Broadband underwater source localization by solving basis pursuit de-noising using coordinate descent search

Underwater localization of multiple acoustic sources is important for underwater communication networks. As the number of sources is much smaller than the number of potential source positions, sparse signal representation can be adopted. In this paper, we propose a new sparsity based matched field (MF) processing method to solve the underwater source localization problem. The proposed method is based on solving the basis pursuit de-noising (BPDN) problem for multi-frequency signals. An efficient algorithm based on homotopy approach and coordinate descent optimization is used to find the BPDN solution. The performance is evaluated by applying this technique to real data obtained in the underwater acoustic experiment SWellEx-96. It is shown that this method can accurately locate multiple sources. It outperforms other MF approaches in resolving sources positioned very closely to each other, tolerance to noise and capability of locating multiple sources.