ASAP Workshop 12-Mar-02 Statistical Signal Processing Algorithms for Time-Varying Sensor Arrays David W. Rieken Veridian Systems Division Ann Arbor R&D Facility Ann Arbor, MI [email protected]Daniel R. Fuhrmann Dept. of Electrical Engineering Washington University St. Louis, MO [email protected]
28
Embed
Statistical Signal Processing Algorithms for Time-Varying Sensor Arrays
Statistical Signal Processing Algorithms for Time-Varying Sensor Arrays. Daniel R. Fuhrmann Dept. of Electrical Engineering Washington University St. Louis, MO [email protected]. David W. Rieken Veridian Systems Division Ann Arbor R&D Facility Ann Arbor, MI [email protected]. - PowerPoint PPT Presentation
Welcome message from author
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
• Arrays of linear arrays: Pesavento, Gershman, Wong (2001).
ASAP Workshop12-Mar-0221
MUSIC for Matrix Sequences
.,, sequencematrix Covariance :Given ,1 MNM V RRR
There are K sources. We wish to find the direction of each.
: ofposition Eigendecom mR
H
H
V
UVUR
m
m
Nm
m
mmm
,
1,
0
0
Nmmm ,2,1,
.,, NmmKNN
mKN
m CC IUUVU H
ASAP Workshop12-Mar-0222
MUSIC for Matrix Sequences
ismatrix covarianceth for the spectrum MUSIC The m
1;;
mmNmm ttf aPIaH
.1
Mz
z
z
Define the composite data vector as
HUUP mmm
.col onto from projection theis mNC U
where
ASAP Workshop12-Mar-0223
MUSIC for Matrix Sequences
.
0
01
MU
U
U
The signal is constrained to lie within the intersection of each of the actual data vectors. The signal subspace of the composite data vector is therefore the column span of
The projection onto the composite signal subspace is
MMM P
P
UU
UU
UUPH
H
H
0
0
0
0 111
ASAP Workshop12-Mar-0224
MUSIC for Matrix Sequences
The MUSIC spectrum is given by
1 aPIaH NMf
1
1
1
1
1
1
11
1
;;
;
;
;;
M
mm
M
mmmNm
MM
NMM
f
tt
t
t
ttf
aPIa
a
a
P
P
Iaa
H
HH
ASAP Workshop12-Mar-0225
Computer Simulation - MUSIC Spectra
Projected Sequence Inverse Iterations Sequence
0.0
45.0
90.0
135.0
180.0
225.0
270.0
315.0
0.0 30.0 60.0 90.0 0.0
45.0
90.0
135.0
180.0
225.0
270.0
315.0
0.0 30.0 60.0 90.0
ASAP Workshop12-Mar-0226
Comparison to CRLB
ASAP Workshop12-Mar-0227
Summary
• Covariance estimation is important for many array processing applications.
• Time-varying sensor arrays are becoming more common and require different covariance estimation algorithms than do their time-invariant brethren.
• We have developed an algorithm which estimates the covariance matrix sequence which arises from a time-varying array and demonstrated the application of that covariance estimate in estimating the direction-of-arrival and the spatial spectrum.
• The time-varying nature of an array can be advantageous rather than detrimental.
• Performance in real-world situations still not quantified: e.g. imperfections in array manifold calibration, sensor location estimates, etc.
ASAP Workshop12-Mar-0228
Acknowledgement
• This work supported in part by MIT Lincoln Laboratory and the Boeing Foundation.