous version of 5 .8 GHz Doppler radar system will be discussed. For commercial purpose, the antenna coverage would be a challenge with a fixed beam antenna. It demonstrates the misalignment issue under the test. An adaptive beam steering antenna normally occupies a larger area. Chapter 2 introduces new adaptive beam steering antenna with the focused beam. Better signal quality is achieved without manually alignment of an antenna. The new antenna size is the same as the conventional patch array antenna. Misalignment of either transmitter or receiver antenna would degrade the signal quality. The antenna at Tx has larger effect s on the misalignment due to the short range re radiation effect. In C hapter 3, a new method to detect the distance to the target with continuous wave ( CW ) radar and an adaptive beam steering antenna is introduced. The system can concurre ntly detect the vibration and the distance to the target. To further reduce the size , system architecture with one shared transmitter and receiver antenna is demonstrated in Chapter 4 . Novel miniature three branchline coupler is designed and implemented to save the area. Wider bandwidth and smaller size are the benefit s of this coupler. Chapter 5 integrates the Doppler vital sign radar

PAGE 16

16 system with this new coupler into one printed circuit board ( PCB ) . Only one a ntenna is used in the system to save the area and eliminate the coupling between transmitter and receiver antennas. System package is another possi ble issue to degrade the signal and is designed for our system . In Chapter 6 , two hand held devices integrated with the Bluetooth are designed and implemen ted. Low weight and low power consumption radar provide higher flexibility for mobile detection. In Chapter 7, this chapter summarizes the whole work and describes the future work in this field. This work demonstrates the adaptive beam steering antenna to get better signal quality and the new method to detect the distance to the target besides the vibration rate . The miniature branchline coupler offers more compact size for our system.

PAGE 17

17 CHAPTER 1 INTRODUCTION AND MOTIVATION 1.1 Doppler Non c ontact V ibration Radar A Doppler non contact radar system has recently drawn lots of attention for vibration detection. Wireless detection provides huge potential for lots of applications, such as human vital sign detection, animal health inspection, human rescue and baby monitoring . A lot of researches have been conducted in this field [1] [20] . This technology is first introduced to apply to physiological movement detections in 1970s [2] . The frequency, strength , distance , and speed of the movement could be extracted by analyzing the phase characteristics in real time. In the wired detecting device, lots of annoying wires nee d to be attached on the target , which is inconvenient for the detection setup . Under some special cases, it is also difficult to contact the subject in person. For example, detect ing the vital signs of the fer ocious animals such as lions or tigers is extre mely dangerous . Wireless detection provides a better solution for this scenario . Besides Doppler noncontact vibration radar, t here is other commercial product for human heart rate detection by using a camera . By detect ing and analyzing the color change on a human face , the heart rate can be extracted. O ptimal lighting environment , however, is required for detection accuracy and the equipment is required to directly point to the human face, which limits the detection range . Th is technique is only able to detect human vital signs . On the contrary , Doppler radar has varieties of the applications. It has been successful ly developed to detect the vibration movement of the metal plate [3] , [4] and the human vital signs detection behind the wall [5] . Besides the respiration and heart rate , i t can also detect all the motion s from the human body , which is another human vital sign index. When people or patients suffer the pain, their bodies would react and unconsciously move more than usual. Some researchers try to analyze it in quantitative. Then, the doctor has a way to tell how the medicine works on the patient , which

PAGE 18

18 could be a great potential contribution in the medical field . For the animal vital sign detection, it is also suitable for inspect ing the health condition of the horse. The heart rate of the horse is an index for horse s health. Normally, the heart rate of the horse should be under 40 beats/min. Contact heart rate sensors make the horse uncomfortable and affects the accuracy o f the measurement. Figure 1 1 . Illustration of radar sensor system A Doppler noncontact vital sign radar system can be implemented on the PCB for the low cost or in the integrated circuit (IC) to save the area. T he electromagnetic (EM) wave of the Doppler noncontact vital sign radar system is used within industrial, scientific, and, medical ( ISM ) band s to satisfy the federal communications commission (FCC) regulation. Figure 1 1 shows the basic concept of the Doppler noncontact vital sign radar system. In the transmitter (Tx) of the radar transceiver , an un modulated signal , T(t) , is transmitted with the amplitude normali zed to unity: (1 1 )

PAGE 19

19 Where f and are the carrier frequency and the phase noise. The target is located at the distance of away from the radar transceiver . The target has periodic movement s, x(t) , at the human chest and belly generated from the heart beats and the respiration. x (t) contains the information of the heart rate and respiration rate as follows: (1 2) Where m r is the amplitude of the respiration, m h is the amplitude of the heart rate, r is the angular frequency of the respiration , and h is the angular frequency of the heart rate. T(t) is modulated by x(t) and reflected back to the receiver (Rx) of the radar transceiver . The received signal R(t) is e xpressed as follows: (1 3) W here is the carrier wavelength and c is the speed of the light . Then, R(t) is down converted to the baseband through the mixer in a transceiver . The baseband output , B (t) , can be expressed as : (1 4) Where t is the total accumulated residue phase from the system and the wireless propagation path. (1 5) x(t) is normally in the range from 0.2 mm to 2 mm . If the system is operate d at 5.8 GHz, x(t) is much smaller than the wavelength , 51.7 cm . When t is at odd multiples of /2 , B(t) can be approximate d to by small angle approximation. This scenario happens when the target is located at an optimal detection point. On the country , w hen t i s at even multiples of /2 , B(t) is

PAGE 20

20 close to one. The vital signal information is missing when the target is located at this so called null detection point [6] . Figure 1 2 . Concept of the optimal and the null detection point Figure 1 2 shows th is concept of the optimal and the null detection point s . The vital sign signal, x 1 (t) , in red line represents the target located at the optimal detection point. The signal is linearly transformed to the output without distortion, which is B 1 (t) . B 1 (t) can be demodulated without distortion if x 1 (t) is smaller than the wavelength of the carrier, T(t) . The vital sign signal, x 2 (t) , in black line is the target located at the null detection point. After the vital sign signal is modulated with the carrier signal, the vital sign signal would be distorted or even become the constant value at output. The optimal and null detection point s alterna tely occur every one eig hth wavelength of the carrier. The accuracy of the vital sign detection is limited by the location of the target . Many work s have be en done to fix this issue [7] [9] . Freq uency tuning technique is adopted to avoid the null detection point , but it increases the complexity of the system [7] . Alternatively, quadrature si gnals, I(t) and Q(t) , are introduced to deal with the null detection

PAGE 21

21 issue. When one signal, either I(t) or Q (t) , is located at the null detection point , the other will be at the optimal detection point. Arctangent demodulation [8] and c omplex signal demodulation (CSD) [9] are developed to process the quadrature signals . For arctangent demodulation, the dc offset need s to be calibrated to recover the desired baseband signals. Since the dc offset varies under different detection environment , c alibration is required in di fferent scenarios. For CSD, the dc offset is not an issue. Nevertheless, strong harmonics form the respiration signals may overwhelm the heartbeat signals [10] , [11] . Although both demodulation methods have their own limitations, they can successfully detect the vital sign signals. In order to detect the vital sign at different situations in real time, CSD is used in this paper. The quadrature signals can be further analyzed with CSD in real time: (1 6) When we c onsider t he magnitude of the S(t) , | e j( t ) | becomes a constant . T he total phase residue , t , can be eliminated when |S(t)| is applied to the fast F ourier transform (FFT) in the frequency domain . The vital sign signal can be accurately extracted by this method without considering the total residue phase , which makes the vital sign detection independent of the detection distance . Without dc offset calibration, the vital sign de tection, therefore, can be real time detection. 1.2 System Implementation Doppler noncontact radar systems have been implemented in many researches. Self injectio n lock radar can reduce the noise floor to achieve better signal to noise ratio (SNR) [12] , [13] . Nevertheless, the system structure is more complex. Some works have successfully

PAGE 22

22 implemented the system on IC [14] [16] , but the implementation on IC takes longer design time and higher cost. Most efficient and fast est way is to implement it on the PCB [17] [19] . All the components integrated on the same board can greatly reduce t he area. In addition, 5.8 GHz is one of the ISM bands, which is one of the reasons for choosing it. U nder the linear approximation, a system operating at higher frequency provides higher sensitivity to detect smaller vibration movements [14] , [15] , but it would have more issues with larger vibration movement. The choice of 5. 8 GHz is a good tradeoff for detecting both respiration rate and heart rate. Figure 1 3 shows the photograph of the Doppler noncontact vital sign radar system in the front side and the back side. In the front side, t he layout of the main circuit is printed on it including the ra dio frequency (RF) circuits, baseband circuits, bias networks, analog to digital converter (ADC), and Zigbee wireless communication circuits. All the component s are soldere d on the front side of the board . Zigbee is the wireless chip based on IEEE 802.1 5 s tandard to transmit the digitized data of the vital signs to the computer, which makes t he user able to read the detection data on your own PC in real time. Two 2× 2 patch array antenna s are connected through subminiature version A ( SMA ) connectors in the b ack side to save the area in horizontal direction s . This hand held device provide s higher mobile flexibility to do the measurement anywhere . Figure 1 3 . P hotograph of the Doppler noncontact vital sign radar system in the front side and back side.

PAGE 23

23 1.3 System Alignment Many studies have demonstrate d the capability of the Doppler noncontact vital sign radar system. A Doppler noncontact vital sign radar system can detect all the movement s from the human body, inclu ding the periodic vital sign signals and annoying random body movements. Lots of research works have been done to improv e the accuracy of the vital sign detection [11] , [17] , [19] , and [20] . Multiple radar systems are used to cancel out the random body motion and maintain the desired vital sign signals. T hese researches , however, are all under the assumption of the good system alignment with the target. A radar system is directly fac ing a target to get the good quality of the signal. Therefore, a system can accurately detect the human vital sign signals with good alignment. M isalignment between the target and the system can degrade the SNR and even fail the random body motion cancellation . People with different heights , however, are required another alignment under each measurement. The accuracy might even degrade when the target under th e detection change s his posture. I t is inconvenient to do the system alignment under every measurement. Figure 1 4 . Measurement setups of the misalignment experiment . The measurement results are shown in time domain.

PAGE 24

24 To demonstrate this issue, Figure 1 4 shows the experiment of the measurement setup . Two 2 2 fixed beam patch array antennas are used in this Doppler noncontact vital sign radar system. One is for Tx and the other is for Rx. A n a ctuator generates a 1 Hz periodic signal with 1 mm peak to peak displacement and is placed at 50 cm away from the Doppl er noncontact vital sign radar system. A 16 cm 14 cm metal plate is attached on the tip of the actuator. There are two experiment scena rios . One is that the actuator is 30 cm higher than the radar system in height and the other is the actuator directly facing the radar system. The alignment angle s are 0 ° and 31° , respectively. The measurement results are shown in the time domain with both scenarios. The red and blue curves represent quadrature baseband signals , I and Q. The red line I signal is at optimal de tection p oint and the blue lin e Q signal is at the null detection point for both cases. The plot clearly shows t he Q signal is distorted and contains bad signal quality when the alignment angle is 0 ° . On the cont rar y , t he red curve shows a clear periodic sign al when the alignment angle is 0 ° . In the second scenario, the target is 31 ° misaligned with the Doppler nonc ontact vital sign radar system. T he result shows nothing but noisy signals. A 2×2 patch array antenna has high antenna gain , but the antenna half power beamwidth (HPBW) is small. A s ingle patch antenna with wider antenna HPBW would provide wider coverage but sacrific ing the antenna gain , which degrades the signal . Furthermore, t he wider antenna HPBW also collect s more surrounding noise from the environment . Standing objects around the target will reflect the signal and cause t he dc offset, which increase s the noise floor. For commercial products, a system would be required to use for a variet y of people at different situations. It is important to make the system adaptive to different scenarios. Figure 1 5 shows another experiment. E ither a Tx antenna or a n Rx antenna is tilted 3 0° when the Doppler noncontact vital sign radar system is 3 1 ° misaligned with the target . The system can receive t he

PAGE 25

25 clear periodic signal s for both cases , but the signal strength is weak er than the system with good alignment on both antennas. Either Tx or Rx antenna s with misalignment would degrade the signal quality. Therefore, it is better to design an adaptive be am steering antenna for the Doppler noncontact vital sign radar system. Figure 1 5 . M easurement result s in time domain when either Tx ant enna or Rx antenna is tilted 30 ° angle.

PAGE 26

26 CHAPTER 2 ADAPTIVE BEAM STEERING ANTENNA DESIGN 2.1 Introduction In wireless communication system, the signal quality is always a major concern. For simplicity, the received signal strength is calculated by radar equation under the far field assumption [21] : (2 1) Where P r is the receiving power, P t is the transmitting power, G t is the antenna gain at Tx , G r is the antenna gain at Rx , and is the radar cross section (RCS). To analyze the received signal, the received power should be higher than the noise level. Larger antenna gain provides the better SNR without increasing power consumption while more focused radiation beamwidth reduces the surrounding noise. For human vital sign detection, any movement close to the target would interfere with the result and reduce the detection accuracy. On the other hand, the RCS depends on the physical area, surface reflectivity, and operating frequency. The RCS of adult human body is in the range of 0.5 m 2 to 3 m 2 [22] , which are smaller than 1 6 cm x 1 4 cm metal plate. Furthermore, RCS of human card iopulmonary activity is even smaller than 0.5 m 2 [23] . In a Doppler noncontac t vital sign radar system, the transmitted signal is reflected and modulated by the periodic movement of human chest due to the respiration and heartbeat. The vital sign signal is very sensitive to the incident angle of antenna radiation due to the focused antenna bea m and small RCS. With a fixed beam antenna, a lignment is required for each measurement, which makes it inconvenient for vital sign detection on people of different height s or posture s . To increase the detecti on coverage, Z. Park introduced a printed antenn a technology with beam steering function [24] . However, its low antenna gain degraded the SNR and thus required

PAGE 27

27 more power in the Tx side to keep the same SNR . The refore, a phase array antenna with beam steering function would be a better solution. A 1 × 4 phase array antenna with passive phase shifters, like Butler matrix, has been studied in [25] . A Butler matrix network, however, is very big at 5.8 GHz and can not provide the 0° (broadside) beam direction. In order to chose the beam direction of the antenna. A single pole four through ( SP4T ) switch is required to connect with the feed line and a 1 × 4 B utler matrix network . It even consumes more area of the PCB. Moreover, other commercial phase shifter s operating at 5.8 GHz are either costly or bulky. A 2×2 patch array antenna with adapt ive phase shifters is implemented to achieve both wider detection range and high antenna gain in this chapter . 2.2 2×2 Patch Array Antenna Design Before we start to design an adaptive beam steering antenna, a traditional 2×2 fixed beam patch array antenna is implemented as a reference. Considering the cost and the performance, Rogers 4350B with the dielectric thickness equal to 3 1 mil s is adopted . The thickness of the board would affect the antenna radiation efficiency and the matching design of the transmission line. The design of the microstrip patch antenna starts from the determination of the width [26] . The practical width leads good radiation efficiency [27] : (2 2) Where W is the ant enna width, f r is the resonant frequency, µ 0 is the permeability of free space, 0 is the permittivity of free space, and r is the dielectric constant . For the microstrip patch structure, the EM wave propagates in the inhomogeneous medium. An effective dielectric constant is calculated as follows [28] :

PAGE 28

28 (2 3) Wh ere h is the dielectric thickness of the board. As W /h >>1 and, r >>1 most of the EM wave would be constrained in the dielectric substrate. The effective dielectric constant would be close to the dielectric constant of the board. T he fringing effect should be considered b ecause the siz e of the patch antenna in width and length is finite . The fringing ef fect makes the microstrip patch electrically wider than physical size. A practical approximate relation is widely used as follows [29] : (2 4) Where is the extended length due to the fringing effect. The length of patch would be extended in both sides. The r ectangular patch antenna is designed as a half wavelength resonator electrically for the dominant TM 01 mode. The physical length would be as follows: (2 5) After designing the patch antenna at desired frequency, the fee dline needs to be designed to 50 . Recessed microstrip feed line is used to save the area instead of the quarter wavelength transformer which occupies lots of area s . (2 6) W here y is the length of the rec essed feedline from the boarder and y 0 is the initial pos ition where there is no recessed feedline. The antenna is designed at 5.8 GHz , which is on e of the ISM band s . Figure 2 1 shows the final design of the 2×2 patch array antenna and the simulation results. The

PAGE 29

29 Figure 2 1 . 2×2 patch array antenna is designed and si mulated in HFSS v13. The antenna gain is 10.8 dB at 5.8 GHz and the reflection coefficient is 21.29 dB at 5.8 GHz. full wave EM simulation is done by ANSYS high frequency structural simulator ( HFSS ) v13. The simulated antenna gain is 10.5 dB at 5.8 GHz an d the resonance of the reflection coefficient , S 11 , occurs at 5.8 GHz. The antenna bandwidth is from 5.76 GHz to 5.86 GHz. The solder mask is used as the protection of the board and needs to be considered in the simulation. It would increase the effective dielectric constant and shift the resonant frequency to the lower value. Furthermore, electro less nickel immersion gold (ENIG) is plated on the copper of the microstrip line. The conductance of the gold is similar to the copper and the gold can prevent the copper

PAGE 30

30 from being rust ed . The nickel is the intermediate metal between the copper and the gold. The conductivity of the nickel is bad and it might degrade the RF performance. The skin depth effect can help us to understand how importance o f the nickel is. The skin depth of the conductor can be calculated as follows: (2 7) Where is the resistivity of the conductor, is the angular frequency of the current, and µ r is the relative magneti c permeability of the con ductor . The skin depth of the gold is approximately 1 um at 5.8 GHz by (2 7) . In ENIG technology, the thickness of gold is around 0.05 u m to 0.1 u m and t he thickness of the Nickel is around 2 u m to 4 u m. The n ickel is beneath of the gold and the thickness of the gold is very thin . Therefore, t he nickel would strongly a ffect RF performance at 5.8 GHz and needs to be considered in the simulation . Figure 2 2 . P hotograph of the 2×2 patch array antenna and the reflection coefficient in the measurement result Figure 2 2 shows the photograph of the 2×2 fixed beam patch array antenna and the reflection coefficient of the measurement result . S parameter s are measured by Agilent vector network analyzer (VNA) E 83 61A. The resonance of the antenna is at 5.8 GHz , which is in good

PAGE 31

31 agreement with the simulation result. The reflection coefficient , however, is only 10. 3 dB at 5.8 GHz and worse than the simulation results. In the simulation, the SMA connector and the sol dering material are not included which will affect the impedance of the antenna. The result, however, is sufficient for the antenna application. 2.3 Antenna Radiation Pattern Effects As describe d in Chapter 1 3 , the SNR degrades when the system is misaligned. The effect of the antenna radiation beamwidth when the system is misaligned, however, has not been studied even though it is a practical scenario when measuring human vital signs. The small vital sign area can be easily misaligned with antenna beam whe n the target's posture changes. Although the Tx and Rx antennas should be reciprocal based on the radar equation (2 1) , the actual measurement results indicate that some other factors might cause different effects on Tx and Rx. This section compares two di fferent antennas (a single patch antenna and a 2×2 patch array antenna ) on Tx and Rx to detect the vibration movement when the antenna and target are misaligned. Figure 2 3 shows the homodyne architecture of the Doppler noncontact vibration detection radar system. A sinusoidal signal generated from the voltage control oscillator (VCO), HMC 431LP4, is amplified by a variable gain control amplifier (VGA), HMC625LP5, and a power amplifier (PA), HMC788LP2E, on the Tx side. VGA is used to control the transmitted power to avoid the signal saturation in baseband when detecting a target at short distance. The signal is equally split to the Tx antenna port and the LO port of the in phase/quadrature phase ( I/Q ) mixer, HMC525LC4, for self mixing. A low noise amplifier (LNA), HMC318MS8G, and two gain blocks, NBB 400, are three amplification stages on the Rx side to reduce noise figure in the baseband [16] . The reflected signal from the Rx antenna is directly down converted to the baseband by the I/Q mixer. The baseband quadrature signals are

PAGE 32

32 sent to an ADC for sampling before sending to the co mputer for the digital signal processing (DSP) and analysis. The range correlation effect significantly reduces the phase noise of the oscillator [6] . The baseband signal, B(t), contains the vibration information shown as follows: (2 8) Where t ) is the vibration movement , A is the vibration displacement, is the accumulated phase residue, and is the phase delay at Rx. The vibration information can be extracted by the CSD method. Figure 2 3 . Homodyne a rchitecture of the Doppler noncontact vibration detection radar system Figure 2 4 shows the simulated radiation patterns of the 2×2 patch array antenna with 42 ° HPBW and the single patch antenna with 91 ° HPBW in ANSYS HFSS. The antenna gain of 2×2 patch array antenna and the single patch antenna are 10.5 dB and 5. 5 dB at 5.8 GHz , respectively . The antennas and the radar transceiver are both implemented on Rogers 4350B

PAGE 33

33 laminates with a thic kness of 30 mils. The antenna input reflection c oefficients are measured by an Agilent V NA E8361. Figure 2 4 . Simulated radiation patterns of the 2×2 array antenna (left) and the single patch antenna (right) in ANSYS HFSS Figure 2 5 . Measured reflection coefficients of the single patch antenna (blue dotted line) and the 2×2 array antenna (solid red line). Figure 2 5 shows the measured reflection coefficients of a single patch antenna in dotted blue line and a 2×2 patch array antenna in solid red line. The antenna bandwidth of the

PAGE 34

34 single patch antenna is from 5.78 GHz to 5.88 GHz and The antenna bandwidth of the 2×2 patch array antenna is reduced to 5.79 5.84 GHz. The degradation is caused by the thickness variation of the solder mask, the nickel, and the gold. The dielectric constant of the solder mask al so varies a lot. All of them will cause the impedance variation and degrade the bandwidth. Nevertheless, the performances of both antennas are sufficient to work at 5.8 GHz. Figure 2 6 shows the measurement setup. A 16 cm × 14 cm metal plate is attached to an actuator. The actuator generates a 1 Hz sinusoidal movement with 1 mm peak to peak displacement. The radar system is 31° vertically misaligned wit h the target and placed at 0.5 m away. Two single patch antennas and two 2×2 array antennas were fabricated. The signals are measured at four different combinations of antennas on Tx and Rx Figure 2 6 . Measurement setup. Figure 2 7 shows the measurement results. Vibration information is extracted by CSD method in the frequency domain. In Figure 2 7 (A) , two 2×2 array antennas are used at both Tx

PAGE 35

35 and Rx. The signal cannot be detected since at 31° the gain of the 2×2 array drops significantly. In Figure 2 7 (B) and Figure 2 7 (C ), a single patch antenna is used at Tx or Rx and a 2×2 array antenna is used on the other. The combined antenna gains at 31° are the same in either case. Nevertheless, Figure 2 7 (B) shows a larger SNR than Figure 2 7 (C ). Several follow up experiments indicate tha t the broader beamwidth or higher antenna gain at the target direction plays a more important role at Tx than at Rx. There are two possible reasons. One possible reason is the multiple reflections between the target and the antenna in short range detection s . Figure 2 7 . Vibration signals measured at four different combinations of antennas on Tx and Rx. The reradiated signal from the antenna depends on the termination impedance. The Tx antenna is directly connected to the PA output which typically has smaller impedance and would reradiate more signal power back to the target. In this case, the broader beamwidth single patch antenna which has the higher gain at target direction is prefer red to be placed on the Tx to improve

PAGE 36

36 overall SNR. In Figure 2 7 (D ), two single patch antennas are used on Tx and Rx and the SNR is the largest due t o the highest combined antenna gain at 31°. Thus, the system with broad beamwidth antennas is more robust for human vital sign detection due to the small human vital sign area. The othe r possible reason is the load effect of the PA. The output power may b e different with two different types of antennas since the impedances of both antennas are not perfectly 50 Figure 2 8 . Impedance of the single patch antenna and the 2×2 fixed beam patch array antenna Figure 2 8 shows the measured impedances of the single patch antenna and the 2×2 fixed beam patch array antenna. Both of them are smaller than 50 have little effect on the load s of PA because the values of two impedances are very similar. In order to verify the multi reflection effect in the short range, there is another experiment: two 2×2 fixed beam antennas are used at Tx and Rx in a Doppler noncontact vibration detection radar

PAGE 37

37 Figure 2 9 . Measured results in frequency domain when (A) only the antenna at Tx tilts 30° and (B) only the antenna at Rx tilts 30°. system. Either a Tx antenna or a n Rx antenna will tilts 30° to face the target. By this experiment, we can eliminate the effect of the different load s of PA. In Figure 2 9 (A) , the antenna at Tx tilts 30° and the antenna at Rx faces 0°. On the contrary, the antenna at Tx faces 0° and the antenna at Rx tilts 30° in Figure 2 9 (B) . The total antenna gain of the entire system is the same for both cases. Better SNR occurs at the case when the antenna at Tx directly faces the target than the antenna at Rx . A 5.8 GHz Doppler noncontact vibration detection radar system using antennas of different beamwidths are experimented. As expected, the antenna with broader beamwidth is more robust in detection when there is a misalignment between t he antenna beam and the target. Unexpectedly, it was observed that placing the antenna with broader beamwidth on Tx achieves better SNR than placing it on Rx. Both experiments verified the effect of the multiple radiation s at the short range. A possible reason is suggested, but further r esearch and verification will be needed . The results provide useful guidelines for designing a radar system for short range human vital sign detection.

PAGE 38

38 2.4 Adaptive Beam Steering Antenna Design After building up a conventional 2×2 patch array antenna, t his pr esents a 2 2 microstrip patch array antenna integrated with two switched line phase shifter s . Each phase shifter consists of two single pole double through (SPDT) switches and two microstrip line s with desired phase difference . Two phase shifters are designed within the patch antenna matching feed line to save the area. A b eam steering functi on can be controlled electronically by changing the dc bias on each switch. A Doppler radar system with this adaptive beam steering antenn a can be fabricated on the same board to save the area and cost . Figure 2 10 shows the schematic of the beam steering phase array antenna . The b eam steering of the ph ase array antenna requires constant phase inc rement for adjacent antenna s . The phase delay , , is calculated as follows [26] : (2 9) w here P , 30 mm, is the pitch of array elements and is the beam steering angle, 22° . The HPBW of 2 2 fixed beam antenna is 41°. A 2 2 patch array antenna with ±22° beam steering angle s could achieve 85° coverage. The phase delay is calculated as 78° by (2 9) . Two phase shifters are required to implement a three way beam steering antenna. Figure 2 10 shows the schematic of the adaptive beam steering antenna. It consists of two antenna arr ay elements, two phase shifters, and one power combiner . Three steering angles can be a chieved by a co mbination of two phase delay values at each antenna. The same phase delay at two antennas can generate the broadside beam direction, which is operated as a conventional fixed beam array antenna. By generating the phase difference between the antenna A d B transmission line ( TL ) 1 and TL 2 or between TL 3 and TL 4 is designed as 78° or 78° . T he

PAGE 39

39 Figure 2 10 . S chematic of the adaptive beam steering antenna. It consists of two antennas, two phase shifters, and one power combiner. The beam steering antenna can steer in three different directions. radiation beam can steer in either 22° or 22°. The size of this antenna system w ill be huge if the discrete phase s hifter chips and po wer combiner chips are used with each antenna element . The power loss might be large as well. Each antenna array element is a linearly polarized two by one (2 1) patch array ant enna. In order to reduce the size of the antenna system, the switched line phas e shifters and the power combiner are integrated with in the antenna feedline network . S1 , S2 , S3 , and S4 are SPDT switches (SKY13348 374LF). The output of this absorptive type switch is terminated to tion for the limited space. The selected microstrip lines which transmit the signal are controlled by SPDT switches. By feeding adjacent antennas with different phase s of signals, the phase array antenna can steer to different angles. This antenna is designed and fabricated on the Rogers 4350B laminate with

PAGE 40

40 3 0 mil thicknesses . Two phase shifter s are designed and optimized to fit in the space among array patches so that the overall size of the antenna remains the same size of the original patch array a ntenna. Phase array patch a ntenna s and phase shifter s are modeled separately in full wave electromagnetic A NSYS HFSS v13 . T he SPDT switch model is extracted from the measurement data on the evaluation board . The whole adaptive beam steering antenna system is then simulated by connecting individual models of patch antennas, phase shifters , and the switches in Agilent Advanced Design System ( ADS ) . Figure 2 11 (A) shows a switched line phase shifter design consisting of t wo 50 microstrip lines, a and b and t wo SPDT switches (SKY13348 374LF ) . Two microstrip lines with different electrical len gth are designed in HFSS. The phase difference between the microstrip line, a, and the microstri p line, b, is designed in 78 °. Two microstrip lines should be for the matching of SPDT switches . Figure 2 11 (B) shows the impedance variance in time domain reflectometry (TDR) simulation, which can track the impedance Figure 2 11 . Phase shifter is a switched line type design. (A) are designed with the 78° phase difference. Two SPDT switches are used to select the desired microstrip line (B) TDR simulation result is done in ANSYS HFSS v13. The impedance v ariation along these microstrip lines is small enough to reduce the reflection loss.

PAGE 41

41 variance along the trace. The impedance of the delay line, b , varies at section I, II, and III along this microstrip line. The smaller impedance variance can reduce the reflection loss. Figure 2 12 (A) shows t he phase delay , 77.6° at 5.8 GHz , between two microstrip a b the simulation. The reflection coefficients at port 1 and port 3 are both below 20 dB at 5.8 GHz for two microstrip lines in Figure 2 12 (B) . The reflection coefficients at port 2 and port 4 are the same as port 1 and port 3 due to the reciprocal of the passive components. Figure 2 12 . Simulated Results between two microstrip lines (A) p hase delay (B) Reflection coefficients Figure 2 13 shows the entire structure of the adaptive beam steering antenna design in ADS. Two 2 1 patch array antennas in red dash dot frame, two switched line phase shifters in in black dash frame are simulated in HFSS. Four SPDT switches in green solid line frame are measured by Agilent VNA E8361. All components are cascaded and simulated in ADS. Figure 2 14 shows the simulated radiation pattern of 2 2 patch array antennas when the beam steers in directions of 22°, 0°, and 22°. It does not include the coupling effect of the p hase shifter , the loss from the SPDT switches and a dc blocking capacitor, and the soldering parasitics . The gain of this antenna at 5.8 GHz is 10.9 dB and the HPBW is 41 ° .

PAGE 42

42 Figure 2 13 . E ntire structure of the adaptive beam steering antenna is cascaded and simulated in ADS. The p atch array antennas, switched line phase shifters, and the power combiner feedline are simulated in HFSS. SPDT switches are measured in VNA . Figure 2 14 . Simula ted radiation pattern of 2 2 patch array antenna when the beam steers in 22°, 0°, and 22°.

PAGE 43

43 The reflection coefficients (S11) of the antenna in the simulation and measurement results are shown in Figure 2 15 . Dot ted lines are the simulation results (S) and the solid lines are the measurement results (M) . The green lines are the results when the beam angle is equal to 22°. The red lines are the 0 ° and the blue lines are 22°. The bandwidth of the antenna is from 5.7 GHz to 5.9 GHz at 22° and 0° and it is reduced to 5.72 5.81 GHz at 22° . The variation is caused by the phase shifters. The isolation between two transmission lines within a phase shift er is not perfect. When t he signal tra vel a phase shifter to steer the beam, the signal might be couple d to the straight line , , and degrades the reflection coefficients . Furthermore, the space is very limited between the 2 nd antenna patch and the 2 nd phase shifter. T here will be another coupling between the delay line, b , of the 2 nd phase shifter and the 2 nd antenna patch when the beam steers to 22° . In addition, the variation of the SPDT switches and soldering parasitics also affect the result. Figure 2 15 . Reflection coefficient (S11) in both simulation and measurement. Three dotted lines are the simulation results (S) with three different beam steering angles a nd three solid lines are the measurement results (M) .

PAGE 44

44 Figure 2 16 shows the adaptive beam steering antenna and the fixed beam antenna. Both sizes are 67 mm x 65 mm, which are approximately half size of a beam steering antenna with the Butler matrix. Four SP DT switches and one dc blocking capacitor are soldered on the front side of the antenna. A dc bias control board is attached to the back side of the antenna through header s . For an antenna performance test, DIP switch es are soldered on S1 to S4 to manually determine the antenna beam steering direction according to the truth table in Table 2 1 . Figure 2 16 . Pho tographs of (A) adaptive beam steering antenna and (B) fixed beam antenna Table 2 1 . T ruth table of beam steering functions Steering angle S1 S2 S3 S4 22° R L R L 0° L L L L 22° L R L R Figure 2 17 shows the photograph of the antenna chambe r, where t he radiation pattern of the adaptive beam steering antenna is measured at three different directions . A battery and DIP switches are attached in the back side of the antenna to reduce the coupling effect on the radiation pattern.

PAGE 45

45 Figure 2 17 . Photograph of antenna chamber

PAGE 46

46 Figure 2 18 . Measured r adiation pattern s including both co polarization and cross polarization in the E plane at different beam steering angles: (A) 20° (B) 0° ( C ) 27°.

PAGE 47

47 Figure 2 18 shows the measured radiation patterns including the co polarization (co pol) and the cross polarization (cross pol) in the E plane when the beam steers to the left (20°), the center (0°), and the right ( 27°). There is a 2° 5° variation due to the process variation, soldering effect, and the variation in SPDT switches. The maximum gain is 6.46 dB at 5.8 GHz, which is 4.5 dB smaller than the simulated gain of the 2 2 patch array antenna. The additional lo ss is from the coupling of the phase shifters, the loss of the SPDT switches and the dc blocking capacitor, and the soldering parasitics. Each SPDT switch w ill c ontribute 1 to 2 dB loss at 5.8 GHz. It is still a good result compared with the commercial phase shifter. The HPBW is 42° which is in good agreement with the simulation result . 2.5 Human Vital Sign Detection and Comparison Figure 2 19 . Photo graph of measurement setup Considering the vital sign detection, reflected modulation signal from the human chest is collected through the antenna. The chest height variation for different people sitting on the chair

PAGE 48

48 could result in misalignment and loss of detection. It is inconvenient to adjust the height of the system when detecting different people. To demonstrate the coverage issue, a subject sit s at a distance of 50 cm in front of the radar system in Figure 2 19 . A 2 2 reference fixed beam antenna and the adaptive 2 2 beam steering antenna are connected with the 5.8 GHz noncontact vital sign radar system, respectively. The elevation of the antenna is positioned 20 cm lower than human chest, which creates a 21.8° angle between the target chest area and broadside direction . The experimental subject holds the breath for 10 sec ond to eliminate the effect from the respiration, which helps to demonstrate the antenna cove rage effect of measuring heartbeat. Figure 2 20 . V ital sign measurement result s in time domain (A) ( C ) from the adaptive 2×2 beam steering antenna and (D ) from the reference 2×2 fixed beam antenna . Figure 2 20 (A) ( C ) show the heart rate measurement result s in time domain with the adaptive 2×2 beam steering antenna switching its beam to different angle s and Figure 2 20 (D ) is the result of the 2×2 reference fixed beam antenna. When the beam steering antenna is switched

PAGE 49

49 to 22 ° (pointing upward) , the result shows clear periodic heartbeat signal. When the beam steering antenna is switched to other angles or when the reference antenna is used, the result s show nothing but noise. Five people are tested with the same scenario and all measurement results show the similar patt erns. Therefore, adaptive beam steering antenna indeed helps to improve SNR when the target area and the antenna broadside direction are misaligned . Also, the dc bias of the phase shifter can be controlled by a micro processor such as MSP 430 for automatic adaptive control.

PAGE 50

50 CHAPTER 3 CONCURRENT DETECTION OF VIBRATION AND DISTANCE USING AN ADAPTIVE BEAM STEERING ANTENNA 3.1 Introduction Our previous work in C hapter 2 has successfully demon strated the better coverage of using a beam steering antenna o n the vital sign radar detection system [30] . In addition to the vibration detection, a radar system is widely used for subject positioning [31] or the distance measurement [32] , [33] . Nevertheless, [31] required multiple radar sensors for detection. [32] and [33] successfully demonstrated using the frequency modulated continuous wave (FMCW) radar system. FMCW r adar, however, is hard to monitor tiny vibrations due to the resolution limited by bandwidth. The hybrid FMCW interferometry radar has been introduced to measure the vibration and distance information alternately [34] , [35] , but the high cost filter and complex architecture were used in the system. On the other hand, Wang et al. [36] implemented a system for concurrent vibration and dis tance detection, but additional tag transmitting modulated signal is required for the system. A conventional unmodulated CW radar system, however, is not able to detect the distance due to the lack of the time information. It was found out that the dc offs et in the baseband signal contains useful information on distance. Lv et al. [37] developed an algorithm of dynamic dc offset tracking for the moti on imaging detection on a digital IF receiver architecture. A high speed ADC is required in the system. In addition, Kim et al. [38] analyzed the phase to extract t he target displacement. The exact position, however, cannot be extracted due to the lack of phase residue information. In this work, by extracting the phase from the dc offsets and processing the phase difference between propagation paths at different ante nna beam steering angles, the distance to a short range target is possible to be accurately calculated with a fast beam switching antenna in a homodyne architecture.

PAGE 51

51 A beam steering antenna with a fast switching rate is required for distance detection in the proposed approach. Many adaptive beam steering antennas have been studied . Tawk et al. [39] demonstrated a reconfigurable antenna design with a rotator attache d to a m otor. Nevertheless, the mechanical switch is not as fast as the electronic switch. Yusuf et al. [40] implemented a beam steering antenna using mutual coupling and reactive loading , but additional control circuit is required to switch the radiation beam in real time. Park et al. [24] introduced a beam steering antenna design. The antenna beam direction can be electronically controlled. Its low gain antenna, however, degrades the SNR for the Doppler noncontact vibration radar system. An antenna array with a Butler matrix exhibits high gain [41] , but additional switch is required for beam steering selection. The size of the entire antenna system at 5.8 GHz w ill be large, and it cannot provide a 0° (broadside) beam direction. A patch array antenna with the beam steering function is a better solution. By using the beam steering a ntenna from our previo us work [30] , increasing the complexity of the architecture , which can be used for structural health monitoring . This chapter is organized as follows. Chapter 3.2 introduces the vibration detection with a beam steering antenna in real time. Chapter 3.3 analyze s the dc offset and identifies the desired dc information for the distance detection. The design and implementation of the entire Doppler noncontact vibration radar system are described . A method using an adaptive beam steering antenna to detect the distance to the target is introduced as well . Chapter 3.4 shows the experimental results and the analysis. 3.2 Vital Sign Detection with a Beam Steering Antenna in Real Time The homodyne architecture of the Doppler noncontact vibration radar system is shown in Figure 2 3 . The photograph of the Doppler noncontact vibration radar transceiver is shown in

PAGE 52

52 Figure 3 1 . The size of the radar transceiver board is 7.3 cm 10.5 cm, which is suitable for a mobile system. An adaptive beam steering antenna at Tx and a fixed beam 2 2 patch array antenna at Rx are connected to the Doppler noncontact vibration radar board through SMA con nectors. 7.3 V is applied in a system and the total current of the system is 0.35 A, which Figure 3 1 . P hotograph of the Doppler noncontact vibration radar transceiver Figure 3 2 . C ontrol signal for adaptive beam steering antenna in LABVIEW (A) the s chematic and (B) the corresponding measured vibration signals.

PAGE 53

53 consumes the power of 2.55 W. In order to steer the beam of the adaptive beam steering antenn a in real time, four 5 V control signals are needed to connect with the antenna. National Instruments (NI) USB 6009 is a 14 bit low cost multifunction data acquisition (DAQ) and used as an ADC. Analog signals can be digitalized and recorded with the softwa re, NI LABVIEW. NI USB 6009 is also able to generate the digital signals to control the adaptive beam steering antenna. Nevertheless, the switching rate of the digital signals can no t be faster than 1 00 Hz . Figure 3 2 (A) shows the schematic of control signal s for an adaptive beam steering antenna in LABVIEW and the corresponding measured vibration signals. This circuit follows the truth table in Table 2 1 to control the antenna steering in two directions fast and alternatively. Figure 3 2 (B) shows the measured results of I and Q signals . Both baseband signals exhibit two different dc offset s corresponding to the switching time in the real time measurement. 100 Hz switching rate, however, may not b e fast enough to extract the distance to the target. Figure 3 3 . Control signals and s ystems b y the field programmable gate array (FPGA) (A) Photograph board (B) four controls signals for the adaptive beam steering antenna in time dom ain

PAGE 54

54 Figure 3 3 (A) shows the photograph of the FPGA board. It can generate four digital output s at high frequency. Figure 3 3 (B) shows the four controls for the adaptive beam steering antenna in the represent the beam steering to the up, center, and down directions, respectiv ely. These four CLK1 CLK2 Figure 3 4 . The switching rate of this adaptive beam steering antenna is 1.2 kHz, which means the vibration detection time at e ach steering angle is 0.83 ms. T he sampling rate is 4.8 kHz. The order of the beam steering angle is 20° ( U ) , 0° ( C ) , and 27° ( D ) . Fast switching rate is required to detect the distance to the target. Figure 3 4 . Control signals of the adaptive beam steering antenna. The switching rate is 1.2 kHz, which means 0.83 ms at each vibration detection period. The order of the beam steering angles is 20°, 0°, and 27 .

PAGE 55

55 3.3 Distance Detection Theory 3.3.1 D C O ffset A nalys is Figure 3 5 . Block diagram of the Doppler noncontact vibration radar system for mechanical vibration detection. Figure 3 5 shows the block diagram of the Doppler noncontact vibration radar system for mechanical vibration detection. After the transmitted signal, T(t) , is modulated with the target movement, x(t) , and reflected, the received signal contains the information of the vibration rate and distance, d 0 . The signal is down converted to the baseband in the radar transceiver system. Then, the baseband signal is digitized through an ADC for the signal process. In this chapter , quadrature signals are used to deal with the null detection point issue. Assuming the target is a single tone movement, they can be expressed by the Bessel function [17] : (3 1) (3 2) w here is a signal of the target, m is the amplitude of the movement, is the angular frequency, C i =J i (4 is the amplitude at each frequency point, is the total

PAGE 56

56 accumulated residual phase shift from the phase delay in the circuit and the wireless transmission path , and DC I and DC Q are the dc components containing the information of the amplitude of the vibration signals and the total residual phase shift: (3 3) (3 4) These are the desired dc information and can be used to recover the vibration signals for the arctangent demodulation [8] . In the constellation plot, the quadrature vibration signal s will form the arc on the unit circle centered at the origin. DC I and DC Q can be estimated to average values of the arc under the l inear approximation. Nevertheless, measured I(t) and Q(t), are not exactly the same as the (3 1 ) and ( 3 2 ): (3 5) (3 6) w here A I and A Q are the amplitudes of the periodic signals in the measurement and DC I0 and DC Q0 are the additional dc offsets. A I and A Q are determined by the power level. They might be slightly different due to the imbalance of the I/Q mixer. DC I0 and DC Q0 are caused by the imperfections of the circuit such as the local oscillator (LO) leakage and the finite isolation between the Tx and the receiver (Rx), and by the clutters reflected from the stationary objects. These are the undesired dc components, which do not contai n any useful information. DC I0 and DC Q0 would shift the center of the circle off the origin while A I and A Q would change the radius of the circle in the constellation plot .

PAGE 57

57 Figure 3 6 . C onstellation plot show ing that the dc offset of the measured data is off the origin and the radius, Ar, is not equal to unity . The circle in red is the unit circle centered at the origin as a reference. Figure 3 6 shows the measurement result of the periodic signal in the constellation plot. A Doppler noncontact vibration radar sensor is located at 1 m away from the actuator which generates a 0.2 Hz sinusoidal movement with 5 mm peak t o peak displacement. The black and small circle with its center at ( 0.2, 0. 12) and a radius, A r , is fitted by the arc in blue which is the measured periodic signal. The large circle in red is the reference unit circle with its center at the origin. As des cribe d in (3 5) and (3 6 ), the measured signal contains the undesired dc offsets and the amplitude of the signal is not equal to unity. A r is determined by A I and A Q in (3 5) and (3 6 ), where DC I0 and DC Q 0 represent the center of the circle, ( 0.2, 0.12). After removing the undesired dc offset and normalizing the measured circle to the unit circle, the perio dic signal is in the form of (3 1 ) and ( 3 2 ). The recovered data will be an arc on a unit circle cent ered at the origin.

PAGE 58

58 3.3.2 Distance D etection with a n A daptive B eam S teering A ntenna The desired dc components in ( 3 3 ) and (3 4 ) contain the information of total accumulated phase residue: (3 7) w here d 0 is the distance between the D oppler vibration radar system and the target , r is the phase delay due to the Rx , and 0 (t) i s the initial phase from the oscillator. The received signal would be self mixed with the transmitted signal. The phase noise of the oscillator can be cancelled out due to the range correlation effect [6] , which means 0 (t) 0 (t 2d 0 / 1 . Therefore, the total phase residue only contains the distance infor mation and the phase delay at Rx . The distance information could be extracted out by using an adaptive beam steering antenna in Figure 3 7 . The Doppler vibration rada r system with a n adaptive beam steering antenna is able to radiate the beams in two directions fast and alternately. Figure 3 7 . Doppler noncontact vibration radar system with a n adaptive beam steering antenna can radiate the beam in two different directions fast and alternately. The angle between two radiation beam s is . Under the assumption of the single beam model, the antenna gain is infinite and the angle of two radiation beams is e qual to . Then, the total phase residue with the angle of the beam w ill be:

PAGE 59

59 (3 8) Where is the phase difference of the antenna from the phase shifter when the antenna beam steers . Again, the last two terms would be much smaller than unity due to the range correlation effect . r i s in (3 7) and (3 8 ) can be cancelled out if the antenna steering rate is fast enough. Thus, t he difference of total phase residues between (3 7) and (3 8 ) is as follows: (3 9) where the distance to the target can be determined as: (3 10) is designed in the phase shifter and is determined by the antenna . After getting the desired dc value, the total phase residue can be calculated by dividing DC I by DC Q in ( 3 3 ) and ( 3 4). (3 11) The distance to the target can be extracted in ( 3 10 ) after 1 and 2 are determined. In order to implement this idea, an adapti ve beam steering antenna is required for the Doppler noncontact vibration radar system.

PAGE 60

60 3.4 Experiment and A nalysis Figure 3 8 . T esting environment of the Doppler vibration radar system. A 15 cm × 15 cm met al plate with an actuator is located at 0.5 m away from the radar system. The actuator generates a 0.2 Hz sinusoidal movement with 5 mm peak to peak displacement . Figure 3 8 shows the testing environment of the Doppler noncontact vibration radar system. The target is a 15 × 15 cm flat metal plate on the actuator. The actuator generates a 0.2 Hz sinusoidal movemen t with a 5 mm peak to peak displacement . The target is located at 0.5 m away from the radar system. The far field boundary can be estimat ed as follows [26] : (3 12) The far field boundary for our antenna is calculated as 0. 1 7 m. Therefore, the target is in the far filed region. An adaptive beam steering antenna is inst alled in the Tx with the beams in up (20°), straight (0°), and down ( 27°) directions. A conventional 2×2 fixed beam antenna is installed at Rx . The center of the antennas is vertically aligned with the center of the metal plate.

PAGE 61

61 Figure 3 9 shows the baseband I(t) and Q(t) signals of the measurement data in the time domain. It shows three different dc offsets for both I(t) and Q(t) signals when the beam steers in different directions, caused by imbalance antenna gain, and different phase due to the phase shifter of the adaptive beam steering antenna. The periodi c vibration frequency can be extracted by complex signal demodulation (CSD) [3] after we separate three signals in each direction. Figure 3 9 . Baseband I(t) and Q(t) signals of the measurement data. Three d ifferent dc offsets are observed for both I(t) and Q(t) channels. Figure 3 10 shows the normalized frequency spectrum of the measurement data. The re d line, the green cross line, and the blue dotted line are the signals when the antenna beam points to the up (20°), the straight (0°), and the down ( 27°) directions, respectively. Strong peaks at 0.2 Hz are clearly measured in three directions. The noise floor at 0° is 5 to 10 dB lower than the noise floors in 20 ° and 27 ° because of the beam steering antenna gain variation at each direction at Tx and the use of fixed beam antenna facing broadside at Rx. The SNR at 0 ° is the best.

PAGE 62

62 Nevertheless , the SNR is sufficient for the three scenarios to extract the vibration rate and the distance to the target. Figure 3 10 . F requency spectrum of the measurement data when an adaptive beam steering antenna steer s in three directions. The periodic movement with 0.2 Hz is accurately detected at three directions. Figure 3 11 (A) shows the measurement results of th e target at 0.5 m away from the Doppler vibration radar in the constellation plot. Three circles are fitted by the arcs to get the dc information. The circle of the beam direction, 20°, is located at (0.21, 0.3) with a radius of 0.94. The circle of the bea m direction, 0°, is located at (0, 0.22) with a radius of 0.39. The circle of the beam direction, 27°, is located at ( 0.22, 0.3) with a radius of 0.5. They all contain undesired dc offsets and should be removed. The distance extraction only requires the desired dc offset. Figure 3 11 (B) shows the results after recovering dc offsets. All three arcs are located on the unit circle centered at the origin. The distance of the target could be extracted by ( 3 7 ) to ( 3 11 )

PAGE 63

63 Figure 3 11 . (A) Meas u red and (B) recovered results in the constellation plot. T he target is at 0.5 m away from the radar system when the adaptive beam steering antenna steers the beam in t hree directions alternately. Recovered data are after removing and normalizing the undesired dc offsets. Three arcs are on the unit circle with its center at the origin.

PAGE 64

64 from these data. Furthermore, the arctangent function in ( 3 11 ) is a periodic function, which means the phase dela y, 1 2 , will have multiple solutions with increment of in ( 3 10 ). Therefore, the extracted distance, d 0 will have multiple solutions separated by : (3 13) Where is determined by the wavelength and the beam steering an gle. TABLE II summarizes the calculation results from the measurement data when the steerin g angles are 20° and 27° at 5.8 angles 20° and 27°. The actual distance to A+ 5 in the beam direction of 27°, respectively. The accurate distance value can be extracted by selecting the correct range section. with the 20° steering angle is larger than with the 27° steering angle. can be extended by using a smaller steering angle and a lower carrier frequency. For example, if the beam steering angle is equal to 20° at 5.8 GHz, is 20 cm. If the steering angle reduces to 10° at 2.4 GHz, can increase to 2.01 m, which should be sufficient for short range Doppler noncontact vibration detection application. Figure 3 12 shows the measured distance at three different detection distances: 0.5 m, 0.7 m, and 1 m with two different angles, 20° and 27°. The solid lines are the measured distance, and the dotted lines are the measured error percentage. The measured distance with both 27° and 20° beam angle s matches the actual distance with an error less than 8%. The errors are mostly from the measurement error and dc value extraction errors. The dc value is extracted by fitting the arc of the periodic signals to the circle under the assumption of balanced a mplitude on I(t) and Q(t) signals.

PAGE 65

65 Table 3 1 . Distance calculation results from the measurement data 2 1 (rad) Distance (m) (20°) Distance ( m ) (27°) A+ 0.274 0.073 A+ 0.475 0.179 A+ 0.677 0.285 A+ 0.878 0.390 A+ 1.080 0.496 ----------------0.201 0.106 Figure 3 12 . Target is measured at three different distances with two beam steering angles. The error of this method is less than 8%. One specific application of this technique is the detection of building deformations due to man made vibrations or natural seismic events. The technique can simultaneously monitor the h might not be noticeable through visual inspection. Figure 3 13 shows the photograph of the seismic experiment by using a seismic simulator (Quanser Shake Table II). A simulator can generate 1 D seismic vibration. A 7 floor architecture is fix ed on the shake table. Figure 3 14 shows the experimental results. The beam steering radar was placed at 0.5 m away from the simulator and beam steering angles of 0° and 27° were used to measure the distance as well as the vibration. During the seismic vibration,

PAGE 66

66 the natural frequencies of the seven floor architecture were successfully measured and recorded. The natural frequencies are determined by the structure of the building instead of the vibration source. It is one of an indicator of the structural health. The nature frequencies shift ed when the building was deformed after the vibration, as shown in Figure 3 14 (A) . The first nature frequency shifts from 1.48 Hz to 1.56 Hz and. The distance variat ion can also be detected from the constellation diagram. Figure 3 14 (B) shows vibration traces at 0.5 m and 0.55 m away, respectively. Using the proposed method, the distances to the target are calculated to be 0.47 m and 0.56 m, respectively. There are possible errors when determining the distance using a ruler. Nevertheless, the radar measurement shows a distinguishable shift of the vibration trace, which Figure 3 13 . Photograph of the seismic simulator experiment

PAGE 67

67 Figure 3 14 . Measured results before and after the seismic event emulated by the shaker (A) in f requency spectrum (B) Recovered vibration traces when the distance changes.

PAGE 68

68 A new method of detecting the distance to the target using CW radar is developed. An unmodulated CW radar system with a fast switching beam steering antenna is able to detect the distance to the target. The periodic movement and the target distance can be accurately detected at the same time. An adaptive beam steering antenna is implemented at 5.8 GHz. With the additional phase shifter integrated on the same b oard, the size of the beam steering antenna is the same as the original 2 2 fixed beam antenna. The vibration rate is accurately detected at three different beam steering angles. T he error of the distance to the target is less than 8 % when the target is l ocated from 0.5 m to 1 m. This technology is suitable for structural health monitoring of buildings affected by man made vibrations or natural seismic events. The detection range section, , can be further extended by using a beam steering antenna with a smaller beam steering angle and operating at the lower carrier frequency.

PAGE 69

69 CHAPTER 4 MINIATURE DIRECTIONAL COUPLER DESIGN 4.1 Introduction In conventional transceiver system, two antennas are normally used in both Tx and Rx . T wo antennas , however, should be designed in the same polarization to maximize the received signal and the i solation between two antennas also need s to be taken into consideration. Chapter 1 has demonstrated t he effect of the misalignm ent for either Tx or Rx antenna s . In Dopple r vital sign radar system, the size is important for commercial mobile device. A patch array microstrip antenna is widely used due to its cost advantage. Nevertheless , the size s of two patch array antennas at 5.8 GHz are large . The mis alignment between bot h antennas also degrades the SNR because the vital sign RCS of the human body is small . Therefore, o ne antenna with the branchline coupler architecture is proposed for Doppler vital sign radar sensor. Figure 4 1 . Photograph of t he Doppler radar system architecture with one branchline coupler and one antenna Figure 4 1 shows the photo graph of thi s architecture. A c ommercial 3 dB hybrid direct ional coupler is used for this experiment . There are four ports of th is coupler: an input port, an isolated port, a through port, and a coupled port. An inpu t port is connected with the Tx and an isolated port is connected with the Rx. One antenna shared between the Tx and Rx is connected with a

PAGE 70

70 through port. A c oupled port is connected with the LO input of the I / Q mixer for self mixing [7] . The power from an input port is equally divided into a through port and a coupled port. Both ports w ill receive the power 4 dB lower than the power at an input port at 5.8 GHz. The p ower level at an isolated p ort is 2 0 dB lower than the power level at an input port. Therefore, an u n modulated signal at Tx from the oscillator will be equally divided and fed into an antenna and a LO input port of the mixer through th is coupler. After the transmitted signal radiates out and reflected by the target through an antenna, t he signal w ill be received from the same antenna. The received signal w ill be equally divided and fed into the Tx and Rx again . The received power is much lower than the power at Tx, so it does not affect the signal at Tx ve ry much . The modulated signal at Rx w ill be down converted with the LO signal into the baseband through the I/Q mixer. The heart rate and respiration rate signal s w ill be extracted and analyzed in the baseband signal. The vital sign signal s are also shown in Figure 4 1 . Figure 4 2 . S tructure s and their ideal simulation results of the conventional two branch and t hree branchline couplers .

PAGE 71

71 A microstrip branchline coupler can be processed and integrated with a Doppler vital si gn radar system in the same PCB to save the cost . Figure 4 2 shows both the structure and the ideal simulation results of the conventiona l two branchline and three branchline couplers. A t hree branchline coupler exhibits the wider bandwidth and better isolation than a two branchline coupler. Considering the process variation and avoiding the T x leakage, a three branchline coupler is a better choice for the system integration . Nevertheless , the size of the three branchline coupler is two time s larger tha n a two branchline coupler. A two branchline coupler consists of four microstrip lines: two series arms and two shunt arms. The length of each b ranchline of coupler is a quarter wavelength. The system operating at 5.8 GHz is integrated on the board of Roge rs 4350B with 30 mil thicknesses . The length is around 6.8 mm since the dielectric constant of the board is 3.6. The impedance of the series arm is designed as Z 0 and the impedance of the shunt arm is Z 0 . Lots of researches have been done to minimize the size and enhance the bandwidth of the couplers [42] [51] . [42] [45] used the lump model to exact the effect ive inductor s and capacitors for the coupler design. T his method , however, is only suitable for the design at low frequency. [46] , [47] , and [48] implemented open stubs to enhance the bandwidth , but thos e designs occupied more space than the conventional coupler. [49] , [50] , and [51] successfully miniaturize the size of the two branchline couplers at high frequency. Here, w e propose to extend this miniature technique to a three branch line coupler. A wideband branchline coupler with the acceptable size is designed and implement ed for a Doppler v ital sign radar system. A three branch line coupler looks like two cascaded two branch line couplers and con sists of five microstrip lines. The impedance s of two series arms and the center of the shunt arm are equal to Z 0 and the impedance s of two side shunt arms are 2.4 ×Z 0

PAGE 72

72 4.2 Miniature Three Branchline Coupler Design A three branchline directional cou pler is twice longer than a two branchline directional coupler. In order to reduce the size of the coupler, the equivalent quarter wavelength microstrip line is designed. A microstrip line can be modeled in T shape which consists of three microstrip lines in Figure 4 3 . Z a , Z b , a , and b are the characteristic impedances and electrical lengths of the microstrip lines in T model structures. Figure 4 3 . E quivalent model of the microstrip line A quarter wavelength microstrip line can be expressed in ABCD matrices: (4 1) Meanwhile, ABCD matrices of the T model microstrip line structures are as follows: (4 2) From equation s (4 1) an d (4 2), compare each sub matrices of two matrices:

PAGE 73

73 (4 3) The relationship between Z a , Z b , a , 0 and b can be determined: (4 4) In T model design, if we can design a two times smaller than the quarter wavelength, we can miniaturize the size of the coupler . The impedance of the series arm in three branchline coupler is 35.5 . By designing Z a = Z b = , the length of the series arm can be reduced to two times of a , which is 57.2 ° . All design parameters are summarized in Table 4 1 . Therefore, the size of this mini ature coupler can be reduced to half of its conventional type of three branchline coupler. Table 4 1 . P arameter summary of miniature microstrip line The impedance ( ) The electrical length ( ° ) Microstrip line Z0=35.5 Miniature microstrip line Za=65, Zb=45 To demonstrate the accuracy of the formula (4 4), the simulation results of an ideal microstrip line and an ideal T model microstrip line are shown in Figure 4 4 . The red dotted line is the ideal quarter the T model microstrip line. The equivalent electrical length of the T model microstrip line is in good agreement with the microstrip line at 5.8 GHz and the impedances of both transmission model microstrip line can be perfec tly equivalent to a conventional microstrip line. After determining the parameter of each block in a T model microstrip line, EM simulation is required to determine the actual size of each

PAGE 74

74 Figure 4 4 . Simul ation result s of an ideal single microstrip line and an ideal T model microstrip line structure in ADS microstrip line. The Rogers 4350B board with 30 mil thickness is used to integrate with the Doppler radar system and the coupler . The design and the simulation result s of the T model microstrip line are shown in Figure 4 5 . The equivalent impedance is 34.2 and the equivalent electrical length is 89 ° at 5.8 GHz. When designing a T model microstrip line, th ere are four parameters which will affect the effective impedance and the phase of the transmission line. The design procedure is to firstly determine the effective impedance. Design two microstrip lines , Z b and Z a , separately. Then, build the T model microstrip line with these two microstrip lines and Figure 4 5 model microstrip line

PAGE 75

75 fine tune with it. The effect of b on the effective impedance is larger than a . Larger b will lower the effective impedance while reducing a can increase the effective impedance. Tune these parameters back and forth to get the desired impedance. It is not necessary to tune to the T model microstrip line to the coupler. Figure 4 6 shows the size comparison of the conventional and T model two branchline couplers in HFSS. The length of the shunt arms is usually extended 10% to 20% at the junction of the branchline coupler in a rule of thumb . The area of this T model type branch line coupler ca n be reduced by 19%. It is possible to further minimize the size by implement ing T model microstrip lines in shunt arms. Here just shows the concept of the miniature of th e directional branchline coupler. Figure 4 6 . S ize comparison between the conventional and T model two branchline coupler s in HFSS. A three branchline coupler is designed to achieve a wider bandwidth. The size comparison between the conventional and T model three branch line couplers is shown in Figure 4 7 . The size reductio n is around 36%. A T model three branchline coupler is only 18% larger than a conventional two branchline coupler. It is almost impossible to implement T model microstrip lines on two side shunt arms because the impedance s of both transmission lines are as

PAGE 76

76 high as 120 . The width of both microstrip line s is about 0.18 mm. A T model microstri p line consist s of one high impedance series arm and one low impedance shunt arm. 0.2 mm is the smallest width in our process. In addition , the length of the center shunt arm can not be reduced. The final layout with the size information is shown in F igure 4 8 . Figure 4 7 . S ize comparison be tween the conventional and T model three branchline couplers in HFSS. F igure 4 8 . F inal layout with the size i nformation of the T model three branchline coupler The EM simulation is implemented in HFSS. The E field plot shows the energy distribution at 5.8 GHz in Figure 4 9 . Most energy is transmitting to the direct port and the c oupled port from the input port. The isolated port received sm all amount of the energy. The S parameter plot exh ibits the similar result s . The isolation can be achieved as good as 31.9 dB at 5.8 GHz. Th e isolation is better than 20 dB from 5.15 GHz to 6.26 GHz. The bandwidth is about 1 GHz. To consider the process variation of the board, the reliability test is performed. The

PAGE 77

77 thickness of the solder mask changes from information, the thickness varies from 10 um to 11 0 u m while the default value is 22 um. Furthermore , a coupler will be integrated into the Doppler radar system. Four mic rostrip Figure 4 9 . EM simulation of the T model three branchline coupler in HFSS. Th e E field pattern shows the energy distribution at e ach port and the S parameter plot shows the bandwidth of the coupler is 1 GHz. Figure 4 10 . EM simulation result when the solder mask thickness is 110 um and the input microstrip line varies 10%. lines at each port will not be the straight lines due to the layout of the system. The impedance of the microstrip line will vary if the microstrip line is bent or the microstrip line width changes due to the proce ss variation. Figure 4 10 shows the simulation result when the solder mask thickness

PAGE 78

78 changes to 110 µ m or the width of the input microstrip line vari es 10 %. The varia tion of the S parameter s is not huge. Neverth e less, one thing worth to mention is that the isolation and the return loss are very sensitive to the impedance at each port. 20 dB is the realistic target for both the isolation and the return loss. 4.3 Measurement Results and Analysis A miniature three branchline coupler is designed in the previous section. Figure 4 11 shows t he photograph of this coupler. A miniature three b ranchline coupler is covered with the green solder mask. Four SMA connectors are soldered on a miniature three branchline coupler. The branchline coupler in the left is surrounded by the dummy ground. The branchline coupler in the right removes the dummy g round which is the same design as the simulation. Figure 4 12 shows the measurement results of both couplers. The plot in the left is the coupler with the dummy ground while the other plot in the right is the coupler without the dummy ground. The dummy ground has huge effect on the coupler design. The measurement result of the coupler without the dummy ground shows a good agreement with the simulation result. The dummy ground should be taken into consideration when you design the coupler in the system. Figure 4 11 . P hotograph of the miniature three branchline couplers Dummy Gnd

PAGE 79

79 Figure 4 12 . M easurement results of both coupler s with the dummy ground (left) and without the dummy ground (right). Figure 4 13 . Basic structure of the two branchline coupler and the even odd analysis The return loss and the isolation are not as good as the simulation result. It is caused by the bad matching at the ports of the coupler. For simplification, this ef fect can be analyzed in the two branchline coupler since they both belong to the symmetric branchline coupler structure. It can be evaluated by the even odd mode analysis due to the symmetric structure [21] . Figure 4 13 shows the basic structure of the two branch line coupler and the even odd analysis. It con sists of two series a rms and shunt arms with quarter wavelength s . The structure is divided into two sections horizontally. Shunt arms become open circuit stubs with one eighth wavelength in even mode. Both ports 1 and 4 are fed into 0.5 V, respectively. T he ABCD matrix is derived for the transmitting and reflecting characteristics of the coupler . The shunt arms are set into an open circuit in even mode:

PAGE 80

80 (4 5) The shunt arms are set into short circuit stubs in odd mode. The port 1 is fed with 0.5 V while the port 4 is fed with 0.5 V. The ABCD matrix is as follows: (4 6) The reflection and transmission coefficients can be derived by the ABCD matr ix. The reflection coefficients in even and odd modes are shown as follows, respectively: (4 7) (4 8) (4 9) and (4 10) show the transmission co efficient in even and odd modes: (4 9)

PAGE 81

81 (4 10 ) Transmission and reflection coefficients in even and odd modes can be further converted into s parameter s at each port. Then, the characteristics of the coupler can be analyzed, including the insertion loss, the isolation, the reflection and so on: (4 11) The return loss, S11, must be zero under the matching conditions. According to the equation (4 11) , the matching criterion is shown as follows: (4 12) The isolation is also equal to zero under th is condition. Therefore, the degradation of the isolation and the return loss in the measurement might be caused by the impedance mismatch at each port. The soldering connection of the SMA connector is the coplanar waveguide structure for the mechanical ro bustness , which is not considered in the EM simulation. The soldering metal and conditions might also change the impedance. Figure 4 14 shows the simulation results of the branchline coupler when the simulation changes from 50 to 40 . 10 variation s are very common when you solder SMA connectors on the port of the coupler. The reflection, S11, and isolation, S13, degrade from about 24 dB to about 16 dB. This effect makes it difficult to integrate the coupler with the system because different components need to be soldered at each port. The output impedance of PA is usually much lower than 50 Also, c o nsidering the soldering condition and process variation, it should be careful to design the coupler.

PAGE 82

82 Figure 4 14 . Simulation results of the branchline coupler when the impedance termination changes from 50 to 40 .

PAGE 83

83 CHAPTER 5 RADAR SENSOR SYSTEM WITH MINIATURE COUPLER DESIGN 5.1 Introduction A previous Doppler vital sign radar system shows a transceiver with two antennas. Alignment between two antennas could be an issue as shown in Chapter 1. The sizes of the two patch array a ntennas are big at 5.8 GHz. A miniature directional branchline coupler i s designed and implemented in C hapter 4 . The size is much smaller than a 2×2 patch array antenna. The Doppler noncontact vital sign radar architecture with one anten na is demonstrated in Chapter 4.1 . Therefore, t he size of the system can be greatly reduced by applying the coupler with one antenna. The branchline coupler built on PCB is much smaller than the commercial directional coupler. In this chapter, The Doppler radar system will be integrated with a miniature directional branchline. It could further help the size reduction of the system. 5.2 Link Budget and System Design Figure 5 1 shows the schematic of the new architecture of the Doppler vital sign radar system. The carrier frequency , 5.8 GHz , is generated from VCO. The power level at VCO output is 2 dBm. Then, the carrier is amplified by VGA, HMC625LP5. Two amplification modes are designed and used: low power mode and high power mode. As VGA operates in low power mode, VGA will generate the 2 dB gain . On the contrary, It w ill generate the 9 dB gain in high power mode. The system can c hoose any power mode at different situations. If there is only high power mode, the signal might be saturated when the subject under the test is with random body movement s . The saturated signal w ill cause non linear effect s and lose the important informati on. On the contrary, if there is only low power mode, the signal strength is low when the target is placed far away from the system. The T model miniature three branchline coupler in Chapter 4

PAGE 84

84 Figure 5 1 . S chematic of the new architecture of Doppler vital sign radar sensor is used in the system. The signal power fed into port 1 w ill be equally divided into port 2 and port 4 with 4.5 dB loss each. The signal at port 4 will be amplified by a gain block, NBB 3 00, with 9.5 dB gain. An un modulated signal w ill be sent into LO port of the I / Q mixer, HMC525LC4. Another unmodulated signal at port 2 w ill be transmitting out through an antenna. When a signal hits a subject , such as a human body, the signal w ill be refl ected and modulated by the periodic movement of the human chest. The signal w ill be received by the same antenna into the system. One antenna is used and shared at both Tx and Rx, which saves lots of area s . The receipted power, P r , can be estimated by the radar equation in (2 1). In our cases, G t is equal to G r , 10 dB, because the Tx and Rx share with the same antenna. RCS is equal to 0.01 m 2 . The r ece ived power levels at the input of the LNA are approximately 53 dBm / 46 dBm at l ow/ high power modes when the subjec t is 1 m away from the system . The received signal fed into port 2 w ill be equally divided into port 1 and port 3 with another 4.5 dB loss each . It will not cause

PAGE 85

85 many effect s at Tx because the rece ived power is very small. In Rx, the received si gnal w ill be amplified through a LNA, HMC320MS8G, with 13 dB gain and 2.6 dB noise figure . After a LNA, there are two cascaded gain blocks, NBB 300, with 9.5 dB gain and 5.1 dB noise figure each. The received signal and the LO signal are originally created from the same VCO, which w ill c ancel out the phase noise in an I/Q mixer due to the short range correlation effect. Quadrature signals in the baseband are both amplified in a baseband amplifie r , MAX4078ESD+. There are four amplification stages in total: 1 0X, 25X, 50X, and 100X. All amplified signals at four stages will be sent into a multiplexer, Max4692ETE+. The biggest amplified signal without saturation w i ll be chose n to ADC from the microcontroller unit ( MCU ) . In the Doppler radar system, the receiver sensitivity is shown as follows : (5 1) Where RBW is the resolution bandwidth, B is bandwidth, f s is sampling rate of Doppler radar , F flicker is the noise figure from the flicker noise, F white is the noise figure from the white n oise, e ADC is the ADC quantization error referred at the input, and kT is the input thermal noise floor per unit bandwidth. RBW is determined by the time domain window size of the Fourier transform. The longer observation time window, the narrower RBW is. It causes less noise energy for a single tone signal in the periodogram. The vital sign signal is in the range between DC to 2 Hz. The flicker noise has a large impact. If F flicker is 36 dB at 1Hz, Fwhite is 6 dB, B is 1MHz, RBW is 0.1 Hz, and f s is 20 Hz, the input refer referred white noise is 1 31 dBm and the input referred flicker noise is 14 8 dBm. The white noise dominates in this case. The power level of the received signal is sufficient for our design. It can also improve the white noise b y

PAGE 86

86 increasin g the sampling rate. Furthermor e , the noise figure in the receiver chain can be calculated as: (5 2) Where F GB1 and F GB 2 are the noise figure s of the gain block 1 and 2 , F mixer is the noise figure of the I/Q mixer, and F BB is the noise figure of the baseband amplifier. G LNA , G GB1, G GB2, and G mixer are the gains of the LNA, the first gain block, the second gain block, and the I/Q mixer, respectively. The t otal gain in Rx is important. It hel ps to reduce the noise figure in a baseband amplifier in (5 2) . An I/Q mixer here is a passive device. The conversion loss, 8 dB, is equal to its noise figure. The flicker noise of the baseband amplifier at 1 Hz is large. High total g ain and Figure 5 2 . Layout of the Doppler vital sign radar system with single antenna in top, bottom, and side view

PAGE 87

87 low noise figure in an I/Q mixer and baseband amplifiers are required for a Doppler radar system. If the flicker noise at 1 Hz is 60 dB [16] , the white noise of noise figure for our system is 5.8 dB and the flicker noise of noise figure is 35.6 dB. Figure 5 2 shows the layout of this Doppler vital sign radar system with a single antenna in top, bottom, and side view s . The board size is 12.6 cm × 7.2 cm. The branchline directional coupler is designed in the left up corner. An antenna can be co nnected with it in the bo rder. Tx and Rx are separated by the coupler. Four regulators are designed to provide either 3.3 V or 5.0 V for different components. Digital signals including the MCU and Zigbee are isolated from the analog signals in the right bo ttom corner. The analog and digital powers are separated by two different regulators to avoid the digital noise contamination. Analog and digital ground s are separated with 3.5 mm gap. In the baseband region, there are two baseband amplifiers and one multi plexer. By default, the maximum amplification of quadrature signals w ill be sent into MCU, Figure 5 3 . Flow diagram of automatic gain control in a Doppler radar system

PAGE 88

88 MSP 430 firstly. If the quadrature si gnals saturate the ADC of MSP 430, it will send a controlling signal to the multiplexer to select 2 nd maximum amplification quadrature signals. The system will iterate this procedure until the optimal quadrature signals are determined. Figure 5 3 shows the flow diagram of this automatic gain co ntrol in a Doppler ra dar system. Analog signals are firstly digitized, processed in MSP430 , and then sent into Zigbee. The computer with a Zigb ee base station can wirelessly receive the final analysis result s and show on the screen. A Zigbee chip periodically transmits package data to a base station. The chip will draw lots of current and cause the digital noise. This digital noise contamination occurs in the previous design. A filter is use d to isolate the analog signal from the digital noise through the bias network in Figure 5 4 . The digital noise w ill go through the two stage filters before going to the analog power plane . The ground planes of the digital and analog are also separated to avoid the digital noise contamination . Figure 5 4 . S chematic and the layout of the filter to isolate the digital noise from the vital sign signal 5.3 Tx Leakage Consideration The system shared with one antenn a at the Tx and the Rx by a miniature three branchline coupler has been introduced in the previous section . One possible issue is the leakage from the Tx to the Rx . The leakage of the high transmitted power might saturate the LNA and cause the failure of the amplification on the received signal. Figure 5 5 shows the link budget of Tx

PAGE 89

89 leakage for both low and high power modes. The power level at each stage is measured by the evaluation board of each component. The power level at the output of the VGA is 9 dBm in high power mode. The isola tion of the coupler is designed for 25 dB at 5.8 GHz. The saturation issue might occur at the last stage, which is the output power level of the 2 nd gain block. According to the data sheet, the input of 1 dB gain compression point ( P 1dBin ) is 4.3 dBm at 6 GHz. The power level in high power mode exceeds this value . Nevertheless , the output power level of the gain block is measured as 16 dBm at 5. 8 GHz. It still works properly if the power does not exceed t he maximum input power of the gain block, 20 dBm. The input power level over P 1dBin might cause harmonics , but it will not affect the vital sign detection. The de sired frequency range is from dc to 2 Hz after down conversion. Figure 5 5 . L ink budget of Tx leakage for low/high power modes In Chapter 4 .3, it shows that the isolation of the coupler is very sensitive to the impedance s at every port. Different type s of SMA connectors will have a huge impact for the characteristics of the coupler. Figure 5 6 shows the isolation comparison between different SMA connectors and loads. The power level is measured by Agilent Spectrum Analyzer E4448A . The measured isolation at different scenarios has a huge difference. There are two different types of the connectors at the port to connect the antenna. One is a straight angle type of the SMA connector and the other is a 90 ° SMA connector. Each connector is connected with differen t

PAGE 90

90 type load, an antenna with a straight angle SMA connector, and an antenna with a 90° angle SMA connector. A coupling port is terminated with a 50 load under the measurement. Although each connector is marked as 50 in the data sheet, the real impedance is different due to the process variation and the soldering condition . Each of them will contribute considerable effects on the isolation of the coupler. The best isolation occur s when the coupler with a straight SMA connector connects with an antenna with a straight SMA con nector. The isolation is 23.16 dB, which makes the power level exceeding the P 1dBin of the 2 nd gain block. Nevertheless, t he gain block can still amplify the signal if the input power is lower than maximum input power . The harmonics will come out when the input power exceeds the P 1dBin , but it will not affect vital sign signal s . Figure 5 6 . I solation comparison between different SMA connectors and loads.

PAGE 91

91 5.4 Measurement Results and A nalysis Figure 5 7 shows the photograph of the bare board s in both the top view and the bottom view. A Zigbee chip and an antenna are instal led in the backside of the board to save the horizontal area. The final system after soldering all the components and installing the antenna shows in Figure 5 8 . The size of the current design with the filter is smaller than the previous one. The filter is to isolate the digital noise from the analog signal as described before . Figure 5 9 shows the comparison between the previous design and the current design of the vital sign signals contaminated by the digital noise. In the previous system, the analog and digital grounds are connected in the entire system. In the new design, the analog and digital grounds are separated by 3.5 mm in the system. When Zigbee transmitted the processed data wirelessly, it will draw lots of the current, which causes the digital noises and contaminates the vital sign signals through the power planes. The first plot in the up left corner shows this phenomenon. The periodic noise peaks represent the frequency of the transmitting rate of the Zigbee. Although the Figure 5 7 . P hotograph of the board in both the top view and the bottom view. The filter and bias board is to isolate the digital noise from the vital sign signal

PAGE 92

92 Figure 5 8 . F inal design comparison between the previous and current design. The size is greatly reduced. noise frequency is much larger than the vital sign frequencies, 4 V peak to peak voltage of the noise may cause large harmonics and affect our results. Furthermore, the plot in the bottom left corner shows that the previous system with the filter and the plots in the upper right corner shows that the new design system without the filter. Either only separating the ground plane or using the filter helps reduce some the digital noise, but none of them can eliminate all of the digital noise s . In the plot of the bottom right corner, the system with both the separated ground and the noise filter can eliminate the digital noise completely. Therefore, the clearer vital signs can be detected. Figure 5 10 shows the function of the automatic gain control in the measurement result. The gain is automatically increas ed when the voltage of the detected signal is too small which

PAGE 93

93 Figure 5 9 . C omparison of the previous design and current design on the vital sign signal contaminated by the digital noise Figure 5 10 . F unction of the automatic gain control in the measu rement result. Gain is automatically shifted to the lower value when the detected signal is saturated Gain is automatically shifted to the higher value when the detected signal is too small

PAGE 94

94 occurs after the red dotted line. On the contrary, the gain is automatically reduced to the lower value when the signal voltage exceeds the A DC dynamic range, which occurs after the blue dotted line. 5.5 System Package In order to deliver to the customer, the package is require d for the system. Nevertheless , the package of the box will change the effective dielectric constant around the antenna and affect the performance of the antenna. Thus, the package needs to be cons idered in the EM simulation. Figure 5 11 show s the EM simulation of an antenna with the package box in HFSS . An Acrylonitrile butadiene styrene (ABS) plastic box and a 5.8 GHz 2×2 patch array antenna are used in the simulation. Sweep the gap between the antenna and the bo rder of the box from 0 mm to 11 mm. T he effective dielectric constants above the antenna will change with the position of the box . It causes the shift of the antenna resonance when the antenna is very close to the box. There is an optimal gap value, which shows the same antenna resonance as it without the box. Figure 5 12 shows the antenna measurement data when the gap between an antenna and the box s weeps from 5 mm to 50 mm. The box does not include the back lid due to the limitation of the measurement. It shows that there is also an optimal design for the gap in the measurement results . The simulation result shows that the optimal gaps are from 6 mm to 11 mm and the measurement result shows that the optimal gaps are from 10 mm to 25 mm. The discrepancy is due to the measur ement error, the lack of the back lid of the box, and the lack of the accurate ABS plastic material information. The ABS plastic material is a compound material and the electrical characteristics change case by case. It varies with the different process an d the vendors. Nevertheless, the parameters of this material in the simulation are set by the typical value. Figure 5 13 shows the photograph of the final system with in the package, the front side view of the

PAGE 95

95 package, and the back side view of the package. The box is OKW A9408341 and the size of the box is 158 mm × 95 mm × 45 mm. The previous box of the system was OKW A9040065 and the size is 189 mm × 110 m m x 60 mm. T he size of the new system is 46% smaller than the Figure 5 11 . A ntenna performance at different gap between the package box and the antenna in the simulation result Figure 5 12 . A ntenna measurement data at different values of the gaps

PAGE 96

96 Figure 5 13 . P hotograph of the final system within the package previous version. The entire system including the bat tery is nicely fit into the small and pretty box. A toggle switch, a tripod mount, and a rechargeable power jack are designed on a box for the customer s.

PAGE 97

97 CHAPTER 6 HAND HELD DOPPLER RADAR TRANSCEIVER 6.1 Introduction In the previous chapter, we implemented a mobile Doppler vital sign radar system. The patients to monitor their vital sign at a hospital or a house. Recently, a hand held device such as a smart watch is more and more popular. People can take the device with them all the time. Weight, size, and power consumption become more important than the detection range. This chapter will design and implement a Doppler vita l sign radar system with the size small than a normal business card. This small and light device can make people monitor their vital signs without feeling it. When you walk in a street, you can put them in the pocket of your shirt. When you ride a bicycle, you can mount your device in front of your chest to monitor your respiration and heart rate. You can manage your healthy plan by tracking and analyzing your vital sign logs. To further reduce the size and the weight of the Doppler vibration radar system, we firstly need to reduce the weight of our system. Previous board laminate, Rogers 4350B, is too heavy for a hand held device. By replacing it with Rogers 5880 laminate, a lighter weight of the system can be achieved. In addition, the electrical quality of this material is better than Rogers 4350B. Lower dielectric constant, r = 2.2, can improve the radiation efficiency of the patch antenna and 10 mm thickness of the laminate is chosen to reduce the weight. Secondly, Zigbee consumes lots of area s and power for the wireless transmission. For hand held device, the requirement of the transmission distance between the sensor and the personal terminal device, like t he smart phone or the laptop, is low. Therefore, we substitute Zigbee with Bluetooth 4.0 for wireless data transmission. Lower power consumption can be achieved with a sufficient transmission distance.

PAGE 98

98 In addition, The Bluetooth 4.0 chip, TI CC2541, also i ntegrate the ADC and MCU function. Previous MCU, MSP 430, can be discarded to further reduce the power consumption and the area. Considering the cost of the components for Doppler radar, the system is still designed at 5.8 GHz. Bluetooth, however, operates at 2.45 GHz. Another antenna is required for the system . A meander type of the Inverted F antenna (IFA) is designed and implemented for the Bluetooth chip to reduce the size. Last but not least, both 3.3 V and 5 V supply voltages are used in the previous system. All the components are chosen to be operated at 3.3 V in this version, which can reduce the number of the power regulator and the power consumption. 6.2 System Design An actuator can provide a constant, stable, and repeatable source for the experiment . requency is between 0.2 Hz to 1 Hz with the peak to peak displacement from 0.5 mm to 2 mm. A 12 cm × 8 cm metal plate is used and attached with an actuator. They are placed at 1 m away from the radar. As introduced in Chapter 5.2, we need to calculate the link budget for this system. In Tx, Only one gain block, Mini Circuit ERA 2SM+, is used with 10.7 dB gain at 6 GHz . a single patch antenna is sufficient and is able to save the area. Figure 6 1 shows the simulated return loss and the radiation of the single patch antenna in HFSS. The bandwidth of the antenna is from 5.7 7 GHz to 5. 83 GHz . Antenna gain at 5.8 GHz is 6.58 dB, which is about 1 dB higher than the previous version due to the better electrical performance of the laminate. On the other hand, an IFA is designed for Bl uetooth communication as describ ed earlier. It is one of the aperture antennas. T his type of antenna has the advantage of the isotropic gain and the small size [54] , [55] . Isotropic gain is suitable for Bluetooth application because the terminal device may not directly point at the radar sensor. The length of the antenna is quarter wavelength, which will occupy large size at low

PAGE 99

99 Figure 6 1 . Simulated return loss and Radiation pattern of the single patch antenna in HFSS Figure 6 2 . Design layout and the simulated antenna gain of the IFA at 2.45 GHz. frequency. The quarter wavelength of 2.45 GHz is 30.6 mm. A meander type of the IFA is designed to further reduce the size. Figure 6 2 shows the design layout of the IFA and the simulated antenna gain at 2 .45 GHz. The antenna exhibits isotropic characteristics and the

PAGE 100

100 antenna gain is 4.0 dB. The mean der line will sacrifice some isotropic characteristics. The length of this meander IFA , however, is only 19 .2 mm, which is 37.3 % shorter than quarter wavelength. Figure 6 3 shows the simulated reflection coefficient of IFA. The resonance of the antenna is at 2.45 GHz and the bandwidth of the antenna is from 2.34 GHz to 2.55 GHz, which can cover the Bluetooth bandwidth. Figure 6 3 . Simulated reflection coefficient of IFA In Rx, previous LNA is replaced by RFMD RF 5515 with 11 dB gain and lower noise figure, 1.6 dB. Only one stage of gain block, Mini Circuit ERA 2SM+, is used. In the baseband, OP amp, TI LMV 2011, is used due to low flicker noise and 3.3 V supply voltage. There are two types of the radar designed this time. One is two antennas at either Tx or Ex, and the other is a miniature two branchline coupler with o ne antenna shared by Tx and Rx. Figure 6 4 shows the layout of the miniature two branchline coupler with a single patch antenna. Another T model two branchline coupler is designed in this system. The size of this miniature two branchline coupler is 36% smaller than conventional two branchline coup ler. In Chapter 4, we knew that the

PAGE 101

101 Figure 6 4 . Layout of the miniature two branchline coupler with a single patch antenna . Figure 6 5 . S parameter of the miniature two branchline coupler with a single patch antenna . performance of the branchline coupler is very sensitive to the impedance at each port. Therefore, the output of the impedance of the gain block at Tx, the LO input of the I/Q mixer, and the inpu t of the LNA have been considered in the simulation. Figure 6 5 shows the S parameter of the miniature two branchline coupler with a single patch antenna. The bandwidth of the coupler is much smaller than it should be. Because the coupler characteristics are highly related to the

PAGE 102

102 impedance of the port, the bandwidth is limited to the bandwidth of the single patch antenna, which is a narrow ban dwidth. For two antenna systems, A Wilkinson power divider [21] is designed in the system to equally separate the power into the Tx antenna and I/Q mixer in Figure 6 6 . Figure 6 7 shows the S parameters of the Wilkinson power divider with the Tx antenna. The antenna resonates at 5.8 GHz. The insertion loss of the port to the I/Q mixer is around 3 dB. Figure 6 6 . Layout of the Wilkinson power divider with the Tx antenna. Figure 6 8 shows the design layout of two Doppler vibration radar systems. One is the system architecture with two antennas and the other is the architecture with one antenna and one cou pler. The sizes of both systems are smaller than a normal business card. The final shape of an entire system is not the rectangular after cutting the empty space. The size will be even smaller. The architecture with one antenna and a coupler has the smalle st area among them. A Bluetooth chip and its peripheral circuit are designed in the bottom part of both systems. The Bluetooth chip can be used as an ADC. Then, the digitalized signal will be transmitted through IFA to the laptop. The power consumption of Bluetooth chip is very low, so the total power consumption can be greatly reduced.

PAGE 103

103 Figure 6 7 . S parameter of the Wilkinson power divider with the Tx antenna. Figure 6 8 . Design l ayout of two Doppler vibration radar systems . Both systems are smaller than a normal business card.

PAGE 104

104 6.3 Implementation and Experiment Results Figure 6 9 shows the photograph of the meander IFA and the single patch antenna. The IFA is operating at 2.45 GHz and the single patch antenna is operating at 5.8 GHz. The size of the conventional IFA at 2.45 GHz is larger than the single patch antenna at 5.8 GHz. Nevertheless, the size of the meander IFA operating at 2.45 GHz here is smaller than the single patch antenna at 5.8 GHz. Figure 6 9 . Photograph of the IFA and the single patch antenna. Figure 6 10 shows the reflection coefficient of the meander IFA and the single patch antenna . The resonance of the meander IFA shifts to 2.35 GHz from 2.45 GHz due t o the process variation and the bandwidth of i t is from 2.26 GHz to 2.44 GHz. Th e resonance of the single patch antenna shifts from 5.8 GHz to 5.73 GHz and the bandwidth of it is from 5.68 GHz to 5. 8 GHz. Nevertheless, both performances are sufficient at desired frequency. Figure 6 1 1 shows the photograph of both systems and the weight of the system with two

PAGE 105

105 Figure 6 10 . Measured reflection coefficient s (A) the meander IFA and (B) the single patch antenna antennas. The solder mask is the compound material. The electrical characteristics are not consistent, especially at high frequency. In order to get rid of the process variation of the solder mask , t here are no solder mask s on the single patch antenna, the meander IFA, the microstrip lines and the miniature two branchline coupler. The weight of the system with two antennas is only 8 g bef ore cutting the empty space and soldering the components . Figure 6 12 shows the weight of both systems after soldering all the components and cutting the redundant space.

PAGE 106

106 Figure 6 1 1 . Photograph of both systems and the weight of the system with two antennas. Figure 6 12 . Photograph of radar systems after soldering all the components an d cutting the redundant space (A) Single antenna version (B) Dual antenna version

PAGE 107

107 It includes the radar system and the Bluetooth. The weight of the one couple r with s ingle antenna version is only 6 g and the weight of dual antenna version is only 8 g. Figure 6 13 shows the testing environment of the Doppler vibration radar system. A 12 cm × 8 cm metal plate is connected with an actuator and is placed at 1.2 m away from the Doppler vibration radar sy st em. An actuator can generate 0.2 Hz to 2 Hz sinusoidal sign al with the peak to peak 0.5 mm to 2 mm displacement. Both systems are tested in the experiment. Firstly, the frequency of the vibration source at the actuator is set at 0.4 Hz. Figure 6 14 (A) shows the measured results after the Fourier transform in the frequency domain for the system with two antennas. The signal at 0.4 Hz is clearly and accu rately detected. Then, the frequency of the vibration source at the actuator changes to 0.2 Hz. Figure 6 14 (B) shows the measured results after the Fo urier transform in the frequency domain for the system with one antenna and one miniature two branchline coupler. The signal at 0.2 Hz is also clearly and accurately detected. Nevertheless, The CSD magnitude of the single antenna version in Figure 6 14 (B) is lower than that of the dual antenna version in Figure 6 14 (A) . It is caused b y additional round trip loss in the coupler. In Tx, the power will be equally split into the Tx anten na and the LO port of I/Q mixer in the system with a coupler. It is the same mechanism as the Wilkinson p ower divider in the system with two antennas . Neve rtheless, the signal will also be equally split in the system with a coupler when the signal is received i n Rx. Only half received power will be sent into an LNA in Rx. Figure 6 13 . Testing environment of t he Doppler vibration radar system.

PAGE 108

108 Figure 6 14 . Measured results after the Fourier transform in the frequency domain. (A) the system with two antennas and (B) the system with an antenna and a miniature two branchline coupler. On the contrary, the total received power will be sent into a n LNA in RX of the system with two antennas. The tradeoff between using a coupler design or not is the signal strength and the occupied area. The signal strengths of both systems, however, are sufficient to detect the vibration signal here. The supply volta ge of both systems is 3.5 V and the total current is only 97

PAGE 109

109 mA, so the power consumption of the entire system is only 340 mW. The hand held device of Doppler vibration radar system has been successfully implemented, which provide a great opportunity for s elf monitoring and managing human health conditions.

PAGE 110

110 SUMMARY In order to make this Doppler vibration system more practical for the detection, several designs are implemented. The experiment demonstrate s that the antenna at Tx plays a more impor tant role than the antenna at Rx for the Doppler vibration radar due to the re radiation effect in the short range detection. An adaptive beam steering antenna is designed and implemented to achieve wide coverage of 85°, by steering the beam direction from 27° to 20°, without sacrificing the area and the antenna gain . Focused beam antenna with the steering function is demonstrated to help the Doppler vibration radar system maintain good signal qualit ies when the system misalignment occurs . Conventional CW radar provides a simple and inexpensive solution for tiny vibration detection, but it cannot detect the distance to the target due to the lack of the time information. This work develops a new method to concurrently measure the distance to the target an d the vibration rate. By analyzing and processing the phase at different beam steering angles, CW radar can detect the distance to the target by analyzing the dc offset besides the vibration detection . This system successfully detects the building deformat ion by observing t he shifts of t he natural frequenc ies and monitoring the phase traces . The distance is also calculated as an indicator of the building conditions . This technology provides a great solution for monitoring structural health in the seismic ar ea. Conventional CW radar with two antennas at 5.8 GHz occupies large areas and is not suitable for mobile devices. A Doppler vibration radar system with one antenna and a branchline coupler is implemented and integrated on a PCB. A new miniat ure branchli ne couple r with wide bandwidths saves 36 % areas of traditional printed branc hline coupler . This system architecture helps not only save the 46% space s but also remove the alignment and coupling issue s between

PAGE 111

111 Tx and Rx antennas. EM effects on the package a re well designed to avoid the previous performance degradation, and the system is well packed in the box as a prototype. For the application to monitor and manage human daily healthy activities, a hand held Doppler radar system at 5.8 GHz is designed and implemented . This comple tely wireless system only weigh s 6 g and consumes 0.34 W. It can detect the target with a 0.5 mm peak to peak displacement at 1.2 m away. The data can be transmitted and displayed in the laptop within 5 m.

PAGE 112

112 LIST OF RE FERENCES [1] A. D. Droitcour, V. M. Lubecke, J. Lin, and O. Boric IEEE MTT S Int. Microw. Symp. Digest , pp. 17 6 1 78, May, 2001. [2] J. C. Lin, Noninvasive Microwave Measurement of Respiration, Proc. IEEE , vol. 63, no. 10, p. 1530, Oct. 1975. [3] IEEE Trans. Microw. Theory Tech n ., vol. 59, no. 12, pp. 3556 3566, Mar. 2011. [4] C. Li , ovements by a 22 40 GHz radar sensor using nonlinear phase m odulation, IEEE MTT S Int. Microw. Symp . , Honolulu, HI, June 2007 , pp. 579 582 . [5] F. K. Wang, T. S. Horng, K. C. Peng, J. K. Jau , J. Y. Li, C. concealed individuals based on their vital signs by using a see through wall imaging system with a self injection IEEE Trans. Microw. Theory Tech n . , vol. 61, no 1, pp. 696 704, Dec. 201 2 . [6] A. D. Droitcour , O. Boric chip silicon Doppler radars for IEEE Trans. Microw. Theory Tech n ., vol. 52, no. 3, pp. 838 848, Mar. 2004. [7] Y. Xiao, J. Lin, O. Boric remote detection of heartbeat and respiration using low power double sideband transmission in Ka IEEE Trans. Microw. Theory Tech. , vol. 54, pp. 2023 203 2, May 2006. [8] B. Park, O. Boric IEEE Trans. Microw. Theory Tech n ., vol. 55, no. 5 , pp. 1073 1079, May 2007. [9] C. Li, and J. Lin, C omplex signal demodulation and random body movement cancellation techniques for noncontact vital sign d etection, in IEEE MTT S Int . Microw . Symp . Dig. , Atlanta, June, 2008 , pp. 567 570 . [10] noncontact vital Proc. IEEE Radio Wireless Symp. , Jan. 2007, pp.281 284 . [11] in non IEEE MTT S Int. Microw. Symp. Dig., Seattle, WA, Jun. 2013 , pp. 1 3 .

PAGE 113

113 [12] F. K. Wang, C. J. Li, C. H. Hsiao, T. S. Horng, J. Lin, K. C. Peng, J. K. Jau, J. Y. Li, and C. sign sensor based on a self injection IEEE Trans. Microw. Theory Techn., vol. 58, no . 12, pp. 4 112 4120, Dec. 2010. [13] P. H. Wu, J. K. Jau, C. J Li, T. S. Horng, and P. and self injection locked IEEE Trans. Microw. Theory Tech n . , vol. 61, no. 1, pp. 6 85 695, Jan. 2013. [14] T. M. Shen, T.J. Kao, T. Y. Huang, J. Tu, J. Lin, R. B. Wu, Antenna d esign of 60 GHz m icro r adar s ystem i n p ackage for n oncontact v ital sign d etection, IEEE Antennas and Wireless Propag . Lett . , vol.11, pp. 1702 1705, 2012. [15] T. Y. J. Kao, Y. Yan, T. M. Shen, A.Y. 60 GHz CMOS Doppler micro radar system in package for vital sign and vibration IEEE Trans. Microw. Theory Tech n . , vol. 61, no 4, pp. 1649 1659, March 201 3 . [16] C. Li, sensitivity software configurable 5.8 GHz noncontact vital sign d IEEE Trans. Microw. Theory Techn., vol. 58, no 5, pp. 1410 1419, May 2010. [17] C. Li, Y. Xiao, a n power Ka band IEEE Trans. Microw. Theory Tech n . , vol. 54, no. 12, pp. 446 5 4471, Dec. 2006. [18] of handheld wireless noncontact vital sign Proceedings of IEEE Radio and Wireless Symp. , San Diego, Jan. 2009 , pp. 514 517 . [19] detection . Theory Techn., vol. 54, no.12, pp. 4464 4471 , Dec 2008 . [20] dimensional noncontact vital sign detection using Doppler ," in IEEE MTT S Int. Microw. Symp. Dig. , Baltimore, MD , June, 20 11, pp. 1 4 . [21] D. M Pozar, M icrowave Engineering , 4th ed. N ew York: John Wiley & Sons, 20 12 . [22] in Proc. IEEE Instru. and Meas. Tech nol . Conf. , May 2012, pp. 1290 1293. [23] J. Kiriazi, O. Boric Proc. 31st Annu. Int. Conf. IEEE EMBS, Minneapolis, MN, 2009, pp. 4808 4811.

PAGE 114

114 [24] IEEE Antennas Wireless Propag. Lett. , vol. 8, pp. 939 942, Aug . 2009. [25] C. H. Tseng, C. J. Chen, T. cost 60 GHz switched beam patch antenna IEEE Antennas Wireless Propag. Lett., pp.432 435, July 2008. [26] C. Balanis, Antenna Theory Analysis and Design. Hoboken , NJ: John Wiley & Sons, Inc., 2005. [27] I. J. Bahl and P. Bhartia, Microstrip Antennas , A rtech House, Dedham, MA, 1980. [28] C. A. Balanis, Advanced Engineering Electroma gne tics , John Wiley & Sons, New York, 1989. [29] Proc. Fifth European Microwave Conf. , pp. 268 272, September 1975. [30] C. M Nieh steering antenna for improved coverage of noncontact IEEE MTT S Int. Microw. Symp. Dig., Tempa, FL, Jun. 2014 , pp. 1 3 . [31] Y. S. Su, C. C. Chang, J. J. Guo, and S. F. Chang, 2 D wireless human subje cts positioning system based on respiration detections, in IEEE MTT S Int. Microw. Symp. Dig. , Montreal, QC, Canada, June 2012, pp. 1 3. [32] nm CMOS Transceiver for FMCW radar IEEE J. Solid State Circuits , vol. 45, no. 4, pp. 928 937, Aprl 2010. [33] J. Lee, Y. A. Li, and M. H. Hung, and S. integrated 77 GHz FMCW radar transceiver in 65 IEEE J. Solid State Circuits , vol. 45, no . 12, pp. 2746 2756, Oct. 2010. [34] G. Wang, C. Gu, T. Inoue, and C. Li, Hybrid FMCW interferometry radar system in the 5.8 GHz ISM band for indoor precise position and motion detection, in IEEE MTT S Int. Microw. Symp. Di g., Seattle, WA, June 2013 , pp. 1 4 . [35] interferometry radar for indoor IEEE Trans. Microw. Theory Techn. , vol. pp, no. 99, pp.1 11, Sep. 2014. [36] F . K . Wang , T S. Horng, K C. Peng, J. Y. Li, and C. C Chen . Combining SIL tag and IL receiver for concurrent vital sign and position sensing in IEEE MTT S Int. Microw. Symp. Dig . , Seattle, WA . June 2013 , pp. 1 4 .

PAGE 115

115 [37] ic range IEEE Trans. Microw. Theory Techn. , vol. 62, no. 9, pp.1837 1846, Sep. 2014. [38] wave IEEE Trans. Microw. Theory Techn. , vol. 51, no.6, pp.1724 1728, June. 2003. [39] cognitive radio front IEEE Trans. Antennas Propag., v ol. 59, no. 5, pp. 1773 1778, May 2011. [40] cost patch antenna phased array with analog beam IEEE Antennas Wireless Propag. Lett., vol. 7, pp. 81 84, 2008. [41] C. C. Chang, T. Y. Ch in, J. C. Wu, and S. GHz fully integrated CMOS Butler matrix for smart IEEE Trans. Microw. Theory Techn. , vol. 56, no. 8, pp. 1757 1763, Aug. 2008. [42] nsmission line and its IEEE Trans. Microw. Theory Tech n . , vol. 5 5 , no. 12 , pp. 2792 2801, Dec. 2007. [43] line coupler using discontinuous IEEE Trans . Microw. Wireless Compon. Lett. , vol. 15 , no. 8 , pp. 519 520, Dec. 2005. [44] C . Mou , an d C . enhanced 3 dB lumped element branch line coupler based on asymmetrical E equivalent sections in Proc . Asia Pacific Microwave. Conf. , Melbourne, VIC, 2011 , pp. 29 32. [45] branchline in Proc. Asia Pacific Microwave. Conf. , Kaohsiung, pp. 184 186, 2012 . [46] IEEE Trans. Microw. Theory Tech n . , vol. 26 , no. 2 , pp. 70 74, Feb. 1978. [47] 90° 3 dB two Proc. International conf. microwave and millimeter wave tech. , Nanjing, pp.116 119, 2008. [48] line couplers with open stubs n . , vol. 57 , no. 1 , pp. 196 204, Jan. 2009. [49] el compact size branch line coupler Microw. Wireless Compon. Lett. , vol. 15, No. 9, pp. 588 590, Sep. 2005.

PAGE 116

116 [50] line and rat race couplers with asymmetrical T structures IEEE Trans. Microw. Theory Tech n . , vol. 60 , no. 7 , pp. 2085 2092, May. 2012. [51] line couplers with predetermined IEEE Trans. Microw. Theory Tech n , vol. 5 5 , no. 9 , pp. 1926 1934, Sept. 2007. [52] A. D. Droitcoure, O. Boric CMOS and BiCMOS single chip direct conversion Doppler radars for remote sensing of Int. Solid State Circuits Conf. Dig., vol. 1, San Francisco, CA, 2002, p.348. [53] Antenna Theory Analysis and Design , 3rd ed. New York: John Wiley & Sons, 2005. [54] mode of a printed inverted IEEE Trans. Antennas Propag. , vol. 61, no. 7, pp. 3420 3426, April 2013. [55] H. W. Liu, S. Y. Lin, and C, F antenna with meander IEEE Antennas Wireless Propag. Lett . , vol. 10, pp. 540 543, May 2011.

PAGE 117

117 BIOGRAPHICAL SKETCH Mr. Chien Ming (Jemmy) Nieh received the B.S. degree in the D epartment of the E lectrophysics from National Chiao Tung University, Hsin Chu, Taiwan R.O.C. in 2005 and M.S. degree in the D epartment of the P hotonics from National Chiao Tung University in 2007. He received his Ph.D. degree in electrical and computer engineering at the Uni versity of Florida in the fall of 2014. He joined AUO Corporation as an active matrix organic LED (AMOLED) panel designer in 2009. Three patents were published in this R&D position. Then, he started to pursue his PhD degree in electrical and computer engineering at the Uni versity of Florida in the fall of 2010. From February to November 2012, he worked as a graduate intern technical at Intel Corporation in San Jose, CA. He worked on signal integrity for high speed input/output ( I / O ) circuits including DDR4 interface. He received Intel innovation award during this internship. In the Ph.D. study, h is research interests include RF and adaptive beam steering antenna, passive component design and EM modeling, miniature coupler design, Doppler radar sensors, and bio medical applications of RF systems. In IMS 2014, he led five people team and won the 3 rd place of high sensitivity radar at student competition. He received the outstanding international student award at UF in 2014.