Low complexity OMP based matched-phase coherent shallow water broadband multi-source localization

This paper proposes a broadband coherent matched-field processing algorithm for multi-source localization in shallow water scenarios. The proposal is based on the OMP (orthogonal matching pursuit) algorithm with phase descent search (PDS). In the sparse recovery procedure, the phase differences between different frequencies are estimated and compensated, and then, in each iteration, by coherently combining different frequency components, the support is updated. The proposed method improves the ability of the matched-phase coherent algorithm in locating more than one source, and it avoids performance loss due to the incoherent method used by the basis pursuit de-noising (BPDN) technique. The proposal is applied to the data collected in a shallow water acoustic source localization experiment, and gives sparse localization results which match the source trajectories calculated from GPS information.