Performance of a computational Bayesian approach for active localization of a mobile scatterer in a refractive ocean environment

Underwater active acoustic systems that employ limited aperture arrays suffer from reduced processing gain and poor angular resolution, thereby hindering inferential objectives such as localization and tracking. Accurate localization is further challenged by the short conventional coherence-time associated with mobile bodies and dynamic environments. A computational Bayesian approach is presented and expanded here, for joint inference of range, depth, and speed of a submerged mobile scatterer in a refractive and multipath environment Barros and Gendron, JASA-EL, 2019. The Gibbs-sampling based approach infers the joint posterior probability density (PPD) of the pressure field wave vectors associated with the angle/Doppler spread arrivals and then maps their joint PPD to the target state joint PPD through acoustic ray interpolation. Performance analysis at high frequencies and relevant ranges using simulated acoustic fields are presented. The posterior uncertainty is investigated as a function of aperture and SNR, and we summarize the PPD using posterior credible intervals. A bound on posterior variance due to uncertainty in sound speed is provided and explored using profiles from the HYCOM database.