study of human life sign detection under debris or behind ... of human... · detection of human...

Study of Human Life Sign Detection under

Debris or behind Barrier using Multi Frequency

Radar System

Submission of Final Report

of

Major Research Project

to

UGC, New Delhi

Project file

and Date

: MRP F.No.43-306/2014 (SR), 05 Sep. 2015

Tenure of

Project : 01/07/2015 to 30/06/2018

Grant

Sanctioned : 8.59 Lakhs

Name of

Investigator : Prof. Dr. Abhay N Gaikwad

Name of Co-

Investigator : Assoc.Prof K S Dongre

Department : Electronics and Telecommunication

Engineering Name of the

Institute : Janata Shikshan Prasarak Mandal‟s Babasaheb

Naik College of Engineering Pusad, 445215,

Dist. Yavatmal, Maharashtra, India.

Declaration

We hereby certify that the work which is being presented in this research project report

titled Study of Human Life Sign Detection under Debris or behind Barrier using Multi

Frequency Radar System submitted to UGC New Delhi under MRP scheme is an

authentic record of our own work carried out during the period 01/07/2015 to

30/06/2018.

Prof Kalpesh S Dongre Prof. Dr. A.N.Gaikwad

Co-Investigator Principal Investigator

This is to certify that the above statement made by investigators is correct to the

best of our knowledge.

Principal

Acknowledgements

We take this opportunity to thanks UGC, New Delhi, Govt. of India, for financial

support under Major Research Project (MRP F. No. 43-306/2014 (SR) dated 05 Sept.

2015.

We also owe our sincere gratitude to Head of the Department, Electronics and

Telecommunication, Dr. N P Jawarkar, Principal Dr. H B Nanvala of our Institute,

Babasaheb Naik College of Engineering Pusad and the Management of Janata Shikshan

Prasarak Mandal, Pusad for encouraging and providing the necessary facilities. We

would like to acknowledge the efforts of students of final year project batches for their

valuable assistance in data collection. Thanks are also extended to Mr. Utkarsh

Verulkar, PG student for assisting in work in initial stages of project. We thanks all the

committee members related to the project for there support and giving valauable inputs

for the improvements in the work.

Last but not least, we would like to thanks all the faculty members, supporting staff,

friends and our family members for their support.

Abstract

In recent years, there is need of new research techniques directed towards developments

of systems for searching and rescuing human victims buried under piles of rubble due

to natural disasters or manmade disasters. Existing methods for rescue operations are

utilizations of dogs, or seismic or optical devices which are not effective if the rubble or

debris covering the human victims is thicker than a few feet, especially for the case

when the victims are completely trapped or too weak to respond to the signal sent by

the rescuers. Nowadays, microwave based radar sensor which allows seeing through

visually opaque material is attracting attention of researchers. It has numerous civilian,

law enforcement and military applications. Microwave based sensor usage can be

extended in search and rescue operations to detect buried people after disasters.

The radar techniques often employed are impulse radar, frequency modulation

continuous wave (FMCW), step frequency continuous wave (SFCW) and ultra wide

band (UWB) noise radar for detection of hidden targets. SFCW in UWB range

possesses several advantages over impulse radar systems. The challenging aspect of

radar is to produce high quality images. For high quality, high resolution and low

attenuation of signal is required. Both requirements are contradictory to each other thus

there is tradeoff between resolution and attenuation. UWB based SFCW radar not only

allows detection of closely positioned targets, but also provides the target information

about shape, position and material content. The other advantage of UWB based SFCW

radar over impulse radar is greater measurement accuracy because it is much easier to

synthesize a pure tone at a frequency than to measure a time delay. SFCW has greater

dynamic range and lower noise because it can transmit at higher power and uses a very

narrow IF bandwidth. High average transmitting power can be obtained due to use of

continuous wave signal which helps in detecting trapped victims in longer vicinity.

Researchers are analyzing factors influencing life sign detection, different algorithms

for detection and discriminations and developing prototype. Both heartbeat and

breathing motions may cause changes in frequency, phase, amplitude and arrival time

of returned signal from a living human body. These responses however may attenuate

drastically due to the thickness and the electrical properties of the obstructions.

Therefore getting successful radar based detection of life sign is usually considered to

be a challenging task.

The problems posed for detection with search and rescue sensor system are strong

interfering signal due to radiation of radar transmitter and reflections from obstructing

medium and interference signals due to multiple reflections from other objects.

Obstructing medium like rubble can be a very attenuating medium, particularly if

metallic grids are embedded. Furthermore as rubble is very inhomogeneous medium,

local discontinuities can act as back scatterers and therefore can irradiate the operator,

thus preventing the detector from being able to distinguish between the operators signal

and that of the survivor. For these reasons, the application of microwave transceivers in

detecting human beings trapped under rubble has been disappointing.

Thus there is great need for developing a new sensitive life detection system which can

be used to locate human victims trapped deep under earthquake rubble or collapsed

building debris. Especially the system needs to be sensitive enough to detect breathing

signal of stationary victims who are completely trapped or too weak to respond to the

existing detection system. Considerable amount of attention and research is required

for dealing problems of search and rescue sensor systems.

The objective of this research project is detection of human life signs through visually

opaque materials like different types of walls. The detection of life sign becomes more

challenging when no apriori information of walls and human targets is available. The

scene may consist of various types of targets with different shapes and material

properties (dielectric) along with human being. Thus the objectives are to detect, locate

and extract the life sign signals which will be useful to the end user for interpretation.

To extract life sign, various signal processing techniques are used. Comparison of

signal processing techniques which are used to obtain breathing frequency of a human

target hiding behind a wall is addressed in chapter 3. Two different signal processing

techniques Fast Fourier transform (FFT) and Hilbert Huang transform (HHT) have been

applied on experimental data. After obtaining location of breathing signal using

standard deviation (SD), breathing frequency is obtained by FFT and HHT methods.

The values of breathing frequencies obtained in the results are in acceptable range.

In chapter 4, instead of reflection parameter, transmission parameter is measured using

single antenna i.e., monostatic radar system with the help of circulator. The

experimental results shows that proposed system is useful when the distance between

radar and target is increased compared to single antenna system without circulator.

Since the human target is same for data collected when plywood wall and brick wall is

used, the respiration frequency values remains same as observed from results.

Detection of human being life signs from behind the wall is addressed in chapter 5

using step frequency continuous wave (SFCW) based radar system. The major problem

in life sign detection is the reflection due to wall which amounts for substantial loss of

energy. The remaining energy signal passes through wall and propagated towards

human being as target. Finally the weak reflected signal from target reach to the

receiving antenna after passing through wall again. To improve the signal strength of

the target and hence detection, clutter reduction technique is proposed. It is observed

that after application of clutter reduction technique, the microwave radar system can

detect human life sign. The performance of Singular Value Decomposition (SVD)

based clutter reduction technique is compared with moving average clutter reduction

technique. It is observed that SVD technique outperforms moving average technique in

removing clutter. After improving signal to clutter ratio, location of life sign signal i.e.,

breathing signal is obtained using standard deviation and frequency of breathing signal

is extracted using Fast Fourier transform (FFT) successfully.

Chapter 6 demonstrates the extraction of human breathing frequency when the human

is standing behind brick wall of thickness 32 cm and 22 cm. Data were collected by

changing distance between antenna and wall, as well as distance between wall and

target was varied . It is observed from results, as the thickness of wall increases,

distance between antennas and target increases, detection of target peak becomes

difficult due to very low amplitude value. But still breathing frequency was extracted

successfully after processing the data. Also it is observed from the results that the

number of reflections increases due to multi paths when the distance between antenna

and target increases.

Chapter 7 presents the summary of contributions made in the research project and

future scope of work.

Contents

Contents......................... ................................................................................................... xi

List of Figures.................................................................................................................. xv

List of Tables ................................................................................................................ xvii

Nomenclature ................................................................................................................. xix

Chapter 1 Introduction ................................................................................................. 1

1.1 Motivation .......................................................................................................... 1

1.2 Basic and Operating Principle of SFCW Radar................................................. 4

1.3 SFCW Radar Parameters ................................................................................... 6

Chapter 2 Literature Survey ....................................................................................... 13

2.1 Introduction ...................................................................................................... 13

2.2 Review of work related to type of Radar used................................................. 13

2.3 Review of work related to clutter reduction .................................................... 14

2.4 Review of work related to Signal Processing .................................................. 15

Chapter 3 Extraction of breathing frequency of human being hidden behind the wall

using different signal processing techniques ................................................................... 21

3.1 Introduction ...................................................................................................... 21

3.2 Methodology .................................................................................................... 21

3.2.1 Experimental setup ................................................................................... 22

3.2.2 Signal Processing...................................................................................... 23

3.3 Result ............................................................................................................... 27

3.3.1 Processing for presence of target behind wall .......................................... 27

3.3.2 Stacking all 1024 traces ............................................................................ 28

3.3.3 Location determination using standard deviation..................................... 28

3.3.4 FFT based results ...................................................................................... 28

3.3.5 HHT based results .................................................................................... 30

3.4 Conclusion ....................................................................................................... 32

Chapter 4 Detection of location and breathing signal of human standing behind

brick wall using monostatic radar system ....................................................................... 33

4.1 Introduction ...................................................................................................... 33

4.2 Methodology ................................................................................................... 34

4.2.1 Development of Experimental Setup ....................................................... 35

4.2.2 Data collection ......................................................................................... 41

4.2.3 Signal Processing ..................................................................................... 43

4.3 Result and Discussion ..................................................................................... 46

4.3.1 Metal Calibration ..................................................................................... 46

4.3.2 Determination of location for data taken in corridor with plywood ........ 47

4.3.3 Determination of frequencies of life sign for data taken in corridor with

plywood wall .......................................................................................................... 51

4.3.4 Determination of location for data taken in room with Brick wall .......... 53

4.3.5 Determination of frequencies of life sign for data taken in room with

Brick wall ............................................................................................................... 55

4.4 Conclusion ....................................................................................................... 57

Chapter 5 Detection of location of target and breathing frequency of human being

standing behind brick wall using two antenna systems .................................................. 58

5.1 Experimental setup with 10 cm brick wall ...................................................... 58

5.2 Data collection with 10 cm thick brick wall ................................................... 58

5.3 Signal Processing Algorithm ........................................................................... 60


5.4.1 Absence of target behind brick wall ........................................................ 66

5.4.2 Presence of target behind Brick wall ....................................................... 67

5.4.3 Raw Image (Amplitude versus Slow time variation) ............................... 68

5.4.4 Improvement in the detection using Clutter reduction technique ............ 68

5.4.5 Location Determination using standard deviation ................................... 73

5.4.6 FFT based Results .................................................................................... 74

5.5 CONCLUSION ............................................................................................... 75

Chapter 6 Effect of Thickness of wall on detection of location and breathing

frequency of human being standing behind the brick wall ............................................. 77

6.1 Experimental Setup ......................................................................................... 77

6.2 Data collection................................................................................................. 78

6.2.1 When thickness of brick wall is 0.32 m ................................................... 78


6.3 Signal processing technique ............................................................................ 80


6.4.1 External calibration .................................................................................. 82


6.4.3 When the thickness of brick wall is 0.22 m ............................................. 95

6.5 Conclusion ..................................................................................................... 106

Chapter 7 Conclusion ............................................................................................... 109

7.1 Summary of contributions ............................................................................. 109

7.2 Below is a list of the major contributions in this research: ............................ 110

7.3 Suggestions for Future Work ......................................................................... 111

References ................................................................................................................113

Appendix A :Authors Contributions ........................................................................... A-1

A.1 Conference Publications ................................................................................A-1

A.2 Journal Contributions .....................................................................................A-1

Appendix B :Photographs ............................................................................................ B-1

List of Figures

Fig. 1.1A SFCW radar waveform ................................................................................. 5

Fig. 1.2 Resolution (a) Range resolution (b) Cross range resolution ......................... 9

Fig. 3.1 Experimental Setup ......................................................................................... 22

Fig. 3.2 Flowchart for Signal processing steps ........................................................... 25

Fig. 3.3 Range profiles ................................................................................................. 29

Fig. 3.4 Raw Image ..................................................................................................... 29

Fig. 3.5 Amplitude variation at target location ......................................................... 30

Fig. 3.6 Frequency spectrum using FFT ..................................................................... 31

Fig. 3.7 Time-frequency plot after HHT ..................................................................... 31

Fig. 4.1 Experimental Setup using Monostatic Radar............................................... 36

Fig. 4.2 UWB Ridge Horn Antenna ............................................................................ 40

Fig. 4.3 Connection with Circulator for calculation of insertion loss ...................... 41

Fig. 4.4 Experimental Setup with plywood wall......................................................... 43

Fig. 4.5 Experimental Setup with brick wall .............................................................. 43

Fig. 4.6 Flowcharts for Signal Processing ................................................................... 45

Fig. 4.7 Metal calibration (a) When Antenna to Metal distance is 0.5m, ................ 47

Fig. 4.8 Distance between Antenna and wall is 0.5m and between wall and target is

1.5m .................................................................................................................. 48

Fig. 4.9 Distance between Antenna and wall is 1m .................................................... 49

Fig. 4.10 Distance between Antenna and wall is 1.5m ............................................... 50

Fig. 4.11 Frequency Spectrum of human being standing behind plywood wall at

different distance ............................................................................................ 52

Fig. 4.12 Distance between antenna and Brick wall =0.5m ....................................... 54

Fig. 4.13 Distance between Brick wall and Human target 0.5m ............................. 54

Fig. 4.14 Frequency Spectrum of human being standing behind plywood wall at

different distance ............................................................................................ 56

Fig. 5.1 Experimental Setup with VNA, two antennas and human subject standing

behind brick wall ............................................................................................ 59

Fig. 5.2 Flow-chart for processing steps ..................................................................... 64

Fig. 5.3 Range profiles for all 1024 traces in absence of target ................................ 69

Fig. 5.4 Range profiles for all 1024 traces in the presence of target ........................ 70

Fig. 5.5 Raw Image ....................................................................................................... 71

Fig. 5.6 Image after clutter reduction......................................................................... 72

Fig. 5.7 Image formed using Target Singular Components ..................................... 73

Fig. 5.8 Frequency Spectrum ...................................................................................... 75

Fig. 6.1 Experimental Setup with two antenna system ............................................. 79

Fig. 6.2 Flow chart of Signal processing technique ................................................... 81

Fig. 6.3 External calibration using Metal sheet ......................................................... 82

Fig. 6.4 Range profile for external calibration .......................................................... 83

Fig. 6.5 For case 1:Distance between wall and human target is 0.5m .................... 84



Fig. 6.8 Plot of amplitude variation for case 1 data ................................................. 91

Fig. 6.9 Plot of amplitude variation for case 2 data ................................................. 91

Fig. 6.10 Plot of amplitude variation for case 3 data ............................................... 92

Fig. 6.11 Breathing frequency extraction using FFT for case 1 dataset.................. 93

Fig. 6.12 Breathing frequency extraction using FFT for case 2 dataset................. 94

Fig. 6.13 Breathing frequency extraction using FFT for case 3 dataset................. 95

Fig. 6.14 For case 4: Distance between wall and human target is 0.5m ................. 96

Fig. 6.15 For case 5: Distance between wall and human target is 0.5m ................. 97

Fig. 6.16 For case 6: Distance between wall and human target is 0.5m ................ 99

Fig. 6.17 Breathing frequency extraction using FFT for case 4 dataset............... 102

Fig. 6.18 Amplitude extraction using SD for case 5 dataset .................................. 103

Fig. 6.19 Amplitude extraction using SD for case 6 dataset .................................. 103




List of Tables

Table 3.1 – Radar Parameter for Experimentation ................................................... 22

Table 3.2 SD values at Different location .................................................................... 28

Table 4.1 Mono-static Radar Parameters ................................................................... 37

Table 4.2 Data collected with various distance between Antenna and plywood wall

and also variation in distance between plywood wall and target. .............. 42

Table 4.3 Data collected with various distance between antenna and brick wall and

also variation in distance between brick wall and target .......................... 42

Table 4.4 Location determination when observation taken in corridor with

Plywood Wall .................................................................................................. 51

Table 4.5 Frequency determination when observation taken in corridor with

Plywood Wall .................................................................................................. 52

Table 4.6 Location determination when observation taken in room with Brick Wall

........................................................................................................................................ 55

Table 4.7 Respiration Frequency determination when observation taken in room

with brick wall ................................................................................................ 57

Table 5.1 SFCW radar parameters with two antennas ............................................. 60

Table 5.2 Matrix representing organization of data collected .................................. 60

Table 5.3 Location of target obtained using SD ......................................................... 73

Table 6.1 Data Collection details for 0.32 m Brick wall taken with two antenna

system .............................................................................................................. 79

Table 6.2. Data Collection details for 0.22 m Brick wall taken with two antenna

system .............................................................................................................. 80

Table 6.3 Position and amplitude when distance between Antenna and Brick-wall

is 0.5m .............................................................................................................. 84

Table 6.4 When distance between Antenna and Brick-wall is 1m ........................... 86

Table 6.5 When distance between Antenna and Brick-wall is 1.5m ........................ 87

Table 6.6 Location obtained from Experimental data after velocity correction ..... 89

Table 6.7 Breathing frequency values for different datasets ................................... 95

Table 6.8 Position and amplitude when distance between wall and target is 0.5m 96

Table 6.9 Position and amplitude when distance between wall and target is 1m... 98

Table 6.10 Position and amplitude when distance between wall and target is 1.5m

........................................................................................................................................ 99

Table 6.11 Location obtained from Experimental data ......................................... 100

Table 6.12 Breathing frequency values for different datasets ............................... 106

Nomenclature

2D Two dimensional

ACC Adaptive clutter cancellation

AGC Automatic Gain Controller

BW Band width

CFAR Constant false alarm ratio

CPI Coherent processing interval

CR Cross Range

CT Curvelet Transform

CW Continuous wave

EM Electromagnetic

EMD Emperical mode decomposition

FAR False Alarm Rate

FDTD Finite difference time domain

FFT Fast Fourier Transform

FIR Finite impulse response

FM Frequency modulation

FMCW Frequency Modulation Continuous Wave

GPR Ground Penetrating Radar

HPBW Half Power Beam width

HHT Hilbert Huang Transform

HOS Higher order Statistics

ICA Independent Component Analysis

IFFT Inverse Fast Fourier Transform

IMF Intrinsic mode function

MA Moving average

MUSIC Multiple Signal Classification

MSE Mean Square Error

MTI Moving target indicator

PC Principle Component

PDF Probability Density Function

PSD Power Spectral Density

PSNR Peak Signal to Noise Ratio

Radar Radar Ranging and Detection

RCS Radar Cross Section

RF Radio Frequency

RGK Radial Gaussian kernel

RPF Recursive pixel finding

SAR Synthetic Aperture Radar

SFCW Step Frequency Continuous Wave

SNCR Signal to Noise clutter ratio

SNR Signal to Noise Ratio

SP Specificity

STFT Short time Fourier Transform

SVD Singular Value decomposition

TOSM Though open short and Match

TWI Through wall imaging

TWRI Through-the-Wall Radar Imaging

UWB Ultra wide band

VNA Vector Network Analyzer

Chapter 1 Introduction

1.1 Motivation

A human kind was always interested to find out the unknown from the very beginning

of the mankind history. Our eyes help us to investigate our environment by reflection of

light. However, a wavelength of visible light allows a transparent view through only

non opaque materials such as glass and not through opaque materials. On the other

hand, Electromagnetic (EM) waves in microwave frequency range, with the exception

of metal are able to penetrate through almost all types of materials around us such as

plywood, plastic, brick wall, concrete wall, etc.

Through-the-Wall Radar Imaging (TWRI) with EM waves is an emerging technology,

allowing to “seeing” through visually opaque material such as walls. Identification of

living and non-living targets using TWRI has lots of applications in various field of

healthcare, defence, monitoring human activities backside of walls in connection with

law enforcement, surveillance, search-and-rescue mission operations and natural

disasters.

Especially when the person is trapped under rubble due to earth quake or due to

collapsed buildings, remote and contactless detection of human vital life signs via radar

sensing is very useful for search and rescue operations.

The trapped victim under such circumstances cannot make movement and the only way

to detect the presence of life is to detect life sign signals as early as possible to reduce

the loss of life. The life signs which are normally used for detection are respiration rate

and heart rate.

In recent years, the research is mainly directed toward sensors based on microwave

radar system for the detection of life sign [1-12]. Ultrasound, millimeter wave

2 Introduction

radiometry, infrared, and X-rays can be used for through wall imaging but RADAR

sensor is the most suitable due to the following reasons. Ultrasound technique can be

used to detect and find target behind a metallic or non-metallic wall but cannot be used

for imaging due to the high resolution requirement in TWRI. Millimeter wave

radiometer uses energies radiated by bodies of targets for detection but limitation is that

it works only up to very short distances [13]. Infrared can be used to image the target

through wall at very short distance as attenuation through a wall is very high. X-ray

based sensors provide good imaging quality but are limited due to their high cost and

lack of medical safety. Researchers are using either impulse based radar system or

Continuous wave radar system [14]. This work relates to the detection of human being

life signs from behind the wall using step frequency continuous wave (SFCW) based

radar system because of its advantages over impulse radar [15]. The TWI Radar system

is used for scanning of the living objects behind a wall. An electromagnetic wave is

transmitted via antenna system, penetrates through the wall, it is reflected by the

investigated object, penetrates again through the wall, and is received back via receiver

antenna. The reflected signal is received by the radar and corresponding data are

collected. Then data processing is carried out to find out information about the living

targets. These targets are located behind a wall or any other visually opaque medium.

The simplest problem is just to find out the location of targets behind the wall. One may

further classify living targets and others on the basis of micro-Doppler characteristics

(breathing and heartbeat). After that one may further generate a 2D image of the target

which contains information about the target‟s lateral extent and its location.

One of the main problems in detection of life sign is strong reflection of brick wall and

antenna air coupling. These reflections are unwanted and are called as Clutter. Different

clutter reductions techniques are used like background subtraction method, range

gating, moving average reduction technique, notch filtering and singular value

decomposition (SVD) [16-18]. SVD has been used by researchers working in through

wall radar imaging to remove the wall clutter and to detect behind-the wall targets [16].

In [17], SVD is applied on the data collected by taking measurements at different

1.1 Motivation 3

antenna positions, whereas in this work; we have applied SVD on data measured at

same antenna location. Though wall and target reflections reside in many Eigen images,

the significant or dominant Eigen images are obtained from first few Eigen values.

Target signal strength can be improved by removal of significant wall reflection using

most significant singular values. In [9], SVD has been used for life sign extraction of

more than one target but it is used after taking Fast Fourier Transform (FFT) of

correlated signal. In our work we are proposing use of SVD for clutter reduction in

earlier stage i.e., before applying FFT.

Other effects due to presence of wall are refraction; attenuation and change in velocity

of the signal which poses challenges for detecting and localization of life sign of human

being. Researchers have carried out work on detecting and extracting life sign of human

being using different signal processing techniques which is briefly given here.

Various types of signal processing algorithms or techniques are used for life sign

extraction [17]. Change detection algorithm was used for detecting life sign of human

[21]. Life detection algorithm using step frequency continuous wave (SFCW) radar is

developed in [22-23]. In [24], they have proposed the life sign detection algorithm for

SFCW radar in which multi-periods and filters are used to extract the micro-vibrations

parameter of the life targets. Remote detection of human vital sign is investigated using

SFCW based radar system [15]. In this paper various factors such as effect of varying

thickness of the obstacles, human subject postures, status of breathing, position of radar

antenna relative to human subject‟s chest, as well as the length of survey times are

studied. Empirical mode decomposition (EMD)-Hilbert Huang transform (HHT)

method for detection of breathing as well as other motions is used [25-26]. On the basis

of specific literature survey on SFCW based radar; it is observed that no one has

applied EMD- HHT technique for extraction of life sign signal on experimental data.

In a nutshell, considerable amount of attention and research work is required for

dealing with different issues related to life sign detection. This motivated us to develop

signal processing techniques to improve detection, localization and extraction of life

sign of the living targets.

4 Introduction

1.2 Basic and Operating Principle of SFCW Radar

Ferris and Currie [13] had carried out a survey of technologies applied for through the

wall surveillance. The technologies used for through wall imaging are classified as

active or passive imaging. In active imaging, RF, acoustic, optical or X ray energy is

used to estimate the reflectivity distribution of a remote scene and in passive,

millimeter wave imaging radiometer is used which uses energies that are radiated by

the bodies of persons within a building for detection [14].

The Stepped Frequency Continuous Wave (SFCW) radar was not included in the table.

Due to complexity of system and higher component cost, SFCW radars were not

popular earlier. However, the cost of RF technologies has been decreasing

considerably, making it more feasible to use SFCW radars. Now a days, SFCW and

short pulse radar represent two different techniques which are used to generate a wide

band of frequencies for detection of hidden objects under some barrier. SFCW radar is

a frequency domain system whereas pulse radar is time domain System. The SFCW

radar system has several advantages over time domain systems [15].

SFCW is a frequency-domain pulse synthesis method that has been widely applied for

radar systems. The SFCW radar operates from a low frequency limit to a high

frequency limit and the digital radio technology is used to provide this stepping up. The

step frequency waveform can be described as an intra pulse version of the common

linear FM pulse compression waveform. A series of N coherent pulses are transmitted

whose frequencies are monotonically increased from pulse to pulse by a fixed

frequency increment Δf. The frequency of the nth (and is varied from 0 to N-1) pulse

can be written as Eq.(1.1). A stepped frequency continuous wave is shown in Figure

1.1.

fnff on (1.1)

where fo is the starting carrier frequency of system and Δf is the frequency step

size, that is, the change in frequency from pulse to pulse. Each pulse is seconds wide,

and the time interval T between the pulses is adjusted for ambiguous or unambiguous

range. Note that the frequency remains constant within each pulse. Groups of N pulses

1.2 Basic and Operating Principle of SFCW Radar 5

(burst) are transmitted and received before any processing is initiated to realize the

high-resolution potential of this waveform.

Fig. 1.1A SFCW radar waveform

The burst time is defined as: “The time corresponding to transmission of N pulses will

be called the coherent processing interval (CPI)” [19].

The instantaneous bandwidth of this waveform is nearly equal to the inverse of the

pulse width and it is much less than the effective bandwidth. Pulses of typical time

duration have narrow bandwidths, thus making the instantaneous bandwidth of the

radar narrow. However, effective large bandwidth can be realized by appropriately

processing the N pulses in a CPI. The waveform's effective bandwidth, denoted as in

contrast, to the instantaneous bandwidth is determined by the product of the number of

pulses N and frequency step size Δf as in Eq. (1.2),

fNBeffec (1.2)

The range resolution for any waveform depends on this effective bandwidth of the

waveform. The range resolution of the SFCW, in meter is given as;

effecB

CR

2 (1.3)

The range resolution can be made finer by either increasing the number of pulses and/or

increasing the frequency step size. Thus, two objects separated in the down range by a

distance greater than can be detected as two distinct objects.

One of the facts about step-frequency radar is that its resolution does not depend on the

instantaneous bandwidth, and that resolution can be increased arbitrarily by increasing

NΔf. There is a constraint on selection of Δf (i.e. 1f ); however, N can be

increased to achieved the very high range resolution. It should be noted that, fine range

6 Introduction

resolution is achieved by large bandwidth irrespective of the waveform and the

compression method used. So large bandwidth is achieved by sequentially over many

pulses, by inter pulse frequency modulation whereas, for conventional radars, it is

achieved in a single pulse by intrapulse phase or frequency modulation.

A step-frequency waveform achieves wide bandwidth (NΔf) sequentially (over a burst

of many pulses) but has a narrow instantaneous bandwidth of 1/ . It provides the high

range resolution of wideband radar systems with some of the advantages over

narrowband radar systems. Step-frequency radar achieves range resolution of c/2NΔf

(equivalent to a bandwidth of NΔf) as compared with range resolution of c/2 for

constant-frequency waveforms.

The SFCW signal is transmitted through an antenna and scattered signals from the wall

and the target can be collected using antenna or antenna array. This raw frequency data

is then sent through inverse fast Fourier transform (IFFT) to yield the scattering from

individual targets. Using electromagnetic theory and signal processing these signals can

be analyzed and an image of the scene behind the wall can be produced.

The main advantage of the stepped frequency technique is the need of low speed analog

to digital converters making it relatively easy with current technologies to efficiently

sample ultra wideband signals. Measuring an object reflection in the frequency domain

and applying an inversion technique to obtain the time information of physical space

removes the requirements for wide instantaneous bandwidth and high sampling rate,

leading to reductions in physical size, weight, complexity and cost of the radar

hardware [20]. The other advantages are high dynamic range, high signal to clutter ratio

and low power consumption.

1.3 SFCW Radar Parameters

The SFCW radar parameters for the life sign detections through barrier should be

chosen carefully which are described below.

Number of frequency points

1.3 SFCW Radar Parameters 7

SFCW radar illuminates target with consecutive train of number of frequencies and

process it coherently after receiving them. Thus the process gain will be high, if number

of frequency points is high. The choice of high frequency points result in small

frequency step size for better resolution. If the number of points is chosen smaller then

data acquisition time is reduced.

Unambiguous range

The maximum range for receiving transmitted radar signal after reflection prior to next

transmittion of pulse is,

i.e. the unambiguous range is given by

fu

cR

2 (1.4)

From equation (1.4) if the frequency step is narrow then ambiguous range will be

greater. Frequency step size ∆f is also calculated as

)1(

N

BWf (1.5)

Beam width of antenna

In monostatic radar system, with synthesize aperture techniques; the beam width of

single antenna should be narrow. If the antenna beam is narrow it is easy to pick line of

sight target signal.

Wall parameters

The wall through which the signal is penetrating plays an important role in detection.

To ensure signal penetration through wall it is desirable to have minimum attenuation at

the working frequencies. The other parameters of wall which are important to combat

the effect of wall such as shift in target position and blurriness in image are thickness of

wall and dielectric constant. These parameters should be known before hand for

processing.

Target parameters

8 Introduction

Behind the wall there are two possibilities of the targets i.e., moving or stationary

human being. Microwave signal both penetrate and get reflected off from target

material. The composition and thickness of the targets are prime factors for receiving

the reflected signal. If the dielectric constant of target is high then reflection will be

high whereas for low dielectric, reflections seem to be very poor. In this report, the

stationary human being is considered since project is about detecting trapped human

being who cannot make movement and is live. The life signs are gauge of survival of

human being. The vital signs which are mainly used are respiration rate and heart rate.

The sensor system should aim at detecting and extracting the heartbeat and the

respiration rate of a human subject. Average human being breathes about 15 to 20 times

per minute and has 60 to 72 heartbeats per minute. This corresponds to the chest-wall

motion with a frequency of 0.25 to 0.33 Hz for respiration and frequency due to beating

of the heart will be 1 to 1.2 Hz.

Down Range Resolution

Down range resolution is the capacity of the radar to discriminate individual elements

that are close to each other in down range as shown in Figure 1.2 (a). High down range

resolution is obtained by using wide bandwidth and is given as equation (1.3) [19].

The effective bandwidth is determined by the total frequency excursion, i.e., N*∆f. The

down range resolution of step frequency radar is given by equation (1.6)

fN

cR

2 (1.6)

where N is number of frequency points and ∆f is step size.

The required bandwidth must be greater than 1 GHz to obtain range resolution in order

to detect object size of few centimeters. The actual value is taken more than theoretical

value.

Cross Range Resolution

Cross range resolution is the capacity of the radar to discriminate individual elements

that are close to each other in cross range as shown in Figure 1.2(b).


R

∆R

Antenna

R

Antenna

ΔCR

(a) (b)

Fig. 1.2 Resolution (a) Range resolution (b) Cross range resolution

Resolution in cross range is a function of wavelength at the lowest operating frequency,

the length of physical antenna aperture and distance to target. Cross range resolution is

defined as

D

RCR

(1.7)

where is wavelength, R is distance to target in far-field from antenna and D is

physical aperture of antenna. For a real antenna, cross range resolution degrades with

increasing target distance. To achieve high cross range resolution, narrow beam width

is required for which the antenna aperture should be quite large which is physically

unmanageable. Another approach is to introduce the concept of a fixed array or

synthetic array. The idea of synthetic array is that a physical antenna moves to each

point. Processing the data allows us to synthesize an effective aperture many times the

size of a real aperture. Thus, the distance travelled during data observation determines

aperture size, limited by time required to scanning. If fixed array is used scanning time

is reduced with increase in complexity of processing signals.

High frequency range is chosen at which the narrow beam width of antenna is achieved.

Thus high cross range resolution requirement leads to the selection of higher frequency.

But at high frequencies, penetration through the wall is low. Thus there are inherent

tradeoffs between resolution and penetration. Better resolution and penetration are the

10 Introduction

major challenges being faced in detection of life sign through wall. Various types of

wall materials are used in different parts of the world like wood, asbestos, brick,

concrete and so on. The characterization of common types of wall is described by [27].

The walls made of wood are approximately transparent to radar frequencies. Thus

frequencies above 10 GHz can be used for imaging. On the other hand in brick wall

attenuation is more. In brick wall, one way attenuation is reported as 5 dB/cm at 5 GHz

and in concrete it is 10 dB/cm at 3 GHz [14].

Selection of parameter for assembling radar system

Selection of frequency range up to approximately 3 GHz can be used as attenuation is

within acceptable range. On other hand bandwidth is chosen so as to resolve the targets

in down range in tenth of centimetres. So the selected frequency range of 1 GHz to 3

GHz with bandwidth of 2 GHz, number of frequency points as 201, gives range

resolution of 7.5 cm in air according to equation (1.6). The maximum distance to target

is calculated using radar range equation (1.8)

22

4max2

3)4(

a

win

G

RLSNRBTekPt (1.8)

where,

Pt=minimum required transmitted power to fulfil the detection criteria,

K=Boltzmanns constant,

Te =Equivalent noise temperature referred to the receiver input,

B=Transmitted bandwidth,

SNRin=Signal to noise ratio at the input of the receiver,

L2w=Two way attenuation due to propagation through the wall,

Rmax=Range of target to be detected ,

Ga=transmitter receiver antenna gain,

Wavelength,

Radar cross section of the target to be detected,


With the help of above equation (1.8), keeping the transmitted power constant, range is

calculated for different types of barriers having variable values of attenuation. While

calculating range, the following values of various parameters are taken.

Pt= 10dB, Ga = 9.9db , kTe=200dB, 2 -18.4dBsm for 2.5 GHz center frequency,

=1sq.m, the two way losses due to wall are taken to be, 15dB, 18dB, 20dB and 23dB

respectively for wood wall, brick, stone and concrete wall. The range is obtained as

2.93m, 2.89m, 2.88m, and 2.84m respectively. Thus as attenuation due to wall

increases, range reduces.

Organization of report is as follows. Brief Review of Literature is given in chapter 2.

Chapter 3 describes about experimental work in which S11 (reflection coefficient) is

measured using single antenna to detect and extract breathing frequency of human

being hidden behind plywood wall. In Chapter 4, instead of S11, S21 is measured using

single antenna with the help of circulator to detect and extract breathing frequency of

human being hidden behind plywood and brick wall. Chapter 5 describes about work in

which S21 is measured using two antennas i.e., transmitting and receiving to detect and

extract breathing frequency of human. The effect of thickness of brick wall on detection

of life sign is carried out in chapter 6. Finally concluding remarks are given in Chapter

7.

Chapter 2 Literature Survey

2.1 Introduction

Non contact Detection and monitoring of Human life sign is cardinal issue in many

applications such as remote non contact health monitoring, homeland and military

security, localization and rescue of living survivor buried under rubbles and debris of

building in post earthquake disaster scenario, localization or detection of invisible

human being hiding behind the wall etc. With the significance of this topic there has

been growing interest of researchers in this field in recent years which has resulted in

large improvement in the technology used.

2.2 Review of work related to type of Radar used

Radar signals reflected from the living target bears the information about their

vital sign. In this technology living human body micro motion because of heartbeat and

respiration results in modulation of incident radar signals which are demodulated to

extract vital sign. These vital sign detection can be achieved by using continuous wave

(CW) [15],[28]–[32] and ultra wide band (UWB) [4][5][6][28],[31]–[49] techniques.

Frequency domain signals techniques such as CW radar [28],[31],[32] and frequency

modulated continuous wave (FMCW) radar [29],[30] and time domain signal

techniques like UWB pulse radar [4],[5],[31],[33],[34],[37],[38],[42],[43],[45]–

[46],[48] and UWB noise radar[35],[44] and step frequency continuous wave (SFCW)

radar [6],[40],[41],[47] have been used in recent times by different research groups.

CW radar has reported detection life sign but could not locate it [34],[48] though it has

been demonstrated by using FMCW radar system to locate the target using background

subtraction method [15] with resolution of lower value [38] since it cannot easily

separate multipath signals or strong reflections from static barrier such as wall.

14 Literature Survey

Compared to this, pulsed radar can detect and locate multiple vital signs [40] and have

better resolution because of large bandwidth under low SNR [4],[31],[33],[42],[43].

Impulse radar has fast data acquisition though dynamic range is not good as SFCW

radar moreover later has good resolution of range and power [37],[40].These key

advantages and features associated with SFCW radar led the researchers

[6],[40],[41],[49] to utilize SFCW radar for life sign detection behind wall.

2.3 Review of work related to clutter reduction

While detecting and locating vital signs such as breathing or respiration and

heart beat, signals reflected from the target might be completely overridden by noise

and clutter and thus necessitates processing of this signal to improve signal to clutter

ratio and SNR. With typical breathing frequency of 0.1 to 1 Hz and heart beat

frequency of 1 to 3 Hz Band pass filtering suitable to pass 0.1 Hz to 3 Hz and to

eliminate DC components present in signal because most of the clutter power is

concentrated about the zero frequency band i.e. DC component removal of these low-

frequency noise by using Band pass filtering have been proposed [6],[31],[32],[40],

[42],[45],[49].Reflected signals from the target from behind the wall are highly

contaminated with signals reflected from the wall and produce stationary clutter .To

reduce the effect of this stationary clutter and improve signal to clutter ratio background

subtraction method has been illustrated [30], [44].In this scans of received signal

without target behind barrier are subtracted from scan with target behind the barrier to

mitigate the effect of barrier whereas it is impossible in practical situation to get data

without target condition. Further Spatial filters such as Moving Average (MA) filter

background subtraction which notches out the zero spatial frequency component has

been effectively used for removal of clutter signal from wall [46], [48]. Meanwhile the

problem of ripples in MA has not been discussed. Clutter suppression has been

achieved by simple method of subtracting each of the range profile from previous to

eliminate components that have not changed [28].A simple low pass FIR filter called

Moving Average Filter has been proposed [42],[43]which is used to smooth the data by

taking an average of samples of input data to produce single output thus eliminating

2.4 Review of work related to Signal Processing 15

noise. Most of the clutter is associated with zero frequency or DC term and its multiple

integers of pulse repetition frequency thus suppression clutter by offering deep stop

band to these frequencies have been reported by using MTI filtering [6],[41],[49]where

mean of all range profile is subtracted from current one to get moving part. Instability

of time base caused by imperfections in the triggering unit of an impulse radar causes

amplitude instability of impulses and this gradual change in amplitude with time is

linear trend which are actually unwanted amplitude variations as they are not from the

human target and should be removed to improve performance. Linear Trend

Subtraction LTS method have been demonstrated to suppress this pulse amplitude

variation in [4],[42],[43],[45] . However all the signal processing techniques above are

suitable for static clutter suppression and cannot be used for non static clutter

cancellation.

Improvement in signal to clutter ratio have been reported by using an Eigen

structure technique SVD (singular value decomposition) which decompose the received

radar signal into subspaces of clutter and target [4],[39].But in SVD only fundamental

harmonics of signal are considered overlooking signals multiple harmonic components.

Curvelet transform (CT) which is a multi-scale analysis algorithm suitable for 2D data

and which has good analyzing capability of line and curve edges has been used for

clutter and source receiver direct coupling wave noise suppression[4]. A linear phase

filter with constant phase delay has been applied to separate out noise components

[45].In [46] adaptive clutter cancellation (ACC) algorithm has been proposed to remove

respiration like clutter.

2.4 Review of work related to Signal Processing

To detect the breath or respiration frequency and heart beat frequency and to

locate the same under severe clutter contamination after pre-processing of received

echo signal from the target laying behind the barrier researchers have proposed many

signal processing algorithms.


FFT gives data about human micro and macro movement‟s frequency and amplitude. In

micro motion lies vital sign information. Extraction of vital sign has been demonstrated

with FFT algorithm [4],[6],[28],[29],[31],[36],[37],[39]–[41],[43],[49]. In [40]

respiration frequency has been detected for different positions of human target behind

wall of 5 cm to 20 cm thickness but there is further scope in developing a method for

faster data collection and improvement in signal processing efficiency. Ellipse-cross

localization has been illustrated to locate human target [49].FFT used in

[28],[29],[40],[49] doesn‟t give time dependent information for non-stationary signal

also gives distorted peaks in frequency domain. Thus it is not suitable to analyze a

narrow frequency band of a non stationary signal.

In [41] it has been demonstrated to detect life sign behind the corner of wall

considering diffraction of signals at corner with SFCW radar of 10 GHz center

frequency and STFT signal processing which maps a signal into a function of time and

frequency .Breathing signal has been detected with STFT [31],[39],[48].However the

results are from ideal conditions and multipath data has not been addressed which can

be possibly used for positioning by combining multiple detections of a target,

corresponding to different propagation paths[31],[39]. In [39]experiments has been

performed on Gypsum wall, Brick wall, wooden door, concrete wall with UWB pulse

radar with center frequency of 4.3 GHz. Heart beat frequency and breathing frequency

has been detected in all cases except concrete wall and gypsum when SVD and STFT

was used but monostatic mode limits the use of spatial filtering to reduce clutter also

STFT has got fixed resolution.

S-transform is the signal processing technique whose performance is dynamic in the

sense that it doesn‟t have fixed resolution as was reported with STFT and thus [36]

have described life sign detection and location using ST. In [36] life sign extraction and

location of human target has been achieved by 2-D FFT method to analyze the

frequency of the cross range data and ST with an experiment in which a brick of wall

thickness 0.2 m and 1 to 4 GHz frequency UWB radar but better performance could

have been achieved with better accuracy if noise signal have been avoided.


FFT and HHT has been utilized over the data collected from physical experiment

performed in laboratory with 20 cm concrete cinderblock wall and 1GHz signal source

and by the FDTD simulation data [37].With use of HHT non linear and non stationary

signals have been decomposed to extract respiration and heart beat signal considering

the condition of two human targets with normal breathing, holding the breath and

repeatedly speaking the word “one, two, three”. Though respiration frequency has been

detected it hardly detects heart beat frequency. It has also reported that the subject

under test should be steady while taking readings or its motion frequency will override

vital sign frequency [37].Use of HHT has also been proposed in[5],[35], [44].In [5]an

experiment using pulsed radar with 1GHz frequency and cinder block wall about 1.5 m

was used and results have been validated with FDTD data.

With same experimental values author of [36] have demonstrated the extraction and

location of breathing and heart beat signal in complex condition as of earthquake rubble

in [47] with FFT and Hilbert-Huang transform also with FFT and RGK methods(radial

Gaussian kernel) but here is scope of improving resolution i.e. better detection.

Nonlinear and non-stationary data analysis has been achieved using HHT [37] to detect

respiration signal and results are validated with FDTD simulation data but it has failed

to detect heart beat signal.

The HHT is based on the principle of empirical mode decomposition (EMD) which

decomposes signal into a collection of intrinsic mode functions (IMF) and the Hilbert

Transform which gives energy of each instantaneous frequency. At the low-frequency

region in HHT method the EMD will generate undesirable IMFs that may cause

misinterpretation to the result. Also the first obtained IMF may cover very wide

frequency range and prevents it from achieving mono component property and the

EMD operation cannot separate signals that contain low-energy components.

Authors of [50] have described a method based on MUSIC algorithm applied to echo

data collected from CW radar with 10 GHz centre frequency to detect breathing signal.

MUSIC detects frequencies in a signal by performing Eigen decomposition on the

covariance matrix of a data vector of M samples obtained from the samples of the

received signal. This has been further modified by authors in [32] where SNR is


improved compared to previous method utilizing smoothing procedures referred to as

temporal and sample de-correlation considering no perfect periodicity of received

signal. Moreover, this method requires less number of computations. Compared to the

old method significant improvement in the value of Dynamic range and noise rejection

has been reported in [32]. This algorithm detects only main harmonic components of

spectral noise since it can add or delete close harmonics placed in not pre known

position. Thus this can be used to detect particular harmonic component with presumed

frequency. However, this method fails to work in case of coloured noise to determine

the number of sinusoids considering colored noise as additional sinusoids. Higher-order

statistics have been applied [33] to this problem as they show robustness to coloured

Gaussian noise.

Gaussian noise and its harmonics have been suppressed by taking FFT of FOC to detect

and locate breathing frequency behind 0.28m wall [6].Method of detecting and locating

vital sign using multiple higher order cumulant has been presented in [33].It has

produced results with high value of SNR and also suppressed higher harmonics to

automatically detect vital sign. Further in [33] clutter with same frequency as

respiration and heartbeat were found. The second-order measures[32], [50] work fine if

the signal has a Gaussian (Normal) probability density function, but many real-life echo

signals are non-Gaussian and this Second-order measures (such as the power spectrum

and autocorrelation functions) contain no phase information. Complete characterization

of Gaussian signal can be achieved by its mean and its variance. Higher order statistics

(HOS) of Gaussian signals are either zero or contain redundant information and many

of echo signals have non-zero HOS and many noises are Gaussian thus in principle the

HOS are less affected by Gaussian background noise than the 2nd order measures like

MUSIC.

A method based on K-mean clustering algorithm and constant false alarm ratio to detect

and locate human vital sign in very low SNCR condition has been proposed [4].

However in an experiment to detect two human subjects this method failed to detect

farther subject also this algorithm works on condition that number of subjects are

known and CFAR window in fast time cannot detect life sign near start and stop point.


These issues have been resolved in [43].Here Automatic gain control (AGC) method

which suppresses signals with strong power and enhances signals with low power and

Recursive Pixels Finding (RPF) algorithm has been suggested which calculates number

of life signs and also removes clutter signals with same frequency and energy as that of

respiration frequency further there is scope to suppress all type of clutter in complex

real environment. In these works however authors have focused only on fundamental

harmonics of respiratory signals rather than its rest of harmonics with which better

detection could have been achieved.

3.1 Introduction 21

Chapter 3 Extraction of breathing frequency of

human being hidden behind the wall using

different signal processing techniques

3.1 Introduction

The objective of this chapter is to extract life sign signal frequency using FFT and HHT

signal processing algorithm from the data which is collected experimentally using

SFCW radar system.

Organization of chapter is as follows. Section 3.2 describes the methodology in which

experimental setup and radar system parameters used for data collection is included. It

also describes the theory in signal processing algorithms i.e., FFT and HHT methods

which are used for extraction of life sign. Results obtained through Experiments are

presented in Section 3.3. Section 3.4 gives final concluding remarks.

3.2 Methodology

The methodology adopted in this chapter is to collect the data using SFCW radar first

for the person behind the opaque material. After data collection, analysis using signal

processing techniques i.e., FFT and HHT for detecting life sign is carried out. For data

collection, experimental setup is explained in section 3.2.1 and signal processing

techniques are described in section 3.2.2

22 Extraction of breathing frequency of human being hidden behind the wall

using different signal processing techniques

3.2.1 Experimental setup

To extract life sign of human subject hiding behind plywood wall an experimental setup

is developed which is shown in Fig.3.1. SFCW based radar system in a mono-static

mode was used with the help of Vector Network Analyzer (VNA) and single antenna as

shown in Fig.3.1. Calibration of VNA is done by standard one port calibration process

i.e., Open Short Matched before reflection parameter S11 is measured in frequency

range from 1 GHz to 3 GHz. The total numbers of scans or observations that were

carried out are 1024. Data is collected by using VBA Macro program and is transferred

to PC for further processing using MATLAB software.

Fig. 3.1 Experimental Setup

Table 3.1 – Radar Parameter for Experimentation

The obstacle is plywood wall of 12 mm thickness. Distance between antenna and wall

is maintained at 46 cm while distance between wall and human being is fixed at 58 cm.

Sr. No. Radar Parameter and specifications

Parameter Specification

1 Operating frequency range 1 GHz to 3 GHz

2 Radiated Power 0 dBm

3 Number of Frequency points 201

4 Number of traces 1024

5 Horn Antenna Gain 20 dB

6 Antenna Beamwidth 49.68(H Plane),E Plane)

3.2 Methodology 23

The total distance from antenna to target is 105.2 cm. The other radar parameters set for

experimentation are shown in Table 3.1.

3.2.2 Signal Processing

After data collection, processing steps given in the flow chart as shown in Fig. 3.2 are

applied to extract breathing frequency representing life sign. Micro-vibration activities

are observed in every scanning data which can be used to differentiate them from the

static targets. The biologic and static object have much difference in the same range, so

we can extract the vital information and remove the static objects. As to human, the

heartbeat frequency is around 1-2 Hz and the respiration/breathing frequency 0.2-0.5

Hz. The static object's frequency is around 0 Hz. Thus, if we find the frequency in 0.2

to 2 Hz, we can detect/extract the vital information.

Step 1. Read trace data in frequency domain

Data collected by placing an antenna just above the chest of human being is considered.

SFCW radar received the data in frequency domain, and stored in matrix form of

dimension 2011024. To read the first trace, column 1 of data matrix is pick up and so

on for all 1024 traces.

Step 2. Frequency domain to Time domain

. The first trace data is converted into time domain by using Inverse Fast Fourier

Transform (IFFT) [22]. The signal due to single trace after IFFT is given by;

)2exp()()(1

tfjfSts k

K

kk

(3.1)

where t varies from 0 to (K-1)/BW with step interval of 1/BW, BW is bandwidth of the

system , K is maximum number of frequency points and S(fk) is the received reflected

signal in frequency domain at kth

frequency.

Step 3. Time domain to spatial domain



The time domain signal is converted into spatial domain. The location of human being

is determined by constructing a range profile in the spatial domain. Range profile is one

dimensional information and given by expression as;

max

1

0)/2(2exp()()( zzczfjfSzs k

K

kk

(3.2)

where c is velocity of light and z is down range distance which is given as

2)( tc

z

(3.3)

Maximum distance calculated using equation (3.4) gives the values as

BWKc

z2

)1(max

(3.4)

But for analysis, maximum distance is taken as per the room dimension i.e., up to 5 m.

Step 4. Stacking all traces

A single range profile will give information about presence of target only. It does not

indicate amplitude variations due to breathing and respiration. For this radar must

illuminate more number of traces. To see the amplitude variations all the range profiles

are stacked one over other. Thus the steps from 1 to 3 are repeated to collect

information from all 1024 traces.

3.2 Methodology 25

Step 1.Read Trace data in Frequency Domain

Step 2. Convert Frequency Domain Data into Time Domain

Step 3.Convert Time Domain Data into Spatial Domain

Step 4.Stack all 1024 traces to Form data matrix

Step 5. Determination of location based on SD

Repeat for all 1024 Trace Data

Step 6. Find Frequency using (a) FFT or (b) HHT

Fig. 3.2 Flowchart for Signal processing steps

Step 5. Determination of location based on standard deviation (SD)

The purpose of processing should be to help the radar operator to clearly understand

whether life sign of a person hiding behind wall is present or absent. The amplitude

value in all the range profiles at different distances should be observed. If there is no

amplitude variation then target is absent.



To verify that if amplitude variation is present then target is present, standard deviation

(SD) is calculated at all the distances over all the range profiles using equation (3.5).

1

0

)(1 N

iin x

NSD (3.5)

where n is distances which varies up to 5 m with step size of 0.075 m and N is total

number of traces.

The value of SD at human being location will be highest compared to static objects.

Step 6. Find Frequency using (a) FFT (b)HHT

(a)Fast Fourier Transform Method

The steps below shows the algorithm based on Fast Fourier transform (FFT) method for

extraction of life sign from signal after clutter reduction technique is applied.

Step i: Search for peaks in range profile of each trace and then observe the peak

variations in all the traces.

Step ii: Note the peak location at which we get variations i.e where SD is highest.

Step iii: Extract the signal amplitude from all traces for the peak location obtained in

step ii.

Step iv: Convert the extracted signal in step (iii) (amplitude versus number of traces)

into frequency domain by applying FFT.

In frequency spectrum if we get presence of frequencies in range of 0.2 to 2 Hz, we can

say that a human being life sign is extracted. At the same time, the location of human

being is also obtained.

(b) HHT Method

3.3 Result 27

The HHT is a nonlinear and non-stationary signal analysis technique based on the

combination of the two processes i.e., Empirical Mode Decomposition (EMD) and

Hilbert spectral analysis (HSA) [23]-[25].

The EMD algorithm can be understood from the following steps:

Determine the local extrema (maxima, minima) of the signal and connect the maxima

and minima with an interpolation function to create an upper and lower envelope

respectively.

Calculate the local mean as half the difference between the upper and lower envelopes.

Subtract the local mean from the signal to form the residue.

Iterate step (a) to (c) on the residual until the signal becomes Intrinsic Mode Functions

(IMF). This is repeated until the final residue is a monotonic function.

The signal of interest from the human target can be clearly displayed in the time–

frequency domain. The spectrum is obtained by taking HHT to every IMF, which is

known as Hilbert Spectrum Analysis. Most of the energy is confined in IMF which is

having highest peak power.

3.3 Result

3.3.1 Processing for presence of target behind wall

Data is process according to flowchart described in Section 3.2. Figure 3.3 show range

profiles for all the 1024 traces obtained. The figure is plotted between distances versus

magnitude. From the figure it is observed that first peak is due to weak isolation

between transmitting and receiving antenna, second peak is due to wall reflection and

third peak is due to reflection from human being. The encircled shaded portion

represents presence of target. Since the distance between antenna and target is 1.05 m,

fluctuation in reflections is observed at 1.05 m marked by data tip as shown in Fig. 3.3.

We can observe from Fig. 3.3 that the amplitude of the clutter (i.e., reflection due to



antenna air interface, wall reflections, and multiple reflections is higher than reflection

due to human being

3.3.2 Stacking all 1024 traces

All the range profiles acquired are stacked one after another so that two dimensional

images is obtained in which on x-axis, number of traces and on y-axis, down range

distance is plotted. This set of traces is assembled together in a two dimensional data

matrix and visualized as a raw image as shown in Fig. 3.4.

3.3.3 Location determination using standard deviation

To obtain the location of human being, SD is calculated at all locations using equation

(3.5). It is observed from Table 3.2, that at human being location, SD value is high

compared to static object locations like wall due to amplitude variations. Amplitude

variations occur due to respiration and heart beats. The location at which maximum

value of SD is obtained is verified and same as with the location of human being set

during experimental setup.

Table 3.2 SD values at Different location

SD value at Wall location SD value at human being Location

0.0001 0.0055

3.3.4 FFT based results

For better understanding, the amplitude values for detected peaks i.e. peak due to

antenna-air interface, due to wall surface and due to human target are observed from all

1024 traces. It is clearly observed that for the first two peaks i.e. peak due to antenna-

air interface and peak due to wall surface, there is no variation in amplitude for all 1024

traces, while, there is significant amplitude variation at human target location. So in the

next step, only amplitude variation portion is extracted with the help of SD value which

is shown in Fig. 3.5.

3.3 Result 29

Fig. 3.3 Range profiles

Fig. 3.4 Raw Image



Fig. 3.5 Amplitude variation at target location

After applying the FFT to the amplitude variations, the result is obtained and is shown

in Fig.3.6. In frequency spectrum, peak is detected at 0.3125 Hz which is nothing but

the breathing frequency (0.2-0.5 Hz) for human. In this heartbeat frequency (1.2-1.7

Hz) was not detected because of small amplitude variation. The FFT result also

includes other clutter frequencies present in signal.

3.3.5 HHT based results

The extracted amplitude variation from peak locations for all the traces is obtained in

the same way as explain in above. The amplitude variation at target location as shown

in Fig. 3.5 is taken as input for EMD. Now, on this result, we had applied the Empirical

Mode Decomposition (EMD) algorithm to decompose it into a finite set of IMF, until

the residue become monotonic as described in algorithm above.

The signal is decomposed into five IMF components. The total energy is decomposed

into different component. Here stoppage criteria used for EMD is squared difference

(SD) [24]. The threshold value 𝜀 is set to 0.3 (typical value). If the squared difference

(SD) is smaller than threshold, the shifting process will be stopped. Hilbert transfer of

IMF gives us instantaneous frequency contained in IMF component. The time-

3.4 Conclusion 31

frequency plot obtained after applying Hilbert transform on IMF 1 is shown in Fig. 3.7.

In time-frequency plot, we got frequency nearly at 0.3998 Hz which is slightly different

from result obtained using FFT method. It is also observed from result of HHT, that all

the energy is concentrated around 0.3998 Hz which is not the case in FFT method.

Hence compared to result obtained using FFT, the result obtained using HHT shows

significant improvement in extraction of breathing frequency.

Fig. 3.6 Frequency spectrum using FFT

Fig. 3.7 Time-frequency plot after HHT



3.4 Conclusion

This chapter demonstrates the extraction of human breathing using two different signal

processing techniques i.e., FFT and HHT. The chapter also described about how the

location of human being is estimated using SD. The experimental results show that FFT

and HHT methods have successfully extracted the breathing frequency of human i.e.,

0.3125 Hz and 0.3998 Hz respectively. Compared to FFT, HHT produce less harmonic

distortions. We have successfully extracted breathing frequency, but heartbeat signal is

not detected. Advanced signal processing technique can be developed to improve the

performance of extraction of heartbeat signal.

4.1 Introduction 33

Chapter 4 Detection of location and breathing

signal of human standing behind brick wall using

monostatic radar system

4.1 Introduction

Quick detection of trapped victims buried under rubble and their subsequent rescue are

big challenges to scientists and technologists. The current study in life sign detection

focuses on i) radar hardware design and ii) developing the SFCW radar signal

processing technique for human vital sign detections in a number of scenarios that

might be pertinent to efficiencies and reliability in earthquake disaster victims search

and rescues. Ultra-wideband (UWB) radar plays an important role in search and rescue

at disaster relief sites. Various types of UWB radar or continuous wave radar are used

for this purpose but SFCW radar technique offers substantial benefits over other radar

systems.

General types of radars systems can be characterized by the relative locations of

the transmitter and the receiver. All radars can be classified as monostatic, bi-static, or

multistatic. A monostatic radar is one in which the transmitter is collocated with the

receiver. In a technical sense, to be truly mono static, the radar must transmit and

receive with the same antenna to maintain the same aperture position and phase center

of the antenna pattern.

Bi-static radar is one in which the transmitter and the receiver are separated by

some distance. In practice, almost all ground penetrating radars are technically bi-static

34 Detection of location and breathing signal of human standing behind brick

wall using monostatic radar system

since the transmit and receiving antennas are usually separated to provide some

isolation due to the power levels involved. These radars are generally regarded as being

monostatic or pseduo-monostatic since the separation of transmitter and receiver are

small fraction of the range to target. Because the separation of the transmitter and

receiver is small, the angle subtended between the transmitter to target and the receiver

to target is close to zero, a straight, single-shot path.

The objective of this chapter is to use monostatic radar system for the life sign detection

of human being hidden behind wall. The monostatic radar system uses single antenna

for transmission and reception. In this chapter the focus is on measuring transmission

coefficient using VNA in monostatic mode with the help of circulator.

In Section 4.2, description about the methodology adopted is given which consist of

development of experimental setup, data collection and signal processing. The

experimental setup describes how circulator is used in the setup. Data collections

details like number of observations, types of wall used etc are given. The signal

processing algorithm is also described. Section 4.3, describes the results and discussion

obtained on experimental data. Finally Section 4.4 gives conclusions.

4.2 Methodology

The methodology adopted to implement the use of monostatic radar using circulator is

given in steps below.

Step 1 Development of Experimental Setup

Step 2 Data Collection

Step 3 Signal Processing

The detail explanations of each step are given below

4.2 Methodology 35

4.2.1 Development of Experimental Setup

SFCW based radar system in a monostatic mode was assembled with the help of Vector

Network Analyzer (VNA), circulator and antenna as shown in Figure 4.1. In this radar

system, single horn antenna is used for transmission and reception purpose. VNA is

used in frequency range of 2 GHz to 4 GHz with 10 dBm output power. Frequency

range is decided based on the availability of circulator in the same range. The number

of frequency points set are 201 and number of traces that are collected are 1024. Data is

collected by using software VEE pro controlled by PC and VBA macro program and

transferred to PC for further processing. MATLAB software is used for processing.

After calibrating VNA by standard two port calibration process i.e., Through Open

Short Matched (TOSM), the scattering parameters S21 was measured in frequency

domain for all the observations.

The SFCW radar system parameters are given in Table 4.1. The connection

between VNA, circulator and antenna is described here.

Port 1 of VNA is connected to port 1 of circulator. Port 2 of circulator is

connected to Antenna and port 3 of circulator is connected to port 2 of VNA. The

antenna is oriented in vertical polarization for data collection. The antenna was

mounted on tripod. Height of tripod on which antenna is kept is same as chest height of

human being who is standing behind the wall as a target. Types of walls that are used

in the experiments are 2 mm thick plywood and brick wall of thickness 10 cm.

In performing the measurements, single antenna which is used to transmit and receive

is kept at distance away from wall and aligned in line with target for maximum signal

reception. During the experiments, the antenna is placed at different distances away

from wall. The details of main components of the experimental setup are given below.

4.2.1.1 Vector Network Analyzer

In this section, a general background is discussed and the theory of operation of Vector

Network Analyzers (VNA) is explored. In general, network analyzer is a device that is

used to characterize the response of a circuit specifically a linear network to signals of



various frequencies. A network analyzer can be a scalar type which measures only the

magnitude response of a circuit to stimuli of various frequencies. In practice, scalar

network analyzers are rarely used in modern test and measurement. VNA measure both

the magnitude and phase of the response of a network to stimuli. The measurements are

made of both the reflection of the signal looking into a port and the transmission of

signal through that port. A VNA can be of any arbitrary number of ports. Common

types of VNA are a two-port and four-port. The transmit/receive can be thought of as a

full two-port VNA with the condition that both the ports can work as stimulus.

In the general case, the VNA is used to measure the reflection and transmission

coefficients of a device under test (DUT). For a one-port DUT, there is an incident

wave „a‟ that is generated by the VNA and supplied to the DUT and a reflected wave

„b’ returning to the VNA.

Fig. 4.1 Experimental Setup using Monostatic Radar

Table 4.1 Mono-static Radar Parameters

Parameters Specification

Operating frequency range 2 GHz to 4GHz

Radiated Power 10 dBm

No of frequency 201

Number of traces 1024

Antenna Gain 10 dBi

Antenna Beam width at centre frequency H-plane (34.560) and E-plane (35.94

0)

Antenna , VSWR 3:1

Circulator a) Insertion loss and Isolation loss 18 dB, 50 dB

For a one-port DUT, the incident and reflected waves are related by the reflection

coefficient which is defined by

a

b

(4.1)

If the normalized complex characteristic impedance z is defined by

oZ

Zz

(4.2)

where Z is the characteristics impedance of the system and Zo is arbitrary reference

impedance. Equation (4.1) can be rewritten as

1

1

z

z

(4.3)

This means that the ratio of reflected power to transmitted power can be completely

determined by the relationship between the complex impedance of the DUT and the

system reference impedance for a one port system. When this concept is extended to a

two-port DUT, a complete set of scattering parameter measurements is made by setting

the incident waves at Port 1 and then port two to zero and observing the system

responses in each case due to a stimulus at one port only. In the case of the VNA radar

system, there will be no stimulus at Port 2. The subscripts of the S-parameters denote

the port at which the response is observed, i and the port at which the stimulus is

presented, j, in the form Sij. This gives,



021

111 a

a

bS

(4.4)

and

021

221 a

a

bS

(4.5)

And, similarly, by considering there to be a perfect match at port one ( Γ1 = 0 )

and no incident stimulus ( a1 = 0 ) the reverse terms can be obtained as

02

2122 a

a

bS

(4.6)

and

02

1112 a

a

bS

(4.7)

Typical VNA measurements derived for magnitude & phase data are all S-parameters,

input and output impedance, Reflection Coefficient, Transmission Coefficient, Return

Loss, Voltage Standing Wave Ratio (VSWR), Group Delay, Phase Delay, Isolation

loss, insertion loss etc.

There are several key parameters for the VNA. This includes the number of

ports, power level, input power range, dynamic range, IF frequency, number of

frequency points, frequency range, averaging etc. The output power range for the VNA

simply specifies the minimum output, maximum output, and the step size of power

level available for the stimulus signal provided by the VNA to the DUT.

In case of radar, the system designer is able to determine the output power level

required to ensure a measurable return signal from the scene where there are lots of

discontinuities. Power levels in VNA are almost always specified in units of dBm by

convention.

4.2 Methodology 39

4.2.1.2 Horn Antenna

A horn antenna or microwave horn is an antenna that consists of a flaring

metal waveguide shaped like a horn to direct radio waves in a beam. A UWB pyramidal

horn antenna with 20 dB gain having bandwidth of 18 GHz is used for transmitting and

receiving signal. A coaxial cable feed line attaches to the connector visible at top. This

type is called as a ridged horn, the curving fins visible inside the mouth of horn

increases the antenna’s bandwidth.

The Half Power Beam width (HPBW) of antenna at centre frequency in H-

plane and E-plane are found to be 34.56о and 35.94

о respectively. The antenna is

oriented in vertical polarization for data collection. They are used as feed

antennas (called feed horns) for larger antenna structures such as parabolic antennas, as

standard calibration antennas to measure the gain of other antennas, and as directive

antennas for radar and microwave radiometers. Their advantages are

moderate directivity, low voltage standing wave ratio (VSWR), broad bandwidth, and

simple construction.

An advantage of horn antennas is that since they have no resonant elements,

they can operate over a wide range of frequencies, a wide bandwidth. The usable

bandwidth of horn antennas is typically of the order of 10:1, and can be up to 20:1 (for

example allowing it to operate from 1 GHz to 18 GHz). The input impedance is slowly

varying over this wide frequency range, allowing low voltage standing wave

ratio (VSWR) over the bandwidth. The gain of horn antennas ranges up to 25 dBi, with

10 - 20 dBi being typical. For the antenna which is used in the experiment, S11 is

measured from which return loss is obtained in the frequency range of 2 to 4 GHz.

The return loss R is given as,

)log(20 11SR (4.8)

The return losses are 18.84 dBm and 40.02 dBm at frequency 2.01 GHz and 3.78 GHz

respectively.

https://en.wikipedia.org/wiki/Antenna_(radio)

https://en.wikipedia.org/wiki/Waveguide

https://en.wikipedia.org/wiki/Horn_(acoustic)

https://en.wikipedia.org/wiki/Antenna_feed



https://en.wikipedia.org/wiki/Feed_horn

https://en.wikipedia.org/wiki/Parabolic_antenna

https://en.wikipedia.org/wiki/Antenna_gain

https://en.wikipedia.org/wiki/Radar_gun

https://en.wikipedia.org/wiki/Microwave_radiometer

https://en.wikipedia.org/wiki/Directivity

https://en.wikipedia.org/wiki/Standing_wave_ratio

https://en.wikipedia.org/wiki/Bandwidth_(signal_processing)

https://en.wikipedia.org/wiki/Resonant

https://en.wikipedia.org/wiki/Frequency

https://en.wikipedia.org/wiki/Bandwidth_(signal_processing)




https://en.wikipedia.org/wiki/Decibel



Fig. 4.2 UWB Ridge Horn Antenna

4.2.1.3 Circulator

A circulator is a passive non-reciprocal three- or four-port device, in which a

microwave or radio frequency signal entering any port is transmitted to the next port in

rotation only. A port in this context is a point where an external waveguide or

transmission line such as a coaxial cable connects to the device. For a three-port

circulator, a signal applied to port 1 only comes out of port 2; a signal applied to port 2

only comes out of port 3; a signal applied to port 3 only comes out of port 1. The

scattering matrix for an ideal three-port circulator is,

010

001

100

S

(4.9)

The important parameters of circulator are obtained by measurements. The insertion

loss is observed when port one is connected to input port of VNA, port 3 of circulator is

matched terminated and port 2 of circulator is connected to port two of VNA. The

insertion loss obtained for frequencies 2.91 GHz and 3.78 GHz are 38.75 dB and 18.48

dB respectively.

https://en.wikipedia.org/wiki/Passivity_(engineering)

https://en.wikipedia.org/wiki/Reciprocity_(electromagnetism)

https://en.wikipedia.org/wiki/Port_(circuit_theory)

https://en.wikipedia.org/wiki/Microwave

https://en.wikipedia.org/wiki/Radio_frequency

https://en.wikipedia.org/wiki/Waveguide_(electromagnetism)

https://en.wikipedia.org/wiki/Transmission_line

https://en.wikipedia.org/wiki/Coaxial_cable

https://en.wikipedia.org/wiki/Scattering_parameters

https://en.wikipedia.org/wiki/File:Schwarzbeck_BBHA_9120_D.jpg

4.2 Methodology 41

Fig. 4.3 Connection with Circulator for calculation of insertion loss

The isolation loss is observed when port one is connected to input port of VNA, port 2

of circulator is matched terminated and port 3 of circulator is connected to port two of

VNA. It is desired that very negligible power to reach to port 3 of circulator. The

obtained values frequencies 2.11 GHz and 3.31 GHz are 48.15 dB and 53.39 dB.

4.2.2 Data collection

The data were acquired with a total number of 201 frequency steps. One sweep of this

entire band needs roughly 0.30 s to form one trace in time domain. For each data set,

we have collected at least 1024 traces. Data were collected using experimental system

as discussed earlier. The data collection is carried out in three stages. In the first stage

external calibration is done with the help of metal sheet. The second and third stages of

data collection are about change in type of wall. In the second stage plywood wall of 2

mm thickness is used and in third stage brick wall of 10 cm thickness is used. The

detail of data collected in Radar system given below.

4.2.2.1 Calibration using metal sheet with antenna

In this experiment, metal sheet of size 46 feet is kept in front of antenna at distance

0.5m and 1.5m. In this test scenario, we get all signal reflected back, from which we

can obtained the correction required due to connection of circulator



4.2.2.2 Data collection for plywood 2 mm wall

The experimental setup shown in Fig. 4.4 with single antenna and circulator is used for

data collection. The distance between antenna and plywood wall is varied as well as the

distance between plywood wall and target is varied. Table 4.2 shows the details of data

collection for different distances

Table 4.2 Data collected with various distance between Antenna and plywood wall and also

variation in distance between plywood wall and target.

Sr. No. Distance between

Antenna to wall (m)

Distance between Wall to

target (m)

Total distance including

plywood wall thickness (m)

1 0.5 1.5 2.02

2 1 1 2.02

3 1 1.5 2.52

4 1.5 0.5 2.02

5 1.5 1 2.52

4.2.2.3 Data collection for 10cm Brick wall

The data collection details for 10 cm brick wall are as given below in Table 4.3. Fig.

4.5 shows the experimental setup with antenna and circulator, the distance between

antenna and wall is 1m and wall to target is 0.5m

Table 4.3 Data collected with various distance between antenna and brick wall and also variation

in distance between brick wall and target

Sr.

No.

Distance between Antenna to

brick wall (m)

Distance between Brick Wall

to target (m)

Total distance including brick

wall thickness (m)

1 0.5 0.5 1.1

2 1 0.5 1.6

3 1.5 0.5 2.1

4.2 Methodology 43

Fig. 4.4 Experimental Setup with plywood wall

Fig. 4.5 Experimental Setup with brick wall

4.2.3 Signal Processing

The signal processing steps which are followed are shown in flowchart given in Fig.

4.6. The description of steps is as follow:

Step 1 Read trace data in Frequency domain

The data collected with the experimental setup is stored in frequency domain. Since

number of frequencies are 201 and number of traces are 1024, the size of matrix data

stored for one measurement is 201 by 2049. First column indicate frequencies and

second and third column indicates real and imaginary part for first trace, fourth and



fifth for second trace and so on. Hence the first step is to read the data for single trace

which is complex in form.

Step 2 Convert Frequency domain data into time domain data

Since information about location and presence of life sign is required,

processing of data is required to be done in time domain. So the data is converted in

time domain using inverse Fourier transform (IFFT).

Step 3 Convert time domain data into spatial domain data

Since location of target is required, time domain data is converted into spatial

domain. While converting the speed of wave through plywood wall and 10 cm thick

brick wall is considered same as speed of wave in air. For plywood this will not make

any significant difference in location of target that is standing behind it, but definitely it

will make difference for brick wall. That is the target will appear at location slightly far

away than actual as wave propagates through brick wall slowly.

Step 4 Detect locations by plotting all 1024 traces

After plotting the range profile for single trace different peaks due to

discontinuities in the scene are observed. These discontinuities are due to circulator,

antenna air interface, wall, target and multipath also. Since we want to detect the

location of target, and to find location, we need to detect the life sign such as

respiration and heart beat of target.

In a single trace, life sign cannot be detected since changes in phase and amplitude of

reflected signal is required to be detected. So all the traces are plotted and observed for

changes in amplitude at all distances. The distance where amplitude changes occur is

considered as location of target. Since there is displacement due to heart and respiration

of target, amplitude variations may occur at more than one location, provided the radar

range resolution is very good.

4.2 Methodology 45

Fig. 4.6 Flowcharts for Signal Processing

Step 5 Extract amplitude value from target location for all 1024 traces

Once the target location is obtained, the amplitude value at target location is

extracted from all the 1024 traces. Since amplitude variation is observed at more than

one place, three amplitude values are chosen, i.e., adjacent two values (previous and

later). The average of three values is taken and is normalized between plus one to

minus one.

Step 6 Apply FFT algorithm to extract Respiration frequency



Fast Fourier transform is applied on the normalized data obtained in previous

step. The sampling frequency should be chosen following Nyquist criteria.

4.3 Result and Discussion

The data collected as described earlier are used for processing. Data are collected in

three stages. In the first stage data are collected for external calibration. External

calibration is carried out to find out the how much correction is required due to used of

circulator in radar. Since the calibration is done before the point circulator is connected,

the reflection due to circulator will appear in the observations. In the second stage, data

are collected in the corridor where plywood wall is used and in third stage data is

collected in room where 10 cm thick brick-wall is used. The processing of data

collected in different stages is carried out and described below.

4.3.1 Metal Calibration

External calibration involves the measurement of returns from a calibrator with known

reflections, such as a metal plate. External calibration is necessary to characterize the radar

systemic error.

External calibration is carried out for finding external error in reading. In this

metal plate of dimension 34 m is kept in front of antenna. Two distances are taken i.e.,

0.5m and 1.5m. The results obtained are described in case I and II.

Case I: When distance between Antenna and Metal sheet is 0.5 m

It is observed from Fig. 4.7 (a) that, reflection due to metal is observed at 2.475

m. The difference between the actual and measured is 2.475-0.5=1.975m. So the

correction required due to circulator and antenna is 1.975m. . Also observed reflections

due to circulator and antenna air interface marked with text arrow in Fig. 4.7 (a).

Case II: When distance between Antenna and Metal sheet is 1.5 m

4.3 Result and Discussion 47

It is observed from Fig. 4.7 (b) that, reflection due to metal is observed at 3.45

m. The difference between the actual and measured is 3.45-1.5=1.95 m. So the

correction required due to circulator and antenna is 1.95 m. On average from case I and

case II results, 1.9625 m is the correction required.

Fig. 4.7 Metal calibration (a) When Antenna to Metal distance is 0.5m,

(b) When Antenna to Metal distance is 1.5m

4.3.2 Determination of location for data taken in corridor with

plywood

In this experiment, the subject is standing behind the plywood wall in normal breathing

condition. The distance between antenna and wall is varied from 0.5m to 1.5m and

distance between wall and subject (target) is varied from 0.5 m to 1.5m. The data

collection details are described in Section 4.2.2. The results obtained after processing

data for different cases are shown below.

Case I: When distance between Antenna and plywood Wall is 0.5m and distance

between Plywood wall to Human being is 1.5m



(a) Encircled part showing reflections (b) Location of target due to target

Fig. 4.8 Distance between Antenna and wall is 0.5m and between wall and target is 1.5m

Figure 4.8 (a) shows reflections due to target when all 1024 range profiles are plotted

All the 1024 traces are plotted on each other. The encircled area shows the reflection

due to target. The basic idea of detecting the human chest motion is by detecting the

time shift of peak of the reflected signal in subsequent signals which is directly

proportional to the displacement of the chest.

Here we have converted the time domain into range (distance) as shown in Fig. 4.8 (a).

It is identified as reflection due to target because, that portion appears shaded due to

variation in amplitude due to chest movement. Variation in amplitude occurs due to

respiration and heart beat of human being who is standing behind wall. The other

reflections in the Fig. 4.8 (a) appears with no variations in all 1024 range profiles which

indicate that they are from static parts present in between the antenna and human being

medium like circulator, wall, etc.

Figure 4.8 (b) shows the result obtained for the data taken when the distance

between antenna and wall is kept as 0.5m and distance between wall and target is 1.5m.

Since the plywood is of 2 mm thickness, the total distance should be 2.02 m. Figure 4.8

(b) shows location of reflections due to target when all 1024 range profiles are plotted

on each other. Figure 4.8 (b) shows the location of target as marked as 3.9 m. From

external calibration it is observed that 1.9625m distance should be subtracted from 3.9

m for obtaining correct location of target. So the distance obtained is 3.9m -1.9625m


=1.9375m. The error would be 2.021.9375 = 0.0825m which is acceptable as the

resolution is 0.075m. Figure 4.8 (b) also shows reflection due to circulator at location

1.725 m, antenna air interface at 2.1m and reflection due to plywood wall at 2.325m.

Since we are interested in reflection due to target, hereafter reflections due to antenna

air interface and reflection due to wall are not marked in figures.

Case II: When distance between Antenna and plywood Wall is 1m

(a) distance between Plywood wall to Human being is 1m

(b) distance between Plywood wall to Human being is 1.5m

Figure 4.9 shows reflections due to target when all 1024 range profiles are plotted on

each other. All the 1024 traces are plotted on each other.

Figure 4.9 (a) shows the result obtained for the data taken when the distance between

antenna and wall is kept as 1m and distance between wall and target is 1m. Since the

plywood is of 2 mm thickness, the total distance should be 2.02 m.

(a) Distance between wall to target is 1m (b) Distance between wall to target is 1.5m

Fig. 4.9 Distance between Antenna and wall is 1m

Figure 4.9 (a) shows the location of target as marked as 3.975 m. From external

calibration it is observed that 1.9625m distance should be subtracted from 3.975 m for

obtaining correct location of target. So the distance obtained is 3.975m1.9625m

=2.0125m. The error would be 2.0125 2.02=0.0075 m which is acceptable as the

resolution is 0.075m.



Figure 4.9 (b) shows the result obtained for the data taken when the distance

between antenna and wall is kept as 1m and distance between wall and target is 1.5m.

Since the plywood wall is of 2 mm thickness, the total distance should be 2.52 m.

Figure 4.9 (b) shows the location of target as marked as 4.425 m. So the distance

obtained after correction is 4.425 m 1.9625m =2.4625m. The error would be 0.0575 m

which is acceptable as the resolution is 0.075m.

Case III: When distance between Antenna and plywood Wall is 1.5m

(a) distance between Plywood wall to Human being is 0.5m

(b) distance between Plywood wall to Human being is 1m

(a) Distance between wall to target is 1.5m (b) Distance between wall to target is 0.5m

Fig. 4.10 Distance between Antenna and wall is 1.5m

Fig.4.10 shows location of reflections due to target when distance between antenna and

wall is 1.5 m and distance between wall and target is 0.5m. Figure 4.10 (b) shows

location of reflections due to target when distance between antenna and wall is 1.5 m

and distance between wall and target is 1m. Figure 4.10 (a) shows the location of target

as marked as 3.975 m. The error would be 0.0075m. Figure 4.10 (b) shows the result

obtained for the data taken when the distance between antenna and wall is kept as 1.5m

and distance between wall and target is 1m. Figure 4.10 (b) shows the location of target

as marked as 4.425 m. So the distance obtained after correction is 4.425 m 1.9625m

=2.4625m. The error would be 0.0575 m. The error values obtained from all the three

cases are shown in Table 4.4.


Table 4.4 Location determination when observation taken in corridor with Plywood Wall

Measured Observed Error

Distance

between

Antenna to

wall(m)

Distance

between

Wall to

Human

Being(m)

Total distance

including

plywood wall

thickness (m)

Reflection

due to

Circulator

(m)

Reflection

due to

Target (m)

Distance to

Target after

correction of

1.9625 m (m)

Difference

between

Observed and

measured

distance (m)

0.5 1.5 2.02 1.725 3.90 1.9375 0.08

1 1 2.02 1.725 3.975 2.0125 0.0075

1 1.5 2.52 1.725 4.425 2.4625 0.0575

1.5 0.5 2.02 1.725 3.975 2.0125 0.0075

1.5 1 2.52 1.725 4.425 2.4625 0.0575

4.3.3 Determination of frequencies of life sign for data taken in

corridor with plywood wall

Five data sets were processed using algorithm described in Section 4.2.3 to

extract the frequency of life sign signal. It is observed from results, respiratory

frequency is extracted but failed to extract heart beat frequency. It appears that heart

beat signal is harder to detect. If the results shown in Fig 4.11 (a) to (e) are observed,

there are many frequencies with insignificant amplitude values. These frequencies may

be due to harmonics and noise present in the signal. Figure 4.11 (a) is plotted when

distance between antenna and plywood Wall is 0.5m and distance between Plywood

wall to human being is 1.5m.

(a) (b)



(c) (d)

(e)

Fig. 4.11 Frequency Spectrum of human being standing behind plywood wall at different distance

Fig.4.11 (b) is plotted when distance between antenna and plywood Wall is 1m and

distance between Plywood wall to Human being is 1m. Fig. 4.11 (c) is plotted when

distance between antenna and plywood Wall is 1m and distance between Plywood wall

to Human being is 1.5m. Fig. 4.11 (d) is plotted when distance between antenna and

plywood Wall is 1.5m and distance between Plywood wall to Human being is 0.5m.

Fig. 4.11 (e) is plotted when distance between antenna and plywood Wall is 1.5m and

distance between Plywood wall to Human being is 1m. The breathing frequencies

measured for all the data set are tabulated in Table 4.5. The variation in breathing

frequency is from 0.1953 Hz to 0.332 Hz which is acceptable range.

Table 4.5 Frequency determination when observation taken in corridor with Plywood Wall

Distance between Antenna

to wall (m)


Human Being (m)

Respiration Frequency

observed (Hz)

0.5 1.5 0.2734


1 1 0.1953

1 1.5 0.332

1.5 0.5 0.332

1.5 1 0.293

4.3.4 Determination of location for data taken in room with Brick

wall

In this experiment, the subject is standing behind the brick wall in normal breathing

condition. The distance between antenna and wall is varied from 0.5m to 1.5m and

distance between wall and subject (target) is kept constant that is 0.5 m. The data

collection details are described in Section 4.2.2. The results obtained after processing

data for different cases are shown below.

Case I: When distance between antenna and brick wall is 0.5m and distance

between brick wall to human being is 0.5m

Figure 4.12 shows the result obtained for the data taken when the distance between

antenna and wall is kept as 0.5m and distance between wall and target is 0.5m. Since

the brick wall is of 10 cm thick, the total distance should be 1.1m. Figure 4.12 (a)

shows reflections due to target when all 1024 range profiles are plotted on each other.

The encircled area shows the reflection due to target. The basic idea of detecting the

human chest motion is by detecting the time shift of peak of the reflected signal in

subsequent signals which is directly proportional to the displacement of the chest. It is

identified as reflection due to target because, that portion appears shaded due to

variation in amplitude due to chest movement.



(a)Encircled part showing amplitude variation (b) Target location

Fig. 4.12 Distance between antenna and Brick wall =0.5m

Variation in amplitude occurs due to respiration and heart beat of human being who is

standing behind wall. The other reflections in the Fig. 4.12 (a) appears with no

variations in all 1024 range profiles which indicate that they are from static parts

present in between the antenna and human being medium like circulator, wall, etc.

Figure 4.12 (b) shows location of reflections due to target when all 1024 range

profiles are plotted on each other. Figure 4.12 (b) also shows the location of target as

marked as 3.075 m. So the distance obtained after external calibration is 3.075m -

1.9625m = 1.1125m. The error would be 1.11251.1 = 0.0125m which is acceptable as

the resolution is 0.075m.

Case II: When distance between Brick wall to Human being is 0.5m

(a) distance between antenna and brick wall is 1 m

(b) distance between antenna and brick wall is 1.5 m

(a)When Distance between (b) When Distance between

Antenna and wall =1m Antenna and wall =1.5m

Fig. 4.13 Distance between Brick wall and Human target 0.5m

Figure 4.13 (a) shows location of reflections due to target when distance between

antenna and wall is 1m and distance between wall and target is 0.5m.


Figure 4.13 (b) shows location of reflections due to target when distance between

antenna and wall is 1.5m and distance between wall and target is 0.5m. Figure 4.13

(a) shows the location of target as marked as 3.75 m. From external calibration it is

observed that 1.9625m distance should be subtracted from 3.75 m for obtaining correct

location of target. So the distance obtained is 3.75m1.9625m =1.79m. Since the brick

wall is of 10cm thickness, the total distance should be 1.6m. The error would be 1.79-

1.6=0.19m which is acceptable as the resolution is 0.19m.

Figure 4.13(b) shows the result obtained for the data taken when the distance

between antenna and wall is kept as 1.5m and distance between wall and target is 0.5m.

Since the brick wall is of 10cm thickness, the total distance should be 2.1m. Fig.3.7 (b)

shows reflections due to target when all 1024 range profiles are plotted on each other.

All the 1024 traces are plotted on each other. Figure 3.7 (b) shows the location of target

as marked as 4.35 m. So the distance obtained after correction is 4.35 m 1.9625m

=2.39m. The error would be 0.29 m which is acceptable as the resolution is 0.29m. All

the error values obtained are tabulated as shown in Table 4.6.

Table 4.6 Location determination when observation taken in room with Brick Wall

Measured Observed Error

Distance

between

Antenna

to

wall(m)

Distance

between

Wall to

Human

Being(m)

Total

distance

including

Brick

wall

thickness

(m)

Reflection

due to

Circulator

(m)

Reflection

due to

Target

(m)

Distance

to Target

after

correction

of 1.96 m

(m)

Difference

between

Observed

and

measured

distance

(m)

Final

Distance

after

velocity

correction

Difference

between

Observed

and

measured

distance

after

velocity

correction

0.5 0.5 1.1 1.725 3.075 1.1125 0.0125 1.1 0

1 0.5 1.6 1.725 3.75 1.79 0.19 1.54 0.06

1.5 0.5 2.1 1.725 4.35 2.39 0.29 2.14 0.04

4.3.5 Determination of frequencies of life sign for data taken in room

with Brick wall

Three data sets were processed using algorithm described in Section 4.2.3 to

extract the frequency of life sign signal. After applying FFT to the amplitude variation

the highest peak is detected as a dominant frequency present in the spectrum. The



respiration frequencies are noted in Table 4.7 for all data set. If the frequency spectrum

shown in Fig 4.14 (a) to (c) are observed, there are many frequencies with insignificant

amplitude values. These frequencies may be due to harmonics and noise present in the

signal. Figure 4.14 (a) is plotted when distance between antenna and brick Wall is 0.5m

and distance between brick wall to human being is 0.5m.

(a) Distance between antenna & brick wall is 0.5m (b) Distance between antenna & brick wall is 1m

(c) Distance between antenna and brick wall is 1.5 m

Fig. 4.14 Frequency Spectrum of human being standing behind plywood wall at different distance

Figure 4.14 (b) is plotted when distance between antenna and brick wall is 1m

and distance between brick wall to human being is 0.5m. Figure 4.14 (c) is plotted

4.4 Conclusion 57

when distance between antenna and brick wall is 1.5m and distance between brick wall

to human being is 0.5m.

Table 4.7 Respiration Frequency determination when observation taken in room with brick wall

Distance between Antenna to

wall (m)


Human Being (m)

Respiration Frequency observed

(Hz)

0.5 0.5 0.273

1 0.5 0.356

1.5 0.5 0.332

4.4 Conclusion

In this chapter, we have proposed monostatic radar system to measure transmission

parameter with the help of circulator. The experimental results shows that proposed

system is useful when the distance between radar and target is small, since most of the

energy is consumed by circulator. Some clutter reduction technique can be used to

improve the signal strength so that detection of life sign signals is possible.

Since the human target is same for data collected when plywood wall and brick

wall is used, the respiration frequency values remains same as observed from results.

Also due to increase in distance between antenna and human being, the value of

respiration frequency does not changed.

Result shows successful detection of breathing frequency of human. In future,

need to work for detection of heart beat frequency using advance signal processing

technique.

Chapter 5 Detection of location of target

and breathing frequency of human being

standing behind brick wall using two antenna

systems

5.1 Experimental setup with 10 cm brick wall

To extract breathing signal of human subject hiding behind the brick wall,

measurement using an experimental setup is carried out in corridor of college building.

The experimental setup as shown in Fig. 5.1, consists of antennas, Vector Network

analyzer, Laptop, cables to connect them and human being standing behind brick wall.

The SFCW radar system parameters used in the experimental setup is described in

Table 5.1. Two identical pyramidal horn antennas with 20 dBi gain having bandwidth

of 18 GHz are used for transmission and reception. The distance between two antennas

is kept 0.30 m. The antennas are mounted on tripod with height adjusted such that the

antennas are exactly aligned with the chest of human being. The antenna is oriented in

vertical polarization for data collection. The dimension for brick wall is 5 6 feet with

thickness of 10 cm. Observations were carried out with and without target behind the

wall. The total distance between the antenna and target is varied in presence of wall.

5.2 Data collection with 10 cm thick brick wall

Three sets of data are collected. In the first data set, human being (target) is absent,

5.2 Data collection with 10 cm thick brick wall 59

whereas in second and third data set the target is present. Measurements were done on a

26 year old male test subject. The distance between wall and target (human being) is

varied whereas the distance between both antennas and wall is kept constant in all

measurements. In the second set, distance between antenna and target is 1m whereas in

third set, the distance between antenna and target is 1.5m.

Fig. 5.1 Experimental Setup with VNA, two antennas and human subject standing behind brick

wall

Two port standard calibration process i.e., Through Open Short Matched (TOSM) is

done for measurement of transmission parameter S21 in UWB frequency range i.e., 1

GHz to 3 GHz. Bandwidth of 2 GHz is used to achieve range resolution as 7.5 cm so

that reflection from even closely spaced targets can be achieved. The total numbers of

scans or observations that were carried out are 1024. The acquired data is saved in the

laptop for every measurement. Every measurement is organized as a matrix as shown in

Table 5.2, in which each row represents the number of frequency points i.e. N=201, i.e.,

signal for one frequency and column represents number of traces i.e., M=1024 i.e.,

number of times signal received at antenna location. For each trace, complex data in

frequency domain is stored.

60 Detection of location of target and breathing frequency of human being

standing behind brick wall using two antenna systems

Table 5.1 SFCW radar parameters with two antennas

Parameters Values

Frequency Range 1 to 3 GHz

Transmitted Power 10 dBm

Number of Frequency points 201

Number of Traces 1024

Antenna Type Horn

Gain of Antennas 20 dB

3 dB Beam width of Antennas 49.68(H-plane),

59.36 (E-plane)

5.3 Signal Processing Algorithm

To obtain the spectral content of the reflected signal from the human being due to chest

movement caused by respiration and heart beat, signal processing techniques are

applied. The signal processing is carried out in steps described in the flowchart, in Fig.

5.2, in the following subsection described the steps involved in flowchart.

Table 5.2 Matrix representing organization of data collected

Trace No.

Frequency. No.

1 2 3 4 … M

Fo S11 S12 S13 S14 … S1M

Fo+f S21 S22 S23 S24 … S2M

… … … … … … …

Fo+(N-1) f SN1 SN2 SN3 SN4 … SNM

Step 1. Read Trace data in Frequency domain

Signal is obtained by placing both (transmitting and receiving) antennas at a height that

is equal to height of chest of target and in line of sight of target. SFCW radar received

the data in frequency domain, and stored as shown in Table 5.2. To read the first trace,

column 1 data is picked up and filtered by applying a standard windowing function,

5.3 Signal Processing Algorithm 61

such as the Hamming window [17]. By applying hamming window, the side lobes are

reduced which will help to reduce false alarm rate and hence improve dynamic range of

the detection.

Step 2. Convert Frequency domain data to Time domain

After windowing in frequency domain, it is converted into time domain by using the

Inverse Fast Fourier Transform (IFFT). The converted signal is presented as signal

strength vs. time delay. The signal received at one of the measurement, after IFFT is

given by Freundorfer, et. al., [31],

)2exp()()(1

tfjfSts n

N

nn

(5.1)

where N is maximum number of frequency points, S(fn) is the received reflected

signal in frequency domain at nth

frequency and t varies from 0 to (N-1)/BW with step

interval of 1/BW, BW is bandwidth of the system.

Step 3. Convert Time domain data to spatial domain

The distance to targets is determined by constructing a range profile in the spatial

domain. The time domain signal is then converted to spatial domain and is called as

range profile. Range profile is one dimensional information and given by expression

as;

max1

0)/2(2exp()()( zzczfjfSzS n

N

nn

(5.2)

where z is down range distance given as

2/)( tcz (5.3)

and c is velocity of light.



The maximum distance Zmax and range resolution R are determined by following

formulas

BWNcZ 2/)1(max (5.4)

and

fNcR

2 (5.5)

where f is step size.

These are the basic steps that are implemented before processing the signal for further

analysis to retrieve information’s which are described in next section.

Step 4. Stacking all traces to form data matrix

A single range profile will be insufficient to give information about presence of life

sign of target. Radar must illuminate more number of traces. To see the amplitude

variations, all the 1024 traces are stacked one after another to form matrix.

Step 5. Apply Clutter reduction technique

The major problem in life sign detection of human being hiding behind brick wall is

loss of substantial amount of energy of signal due to reflection from wall. Due to high

contrast between brick wall and air, reflection occurs. Thus small amount of energy is

passed through wall and reach to the target. Again at target, due to dielectric contrast

between air and human being, reflections take place. The signal reflected by the target

reaches to receiving antenna after passing through wall again further reducing its signal

strength. As a result, this strong reflection which is called as clutter obscures the target

information. Also static clutter is caused due to contribution from other objects present

in external environment. Thus clutter reduction is inevitable for target detection. For

clutter reduction, it is important to separate target and clutter first from the received


data. Clutter reduction is carried out using the SVD technique. This subspace method

divides the data into target and clutter subspaces.

Let us consider that data are taken when the target is present and represented by the

matrix with dimensions N × M; and (i = 1, 2,…, N; j = 1, 2,…, M), where the indices i

and j are the distance (down range) and number of traces respectively. The SVD can be

represented as:


Step 2. Convert Frequency Domain Data into Time

Domain


Step 4.Stack all 1024 traces to Form data matrix

Step 5. Apply Clutter Reduction

Step 6.Find Location of target using SD


Step 7.Apply FFT to find frequency



Fig. 5.2 Flow-chart for processing steps

TUSVX (5.6)

Or

Tii

N

ii vuX

1

(5.7)

Or

NXXXX ...21 (5.8)

where Xi are called as ith

Eigen image of X. Our aim is to identify which Eigen image

represents target response. When range profile with and without presence of the target

is compared, we conclude that the second column of U (u2) i.e., second singular

component refers to the target signal. The target response (T1) estimated can then be

obtained from:

TvuXT22221

(5.9)

T1 represents the image having focused target response and reduced clutter. Since the

wall reflections are stronger than target reflections, the dominant first singular value

represents wall subspace. Also while taking measurement it is ensured that the antenna

is perpendicular to the wall surface i.e., antenna is not tilted at any angle with respect to

brick wall surface and also the wall thickness remains same for all the measurements

the clutter subspace does not span to multidimensional subspace [20]. However the

target subspaces do not remain single subspace and may span to multidimensional

subspace as there is variation of amplitude values at target location. So instead of

considering only second singular components as used in equation (5.9), all the singular

components related to target should be used. Since all N indices do not contribute to

target subspace, there is need to determine which singular components relates to target

subspace. The distance from range profiles are used to find out the corresponding

singular components. The target subspace obtained using multiple Singular components


are obtained as equation (5.10);

i

Tiii vuT2

(5.10)

where is a set of all indices for target singular vectors.

More details about the clutter reduction technique are described in [17].

Step 6. Find Location of target using standard deviation (SD)

The purpose of processing should be to help the radar operator to clearly understand

whether life sign of target is present or absent. The amplitude value in all the range

profiles at different distances should be observed. If there is no amplitude variation at

all the distances in all 1024 traces, then target is absent. If target is present, then at

target location there will be amplitude variations at all 1024 traces due to chest

movement caused by respiration and heart beat. To observe whether there is amplitude

variation or not, standard deviation (SD) is calculated at all the distances over all the

range profiles using equation (5.11).

1

0

)(1 M

mid x

MSD

(5.11)

where d is distance taken upto 5 m with step size of 0.075 m and M is total number of

traces. The value of SD at human being location will be higher compared to static

objects.

Step 7. Extraction of Breathing Frequency using the Fast Fourier Transform Method

Once the location where SD value is maximum is obtained, the following steps are

applied before the Fast Fourier transform (FFT) method is used for extraction of life

sign.

1. Extract amplitude of signal from all traces for the location obtained in step 6. Since

there is fluctuation, instead of amplitude at one location, succeeding and preceding

locations amplitude are also picked up and their average is taken as amplitude



variations at target location. Since there are 1024 traces, total number of amplitude

values will be 1024.

2. Normalize the amplitude values between 1.

3. Convert the extracted signal in step (ii) (amplitude versus number of traces) into

frequency domain by applying the FFT algorithm.

4. The breathing frequency for normal human being corresponds to frequencies 0.2 to

0.5 Hz while heart rate corresponds to 0.8 to 1.5 Hz. Since we are interested in

frequencies in the range of 0.2 to 1.5 Hz, a second order band pass filter using butter

worth is applied before the FFT.


The aim is to develop algorithm to improve the detection of respiratory motion

response of a human being standing behind the brick wall by reducing stationary clutter

originating from fixed object like wall reflections, reflections due to antenna air

interface, multipath etc.,. The performance of algorithm is investigated by carrying out

experimental work in absence of target and by changing distance between source and

human target. Three sets of data are collected by the experimental work. In the first set

the target is absent, in second set the distance between radar and standing human being

is set as 1m and in third set it is 1.5m. The processing is done according to the

flowchart described in Fig. 5.2. Results for these sets of data are discussed below.

5.4.1 Absence of target behind brick wall

First set of data is considered for processing which is taken without target. Figure 5.3

show range profiles, when all the 1024 traces overlapped on each other. The figure is

plotted for distance versus magnitude. For range profile, maximum distance calculated

using equation 5.4 is 15 m but it is taken up to 5 m.


When the target is absent behind brick wall only three significant peaks are observed as

shown in Fig. 5.3. The first peak is due to weak isolation between transmitting and

receiving antenna, second peak is due to front side of brick wall and third peak is due to

back side of brick wall. Since the radar is used in UWB range, it resolve front and back

side of brick wall as thickness of wall is greater than range resolution. As target (human

being) is absent, amplitude variation in range profiles is not observed. The remaining

small peaks observed thereafter are due to the multipath reflections.

5.4.2 Presence of target behind Brick wall

Detection of target can be obtained from the single range profile by observing reflected

amplitude at target location. To observe amplitude variations due to

breathing/respiration, radar must illuminate more number of traces. To see the

amplitude variations all the range profiles are stacked one over other. Thus the

information in more than one traces are collected. Figure 5.4 (a) shows the range

profiles for all 1024 traces when the second set of data is used in which target (human

being) is present behind the wall. When the target is present, additional peaks compared

to Fig. 3, due to reflection from target (human being) is observed as shown in Fig.5.4

(a) and (b). The shaded portion represents presence of target marked by data tip, which

indicate location as 1.35m and 1.8m for second data set and third set respectively. It is

noted that, location of the target from the antenna system is different from actual due to

presence of brick wall in between them.

The amplitude values for detected peaks i.e. peak due to antenna-air interface, due to

wall surface and due to human target for all 1024 traces are observed from Fig. 5.4 (a)

and (b). From the Fig. 5.4, it is clearly observed that for the first three peaks i.e. peak

due to antenna-air interface and peak due to front and back side of wall surface, there is

no variation in amplitude for all 1024 traces, while, there is significant amplitude

variation at human target location. The variations in amplitude at locations of target are

due to respiration and heart rate. Since the range resolution is 0.075m, the minimum

difference between two adjacent locations will be 0.075 m. We can observe from Fig.

5.4 (a) and (b), that the amplitude of the clutter (i.e., reflection due to antenna air



interface and wall) is higher than reflection due to human being. By using clutter

reduction technique, improvement in the target signal strength can be achieved which is

described in next section.

5.4.3 Raw Image (Amplitude versus Slow time variation)

All the range profiles acquired are stacked one after another so that two dimensional

image is obtained in which on x-axis we have number of range profiles or number of

traces and on y-axis we have down range distance. This set of traces, can thus be

assembled together in a two dimensional structure, and visualized as an image known

as raw image [18]. The raw image as shown in Fig.5.5 (a) and (b) for second and third

data set is obtained. We can observe that the amplitudes of the clutter (i.e., reflection

due to antenna air interface and wall reflections) are higher than target reflection. i.e.,

though target is present, the signal strength is so weak that the reflection due to target is

not significant in Fig.5.5 (a) and (b). The improvement in the target signal strength can

be achieved by using clutter reduction technique which is described in next section.

5.4.4 Improvement in the detection using Clutter reduction technique

The clutter reduction as described in step 5 of Fig. 5.2, is applied on raw image. In

Fig.5.6 (a) and (b), the changes in the amplitude at target location due to human subject

movements can be easily observed i.e., from Fig. 5.6 (a), at 1.35 m in range, there is

amplitude variation as observed with dark and light shaded portion all along the number

of traces.


Fig. 5.3 Range profiles for all 1024 traces in absence of target

(a)Second data set



(b) Third data set

Fig. 5.4 Range profiles for all 1024 traces in the presence of target

Similarly from Fig. 5.6 (b), at 1.8m, the amplitude variations are observed. From

these results one can visualize that target is enhanced with suppression of clutter.

(a) Second data set

(b) Third data set


Fig. 5.5 Raw Image

To measure the performance of clutter reduction technique, average target signal

strength is calculated. For second data set, the average signal strength value before

clutter reduction at target location is 0.045 whereas after clutter reduction using only

second singular components (using equation 9) it is 0.2732. It is further improved when

seven singular components related to target subspace are used to form target image as

described in equation (10) to 0.3777. These are obtained by observing range profiles for

each singular component. If the target reflection is observed then that singular

component is considered for forming target image. All such singular components are

used to form final target image. The image formed after adding all images

corresponding to seven singular components i.e., second, third, fourth, fifth, sixth,

seventh and eigth, is obtained as shown in Fig.5. 7(a). For the third data set, the

average signal strength is 0.0055 before clutter reduction and after clutter reduction

using only second singular component, it is improved to 0.3950. When the three

singular components i.e., second, third and fourth are used the average signal strength is

further increased to 0.4436. The image formed after adding all images corresponding to

three singular components i.e., second, third and fourth, is obtained as shown in Fig.

5.7(b). Here since the distance between radar and target is increased, the number of

singular components related to target subspace has reduced. In Fig.5.7 (a) and (b), the

changes in the amplitude at target location due to human subject movements can be

easily observed at same location as earlier. Thus enhanced value of target signal

strength at target location is observed either using single singular component or

multiple singular components related to target space are taken.



(a) Second data set

(b) Third data set

Fig. 5.6 Image after clutter reduction

(a) Second data set


(b) Third data set

Fig. 5.7 Image formed using Target Singular Components

5.4.5 Location Determination using standard deviation

To obtain the location of human being, SD is calculated at all distance value using

equation (5.11). The location at which maximum value of SD is obtained is verified

with the location of human being set during experimental setup. The Table 5.3, gives

SD values for both the experimenntal data sets. The location at which SD is highest is

obtained by using index value. From Table 5.3, for the second data set, the index value

obtained is 18 which give distance as 1.35m. Index value is converted into distance by

multiplying it by range resolution of 0.075. Similarly for third data set, the index value

is 25 and the distance obtained is 1.8m. If we compare the result obtained in Table 5.3

and result obtained in Fig. 5.4, the target locations in both cases are same. Thus it

confirms that SD can be used to find location and presence of target.

Table 5.3 Location of target obtained using SD

Maximum SD value Index /location at

which maximum SD

occurs

True Location of

target after

correction (m)

Second Set of

Experiment

0.0253/0.0270 18/1.35m 1.05

Third Set of

Experiment

0.0112/0.0120 25/1.8m 1.5



In the second data set, the actual distance between antennas and target is 1m but

distance obtained is 1.35 m marked by data tip as shown in Fig. 5.4 (a). The true

location can be obtained by using equation (5.12).

)1( rwallobsevedtrue ddd (5.12)

where dobserved is observed distance from Fig. 4 (a) which is 1.35m, dwall is thickness of

wall which is 0.1m and r is average dielectric constant which is measured as 9 of the

wall. After using external metal calibration, which gives shift of 0.1m, the calculated

true location has 5cm error for the second data set whereas for third data set the error is

zero. Since the resolution of radar system is 0.075 m the error is within acceptable

range.

5.4.6 FFT based Results

After finding location of target, the steps described in step 6 of Fig. 5.2., are applied. It is

observed that the adjacent position also have significant amplitude variations. So, three

amplitude values are selected i.e., the location at which SD is maximum, previous and

succeding location. The average of three locations amplitude is taken as input to FFT.

Assuming normal breathing frequency range as 0.2-0.5 Hz for human, a band pass filtered is

used . The result obtained after FFT is shown in Fig. 5.8 (a) for second data set and in Fig.

5.8(b) for third data set. The frequency components of interest are chosen by observing largest

amplitude. The frequencies observed are 0.4102 Hz and 0.3125 Hz respectively. Similar results

are obtained if image after clutter reduction using all target subspace is used. So the breaths/min

would be at about 24 and 19 respectively. From Fig. 5.8, other smaller peaks are observed

which are probably due to harmonics of breathing frequencies or clutter.

5.5 CONCLUSION 75

(a) Second data set

(b) Third data set

Fig. 5.8 Frequency Spectrum

5.5 CONCLUSION

The focus of this chapter is to explore the possibility of improvement in the detection and

extraction of breathing frequency of human being positioned behind the brick wall using SFCW

radar system in UWB frequency range. After applying clutter reduction technique using SVD,

clutters are successfully minimized which implies that the technique is powerful. Target signal

strength is improved further by using all Singular components related to target. Due to use of

clutter reduction target signal strength increases which increases the probability of correct

target detection.



Increase in target signal strength is useful for process of extraction of life sign signal.

The proposed methods not only extract the breathing, but also estimate the location of

human being using SD. The determination of higher SD value helps in the automatic

detection of presence of life sign signal of human being along with location. The

experimental results also show that FFT method is useful for extraction of breathing

frequency of human being after using band pass filter. From result obtained using

second and third set data, it is observed that when the distance between antenna and

target is increased, the amplitude of breathing signal reduces.

Extraction of heart beat frequency can be carried out in future using advance signal

processing techniques. The effect of non stationary clutter i.e., other movements in

nearby region of the detection is not considered in this work. In future study, brick wall

can be replaced by different complex type of walls like concrete with metal inside can

be considered and its effect on the detection can be studied. Real time analysis for the

detection and extraction of life sign frequencies is one of the major interests of users.

The developed techniques may be explored to be applied for real time analysis.

6.1 Experimental Setup 77

Chapter 6 Effect of Thickness of wall on

detection of location and breathing frequency of

human being standing behind the brick wall

The purpose of this chapter is to see the effect of thickness of brick wall on detection of

life sign. Two experiments were carried out, one with 0.32 m thick brick wall and

second with 0.22 m thick brick wall.

Swath calculations

6.1 Experimental Setup

The experiment was carried out with bi-static antenna mode as shown in Fig.6.1 where

vector network analyzer, cables, and laptop were used. Standard two port calibration

processes i.e., Open Short Matched and Through was done to calibrate VNA before

transmission parameter S21 is measured in frequency range from 1 GHz to 3 GHz. Total

1024 numbers of scans or observations were carried out. Two identical pyramidal horn

antennas with 10 dB gain having bandwidth of 2 GHz were used for transmission and

reception. The antennas were mounted on tripod with height adjusted such that the

antennas are exactly aligned with the chest of human being with 50 cm distance

between two antennas and facing same direction. The antennas were oriented in vertical

polarization for data collection. The experiments were carried out with a brick wall

having thickness of 32cm. In performing the measurements, transmit and receive

antennas were kept at fixed locations and aligned for maximum signal reception. The

78 Effect of Thickness of wall on detection of location and breathing frequency

of human being standing behind the brick wall

distance between antennas should be sufficiently large such that the wall is in the far

field of antenna. During the experiments, the antennas were placed at different

distances from the brick wall. Data is collected by using software VEE pro controlled

by PC and VBA macro program and transferred to PC for further analysis. MATLAB

software use is for data processing.

6.2 Data collection

6.2.1 When thickness of brick wall is 0.32 m

Measurement was done for 0.32 m thick brick wall keeping distance between antennas

and brick wall as fixed whereas distance between wall and human target is varied.

Table 6.1 describes three different cases utilized for data collection.

In case 1, distance between Antenna and wall is 0.5m and distance between wall and

human target is varied from 0.5 m to 1.5m. In case 2, distance between Antenna and

wall is 1m and distance between wall and human target is varied from 0.5 m to 2 m

whereas in case 3, distance between Antenna and wall is 1.5m and distance between

wall and human target is varied from 0.5 m to 2m.

6.2 Data collection 79

Fig. 6.1 Experimental Setup with two antenna system

Table 6.1 Data Collection details for 0.32 m Brick wall taken with two antenna system

Sr. no. Distance between

Antenna and Wall (m)

Distance between Wall

and Target (m)

Total Distance including

Wall Thickness (m)

Case 1

0.5 0.5 1.32

0.5 1 1.82

0.5 1.5 2.32

0.5 2 2.82

Case 2

1 0.5 1.82

1 1 2.32

1 1.5 2.82

1 2 3.32

Case 3

1.5 0.5 2.32

1.5 1 2.8

1.5 1.5 3.32

1.5 2 3.82


Similar measurements for 0.22 m thick brick wall were carried out as done for 0.32 m

thick brick wall. Table 6.2 describes three different cases utilized for data collection.

Table 6.2. Data Collection details for 0.22 m Brick wall taken with two antenna system

Sr. no. Distance between

Antenna and Wall (m)

Distance between Wall

and Target (m)

Total Distance including

Wall Thickness (m)

Case 4 0.5

0.5 1.22

1 1.72

2 2.72

Case 5 1

0.5 1.72

1 2.22

1.5 2.72

Case 6 1.5

0.5 2.22

1 2.72

2 3.72

In case 4, distance between Antenna and wall is 0.5m and distance between wall and

target is varied from 0.5 m to 2m. In case 5, distance between Antenna and wall is 1m

and distance between wall and target is varied from 0.5 m to 1.5 m whereas in case 6,

distance between Antenna and wall is 1.5m and distance between wall and target is

varied from 0.5 m to 2m.

6.3 Signal processing technique

Processing steps given in the flow chart as shown in Fig. 6.2 are applied to extract

signal representing life sign. IFFT is used to convert data collected in frequency domain

to time domain. To locate the target and extract range information, time domain data is

converted into spatial domain using the relation between time, velocity and distance.

Once the signal is in spatial domain, the location of target is obtained. Thus the location

along with the amplitudes can be obtained from range profile. Aim is to detect the chest

movement of human being. Since single trace won’t give the location of chest

movement, all 1024 traces were used. The procedure is repeated to obtained amplitude

variation at location of human being (target). All the 1024 range profiles obtained are

stack one over other so that amplitude variation can be observed as given in step

4.Amplitude variation indicates presence of human target as obtained in step 5. Once

the location of amplitude variation is obtained, then the amplitude value at that location

is extracted. This is done for all the 1024 traces i.e., we formed the matrix of size

1x1024. Instead of taking amplitude of one location, averages of three locations i.e.

succeeding and preceding the target locations amplitude values are taken as given in


step 6. Due to chest movement, the location of amplitude may vary. The variation in

location depends on the range resolution which is in this case is 0.075 m. Now

according to step number 7, apply FFT algorithm to find frequency of breathing signal.


Step 2. Convert Frequency Domain Data into Time

Domain


Step 4.Stack all 1024 traces one above another to

observe amplitude variation

Step 5.Note the Location of amplitude variation as target

location

Step 6. Take the average of three amplitude values for all

1024 traces


Step 7.Apply FFT to find frequency

Fig. 6.2 Flow chart of Signal processing technique




6.4.1 External calibration

External calibration is carried out with the help of metal. The experimental setup

is as shown in Fig. 6.3 in which metal is kept at a distance of 1m from antenna.

After processing the data collected for external calibration, we observed the

results as shown in Fig. 6.4, in which the first peak is obtain because of reflection

due to antenna coupling at a distance of 0.45m and second peak is obtained at 1.2

m, due to reflection from metal sheet placed in front of antennas. Thus there is

need of correction of 0.2m.

Fig. 6.3 External calibration using Metal sheet


Fig. 6.4 Range profile for external calibration


6.4.2.1 Detection of position of different peaks with amplitude

We have processed the data according to the algorithm as described in section 6.2. Data

collections as given in Table 6.1 and 6.2 are used for processing. The processing of data

is carried out in steps i.e., case 1 to case 3 data is used for analysis and are described

below.

Case 1: When distance between Antenna and Brick-wall is 0.5m

The case 1 data as described in Table 6.1 is taken for analysis. The measurements were

taken when the distance between antennas and brick wall is at a fixed distance of 0.5m

and the distance of target from the wall is varied as 0.5 m, 1 m, 1.5 m, and 2 m. Figure

6.5 show the result obtained when the distance between target and wall is 0.5 m. From

Fig. 6.5, the first peak is due to reflection from antenna, second peak is due to reflection

from front side of brick wall, third peak is due to reflection from back side of brick wall

and the fourth is from target.

When the target is at distance of 0.5 m from wall the actual distance from antenna to

target is 1.32m and observed distance from Fig. 6.5 is 1.95m which is also shown in



Table 6.3. In Fig. 6.5 the first peak due to reflection from antenna is at position 0.45m

whose amplitude is of 0.2463. Second peak is due to reflection from front side of brick

wall is at position 0.75m whose amplitude is 1. The third smaller peak is due to

multipath and is at 1.2m whereas the back side of brick wall is observed at position

1.45m whose amplitude is 0.1187. Next peak is due to the target position at 1.95m

whose amplitude is 0.03959. Table 6.3 shows location and amplitude of various peaks,

in which sr. no. 1 shows the result when target is at the 0.5m from brick wall, sr. no. 2

shows the result when target is at 1m from brick wall, sr. no. 3 is when target is at 1.5m

from brick wall and sr. no. 4 is when target is at 2m from brick wall. It is observed from

the results obtained in Table 6.3 that as distance between antennas and target increases,

the amplitude at target reduces. Also there are other peaks apart from peaks due to

objects present in scene of measurements. The reasons for the presence of this peak

may be due to multipath.

Fig. 6.5 For case 1:Distance between wall and human target is 0.5m

Table 6.3 Position and amplitude when distance between Antenna and Brick-wall is 0.5m

Sr.

No.

Distance

between

First peak due

to antenna

Second peak

due To front side

Third peak due

to back side of

Fourth peak

due to target


Case 2: When distance between Antenna and Brick-wall is 1m

Measurements are taken when the antenna is at a fixed distance of 1m from wall and

the distance (position) of target is changed from the wall as 0.5m, 1m, 1.5m and 2m.

When the target is at distance of 0.5 m from wall the actual measurement from antenna

to target is 1.82 m and observed measurement from Fig. 6.6 is 2.475m and also shown

in Table 6.4. In Fig. 6.6, the first peak due to reflection from antenna is at position

0.45m whose amplitude is of 0.427m. Second significant peak due to reflection from

front side of brick wall is at position 1.2 m whose amplitude is 1. Third peak is due to

back side of brick wall is at position 1.4m whose amplitude is 0.04694. Fourth peak is

due to the target is at position 2.475m whose amplitude is 0.06259. Table 6.4 shows

position and amplitude of various peaks, for the remaining data. It is observed from the

comparison of results obtained from case 1 and case 2, that the number of reflections

increases due to multi paths when the distance between antennas and target increases.


Measurements are taken when antennas are at a fixed distance of 1.5 m from brick wall

and the position of target from the wall are changed from 0.5 m to 2 m in step of 0.5 m.

The results obtained after processing are tabulated in Table 6.5. In below Fig. 6.7, the

encircled peak indicates reflection due to target. The position and amplitude is given in

Table 6.5. Table 6.5 shows position and amplitude of various peak, in which sr. no. 1

shows the result when target is at the 0.5m from brick wall, sr. no. 2 shows the result

when target is at 1m from brick wall, sr. no. 3 is when target is at 1.5m from brick wall

and sr. no. 4 is when target is at 2m from brick wall. It is observed from results shown

in Table 6.5, that as the distance between antennas and target increases, detection of

target peak becomes difficult due to very low amplitude value.

wall and

human

target

of brick wall brick wall

Position Amp. Position Amp. Position Amp. Position Amp.

1. 0.5m 0.45m 0.2463 0.75m 1 1.425m 0.1187 1.95 m 0.03959

2. 1 m 0.45m 0.2468 0.75m 1 1.425m 0.1194 2.4 m 0.02

3. 1.5m 0.45m 0.247 0.75m 1 1.425m 0.1193 2.85 m 0.0147

4. 2 0.45m 0.2467 0.75m 1 1.425m 0.1189 3.45 m 0.00456



Table 6.4 When distance between Antenna and Brick-wall is 1m


Sr.

No.

Distance

between

wall and

human

target (m)

First peak due

to antenna

Second peak

due To front

side of brick

wall

Third peak due

to back side of

brick wall

Fourth peak

due to target


1. 0.5

0.45m 0.427 1.2m 1 1.875m 0.04694 2.475 m 0.06259

2. 1 0.45m 0.4196 1.2m 1 1.875m 0.04741 2.85 m 0.01715

3. 1.5 0.45m 0.4206 1.2m 1 1.875m 0.04664 3.675 m 0.01113

4. 2 0.45m 0.4212 1.2m 1 1.875m 0.04775 3.975 m 0.0099



Table 6.5 When distance between Antenna and Brick-wall is 1.5m

When the results obtained in Table 6.4 and 6.5 are compared for same distance between

Antenna and human target i.e., for example sr. no. 4 from Table 6.4 and sr. no. 3 from

Table 6.5, the distance between Antenna and human target is 3m but the amplitude

value at location of target is different. This reduction in amplitude is due to changes in

swath as the distance between antenna and wall changes. From Table 6.4, the distance

between Antenna and wall is 1m whereas in Table 6.5, the distance between them is

1.5m. As the swath increases, intensity reduces.

Sr.

No

.

Distanc

e

between

wall

and

human

target

(m)

First peak due to

antenna

Second peak

due To front

side of brick

wall

Third peak due to

back side of brick

wall

Fourth peak

due to target

Positio

n Amp.

Positio

n

Amp

.

Positio

n Amp.

Positio

n Amp.

1. 0.5 0.45m 0.00508

3 1.725m 1 2.1m

0.00075

4 2.925m 0.000117

2. 1 0.45m 0.5673 1.725m 1 2.4m 0.06354 3.30m 0.01865

3. 1.5 0.45m 0.5648 1.725m 1 2.4m 0.06227 3.825 0.02117

4. 2 0.45m 0.00505

7 1.725m 1 2.1m

0.00074

6 4.35m

0.000150

5



6.4.2.2 Detection of location of Human being

From flowchart given in Fig.6.2, after observing range profile the distance to target is

noted down as shown in Table 6.6. It is observed from the results that there is

difference between the actual distance between antennas and target and measured

distance at target. The difference is given in last column of Table 6.6. One of the

reasons for difference is due to change in velocity of wave when it propagates through

different medium other than air.

6.4.2.3 Detection of location of Human being after Velocity correction

For example, from Table 6.6, from case 1, if first row is considered, the distance

between antenna and wall is 0.5m and distance between wall and target is 0.5 m. The

total distance between antennas and target by considering wall thickness as 0.32 m will

be 1.32 m. But the observed result is 1.95 m. So the difference is of 0.63 m. Due to

external calibration there is correction of 0.2 m which gives the distance as 1.75m

instead of 1.95m. Hence the difference of 1.75-1.32=0.43 m shows the deviation of

actual location because of decrease in velocity of microwave when the wave propagates

through brick wall. This effect can be minimized by using velocity correction. The

corrected location of human target related to antenna can be calculated by equation

(6.1),

Distance after Correction = Observed distance D ( )1( (6.1)

Where, D is the thickness of brick wall which is 0.32m. Assume as 4.5.

Distance after Correction =1.75 - 0.32(*1.1213)

= 1.75 - (0.3588)

=1.39 m

The distance after correction matches with the actual measurement used for data

collection, thus the error is 0.0004m which falls in acceptable range as range resolution


is 0.075 m. For the remaining data using (6.1), the actual/measured location is

compared with velocity corrected location and difference (error) are calculated which

are shown in Table 6.6. The values of error are acceptable. The Table 6.6 also gives the

maximum values of standard deviation (SD) on the basis of which the location of

human target is obtained.

Table 6.6 Location obtained from Experimental data after velocity correction

Sr no. Brick

wall

thicknes

s

(m)

Distanc

e

betwee

n

Antenn

a and

wall

(m)

Distanc

e

betwee

n wall

and

target

(m)

Measur

ed

Total

distanc

e (m)

Distance

to target

observe

d from

result

(m)

Actual

Distance

after

correction

external and

velocity (m)

(0.4304+0.2

)=0.6304

Differenc

e

Between

Measure

d and

observed

distance

Maximum

Sigma

Value

observed

at human

target

location

Case 1

0.32 0.5 0.5 1.32 1.95 1.3196 0.0004 0.0039

0.32 0.5 1 1.82 2.4 1.7696 0.0504 0.0014

0.32 0.5 1.5 2.32 2.925 2.2946 0.0254 0.0011

0.32 0.5 2 2.82 3.45 2.8194 0.0004 0.00072

Case 2

0.32 1 0.5 1.82 2.475 1.8446 0.0246 0.0011

0.32 1 1 2.32 2.85 2.2196 0.1004 0.0013

0.32 1 1.5 2.82 3.45 2.8196 0.0004 0.0015

0.32 1 2 3.32 4.05 3.4196 0.0999 0.0007

Case 3

0.32 1.5 0.5 2.32 2.925 2.2946 0.0254 0.0036

0.32 1.5 1 2.82 3.485 2.816 0.004 0.0019

0.32 1.5 1.5 3.32 3.9 3.6446 0.054 0.0045

0.32 1.5 2 3.82 4.5 3.8696 0.0496 0.0048

6.4.2.4 Extraction of Amplitude variation from target location

For better understanding, the amplitude values for detected peak i.e., peak due to

antenna coupling, due to front side of wall, due to back side of wall and due to human

target for all 1024 traces were recorded. It is clearly observed that for the first three

peak i.e., peak due to antenna coupling, due to front side of wall, and due to back side

of wall there is no variation in amplitude for all 1024 traces. There is significant

amplitude variation at human target location. So in the next step only amplitude

variation portion is extracted.



Instead of extracting amplitude from single location, succeeding and preceding location

i.e., total three location amplitude value is taken and there average is calculated for all

1024 traces. This average value at all 1024 traces is taken and a plotted. The time

domain plots are obtained for all the cases i.e., case 1, case 2 and case 3 data sets. The

average amplitude variation plot at target location is shown in Fig. 6.8 for different

distances between wall and target for case 1 data set. Similarly for case 2 and case 3

data set, the average amplitude variations at target location is plotted in Fig. 6.9 and

Fig. 6.10 respectively.

a. Average when target is at 0.5m b. Average when target is at 1m

c. Average when target is at 1.5m d. Average when target is at 2m.


Fig. 6.8 Plot of amplitude variation for case 1 data

a. Average when target is at 0.5m b. Average when target is at 1m

c. Average when target is at 1.5m d. Average when target is at 2m


The plots shows that there are multiple frequencies present in the signal i.e., it includes

breathing signal, heartbeat signal and some clutter. It is also observed from the

amplitude variations that The reason of presence of different signals may be due to

presence of brick wall in between radar antennas and human target. Thus it makes

extracting of breathing frequency more complex.



a. Average when at 0.5m b. Average when target is at 1m

c. Average when target is at 1.5m d. Average when target is at 2m


6.4.2.5 Detect breathing frequency using FFT

After applying the FFT to the amplitude variations, the results are obtained and are

rightly shown in Fig. 6.11 for case 1 dataset, in Fig. 6.12 for case 2 dataset and Fig.

6.13 for case 3 data set. In frequency spectrum, the peak is detected and corresponding

breathing frequency (0.2-0.7 Hz) for human are noted as shown in Table 6.7. In this

experimental study heartbeat frequency (1.2-1.7 Hz) was not detected due to its


relatively weak amplitude variation. It is observed from results, that there are multiple

frequencies present which are due to harmonics and clutter.

a.when target is at 0.5m from wall b. when target is at 1m from wall

c. when target is at 1.5m from wall d. when target is at 2m from wall

Fig. 6.11 Breathing frequency extraction using FFT for case 1 dataset



a. when target is at 0.5m from wall b. when target is at 1m from wall

c. when target is at 1.5m from wall d. when target is at 2m from wall


a when target is at 0.5m from wall b when target is at 1m from wall


c when target is at 1.5m from wall d when target is at 2m from wall


Table 6.7 Breathing frequency values for different datasets

Case 1 Case 2 Case 3

0.4492 Hz 0.5078 Hz 0.4883 Hz

0.3516 Hz 0.3516 Hz 0.4492 Hz

0.4688 Hz 0.4883 Hz 0.6641 Hz

0.4492 Hz 0.4883 Hz 0.5273 Hz

6.4.3 When the thickness of brick wall is 0.22 m

We have processed the data which is collected by varying distance between antennas

and wall and distance between wall and human target. As per Table 6.2, we have

varied distance between antenna and wall from 0.5m to 1.5 m in step of 0.5m. These

various steps are taken as case 4 to case 6 for analysis and are described below. In case

4, the distance between wall and human target is varied and data is collected.

6.4.3.1 Detection of position of different peaks with amplitude


The data as described in case number 4 shown in Table 6.2 is taken for analysis. The

measurements were taken when the distance between antennas and brick wall is at a

fixed distance of 0.5m and the distance of target from the wall is varied as 0.5 m, 1 m,

and 2 m. Figure 6.14 show the normalized result obtained when the distance between

target and wall is 0.5 m. From Fig. 6.14, the first peak is due to reflection from

antennas coupling, second peak is due to reflection from front side of brick wall, third

peak is due to reflection from back side of brick wall and the fourth is due to human

target.

When the target is at distance of 0.5 m from wall the actual distance from antenna to

target is 1.22m and observed distance from Fig. 6.14 is 1.725m which is also shown in

Table 6.8. In Fig. 6.14, the first peak due to reflection from antenna coupling is at

position 0.45m whose amplitude is of 0.2749. Second peak is due to reflection from



front side of brick wall is at position 0.75m whose amplitude is 1. The third peak is due

to back side of brick wall is observed at position 1.425m whose amplitude is 0.1149.

Next peak is due to the target position at 1.75m whose amplitude is 0.1834.

Table 6.8 shows location and amplitude of various peaks, in which sr. no. 1 shows the

result when target is at the 0.5m from brick wall, sr. no. 2 shows the result when target

is at 1m from brick wall, sr. no. 3 is when target is at 2m from brick wall. It is observed

from the results obtained in Table 6.8 that as distance between antennas and target

increases, the amplitude at target reduces. Also there are other peaks apart from peaks

due to objects present in scene of measurements. The reasons for the presence of this

peak may be due to multipath.

Fig. 6.14 For case 4: Distance between wall and human target is 0.5m


Sr.

No.

Distance

between

wall and

human

target

First peak due

to antenna

Second peak

due To front

side of brick

wall

Third peak due

to back side of

brick wall

Fourth peak

due to target



Case 5: When distance between Antenna and Brick-wall is 1m

Measurements are taken when the antenna is at a fixed distance of 1m from wall and

the distance (position) of target is changed from the wall as 0.5m, 1m, and 1.5m. When

the target is at distance of 0.5 m from wall the actual measurement from antenna to

target is 1.72 m and observed measurement from Fig. 6.15 is 2.175m and also shown in

Table 6.9. In Fig. 6.15, the first peak due to reflection from antenna is at position 0.45m

whose amplitude is of 0.2815. Second significant peak due to reflection from front side

of brick wall is at position 1.275 m whose amplitude is 1. Third peak is due to back side

of brick wall is at position 1.725m whose amplitude is 0.1167. Fourth peak is due to the

target is at position 2.175m whose amplitude is. Table 6.9 shows position and

amplitude of various peak, in which sr. no. 1 shows the result when target is at the 0.5m

from brick wall, sr. no. 2 shows the result when target is at 1m from brick wall and sr.

no. 3 is when target is at 1.5m from brick wall.


1. 0.5m 0.45m 0.2749 0.75m 1 1.425m 0.1149 1.725 m 0.1834

2. 1 m 0.45m 0.2735 0.75m 1 1.425m 0.1147 2.175 m 0.1002

3. 2 0.45m 0.273 0.75m 1 1.425m 0.1138 3.3 m 0.03445



Table 6.9 Position and amplitude when distance between wall and target is 1m

Case 6. When distance between Antenna and Brick-wall is 1.5m

Measurements are taken when antennas are at a fixed distance of 1.5 m from brick wall

and the position of target from the wall are changed from 0.5 m to 1.5 m in step of 0.5

m. The results obtained after processing are tabulated in Table 6.10. In below Fig. 6.16,

the data tip at 2.7m indicates reflection due to human target. The positions and

amplitudes at different objects present in scene are given in Table 6.10.

Table 6.10 shows position and amplitude of various peaks, in which sr. no. 1 shows the

result when target is at the 0.5m from brick wall, sr. no. 2 shows the result when target

is at 1m from brick wall, and sr. no. 3 is when target is at 1.5m from brick wall. It is

observed from results shown in Table 6.10, that as the distance between antennas and

target increases, detection of target peak becomes difficult due to very low amplitude

value.

Sr.

No.

Distance

between

wall and

human

target

First peak due

to antenna

Second peak

due To front

side of brick

wall

Third peak due

to back side of

brick wall

Fourth peak

due to target


1. 0.5m 0.45m 0.2815 1.275 1 1.725 0.1167 2.175 0.1076

2. 1 m 0.45m 0.2818 1.275 1 1.725 0.117 2.85 0.03321

3. 1.5m 0.45m 0.2817 1.275 1 1.725 0.1165 3.3 0.03425




6.4.3.2 Detection of location of Human being

From flowchart given in Fig.6.2, after observing range profile the distance to target is

noted down as shown in Table 6.5. It is observed from the results that there is

difference between the actual distance between antennas and target and measured

distance at target. The difference is given in last column of Table 6.5. The reasons for

difference are due to change in velocity of wave when it propagates through different

medium other than air.

Sr.

No.

Distance

between

wall and

human

target

First peak due

to antenna

Second peak

due To front

side of brick

wall

Third peak due

to back side of

brick wall

Fourth peak

due to target


1. 0.5m 0.45m 0.4481 1.725 1 2.325 0.1085 2.7 0.0121

2. 1 m 0.45m 0.4473 1.725 1 2.325 0.1086 3.3 0.0616

3. 1.5m 0.45m 0.4478 1.725 1 2.325 0.1087 4.275 0.0261



6.4.3.3 Location of Human being after Velocity correction

For example, from Table 6.11, from case 4, if first row is considered, the distance

between antenna and wall is 0.5m and distance between wall and target is 0.5 m. The

total distance between antennas and target by considering wall thickness as 0.22 m will

be 1.22 m. But the observed result is 1.75 m. Due to external calibration there is

correction of 0.2 m which gives the distance as 1.55m instead of 1.75m. Hence the

difference of 1.55-1.22=0.33 m shows the deviation of actual location because of

decrease in velocity of microwave when the wave propagates through brick wall. This

effect can be minimized by using velocity correction. The corrected location of human

target related to antenna can be calculated by equation (6.1). For the remaining data

using eq. 6.1, the actual location is compared with corrected location and errors are

calculated which are shown in Table 6.11. The distance after correction almost matches

with the actual measurement, with error around 0.07m which falls in acceptable range

as range resolution is 0.075 m. Thus the values of error are acceptable. The Table 6.11

also gives the maximum values of standard deviation (SD) on the basis of which the

location of human target is obtained.

Table 6.11 Location obtained from Experimental data

Sr

no.

Brick

wall

thickn

ess

(m)

Distanc

e

betwee

n

Antenn

a and

wall

(m)

Distan

ce

betwe

en

wall

and

target

(m)

Measur

ed

Total

distanc

e (m)

Distance

to target

observed

from result

(m)

True

Distance

after

correction

external and

velocity (m)

(0.2959+0.2)

=0.4959

Differen

ce

Between

Measure

d and

observe

d

distance

Maximu

m SD

Value

observe

d at

human

target

location

Cas

e

4

0.22 0.5 0.5 1.22 1.725 1.229 0.0091 0.0101

0.22 0.5 1 1.72 2.175 1.6791 0.0409 0.0156

0.22 0.5 2 2.72 3.3 2.8041 0.0841 0.0039

Cas

e

5

0.22 1.0 0.5 1.72 2.25 1.7541 0.0341 0.0139

0.22 1.0 1.0 2.22 2.85 2.3541 0.1341 0.0118

0.22 1.0 1.5 2.72 3.3 2.8041 0.0841 0.0068

Ca

se

6 0.22 1.5 0.5 2.22 2.7 2.2041 0.1341 0.0159

0.22 1.5 1.0 2.72 3.3 2.8041 0.0841 0.0075


0.22 1.5 2.0 3.72 4.275 3.816 0.096 0.0035

6.4.3.4 Extraction of Amplitude variation from target location

For better understanding, the amplitude values for detected peak i.e., peak due to

antenna coupling, due to front side of wall, due to back side of wall and due to human

target for all 1024 traces were recorded. It is clearly observed that for the first three

peak i.e., peak due to antenna coupling, due to front side of wall, and due to back side

of wall there is no variation in amplitude for all 1024 traces. There is significant

amplitude variation at human target location. So in the next step only amplitude

variation portion is extracted.

Instead of extracting amplitude from single location, succeeding and preceding location

i.e., total three location amplitude value is taken and there average is calculated for all

1024 traces. This average value at all 1024 traces is taken and a plotted. The time

domain plots are obtained for all the cases i.e., case 4, case 5 and case 6 data sets. The

average amplitude variation plot at target location is shown in Fig. 6.17 for different

distances between wall and target for case 4 data set. Similarly for case 5 and case 6

data set, the average amplitude variations at target location is plotted in Fig. 6.18 and

Fig. 6.19 respectively.




c. when target is at 2m from wall


(a) when target is at 0.5m from wall (b) when target is at 1m from wall

(c ) when target is at 1.5m


Fig. 6.18 Amplitude extraction using SD for case 5 dataset


(c ) when target is at 2m

Fig. 6.19 Amplitude extraction using SD for case 6 dataset

6.4.3.5 Detect breathing frequency using FFT

After applying the FFT to the amplitude variations, the results are obtained and are

rightly shown in Fig. 6.20 for case 4 dataset, Fig. 6.21 for case 5 dataset and Fig. 6.22

for case 6 data set. In frequency spectrum, the peak is detected and corresponding

breathing frequency in the range of 0.2-0.6 Hz for human are noted as shown in Table

6.12. In this experimental study heartbeat frequency (1.2-1.7 Hz) was not detected due

to its relatively weak amplitude variation. It is observed from results, that there are

multiple frequencies present which are due to harmonics and clutter.


(c ) when target is at 1.5m


a. when target is at 0.5m from wall (b) when target is at 1m from wall





Table 6.12 Breathing frequency values for different datasets

Case 4 Case 5 Case 6

0.4102 0.5078 0.293

0.3125 0.4297 0.293

0.5469 0.4688 0.2734

6.5 Conclusion

This chapter demonstrates the extraction of human breathing frequency when the

human is standing behind brick wall of thickness 32 cm and 22 cm. Data were collected

by changing distance from antenna to wall from 0.5m to 2m. As well as distance

between wall and target was varied from 0.5m to 2m. It is observed from results, as the

distance between antennas and wall increases, detection of target peak becomes

difficult due to very low amplitude value. The reason may be due to the swath of

antenna, which indicates the reduction in power as distance increases. From this it is

concluded that the antenna beam width plays important role in deciding the detection

capability of the target.

From the range profiles, it is observed that there are other peaks apart from peaks due to


may be due to multipath. Also it is observed from the comparison of results obtained

from case 1 and case 2, that the number of reflections increases due to multi paths when

the distance between antenna and target increases.

The location of human target is obtained using standard deviation which gives slightly

different location when compared with actual distance kept during experiments. To

correct the location external calibration and velocity correction is carried out. In spite of

correction there is difference in actual and obtained location. The possible reasons

could be human error while standing as a human target and range resolution error.

6.5 Conclusion 107

The amplitude variation plots described in Section 6.4.2.3 and 6.4.3.4 shows that there

are multiple frequencies present in the signal i.e., it includes breathing signal, heartbeat

signal and some clutter i.e., plots are not sinusoidal. Thus it makes extraction of

breathing frequency more complex. Advanced signal processing technique can be

developed to improve the performance of extraction of heartbeat signal.

The experimental results from all observations show that FFT method has successfully

extracted the breathing frequency of human target. Apart from breathing frequency,

there are many other frequencies or harmonics present in the output whose source is

unknown.

7.1 Summary of contributions 109

Chapter 7 Conclusion

The main objective of this research project was to explore different ways to enhance the

performance of a human breathing detection system. To address areas in which

improvement is required is to increase the signal strength of human target by reducing

clutter due to stationary objects present in the scene of observations, automatic

detection of location of human target and extraction of life sign of human target with

reduced errors. There are number of different techniques which researchers have used

to solve these problems. These techniques were reviewed in Chapter 2.

7.1 Summary of contributions

Signal processing methods are used by human breathing detection radar system for

extraction of breathing frequency. Successful results are obtained using FFT and EMD-

HHT methods. Also due to increase in distance between antenna and human being, the

value of respiration frequency does not changed. The experimental results also show

that FFT method is useful for extraction of breathing frequency of human being after

using band pass filter. It is observed that when the distance between antenna and target

is increased, the amplitude of breathing signal reduces. Successful extraction of

breathing frequency is achieved when human target is standing behind wall of different

types and of different thickness i.e., 0.2 cm thick plywood, brick wall of thickness 10

cm, 22 cm and 32 cm. It is also observed that, as the human target is same for data

collected when plywood wall and brick wall is used, the respiration frequency values

remains same. It is also observed from results that as the distance between antennas

and wall increases, detection of target peak becomes difficult due to very low amplitude

value.

The proposed methods not only extract the breathing frequency, but also estimate the

location of human being using standard deviation (SD). The determination of higher SD

value helps in the automatic detection of presence of life sign signal of human being

along with location. The location of human target is obtained using SD which gives

slightly different location when compared with actual distance kept during experiments.

To correct the location external calibration and velocity correction is carried out. In

spite of correction there is difference in actual and obtained location. The possible

reasons could be human error while standing as a human target and range resolution

error.

Increase in target signal strength is useful for process of extraction of life sign signal.

The focus is to explore the possibility of improvement in the detection and extraction of

breathing frequency of human being positioned behind the brick wall. After applying

clutter reduction technique using SVD, clutters are successfully minimized which

implies that the technique is powerful. Target signal strength is improved further by

using all Singular components related to target. Due to use of clutter reduction target

signal strength increases which increases the probability of correct target detection.

7.2 Below is a list of the major contributions in this

research:

1. Reduced the clutter in human breathing detection which increased the dynamic range

and hence the distance between radar and human target.

2. Human target location is determined using SD which is used for automatic

determination of position of human target behind wall.

3. Breathing frequency of human target is extracted successfully in all the

measurements.

7.3 Suggestions for Future Work 111

7.3 Suggestions for Future Work

The detection of human target hidden behind wall was studied. The proposed human

detection algorithm reduced clutter to enhance target signal strength and extract

breathing frequency.

There are several different ways this research could be extended in the future. Some of

these are as follows:

Chapter 3 demonstrates the extraction of human breathing using two different signal

processing techniques i.e., FFT and HHT. Since EMD-HHT technique was used when

plywood was used as a wall, in future, need to apply EMD-HHT algorithm by using

brick wall. The another problem with HHT method is to determine which IMF should

be selected as many IMFs have a frequency spectrum in the range of the human

breathing frequency range. In future, advanced signal processing techniques will be

explored to extract heart beat frequency which is not detected.

In chapter 4, monostatic radar system is used to measure transmission parameter S21

with the help of circulator. The experimental results are improved compared to

measurement of reflection parameter S11 when the distance between radar and target is

increased. Since the wide bandwidth circulators are not available in the required

frequency range, desired results could not be achieved.

The effect of non stationary clutter i.e., other movements in nearby region of the

detection is not considered in this work. In future study, brick wall can be replaced by

different complex type of walls like concrete with metal inside can be considered and

its effect on the detection can be studied.

Real time analysis for the detection and extraction of life sign frequencies is one of the

major interests of users. The developed techniques may be explored to be applied for

real time analysis.

From the range profiles, it is observed that there are other peaks apart from peaks due to


may be due to multipath. Also it is observed from the comparison of results obtained

that the number of reflections increases due to multi paths when the distance between

antenna and target increases.

The experimental results from all observations show that FFT method has successfully

extracted the breathing frequency of human target. Apart from breathing frequency,

there are many other frequencies or harmonics present in the output whose source is

unknown. This present research could be extended by studying reasons behind the

presence of various frequencies.

The amplitude variation plots described in Section 6.4.2.3 and 6.4.3.4 are not

sinusoidal. Thus it makes extraction of breathing frequency more complex. Signal

processing technique can be developed to improve the performance of extraction of

breathing.

Although the proposed method for human breathing detection can detect single target, a

goal for the future will be to detect multiple targets behind a wall. The main problem in

multi-target detection is the existence of false targets generated by multipath for the

reflected waves.

We have successfully extracted breathing frequency, but heartbeat signal is not

detected. Advanced signal processing technique can be developed to extract heartbeat

signal. This research could be extended by studying the human breathing detection

under rubble.

113

References

1. Chen, K. M., Y. Huang, J. Zhang, and A. Norman, Microwave life-detection

systems for searching human subjects under earthquake rubble and behind

barrier", IEEE Transactions on Biomedical Engineering, Vol. 27, pp.105-114,

2000.

2. Jing Li,Lanbo Liu, Zhaofa Zeng, and Fengshan Liu, Advanced Signal

Processing for Vital Sign Extraction With Applications in UWB Radar

Detection of Trapped Victims in Complex Environments, IEEE Journal of

selected topics in Applied Earth Observations and Remote Sensing, vol.7,

pp.783 -791, 2013.

3. Li, C., J. Cummings, J. Lam, E. Graves, and W. Wu, Radar remote monitoring

of vital signs", IEEE Microwave Magazine, Vol. 10, pp. 47-56, 2009.

4. Y Xu, S Dai, S Wu, J Chen, and G Fang, Vital Sign Detection Method Based

on Multiple Higher Order Cumulant for Ultrawideband Radar, IEEE

Transactions Geoscience and Remote Sensing, vol.50, no.4. April 2012.

5. Y.Xu, S.Wu, C.Chen, J. Chen and G. Fang, A Novel Method for Automatic

Detection of Trapped Victims by Ultrawideband Radar, in IEEE Transactions

on Geoscience and Remote Sensing, vol. 50, no.8, pp.3132-3142, Aug.2012.

6. Y. Xu, J. Shao, J. Chen and G. Fang, "Automatic Detection of Multiple Trapped

Victims by Ultra-Wideband Radar," in IEEE Geoscience and Remote Sensing

Letters, vol. 10, no. 6, pp. 1498-1502, Nov. 2013.

7. A. Nezirovic, A. G. Yarovoy and L. P. Ligthart, "Signal Processing for

Improved Detection of Trapped Victims Using UWB Radar," in IEEE

Transactions on Geoscience and Remote Sensing, vol. 48, no. 4, pp. 2005-2014,

April 2010.

8. F. H. C. Tivive, A. Bouzerdoum and M. G. Amin, "A Subspace Projection

Approach for Wall Clutter Mitigation in Through-the-Wall Radar Imaging,"

in IEEE Transactions on Geoscience and Remote Sensing, vol. 53, no. 4, pp.

2108-2122, April 2015.

9. M. Mabrouk, S. Rajan, M. Bolic, I. Batkin, H. R. Dajani and V. Z. Groza,

"Detection of human targets behind the wall based on singular value

decomposition and skewness variations," 2014 IEEE Radar Conference,

Cincinnati, OH, 2014, pp. 1466-1470.

10. T. C. Chen, J. H. Liu, P. Y. Chao and P. C. Li, "Ultrawideband Synthetic

Aperture Radar for Respiratory Motion Detection," in IEEE Transactions on

Geoscience and Remote Sensing, vol. 53, no. 7, pp. 3749-3763, July 2015.

11. Johnson, J.T., Demir, M.A. and Majurec, N., 2009, Through-wall sensing with

multi-frequency microwave radiometry: a proof of concept demonstration,

IEEE Transactions on Geoscience and Remote Sensing, 47, pp. 1281–1288.

12. Arai,I, Survivor Search Radar System for Persons Trapped under Earthquake

Rubble, in Proceeding of the IEEE Microwave Conference, Vol. 2, pp. 663-

668, December 2001.

13. D.D. Ferris Jr. and N.C. Currie, A survey of current technologies for through-

the-wall surveillance (TWS),” In Proceedings of SPIE, vol. 3375, pp. 62-72,

1998.

14. D. D. Ferris Jr. and N. C. Currie, “Microwave and millimeter wave systems for

wall penetration, in Proceedings of SPIE, vol. 3375, pp. 269-279, 1998.

15. Lanbo Liu and Sixin Liu, Remote Detection of Human Vital Sign With

Stepped-frequency continuous wave radar, IEEE Journal of selected topics in

Applied Earth Observations and Remote Sensing, vol. 7, no. 3, March 2014.

16. P.K. Verma, A.N. Gaikwad, D. Singh, and M.J. Nigam, Analysis of clutter

reduction techniques for through wall imaging in UWB range, Progress In

Electromagnetics Research B, vol. 17, pp. 29-48, 2009.

17. A.N. Gaikwad, D. Singh and M.J. Nigam, Application of clutter reduction

techniques for detection of metallic and low dielectric target behind the brick

wall by stepped frequency continuous wave radar in ultra-wideband range,

IET Radar, Sonar and Navigation, vol. 5, pp. 416-425, 2011.

18. R. Chandra, A. N. Gaikwad, D. Singh and M. J. Nigam, An approach to

remove the clutter and detect the target for ultra-wideband through-wall

imaging, Journal of Geophysics and Engineering, vol. 5, pp. 412-419, 2008.

115

19. Gurnam S. Gill, High-Resolution Step-Frequency Radar, CRC Press

LLC,2001.

20. A. P. Freundorfer and K. Iizuka, "A study on the scattering of radio waves from

buried spherical targets using the step frequency radar," in IEEE Transactions

on Geoscience and Remote Sensing, vol. 31, no. 6, pp. 1253-1255, Nov 1993.

21. M. G. Amin and F. Ahmad, "Change Detection Analysis of Humans Moving

Behind Walls," IEEE Transactions on Aerospace and Electronic Systems, vol.

49, no. 3, pp. 1410-1425, July 2013.

22. Francesco Soldovieri, Ilaria Catapano, Lorenzo Crocco, Lesya N. Anishchenko,

A Feasibility Study for Life signs monitoring via a Continuous-Wave Radar,

International Journal of Antennas and Propagation, 2012.

23. Betul Yilmaz1, Sevket Demirci1, Enes Yigit, and Caner Ozdemir, An

Experimental Study of Through-the-Wall Radar for Life sign Detection,

PIERS Proceedings, pp. 1602-1604, Aug. 2013.

24. Kong Ling-jiang, Su Ting-ting, Cui Guo-long, Yang Jian-yu, Life detection

algorithm for stepped-frequency CW radar, Radar Conference IET

International, pp.1– 4, 20-22 April 2009.

25. D. P. Fairchild and R. M. Narayanan, Classification of human motions using

empirical mode decomposition of human micro-doppler signatures, IET

Radar, Sonar & Navigation, vol. 8, no.5, pp.425434, 2014.

26. Huang, N. E. et al., The empirical mode decomposition and the Hilbert

spectrum for nonlinear and non-stationary time series analysis, Proc. R. Soc.

Lond. A, 454 : pp. 903–995, 1998.

27. Ali Hussein Muqaibel and Ahmad Safaai-Jazi, A New Formulation for

Characterization of Materials Based on Measured Insertion Transfer Function,

IEEE Transactions on Microwave Theory and Techniques, vol. 51, No. 8,

August 2003.

28. M. D‟Urso, F. Gianota, R. Lalli, and L. Infante, Differential approach for

through-the-wall life signs detection, in Radar Conference, 2010 IEEE, 2010,

pp. 1079–1082.

29. M. Loschonsky, C. Feige, O. Rogall, S. Fisun, and L. M. Reindl, Detection

technology for trapped and buried people, in Wireless Sensing, Local

http://ieeexplore.ieee.org/xpl/mostRecentIssue.jsp?punumber=5353247



Positioning, and RFID, 2009. IMWS 2009. IEEE MTT-S International

Microwave Workshop on, , pp. 1–6, 2009.

30. F.-K. Wang, T.-S. Horng, K.-C. Peng, J.-K. Jau, J.-Y. Li, and C.-C. Chen,

Detection of Concealed Individuals Based on Their Vital Signs by Using a

See-Through-Wall Imaging System With a Self-Injection-Locked Radar,

IEEE Trans. Microw. Theory Tech., vol. 61, no. 1, pp. 696–704, Jan. 2013.

31. Y. Wang, Q. Liu, and A. E. Fathy, CW and Pulse, "Doppler Radar Processing

Based on FPGA for Human Sensing Applications, IEEE Trans. Geosci.

Remote Sens., pp. 1–11, 2012.

32. M. Ascione, A. Buonanno, M. D‟Urso, L. Angrisani, and R. Schiano Lo

Moriello, A New Measurement Method Based on Music Algorithm for

Through-the-Wall Detection of Life Signs, IEEE Trans. Instrum. Meas., vol.

62, no. 1, pp. 13–26, Jan. 2013.

33. G Rai Wang, H Gul Han, S Yul Kim and T Wook Kim, Wireless Vital Sign

monitoring using penetrating impulses, IEEE Microwave and Wireless

componentts letters, doi 10.1109/LMWC.2016.2630001.

34. Chi-Wei Wu and Zi-Yu Huang, Using the Phase Change of a Reflected

Microwave to Detect a Human Subject Behind a Barrier, IEEE Trans. Biomed.

Eng., vol. 55, no. 1, pp. 267–272, Jan. 2008.

35. R. M. Narayanan, Through-wall radar imaging using UWB noise waveforms,

J. Frankl. Inst., vol. 345, no. 6, pp. 659–678, 2008.

36. J. Li, Z. Zeng, J. Sun, and F. Liu, Through-Wall Detection of Human Being‟s

Movement by UWB Radar, IEEE Geosci. Remote Sens. Lett., vol. 9, no. 6, pp.

1079–1083, Nov. 2012.

37. L. Liu, Z. Liu, and B. E. Barrowes, Through-Wall Bio-Radiolocation With

UWB Impulse Radar: Observation, Simulation and Signal Extraction, IEEE J.

Sel. Top. Appl. Earth Obs. Remote Sens., vol. 4, no. 4, pp. 791–798, Dec. 2011.

38. Y. Wang, Q. Liu, and A. E. Fathy, Simultaneous localization and respiration

detection of multiple people using low cost UWB biometric pulse Doppler radar

sensor, in Microwave Symposium Digest (MTT), 2012 IEEE MTT-S

International, 2012, pp. 1–3.

39. S. Singh, Q. Liang, D. Chen, and L. Sheng, Sense through wall human

detection using UWB radar,EURASIP J. Wirel. Commun. Netw., vol. 2011, no.

1, p. 20, 2011.

117

40. Y. Zhou, L. Kong, G. Cui and J. Yang, "Remote Sensing of Human Body by

Stepped-Frequency Continuous-Wave," 2009 3rd International Conference on

Bioinformatics and Biomedical Engineering, Beijing, pp. 1-4, 2009.

41. A. Sume et al., Radar Detection of Moving Targets Behind Corners, IEEE

Trans. Geosci. Remote Sens., vol. 49, no. 6, pp. 2259–2267, Jun. 2011.

42. Zhu Zhang, Xiao Zhang, Hao Lv, Guohua Lu, Xijing Jing, and Jianqi Wang,

Human-Target Detection and Surrounding Structure Estimation Under a

Simulated Rubble via UWB Radar, IEEE Geosci. Remote Sens. Lett., vol. 10,

no. 2, pp. 328–331, Mar. 2013.

43. L Ren, Y Seo Koo, H Wang, Y Wang, Q Liu,and A E. Fathy, Noncontact

Multiple Heartbeats Detection and Subject Localization Using UWB Impulse

Doppler Radar, IEEE Microwave And Wireless Components Letters, Vol. 25,

no. 10, pp. 690692, Oct. 2015.

44. P.-H. Chen, M. C. Shastry, C.-P. Lai, and R. M. Narayanan, A Portable Real-

Time Digital Noise Radar System for Through-the-Wall Imaging, IEEE Trans.

Geosci. Remote Sens., vol. 50, no. 10, pp. 4123–4134, Oct. 2012.

45. X. Li, D. Qiao, Y. Li, and H. Dai, A novel through-wall respiration detection

algorithm using uwb radar, in Engineering in Medicine and Biology Society

(EMBC), 2013 35th Annual International Conference of the IEEE, 2013, pp.

1013–1016.

46. Z. Li, W. Li, H. Lv, Y. Zhang, X. Jing, and J. Wang, A Novel Method for

Respiration-Like Clutter Cancellation in Life Detection by Dual-Frequency IR-

UWB Radar, IEEE Trans. Microw. Theory Tech., vol. 61, no. 5, pp. 2086–

2092, May 2013.

47. J Li, C Liu, Z Zeng, and L Chen, GPR Signal Denoising and Target Extraction

With the CEEMD Method, IEEE Geoscinece and Remote Sensing Letters, vol.

12, no. 8, pp. 16151619, Aug. 2015.

48. Y. Zhang et al., A detecting and identifying method for two stationary human

targets using single-channel ultra-wideband radar, EURASIP J. Adv. Signal

Process., vol. 2012, no. 1, p. 202, 2012.

49. Y. Jia, L. Kong, X. Yang and K. Wang, Through-wall-radar Localization for

Stationary Human Based on Life-sign Detection, 2013 IEEE Radar

Conference, Ottawa, ON, 2013, pp. 1-4.

50. M. Ascione, A. Buonanno, M. D‟urso, L. Angrisani, and R. S. L. Moriello,

Distributed sensing in homeland security applications, in Measurements and

Networking Proceedings (M&N), 2011 IEEE International Workshop on, 2011,

pp. 16–21.

Appendix A :Authors Contributions

A.1 Conference Publications

The details of paper presented in conferences are given below.

1. Improvement in detection of human life sign hidden behind the wall using

clutter reduction technique, IEEE International Conference on Emerging trends

in communication technologies, 18-19 Nov. 2016, Graphic Era University,

Dehradun, Uttarakhand India.

2. "Experimental study and analysis of stepped-frequency continuous wave based

radar for through the wall detection of life signs," 2016 IEEE Region 10

Conference (TENCON), Singapore, 22-25 Nov., 2016, pp. 1565-1569.

3. Extraction of Life Sign of Human Being Hidden behind the Wall, 11 th

International Conference on Industrial &Information systems, 03-04 Dec. 2016,

RailTel, IITR centre of Excellence in Telecom. Department of Electronics and

Communication, IIT Roorkee.

A.2 Journal Contributions

1. “Extraction of breathing frequency of human being hidden behind the wall

using different signal processing techniques,” JETIR,Vol. 5, (7), pp.223230,

July 2018.

2. “Improved Detection and Extraction of Human Life Sign

Hidden Behind the Brick Wall Using Stepped-Frequency

A-2 :Authors Contributions

Continuous Wave Radar,” JETIR, Volume 5, Issue 9, pp.471480, September

2018.

Appendix B :Photographs

Photographs taken during experimentations of equipments, components, constructed

brick wall, during measurements.

Handheld Vector Network Analyzer with components

B-2 :Photographs

Constructed Brick walls with thickness 0.22m and 0.32m

Measurements with 0.1 m thick brick wall

study of human life sign detection under debris or behind ... of human... · detection of human...

Documents