Estimation of moving targets located behind reinforced concrete - - PowerPoint PPT Presentation

• Estimation of moving targets

located behind reinforced concrete wall using moving sensor arrays

1University of Belgrade, Serbia 2Washington University in St. Louis, USA

Marija M. Nikoli1, Arye Nehorai2, and Antonije R. Djordjevi1

COST Action IC0603 Workshop Cyprus, April 2008

• Contents

Estimation of personnel hidden in buildings

reinforced with concrete walls

Through-wall vision

Measurements by sensor arrays moving

in front of the building

Recent investigations consider homogeneous walls Parameters of the wall:

Unknown Known

The influence of the wall on the signal propagation

taken into account

2D electromagnetic simulation

• Forward model

We consider reinforcement with metallic bars Simulation model:

Stationary and moving objects

Dielectric and metallic bodies

Sensors

Filament conductors

2-D field simulator

MoM solution Equivalent surface electric and magnetic currents Bars modeled as PEC wires

• Simulation model

Array consists of M sensors Measurements are taken at

N known positions L frequencies

In the 2-D model, induced fields are used

instead of electromotive forces

The array is a multiport network

• M

k l ik l i

w f n E f n E

1

, ,

noise

• M

k l k l ik l i

w f I f n z f n E

1

, , ,

• Time-domain response

Signals reflected from the wall

Homogeneous Reinforced

Sensors transmit short pulses

Reflections are separated in the time domain

• Time-domain response

Reflections from a metallic plate behind the wall Reinforced walls do not introduce additional delays The signal is significantly attenuated

• Estimation of wall permittivity

Signals reflected from the front side contain

information about wall permittivity

We focus the measurements in the time domain

with respect to the front side of the wall

• n

s t n E NM t E

ik N n M i M k ik ik

• 1

1 1 2

, 1

• Estimation of wall permittivity

h(t) - reference waveform h(t-1m/c0) -waveform induced in Rx at a

distance of 1m from the Tx

2 r

ˆ 1 ˆ 1 ˆ

• R

R

2

1 1

ˆ

• T

T

dt t h dt t h t E R

• Estimation of wall thickness

Measurements are

focused with respect to the back side of the wall

• 2

r w

sin / sin n t

• min

arg

• d

a y y x x

i i k

• 2

r r

/ sin / /cos

• c

w y y w

w

• N

n M i M k T ik

dt t h w t n E w I

1 1 1 0

,

• Estimation of targets

Assumptions

Static scene measurements available

Walls, furniture

Dynamic scene measurements available

Static scene altered by the appearance of targets

Difference in measurements associated

with fields scattered from targets

Parameters of bars need not be known

• Estimation of targets

The space in the building is divided

to small cells1

The scattered field is focused with respect

to different cells (x,y)

• 1F. Ahmad, M. G. Amin, and S. A. Kassam, “Synthetic aperture beamformer for imaging through

a dielectric wall,” IEEE Trans. Aerosp. Electron. Syst., vol. 41, no. 1, pp. 271–283, Jan. 2005

• Estimation of targets
• dt

t h y x t E y x I

T

• ,

; ,

• N

n M i M k k i ik

y x y x t E y x t E

1 1 1

, , , ;

• 2

, r r , , ,

/ sin / /cos

• ,

c w w y y y x

k i k i k i k i

• Modeling bar distortion

Bars significantly distort waveform Estimation is improved when distortion is modeled New reference pulse

• f

f T f H F T t t h 2 j exp , rect ~

2 1

• Simulation scenario

Targets

PEC cylinders, radius 0.2m

Walls

Thickness w = 0.2m Permittivity r = 3 Bar diameter Dbar = 2cm Bar period dbar = 15cm

Array

5 sensors Separation between sensors

0.2m

Standoff distance 0.75m Measurements every 0.2m

Frequency

fmax = 2GHz, f = 5MHz

• Wall parameter estimation

Estimated permittivity r = 2.94 There is no unique solution for wall thickness!

• Two targets

• Improved quality when

parameters of bars are known

• Erroneous wall thickness

• Conclusions

Reinforced wall significantly attenuates low

frequencies and distorts transmitted waveforms

Signals reflected from the objects behind the

wall have an oscillating nature

Scene images are blurred because of

the multiple echoes

We assumed we have available

Measurements of the static scene behind the wall Measurements altered by the appearance of people

• Conclusions

Measurements are coherently added with respect to

different time-delays to estimate

Wall permittivity Wall thickness Position and number of targets

The algorithm performed satisfactorily when bar

parameters are unknown

If the bar parameters are known:

Estimation is significantly improved The minimal necessary SNR is lower compared to the case

where the influence of the bars on signal shape is ignored