Yung P. Lee

1
Space-Time Adaptive Matched-field Processing
(STAMP)

• Yung P. Lee
• (ASAP 2001, March 14, 2001)
• Science Applications International Corporation
• 1710 SAIC Drive
• McLean, VA 22102
• Yung_at_osg.saic.com

2
Sonar Signal Processing Background
Fourier Transform Spectral (Frequency)
Content
Spatial Beamforming Direction (Angle) of
Arrival (DOA)
3
Matched Field Processing
Matched Field Processing 3D (Range,depth,
bearing) Localization
Matched Field Tomography Modal Information
Environmental Info.
4
(No Transcript)
5
Synthetic Aperture Matched Field Processing
source at 76 m towed at 2.5 m/s from 9.18 km
6
(No Transcript)
7
Space Time Matched Field Processing
Localization Doppler (velocity) Discrimination
Matched Field Processing
Space Time Matched Field Processing
Phone-Doppler Space
Beam-Doppler Space
8
BACKGROUND/OBJECTIVE

• Space-Time Adaptive Processing (STAP) coherently
  combines signals from the elements of an array
  and the multiple snapshots of signals, to achieve
  large spatial/temporal signal gain, to suppress
  interference, and to provide target detection in
  azimuth and velocity.
• Matched-field processing (MFP) coherently
  combines complex multi-path arrivals, to recover
  signal multi-path spreading loss and to provide
  range/depth localization.
• STAMP combines STAP and MFP to improve detection
  and localization performance for the mobile
  multi-line-towed-array sonar systems.

9
STAP Detect the dot Null the Jammer and the
slanted clutter
STAMP Detect/combine/class/localize the dots
Null the Jammer and the clutter
0 90 180
FWD
Target
Jammer (own-ship)
Jammer
Clutter (Bottom Bounce)
Clutter (Bottom Reverberation)
Azimuth (deg)
Target
Passive Forward-sector processing
AFT
-Dfmax 0
Dfmax
-Dfmax 0
Dfmax
Doppler (Hz)
Doppler (Hz)
10
Multi-path Doppler/Angle Spread
C1 ,Df1
Cm ,Dfm
Dfmf0v/cm Higher Mode (Path,Angle), Larger cm
Larger cm, Higher Angle (off horizontal),
Smaller Doppler
11
OUTLINE

• STAMP Processing
• Simulation scenario for forward-sector processing
• Simulation Results

12
Space-Time Adaptive Matched-field Processing
(STAMP)
Doppler Processing Xr (f)
Conventional Beamforming Br(f)
Br(f0) Beam-space replica (Selected Beams and
Dopplers)
Propagation Code to generate Replica xr(t)
AEL Environ.
Search R,Z,q ,v
Plane-wave STAP
WB/NB Adaptive MFP
Phone 1 Line 1 x11(t)
Doppler Processing X1(f)
Conventional Beamforming B1(f)
Output Ambiguity Surface R,Z,q ,v
B(f) Beam-Space Vector (selected Beams and
Dopplers)
Phone n Line 1 xn1(t)
Forming Covariance Matrix R lt B(f)
B(f)gtf Decomposition
Phone 1 Line k x1k(t)
Doppler Processing Xk(f)
Conventional Beamforming Bk(f)
Phone n Line k xnk(t)
Bk(f) bk(f,q1) bk(f,ql)
B(f) B1(f). B1(fmDf),.., Bk(f). Bk(fmDf)
13
Adaptive Processing
Adaptive Weight Vector
Adaptive Output
A is the steering vector R is the measured
covariance matrix
High resolution Sidelobe suppression Subject to
mismatch Robust Methods (widen the peak)
14
Wideband-Narrowband (WB/NB) Feedback-Loop
White-Noise-Constrained (FLWNC) Adaptive
Processing
Br(f0) Beam-space replica (Selected Beams and
Dopplers)
yes e s
no
Covariance Matrix R lt B(f) B(f)gtf Decomposit
ion
Adaptive weight W
yes e s
no
WB/NB Processing S(f)WB(f)
B(f) is narrowband (single f) R and W are
broadband (averaged over band of f)
15
Simulation Geometry (F200 Hz) target(NB)120
dB, own-ship(BB)120 dB, bottom bounce(BB)115
dB WNL70 dB, 0.1 l random phase error
Single-Line
No environmental mismatch
4-Line-Sequential
4-Line-Vertical
3 kts
10 km
own-ship noise
towed array
188 m
3 kts
bottom bounce
16
Single-Line BTRs of Each Signal Component Forward
Endfire at 0o
17
Single-Line Doppler/Azimuth Responses
integration time 256-sec, Target Range10 km,
Forward Endfire at 0o
18
Single-Line Beam/Cell Spectrograms
19
Adaptive Beam/Cell Spectrograms
20
Array Size Dependence of MFP Range Tracking
search at target depth and target speed
21
Depth Discrimination of Adaptive MFP Range
Tracking 4_Line_Vertical Array
search at target speed
22
Speed Discrimination of Adaptive MFP Range
Tracking 4_Line_Vertical Array
search at target depth
23
SUMMARY

• STAMP processing that combines STAP and MFP has
  been developed.
• Simulations show that STAMP coherently combines
  signal multi-path spread in azimuth and Doppler
  and greatly enhances target detection as well as
  providing target range and depth classification
  and localization.