Вы находитесь на странице: 1из 25

# http://www.girdsystems.

com

## GIRD Systems, Inc.

310 Terrace Ave.
Cincinnati, Ohio 45220

Based on

## R. O. Schmidt, “Multiple emitter location and signal parameter estimation,”

IEEE Trans. Antennas & Propagation, vol. 34, no. 3, March 1986, and
R. Roy and T. Kailath, “ESPRIT – Estimation of signal parameters via rotation invariance
techniques,” IEEE Trans. Acoust., Speech, Signal Proc., vol. 17, no. 7, July 1989
http://www.girdsystems.com

Introduction
• Well known high-resolution DOA algorithms
• Able to find DOAs of multiple sources
• High spatial resolution compared with other alg.
(I.e., a few antennas can result in high accuracy)
• MUSIC stands for Multiple Signal Classifier
• ESPRIT stands for Estimation of Signal
Parameters via Rotational Invariance Technique
• Apply to only narrowband signal sources
http://www.girdsystems.com

## Narrowband Signal Sources

• A complex sinusoid
s(t )   e j e jt   e jt
• A real sinusoid is a sum of two sinusoids
 
 cos(t   )  j
e e jt
 e j e jt  1e jt  2e jt
2 2
• A delay of a sinusoid is a phase shift

## s(t  t0 )  e jt0  e jt  e jt0 s(t )

• Apply approximately to narrowband signals
http://www.girdsystems.com

## • Assume that all frequencies are different

• Assume that all amplitudes are uncorrelated

 i2 ; i  j
E{i  j }  
 0; i  j
http://www.girdsystems.com

## A Uniform Linear Array

A signal source s(t )   e jt

“impinges” on the array
with an angle 
c: propagation speed

## • If the received signal at sensor 1 is x1 (t )  s(t )

(i  1)d sin 
• Then it is delayed at sensor i by i 
c
• Then the received signal at sensor i is
( i 1) d sin 
 j
xi (t )  e ji s1 (t )  e ji s(t )  e c
s(t )
http://www.girdsystems.com

Signal Model
• Put received signals at all N sensors together:

 1 
 x1 (t )    j d sin 
 x (t )   e c 
 2   2 d sin 

x(t )   x3 (t )    e j c  s(t )  a( ) s(t )
   
   
 xN (t )   
  j ( N 1)cd sin  
e 
• a( ) is called a “steering vector”
http://www.girdsystems.com

Signal Model
• If there are I sources signals received by the array, we get
a “signal model”:

## x(t) --- received signal vector ( N by 1 )

s(t) --- source signal vector ( I by 1 )
n(t) --- noise vector ( N by 1 )
( N by I )
s(t )  [ s1 (t ), , sI (t )]T
• Sources are independent, noises are uncorrelated
• Column of A can also be normalized
http://www.girdsystems.com

## The MUSIC Algorithm

• Compute the N x N correlation matrix
R x  E{x(t )x H (t )}  ARs A H   02I
where Rs  E{s(t )s H (t )}  diag.{12 , ,  I2 }
• If the sources are somewhat correlated so R s is not
diagonal, it will still work if R s has full rank.
• If the sources are correlated such that R s is rank deficient,
then it is a problem. A common solution is “spatial
smoothing”.
Q: Why is the rank of R s (being I) so important?
A: It defines the dimension of the signal subspace.
http://www.girdsystems.com

## The MUSIC Algorithm

• For N > I, the matrix ARs A H is singular, i.e.,
det[ARs A H ]  det[R x   02I]  0
• But this implies that  02 is an eigenvalue of R x
• Since the dimension of the null space of ARs A H is N-
I, there are N-I such eigenvalues  02 of R x
• Since both R x and ARs A H are non-negative definite,
 2
there are I other eigenvalues i such that i  2
  0 0
2

## R xui  [ARs A H   02I]ui   i2ui ; i  1, 2, ,N

 i2   02  0, i  1, , I ;  i2   02 , i  I  1, , N
http://www.girdsystems.com

## The MUSIC Algorithm

R xui  [ARs A H   02I]ui   i2ui ; i  1, 2, ,N
• This implies
ARs A H ui  ( i2   02 )ui ; i  1, 2, ,N
 ( 2
  2
0 )ui ; i  1, 2, , I
AR s A ui  
H i

0; i  I  1, , N
• Partition the N-dimensional vector space into the
signal subspace U s and the noise subspace U n
 Us Un   [ u1 uI u I 1 uN ]
Us : ( i2  02 )  0 eigenvalues Un : 0 eigenvalues
http://www.girdsystems.com

## The MUSIC Algorithm

• The steering vector a(i ) is in the signal subspace
• Signal subspace is orthogonal to noise subspace
 ( i   0 )ui ; i  1, 2, , I (1)
2 2
AR s A ui  
H

0; i  I  1, , N (2)
(1) means I linear combinations of columns of A
equal the signal subspace spanned by columns of U s
(2) means the linear combinations of columns of A,
i.e., the signal subspace, is orthogonal to U n
http://www.girdsystems.com

## The MUSIC Algorithm

• The steering vector a(i ) is in the signal subspace
• Signal subspace is orthogonal to noise subspace
• This implies that a H (i )Un  0
• So the MUSIC algorithm searches through all
angles  , and plots the “spatial spectrum”
1
P( ) 
a ( )Un
H

## • Wherever   i , P( ) exhibits a peak

• Peak detection will give spatial angles of all
incident sources
http://www.girdsystems.com

## MUSIC spatial spectrum compared with other methods

http://www.girdsystems.com

## Pros/Cons of The MUSIC Algorithm

• Works for other array shapes, need to know sensor
positions
• Very sensitive to sensor position, gain, and phase
errors, need careful calibration to make it work well
• Searching through all  could be computationally
expensive
• The ESPRIT algorithm overcomes such
shortcomings to some degree
• ESPRIT relaxes the calibration task somewhat
• ESPRIT takes much less computation
• But ESPRIT takes twice as many sensors
http://www.girdsystems.com

## The ESPRIT Algorithm

• Based on “doublets” of sensors, i.e., in each pair of
sensors the two should be identical, and all doubles
should line up completely in the same direction with a
displacement vector Δ having magnitude 
• Otherwise there are no restrictions, the sensor patterns
could be very different from one pair to another
• The positions of the doublets are also arbitrary
• This makes calibration a little easier
• Assume N sets of doublets, i.e., 2N sensors
• Assume I sources, N > I
http://www.girdsystems.com

## The amount of delay

between two sensors in
each doublet for a
given incident signal is
the same for all
doublets, which is
 sin  / c
A sensor array of doublets
http://www.girdsystems.com

## The ESPRIT Algorithm

• This array is consisted of two identical subarrays, Z x
and Z y , displaced from each other by Δ
I
x(t )   a(i ) si (t )  n x (t )  As(t )  n x (t )
i 1
I
y (t )   a(i ) e si (t )  n x (t )  AΦs(t )  n x (t )
j i

i 1

## • The steering vector a( ) depends on the array

geometry, and should be known just like in MUSIC
http://www.girdsystems.com

## The ESPRIT Algorithm

• The objective is to estimate Φ , thereby obtaining  i
• Define
 x(t )   A  n x (t ) 
z(t )      s(t )     As(t )  n z (t )
 y (t )   AΦ n y (t ) 
• Compute the 2N x 2N correlation matrix
R z  E{z(t )z H (t )}  ARs A H   02I
• Since there are I sources, the I eigenvectors of R z
corresponding to the I largest eigenvalues form the
signal subspace U s ; The remaining 2N-I
eigenvectors form the noise subspace U n
http://www.girdsystems.com

## The ESPRIT Algorithm

• U s is 2N x I, and its span is the same as the span of A
• Therefore, there exists a unique nonsingular I x I
matrix T such that (A needs to be known here)
Us  AT
• Partition U s into two N x I submatrices
 U x   AT 
Us      
 y 
U AΦT 
• The columns of both U x and U y are linear
combinations of A , so each of them has a column
rank I
http://www.girdsystems.com

## The ESPRIT Algorithm

• Define an N x 2I matrix, which has rank I
Uxy  U x U y 
• Therefore, U xy has a null space with dimension I,
i.e., there exists a 2I x I matrix F such that
 Fx 
U xy F  0   U x U y     U x Fx  U y Fy  0
Fy 
 ATFx  AΦTFy  0  AΦTFy   ATFx
• Then the above gives, since T has full column rank
AΦT  ATFxFy1  AΦ  ATFxFy1T1  Φ  TFxFy1T1
http://www.girdsystems.com

## The ESPRIT Algorithm

• The final algorithm is
1 1
Φ  TF F T
x y

## • In practice, the measurement could be noisy, and

there could be array calibration errors, so a total least
squares ESPRIT is used
• The estimation is “one shot”, even with matrix
inversions the computation is much less than the
MUSIC search
• We must have N > I for ESPRIT to work with 2N
sensors, so need twice as many sensors as MUSIC
http://www.girdsystems.com

## Simulation results of ESPRIT comparing with MUSIC

http://www.girdsystems.com

Conclusions
• Both methods are high-resolution, much better spatial
resolution than beamforming and other methods
• Both methods are able to detect multiple sources
• If sensors are expensive and few, and if computation
is not of concern, MUSIC is suitable
• If there are plenty of sensors compared with the
number of sources to detect, and if computational
power is limited, ESPRIT is suitable
http://www.girdsystems.com

## • Founded in 2000 - based in Cincinnati

• Specializes in communications and
signal processing, especially
developing novel algorithms to solve
challenging problems
• As of Nov. 2009, won more than 13
Phase I awards and 6 Phase II awards
from Navy, Air Force, Army
• Partnerships with many large
contractors including Northrop
Grumman, L-3 Communications, etc.
http://www.girdsystems.com

• Key Technology Areas
– Interference Mitigation (no reference, in-band)
– Direction Finding (wideband, high-resolution)
– Location/Navigation (Assisted GPS, GPS denied,
signals of opportunity)
– Wireless Network Security (physical layer)
– Power Amplifier Linearization
– Novel Communications systems/modeling
• Contact Information: GIRD Systems, Inc. Phone: (513) 281-2900
310 Terrace Ave. Fax: (513) 281-3494
Cincinnati, OH 45220 E-mail: info@girdsystems.com
USA Web: www.girdsystems.com