multi-sensor measurements for the detection of buried targetslcarin/scott8.3.05.pdf2005/08/03 ·...

Multi-Sensor Measurements for the Detection of Buried Targets

Waymond R. Scott, Jr. and James McClellanSchool of Electrical and Computer Engineering

Georgia Institute of TechnologyAtlanta, GA 30332-0250

[email protected]

MURI Review 8-3-05 Scott/McClellan, Georgia Tech 2

Outline

IntroductionThree Sensor ExperimentMulti-static RadarSeismic ArrayAccomplishments/Plans


Multi-Sensor Cooperation/Adaptation

GPR

EMI

Seismic

Imaging

SigProc

Imaging

Features

Features

Features

DecisionProcess

ExploitCorrelation & Sensitivity

Feedback

Feedback

Controls

Controls

Controls

Controls

Controls

ControlsFrequency BandwidthFocusingRemote ImagingMeasurement Spacing

Frequency BandwidthMeasurement Spacing

Frequency BandwidthArray Elements UsedMeasurement Spacing


Experimental Test BedDevelop a set of experiments to investigate the potential of multi-modal processing

Landmine (Three sensors/six modes)EMIGPR

• Bi-Static• Multi-Static

Seismic• Unfocused• Focused• Remote Detection

Buried Structure (Two sensors/four modes)GPR

• Bi-Static• Multi-Static

Seismic• Bi-Static• Multi-Static


Outline

IntroductionThree-Sensor ExperimentMulti-static RadarSeismic ArrayAccomplishments/Plans


Diagram of the Laboratory Model


Three Sensor Experiment


Three Sensor Experiment

Experimental Scenario #16 Mines> 20 Clutter objectsRelatively uniform distribution

Experimental Scenario #27 Mines> 25 Clutter objectsNon-uniform distribution

Experimental Scenario #3Non-uniform distribution of 7 Mines (2 AT, 5 AP) and 57 clutter objectsRock field surrounding AP mine and canAP mines and clutter objects grouped around and on top of AT mines


Burial Scenario #1

1.8m by 1.8m Scan Region

Rocks (3 and4 cm deep)

Dry Sand(5cm deep)

MINESVS-2.2

(7cm deep)

TS-50(1.5cm deep)

w/ Nail

M-14(0.5cm deep)

VS-50(1cm deep)

PFM-1(1.5 cm deep)

VS-1.6(6.5cm deep)

SeismicSources

Cans (3 and2.5 cm deep)

AssortedMetal Clutter (2 to 4 cm deep)

Shells(4cm deep)

ThreadedRod(3.5cm deep)

Penny(5.5cm deep)

Nails(4cm deep)

Ball Bearing(3.5cm deep)

Shells(5.5cm deep)


Comparison of ImagesSeismic Sensor

Image 30 dB Scale

GPR Sensor Energy Image of Migrated

Data (25 dB scale)

EMI SensorImage ( 90 dB scale)


Outline



Resistive-Vee AntennaWhy resistive-vee antenna?

Performs well in bistatic GPRsRadiates clean, short pulsesHas a low radar cross section

Reduces multiple reflections between the antenna and the ground

Geometry suitable in array applications

y

z

2 =

11.

43cm

A 2=

16.3

3cm

A p

w = 3mm

2 =

1.5

85m

ma

y ae1 = bz

y2 0 1 0 = + ( )c c z-z

y c c z zc z z

2 0 1 0

2 0

= + ( - ) + ( - )2

z = L0 0.2

L = 17.15cm

feed line(coplanar stripline)

Antenna designOptimized resistive loading profileOptimized arm shapeFrequency: up to 8GHz

Optimized Resistive Loading ProfileOptimized Arm Shape


Resistive-Vee Antenna Implementation

Diagram of Measurement Setup

Comparison of Reflections

Photograph of Old Antenna

Photograph of New Antenna


Multistatic GPR SystemComponents

Antenna array (2 Tx’s and 4 Rx’s)Placed in a dielectric frameProvides 8 bistatic apertures – aperture sizes are 12cm to 96cm with 12cm increment

Vector network analyzer (VNA)Provides a source & two samplers

Microwave switchesActivates/deactivates the antennas

3D positionerControl computer (not shown)Microwave absorber

Behind the array frameLowers the reflections from the positioner

Between the antennasLowers the reflection from the frame

Around feed cablesSuppresses radiation due to common mode currentHides electric conductors

VNA

Antenna array

Microwave absorber

3D positioner

Wet sand

Microwave switches

Picture of Multi-Static GPR

48cm 48cm

TTRRRR


Multistatic GPR SystemVNA & antenna connection

VNA has one source and two samplers (A & B)T1 P1 T2 P2R1R2R3R4

VNA & switch operationResponse from 8 bistatic apertures are obtained by the following procedure:

1. Initialize all switches2. Obtain bistatic responses from T1-R1 and T1-R33. Throw internal switch4. Obtain bistatic responses from T2-R1 and T2-R35. Throw internal switch and external switches6. Obtain bistatic responses from T1-R2 and T1-R47. Throw internal switch8. Obtain bistatic responses from T2-R2 and T2-R4

Ext. Switch A sampler A

Ext. Switch B sampler B

Int. Switch source

Diagram of Multi-Static GPR


Multistatic GPR SystemRaw bistatic responses

VNA sweeps 401 frequencies from 60MHz to 8.06GHz (20MHz increment)Control computer stores the raw responses in the frequency domainThe raw responses are calibrated after the measurement

Calibration

Free space response (FREE)Free space is simulated by pointing the array toward absorber-padded cornerSubtraction removes direct coupling in the system

Through response (THRU)Antenna ports are connected by a 5ft cableDivision removes distortion in the cables feeding the antenna

ij

ijjiji THRU

FREERTRT

−=

Picture of FREE Measurement

Picture of THRU Measurement


Multistatic GPR SystemData acquisition

The array is scanned over 1.8m x 1.8m scan regionThe scan region is gridded into 91 x 91 points (∆x = ∆y = 2cm)

Per pointGPR obtains 8 bistatic responses at 401 frequency pointsInternal switch throw: 4 timesExternal switch throw: twiceTime: approximately 6 seconds

Per experimentData size: 8 (pairs) x 91 x 91 (points) x 401 (freq) x 8 (complex single) = 208 MBInternal switch throw: 33 124 timesExternal switch throw: 16 562 timesTime: approximately 14 hours

Picture of Scan Region

xy


3D ExperimentsFree Space (calibration)

Spheres, Mines, Shapes, and No targetBuried targets (with clean surface and with surface covered by rocks)

No TargetsGrid of spheres (calibration)Grid of mines and clutterBuried structure


Free Space Target Measurements -Setup

36.5cm

73 cm

xy


Free Space Target Measurements -Targets

GT Plywood (38.5 x 46.5 x 1.84 cm)

11cm Shotput

1” Metal Sphere

TS-50

VS 1.6


Multistatic Experiment in Free Space

Targets in Free Space A target is placed on top of 36.5cm high Styrofoam supportAntennas are elevated by 73cm from the ground

Radiation center is 90.8cm high from the groundGPR can obtain 6nsec of free space response before it sees the ground

Diagram of Experiment Setup



-90 -45 0y, cm

45 90 -90 -45 0y, cm

45 90 -90 -45 0y, cm

45 90 -90 -45 0y, cm

45 90

-90 -45 0y, cm

45 90 -90 -45 0y, cm

45 90 -90 -45 0y, cm

45 90 -90 -45 0y, cm

45 90

0

2

4

6

8

10

t, ns

ec

0

2

4

6

8

10

t, ns

ec

T1 R1 (12cm) T1 R2 (24cm) T1 R3 (36cm) T1 R4 (48cm)


-60 dB

-80

-70

-90

-100

-60 dB

-80

-70

-90

-100

Bistatic Responses of 11.4cm (4.5”) Diameter Metal Sphere



ω – k migrationTarget: 11.4 cm (4.5”) diameter metal sphereApplied only to T1-R1 (12cm) response(Tx-Rx spacing is ignored)

Raw Response Migrated Image

Picture of Target

0 dB

-10

-5

-15

-25

0 dB

-10

-5

-15

-25

-20-20

2

4

3

5

7

6

t, ns

ec

2

4

3

5

7

6t,

nsec

-70 0-35 35 70y, cm

-70 0-35 35 70y, cm



-90 -45 0y, cm

45 90 -90 -45 0y, cm

45 90 -90 -45 0y, cm

45 90 -90 -45 0y, cm

45 90

-90 -45 0y, cm

45 90 -90 -45 0y, cm

45 90 -90 -45 0y, cm

45 90 -90 -45 0y, cm

45 90

0

2

4

6

8

10

t, ns

ec

0

2

4

6

8

10

t, ns

ec



-60 dB

-80

-70

-90

-100

-60 dB

-80

-70

-90

-100

Bistatic Responses of GT plywood



ω – k migrationTarget: 38.5cm x 46.5cm x 1.8cm GT shape plywoodApplied only to T1-R1 (12cm) response(Tx-Rx spacing is ignored)

Raw Response Migrated Image

Picture of Target

0 dB

-10

-5

-15

-20

0 dB

-10

-5

-15

-20


Multistatic GPR ExperimentTargets in sand

Targets are buried 1cm – 13cm deep when measured from the surface of the sand to the top of the targetAntennas are elevated by 10cm from the sandIn cluttered-surface experiment:

Rocks are scattered over the scan regionRock density is empirically chosen to maximize the clutter effect

Sand with Clean Surface Sand with Cluttered Surface


Multistatic GPR ExperimentBuried targets

Anti-tank mines: VS-1.6, VS-2.2, and TMA-5Anti-personnel mines: TS-50, PFM-1, M-14, and mine simulant Clutter objects: metal sphere, rock, crushed can, and nylon cylinder

Burial MapPicture of Targets


Bistatic Responses of Landmine Targets at x = 0cm

Multistatic GPR Experiment

-90 -45 0y, cm

45 90 -90 -45 0y, cm

45 90 -90 -45 0y, cm

45 90 -90 -45 0y, cm

45 90

-90 -45 0y, cm

45 90 -90 -45 0y, cm

45 90 -90 -45 0y, cm

45 90 -90 -45 0y, cm

45 90

0

2

4

6

8

10

t, ns

ec

0

2

4

6

8

10

t, ns

ec

Clean-Surface Experiment

Cluttered-Surface Experiment



-60 dB

-80

-70

-90

-100

-60 dB

-80

-70

-90

-100

simulant

VS-2.2VS-1.6

PFM-1


Landmine Scenario


Landmine Scenario: Clean Surface

One Pair of antennas: T1R1 Four Pairs of antennas: T1R1, T1R2, T1R3, T1R4

Migrated Images: Time Domain


Multistatic GPR ExperimentEnergy in Migrated Images Summed Over Depth: Time Domain

One Pair of antennas: T1R1

Four Pairs of antennas: T1R1, T1R2, T1R3, T1R4

One Pair of antennas: T1R1

Four Pairs of antennas: T1R1, T1R2, T1R3, T1R4

Shallow Depths Deep Depths


Landmine Scenario: Rock Covered Surface




Buried StructureTarget Location & Depth


x = 0cm-cut

Buried Structure

-90 -45 0y, cm

45 90 -90 -45 0y, cm

45 90 -90 -45 0y, cm

45 90 -90 -45 0y, cm

45 90

-90 -45 0y, cm

45 90 -90 -45 0y, cm

45 90 -90 -45 0y, cm

45 90 -90 -45 0y, cm

45 90

0

2

4

6

8

10

t, ns

ec

0

2

4

6

8

10

t, ns

ec

R1 T1 R3 T1 R2 T2 R4 T2

R1 T1 R3 T1 R2 T2 R4 T2

0 dB

-20

-10

-30

-40

0 dB

-20

-10

-30

-40

rel. max.

rel. max.

Clean Scan

With-Rock Scan


Buried Structure


Buried Structure: Clean Surface




Buried Structure: Rock Covered Surface




Multistatic Inversion Issues

Back-Projected Images of 1” Metal Sphere

Input pulse: Diff. Gaussian; f0 = 2.5GHz

T1-R1 T2-R4

Responses of 4.5” Metal Sphere

T1-R1 T2-R4

Inversion Issues

Shape of response curvesThe response curves reduce to a hyperbola when

a = 0 (monostatic) orh >> a (far-field)

In this work, either a is significant or a > hBack-projected images

Images obtained by the typical delay-sum back-projection algorithm do not line up in depthSimple summation of back-projected images may not work

Complicated responses from a single target

Responses depending on the target shape, aspect angle, the illuminating waveform, etc.May be useful in identifying the target

Strong ground reflection needs to be removed before inversion


Outline



Ground Contacting Seismic Sensor Array Deployed on a Small Robotic Platform

Source Platform

Sensor Platforms


Ground Contacting Seismic Sensor Array

Allowing contact with the ground makes is possible to construct robust and low-cost sensors arrays

Developed such a sensor using COTS componentsSensor weight (6 oz.) is much less than typical landmine detonation weights (more than 10 lbs. for AP landmines, much greater for ATlandmines)Array for this work

3 lines of 10 sensors1.35 inches between sensors in each line (fixed by sensor configuration)Variable spacing between lines

• Currently 4 inches (10 cm)• Maximum of 11 inches (29 cm)

Experiment shown with VS-1.6 AT Landmine Buried 5cm Deep


ADXL103 Accelerometer

ICP Preamplifier

Pipe Insulation Foam Spring

Sorbothane Sensor Foot

19mm O.D. X 46 cm Long Aluminum Tube

SMB Connector


Experimental Setup of Seismic Landmine Detection Measurements

3 by 10 Element Array of Ground-

Contacting Sensors

Seismic Source

VS-1.6 AT Landmine (5cm deep)

50 Tons of Damp, Compacted Sand

Scan Region of 2m by 2m

45 cm

174 cm


Seismic Detection of VS-1.6 AT Landmine Buried 5cm Deep in Experimental Model

Landmine (22.2 cm diameter) buried 110 cm from first measurement location in 171 cm linear scan.

5 measurements with center line of sensor array along a line over the burial location.

Compressional, Rayleigh Surface, and ReflectedWaves.

Resonance of Landmine-Soil System.

Interactions of incident waves with buried landmine evident in measured data.


AccomplishmentsDeveloped three sensor experiment to study multimodal processing

Developed new metal detector and a radarInvestigated three burial scenariosShowed responses for all the sensors over a variety of targetsDemonstrated possible feature for multimodal/cooperative processingDeveloped new 3D quadtree strategy for GPR data

Developed seismic experiments, models, and processing Developed optimal maneuver algorithm to locate targets with a seismic array Demonstrated reverse-time focusing and corresponding enhancement of mine signatureDemonstrated imaging on numerical and experimental data from a clean and a cluttered environmentModified time-reverse imaging algorithms to include near field DOA and range estimates. The algorithms are verified for both numerical and experimental data with and without clutter.Modified wideband RELAX and CLEAN algorithms for the case of passive buried targets. The algorithms are verified for both numerical and experimental data with and without clutter. Developed a vector signal modeling algorithm based on IQML (Iterative Quadratic maximum Likelihood) to estimate the two-dimensional ω-k spectrum for multi-channel seismic data.

Developed multi-static radarDemonstrated radar operation with and without clutter objects for four scenariosInvestigated pre-stack migration imaging of multi-static data

Buried structuresDeveloped numerical model for a buried structureDemonstrated two possible configurations for a sensorMade measurement using multi-static radar


PlansThree sensor experiment (Landmine)

Incorporate reverse-time focusing and imagingIncorporate multi-static radarMore burial scenarios based on inputs from the signal processorsLook for more connections between the sensor responses that can be exploited for multimodal/cooperative imaging/inversion/detection algorithms

Imaging/inversion/detection algorithmsUse reverse-time ideas to characterize the inhomogeneity of the groundInvestigate the time reverse imaging algorithm for multi-static GPR data.Investigate the CLEAN and RELAX algorithms for target imaging from reflected data in the presence of forward waves with limited number of receivers.Develop elimination and end-game strategies for seismic detection with a maneuvering arrayInvestigate joint imaging algorithms for GPR and seismic data.

Buried Structures & TunnelsExperiments with multi-static radarDevelop joint seismic/radar experimentSignal Processing

multi-sensor measurements for the detection of buried targetslcarin/scott8.3.05.pdf2005/08/03 ·...

Documents