Portable Imaging UWB Radar System with Two-Element Receiving Array Anatoliy O. Boryssenko* UMass, Amherst, MA, USA (
[email protected])
Dmitriy L. Sostanovsky Ratio Company, Kiev, Ukraine, (
[email protected])
Elena S. Boryssenko A&E Partnership, Amherst, MA, USA (
[email protected])
1
Talk Highlights UWB Radar Trough-Wall Vision with One & Two Receiving Antennas Advanced UWB Antenna Element
Matlab for Radar Control and Processing Radar Scene Imaging in 1D & 2D in Matlab
2D Scene Imaging with UWB Array (Time-Domain Focusing) 2
1-Rx Radar Trough-Wall Vision Radar - 1 Operational setup Tx & Rx blocks
Control unit
3
1-Rx Radar Trough-Wall Vision Radar - 2 Some Specs Tx PRR 16 kHz Peak-to-Peak Voltage 30 V Equivalent Bandwidth 0.5-1.5 GHz (@-10dB) Rx/Processor Stroboscopic – “GHz→kHz” 20 scans per second (software) 256/512/1024 samples per scan (software) 1-64 Stacks (software) 0-48 dB Time-varying gain (software) Background removing (software) Signal post-processing (1-D & 2-D filters in time- and frequency domain) Antenna: Bowtie : Cavity-Backed R-Loaded
4
A-Scan Presentation of the Received Data Signal: without background compensation
Operational scene Tx-A & Rx-A Direct Coupling
Notebook: -Control -Processing -Imaging
Basic UWBR Hardware
Signal: with background compensation
Tx, Rx & Antennas On Tripod
5
B-Scan Presentation of the Received Data
Operational scene
Scene radar image
30 cm wide reinforced concrete
Range up to 7m
6
Directions of Improvements:: Advanced Antenna Elements & Receiving Arrays New Antenna
Two Receivers
7
2-Rx 1-Rx Trough-Wall Vision UWB Radar - 1
UWBR Hardware
Windows compatible sound card for DAC & Serial (RS231) port for control Stereoacoustic = Stereo-Imaging 8
2-Rx 1-Rx Trough-Wall Vision UWB Radar - 2 Some Specs Tx PRR 1-10 MHz Peak-to-Peak Voltage 30-300 V Equivalent Bandwidth 0.35-3.0 GHz @-10dB Rx/Processor Stroboscopic – “GHz→kHz” 100 scans per second (software) 256/512/1024 samples per scan (software) Arbitrary stacking (software) 0-48 dB Time-varying gain 0-48 dB Time-varying gain (software) Background removing (software) Signal post-processing (1-D & 2-D filters in time- and frequency domain) Antenna: Novel Tongue Element in 1-Tx&2-Rx array (No resistive damping!)
9
2-Rx 1-Rx Trough-Wall Vision UWB Radar - 3
Control PCB
Backside view
Aperture side view
Right Rx PCB
Tx PCB
Left Rx PCB 10
Novel UWB Antenna:: Geometrical MoM Model & Simulated Input Impedance
Nice 50-Ω impedance & 5-6 dB gain bandwidth ! Direct termination to 50-Ω coax – no balun!
11
Novel UWB Antenna:: Simulated Radiation Pattern in 0.50-3.00 GHz
0.50 1.00 1.5 GHz
2.00 2.50 3.00 GHz
E-plane
Good gain and front-to-back ratio ! H-plane
12
Radar Control, Signal Processing and Data Imaging Using Matlab 2 Byte Command Word (through serial port)
•
CONTROL.COM_PORT=1; % 1 / 2 CONTROL.Start=1; %1 CONTROL.GM=4; % Maximum gain factor for AGC 0...7 CONTROL.G0=0; % Initial gain factor 0...7 CONTROL.RESERVE_HIGH=1; …
• • •
Controlling of ADC for Sound Card through Matlab ADC.SamplePerScan=700; %722; ADC.ScanShift=-100/2; ADC.NumberOfScans=10; ADC.SampleRate=44100; ADC.BitsPerSample=16; ….
Radar Hardware Control Start / Stop Range / Range Window Time-Varying Gain, etc.
Data Acquisition Scattered Waveforms are downconverted from UHF/Microwave 0.5-3.0 GHz to Audio Frequencies ~ 0.5-3.0 kHz that can be digitized using a standard compute audio card (very chip and good ADC!)
Processing Specifications for Signal Processing in Matlab
Signal Processing & Data Imaging
PROCESS.AmplitudeNorm=1; PROCESS.ScanStack=1; PROCESS.BckgRemove='NO'; %'YES'/'NO' PROCESS.SkipScans=50; …
A lot of such options in Matlab but a number of smart tricks is required to perform this in nearly “real-time” mode
Image Plot Options …
1-D A-Scan (oscilloscope mode) 2-D B-Scan (wavelets, dense,..) 2-D Stereoscopic View
For this prototype all control/processing/imaging software is written in Matlab !
13
1-D (A-Scan) Imaging = Oscilloscope Mode Background Removing Mode All is quite
Regular Mode A target is invisible
?
A target is moving & becomes visible
Target Tx-A & Rx-A Direct Coupling
14
2-D (B-Scan) Imaging = Stacked Waveforms
Regular Mode A target is stationary
Background Removing Mode A target is moving
Imaging Technique similar to those Used for Ground Penetrating Radars (GPR)
15
2D Scene Imaging in Matlab: “Stereoscopic” View – Time-Domain Focusing Ranging and Angular Tracking
Image Pixel Map
A Naive Time-Domain Focusing Technique: 1.
Equalize time delay to focus on a image “pixel”
2.
Compute inter-channel windowed correlation
s1( 2 ), ij (t ) = S 1( 2 )(t − D 0 c − Di , j 1(2) c )
I i , j = ∫ W (t ) ⋅ s1,ij (t ) ⋅ s2,ij (t )dt TW
Translate ToA into DoA
16
2D Scene Stereoscopic Imaging - 1: Why is the Target Mark so Wide Spread? Rx-2
Rx-1
Because of a small base, viz. the distance between Tx & Rx, the signals in the Rx channels are slightly mutually shifted in time that causes the wide spread mark. 17
2D Scene Stereoscopic Imaging - 2: Why is the Target Mark so Wide Spread?
No Target
With Target
Target mark resolved in range
Background Removing Mode 18
2D Scene Stereoscopic Imaging: How it woks? Simulated
Measured
Target Mark
The target mark is widespread in azimuth = More sophisticated time-domain beam-forming is required ! E.g., a similar problem is solved in stereoacoustic But Matlab cannot handle this in real time 19
2-Rx Radar Trough-Wall Vision Radar: GUI-Based Presentation for 2 Channels W/o Background Removing
With Background Removing
Target Position Area 20
2-Rx Radar Trough-Wall Vision Radar: 2D GUI-Based Scene Imaging
Background Removing #1
Background Removing #2
Target Position Area 21
Directions of Potential Future Efforts
More Elements in UWB Arrays Two-Plane & Dual-Polarization Image Focusing More Sophisticated Imaging Techniques Fast DSP Signal Processing Fast ADC instead of Sow Sound Card Antenna Adaptations to Media Interfaces Tracking and Recognition of Complex Scenes
22