ELEE Research Component of PhD Qualifier Exam By Kenneth T. Bowers
Lester W. Cory Conference Room, Science & Engineering Building (SENG), Room 213A
Paul J. Gendron
5089998510
pgendron@umassd.edu
Topic: Environmentally Aware Bayesian Active Sonar
Abstract: This work introduces a Bayesian framework for active sonar detection in a refractive ocean waveguide associated with horizontally stratified sound speed profile (ssp) and an attendant reverberation field. The detector is constrained by significant depth uncertainty of the scattering body of interest. The proposed scheme effectively quantifies the discriminating information between two composite hypotheses associated with these uncertainties. The framework rigorously accounts for environmental properties regarding the refractive media, integrating either theoretical surface and volume reverberation models or in situ observations. Uncertainty in the reverberation field and the target's depth are systematically and sensibly incorporated through proper marginalization. We consider broadband acoustic array observations over a vertical aperture. The covariance structure associated with the depth dependent target is constructed based on all available environmental information. The resulting Bayes Factor Active Sonar (BFAS) yields a set of range/time dependent quadratic forms operating within beam-delay space, minimizing average risk by optimally balancing the target-waveguide eigen modes against noise and reverberation subspaces. By doing so, BFAS combines multi-path arrivals and attenuates reverberation and noise. This substantially increases signal-to-reverberation-plus-noise ratios. By leveraging the small set of depth-invariant modes and their associated covariance structure we establish lower bounds on processor performance. These bounds demonstrate that even under severe target depth uncertainty, BFAS multi-path combining outperforms a conventional single specular arrival detector even when that specular detector is conditioned on perfect knowledge of the target's depth. Extensive performance validation across various measured refractive and shallow-water environments lends credence to the robustness of the Bayesian framework for active sonar system design. The framework provides a roadmap for sensible treatment of diverse other uncertainties such as through an orthogonal modal decomposition of ssp. [Funded by the Office of Naval Research]
Advisor(s): Dr. Paul J. Gendron, Professor, Dept. of Electrical & Computer Engineering, UMASS Dartmouth
Committee Members: Dr. David A. Brown, Professor, Dept. of Electrical & Computer Engineering, UMASS Dartmouth;Dr. Ana Doblas, Assistant Professor, Dept. of Electrical & Computer Engineering, UMASS Dartmouth; Dr. Dayalan P. Kasilingam, Professor, Dept. of Electrical & Computer Engineering, UMASS Dartmouth
NOTE: All ECE Graduate Students are ENCOURAGED to attend. All interested parties are invited to attend. Open to the public.
*For further information, please contact Dr. Paul J. Gendron via email at pgendron@umassd.edu.