Spatial Array Processing

Signal and Image Processing Seminar

Murat Torlak

Telecommunications & Information Sys. Eng.

The University of Texas at Austin

Introduction

A sensor array is a group of sensors located at

spatially separated points

Sensor array processing focuses on data collected at

the sensors to carry out a given estimation task

Application Areas
Radar
Sonar
Seismic exploration
Anti-jamming communications
YES! Wireless communications

Problem Statement
s1 (t)
s2 (t)
1

x1 (t)

x2 (t)

x3 (t)

x4 (t)

Find
1. Number of sources

2. Their direction-of-arrivals (DOAs)

3. Signal Waveforms

x5 (t)

x6 (t)

Assumptions

Isotropic and nondispersive medium

Uniform propagation in all directions

Far-Field
Radius of propogation

size of array

Plane wave propogation

Zero mean white noise and signal, uncorrelated

No coupling and perfect calibration

Antenna Array

Source

X1

X4

X3

X5

Array Response VectorFar-Field Assumption

Narrowband
- Delay
Phase Shift
=
Assumption

1; ej 2fc 4 sin =c ; : : : ; ej 2fc 44 sin =c T

a =

 xt

Single Source Case =

2
x t
6 x1 t
6 2
6
6
6
.
6
.
6
.
4

xM t

3 2
7 6
7 6
7 6
7=6
7 6
7 6
7 6
5 4

s1 t
s1 t ,

.
.
.

s1 t , M , 1

where

3 2
7 6
7 6
7 6
76
7 6
7 6
7 6
5 4

e,j 2fc
.
.
.

e,j 2fc M ,1

= 4 sin 1 =c.

3
7
7
7
7 s t = a s t
7 1
1 1
7
7
5

General Model

By superposition, for

xt

=
=

d signals,

a1 s1 t +


+ ad sd t

d
X
ak sk t
k=1

Noise

xt

=
=

d
X

ak sk t + nt

k=1
ASt + nt

where

A = a1 ; : : : ; ad 
and

St =

s1 t; : : : ; sd t T :

Low-Resolution Approach:Beamforming

Basic Idea

d
X

i,1j 2fc 4 sin k =c sk t = X sk tejwk i,1

xi t =
=e
k=1
k=1
where wk

= 24 sink =c and i = 1; : : : ; M .

Use DFT (or FFT) to nd the frequencies

F F w



=

1

2
6
6
6
FwM  = 6
6
6
6
4

ejw1

ejw2

.
.
.

.
.
.

ej M ,1w1

ej M ,1w2

.

jF xi tj = jF xtj2

To smooth out noise

N
1 X jF xtj2
B wi  =
N

Look for the peaks in

t=1

fwk g
1

ejwM
.
.
.

ej M ,1wM

3
7
7
7
7
7
7
7
5

Beamforming Algorithm

Algorithm

Rx =

PN

xtx t
t=1
2. Calculate B wi  = F wi Rx Fwi 

1. Estimate

1
N

B wi  for all possible wi s.

Calculate k , i = 1; : : : ; d.

3. Find peaks of
4.

Advantage
- Simple and easy to understand
Disadvantage
- Low resolution

Number of Sources
Detection of number of signals for

Rx

=
=

xt = Ast + nt

E fxtx tg = A E fsts tg A + E fntn tg
|
z
|
z
A |Rzs |Az
| z

M d dd dM

Rs

2
+ nI

2
nI

n is the noise power.

No noise and rank of
where

d M,

Rs is d

Rx = ARs A will be
f1 ; : : : ; d ; 0; : : : ; 0g:
Real positive eigenvalues because Rx is real, Hermition-symmetric

Eigenvalues of

rank

Check the rank of

Rx or its nonzero eigenvalues to

detect the number of signals

2
Noise eigenvalues are shifted by n

f1 +

where 1

:::

n ; : : : ; d + n ; n ; : : : ; n g:

d and 

Detect the number of principal (distinct) eigenvalues

9

MUSIC
Subspace decomposition by performing eigenvalue
decomposition

 + n I = X k ek e
Rx = ARs A 2
k
k=1

ek is the eigenvector of the k eigenvalue

spanfAg = spanfe1 ; : : : ; ed g = spanfEs g
Check which a  spanfEs g or PA a  or
P? a, where PA is a projection matrix
A
Search for all possible  such that
where

jP? aj2 = 0 or M  =
A

After EVD of

1
PAa = 1

Rx
P? = I , EsE = EnE
A
s
n

where the noise eigenvector matrix

En = ed+1; : : : ; eM

10

Root-MUSIC
For a true , j 2fc 4 sin =c is a root of

P z  =

M
X

k=d+1

1; z; : : : ; z M ,1 T ek e 1; z ,1 ; : : : ; z ,M ,1 :
k

After eigenvalue decomposition,

fek gd=1
k
- Form pz 
- Obtain 2M , 2 roots by rooting pz 
- Pick d roots lying on the unit circle
- Solve for fk g
- Obtain

11

Estimation of Signal Parameters via

Rotationally Invariant Techniques (ESPRIT)
Decompose a uniform linear array of
two subarrays with

M , 1 sensors

M sensors into

Note the shift invariance property

2
6
6
2  = 6
6
a
6
6
4

ejw
ej 2w
.
.
.

ej M ,1w

3 2
7 6
7 6
7 6
7=6
7 6
7 6
5 4

ejw
.
.
.

ej M ,1w

3
7
7
7 jw 1 jw
7e = a e
7
7
5

General form relating subarray (1) to subarray (2)

2
ejw
6
A2 = A1 6
6
4

..

ejwd

3
7
7 = A1:
7
5

 contains sufcient information of fk g

12

ESPRIT

spanfEs g = spanfAg and Es = AT

- T is a d  d nonsingular unitary matrix
-

T comes from a Grahm-Schmit orthogonalization

of

Ab in

Rx

EssE + En E
s
n
2
AH RsA + n I
E2 = A2T and E1 = A1T
s
s
=

Es 2 = A2 T = A1 T = Es 1T,1 T

Multiply both sides by the pseudo inverse of

E1
s

E1Es 2 = E1 E1,1 E1E1 T,1 T = T,1 T

s
where  means the pseudo-inverse

A = AsH A,1AsH
Eigenvalues of

T,1T are those of .

13

Superresolution Algorithms

1. Calculate

Rx =

1
N

PN

xkx k
k=1

2. Perform eigenvalue decomposition

3. Based on the distribution of

fk g, determine d

4. Use your favorite diraction-of-arrival estimation

algorithm:
(a) MUSIC: Find the peaks of

180

- Find

M  for  from 0 to

^ k
fk gd=1 corresponding the d peaks of

M 
.

(b) Root-MUSIC: Root the polynomial

pz 

d roots that are closest to the unit

rk c
^
circle frk gd=1 and k = sin,1 2fc
.
k

- Pick the

(c) ESPRIT: Find the eigenvalues of

f kg
k
^
- k = sin,1 2fcc4

14

E1E2,
s
s

Signal Waveform Estimation

Given

A, recover st from xt.

Deterministic Method

wk such that
wk ? ai ; i 6= k; wk 6? ak 

No noise case: nd

A can do the job

Axt = AAst = st
With noise,

nt
Axt = st + Ant
 increased noise

Disadvantage =

15

Stocastic Approach

wk to minimize

min E fjwk xtj2 g =  min wk Rk wk
a k wk =1
a k wk =1

Find

Use the Langrange method

min E fjwk xtj2 g , ;w wk Rk wk + 2a k wk , 1

min 
a k wk =1
k
Differentiating it, we obtain

Rx wk = ak ; orwk = R,1 ak 

x
.

Since

ak wk = ak R,1ak  = 1,

x

Then

 = a k R,1 ak 
x
Capons Beamformer

wk = R,1ak =a k R,1ak 

x
x
16

Subspace Framework for Sinusoid

Detection

xt =

d
P
k=1

k e k +j!k t

M , i.e.,

Let us select a window of

xt

  =

xt; : : : ; xt , M + 1 T

Then

xt

 

2
6
6
6
6
6
6
6
4

3
2
7
6
7
d 6
7 X6
7=
6
7
7 k=1 6
6
7
6
5
4

xt
xt , 1
.
.
.

xt , M + 1

2
6
d 6
X6
6
6
6
k=1 6
4
|

e, k +j!k 
.
.
.

e k +j!k ,M +1

z

.
.
.

k e k +j!k t,M +1

3
7
7
7
7 e k +j!k t
7 k
7|
z
7
5

sk t

a k
d
X
a k sk t As t ;


k+ k t
k e +j!j!t,1
k e k k

k=1

  =

 

where M is the window size, d the number of sinusoids, and

k = e k +j!k .

17

3
7
7
7
7
7
7
7
5

Subspace Framework for Sinusoid

Detection

Therefore, the subspace methods can be applied to

nd

f k + j!k g

Recall

xt =

Then nding

d
X
k=1

 +j! t
ke k k

f k g is a simple least squares problem.

18

Wireless Communications
co

-c

ha

nn

el

in

te

rf

er

en

ce

Multipaths
th

t
ec

Pa

Di

Cellular Telephony
Office Building

Residential Area

Personal Communications
Services (PCS)

Outdoors
To Networks

Di

re

th
t Pa
irec
D

ct

th

th

pa

i
lt

Mu

Wireless LAN

Increasing Demand for Wireless Services

Unique Problems compared to Wired
communications
19

Pa

Problems in Wireless Communications

Scarce Radio Spectrum and Co-channel
Interference
1
4

2
3
1
4

1
2

3
1

Multipath
Multipath

Direct P

ath

Mu

ltip

ath

Base
Station

Time

Desired Signal

Reflected Signal

Coverage/Range
20

Smart Antenna Systems

Employ more than one antenna element and exploit

the spatial dimension in signal processing to improve
some system operating parameter(s):
- Capacity, Quality, Coverage, and Cost.
User One

User Two

Multiple RF Module

Advanced Signal Processing

Algorithms

Conventional
Communication Module

21

Experimental Validation of Smart Uplink

Algorithm

Comparison of constellation before (upper) and after

imaginary axis

smart uplink processing (middle and lower)

imaginary axis

real axis
Antenna Output

imaginary axis

real axis
Equalized Signal 1

real axis
Equalized Signal 2

22

Selective Transmission Using DOAs

Beamforming results for two sources separated by

20

Power Spectrum

1
0.8
0.6
0.4
0.2
0

0.5

1
1.5
Frequency [Hz], User #1

1
1.5
Frequency [Hz], User #2

x 10

Power Spectrum

1
0.8
0.6
0.4
0.2
0

0.5

23

x 10

Selective Transmission Using DOAs

Beamforming results for two sources separated by

Power Spectrum

1
0.8
0.6
0.4
0.2
0

0.5

1
1.5
Frequency [Hz], User #1

1
1.5
Frequency [Hz], User #2

x 10

Power Spectrum

1
0.8
0.6
0.4
0.2
0

0.5

24

x 10

Future Directions

Adapt the theoretical methods to t the particular

demands in specic applications
Smart Antennas
Synthetic aperture radar
Underwater acoustic imaging
Chemical sensor arrays
Bridge the gap between theoretical methods and
real-time applications

25