Sparsity-driven passive tracking of underwater acoustic sources

Abstract

Tracking underwater acoustic sources using passive sonar is a challenging task due to the complex interactions that acoustic signals undergo as they propagate through the undersea environment. Notwithstanding these challenges, methods for tracking acoustic sources, which assume that acoustic environmental information is available, have been proposed. These methods are challenged by their high computational complexity and often do not exploit the temporal structure inherent to the tracking problem. This work proposes a sparsity-driven approach for tracking broadband acoustic sources. Source location maps (SLMs), one per frequency, are sequentially estimated while capturing the temporal dependance between successive SLMs. Coherence across the SLMs' support is enforced to guarantee that the source-location estimates are independent of frequency. An iterative solver based on the proximal gradient method is developed to construct the SLMs. Numerical examples on real data illustrate the performance of the proposed algorithm.