1. Adaptive Waveform Design for Target Localization and Tracking for Cognitive MIMO Sonar
Joseph Tabrikian
Underwater Acoustics Symposium
Tel-Aviv University
June 17, 2013
Research students: W. Huleihel and N. Shraga

2. Outline Introduction Cognitive MIMO radar/sonar configuration
Adaptive waveform for target localization
Adaptive waveform for MIMO sonar – static scenario
Adaptive waveform design for MIMO sonar – dynamic scenario
Conclusions and future work

3. Introduction - Cognitive Radar/Sonar
Adaptive Waveform Design
Environment
Adaptive Receiver
Detection/
Localization/
Tracking/
Classification
Transmit Signal
Receive
Signal
Key point: Transmit waveform is designed at very low SNR’s before the target is detected.

4. Cognitive MIMO Radar/Sonar Configuration

5. Cognitive MIMO Radar/Sonar Configuration

6. Cognitive MIMO Radar/Sonar Configuration
Target dynamic model
Detection/
Estimation/
Tracking
Optimal Adaptive Waveform Design
Optimal Receiver
noise

7. Waveform Design for Optimal Target Localization
Considered criteria:
Bayesian Cramér-Rao bound (BCRB)
Simple, analytic expressions
Ignores large-errors/threshold phenomenon
Reuven-Messer bound (RMB)
Higher complexity
Takes into account large-errors/threshold phenomenon and therefore is able to control the sidelobes

8. Simulations – Cognitive MIMO Radar

9. Simulations – Cognitive MIMO Radar
BCRB-based waveform design
Posterior pdf’s and transmit beampatterns
Auto-focusing effect: Automatic beamforming before detection/estimation.

10. Simulations – Cognitive MIMO Radar
RMB-based waveform design
Posterior pdf’s and transmit beampatterns
Auto-focusing effect: Automatic beamforming before detection/estimation.

11. Simulations – Cognitive MIMO Radar
Single target – direction estimation accuracy:
ASNR=-6dB
k=6

12. Cognitive MIMO Sonar

13. Simulations – Cognitive MIMO Sonar

14. Simulations – Cognitive MIMO Sonar
Single target – posterior pdf:

15. Simulations – Cognitive MIMO Sonar
Single target – beampattern:

16. Waveform Design for Optimal Target Tracking
Dynamic model:
What is the optimal transmit (spatial) waveform for tracking?

17. Simulations – Cognitive MIMO Sonar Target Tracking

18. Simulations – Cognitive MIMO Sonar
Single target – posterior pdf (via Monte-Carlo):

19. Simulations – Cognitive MIMO Sonar
Single target – posterior pdf (via Monte-Carlo):

20. Conclusions and Future Work
A new optimal waveform design approach for cognitive MIMO radar/sonar is proposed based on minimizing the BCRB and RMB at each step using the measurements from previous steps.
The RMB-based algorithm was shown to provide better results, since it is able to control the sidelobes.
This approach provides an automatic focusing array: beamforming before detection or estimation.
The method was adapted to consider dynamic targets, which can be interpreted as track-before-detect in transmission.
Further research will cover the following issues:
Taking into account environmental uncertainties,
Wideband signal model,
Realistic shallow water channel simulations,
Considering other optimization criteria, such as probability of detection.

21. Thank you!

22. Simulations – Cognitive MIMO Radar