Differential Subspace Re-weighted Modelling for UWA Source Localization at Low SNRs

Abstract

The main aim of this work is to develop a robust, optimal sensor based 2-D and 3-D Differential Subspace (DS) re-weighted sparse reconstruction (SR) model for underwater acoustic (UWA) source localization. The DS scheme correctly estimates the spatially contiguous source locations with an optimal number of sensors (ONS) at low SNRs in spatially correlated noise. This is achieved by the Hadamard product of the MUSIC magnitude and differential phase spectrum. A Modified Differential Subspace (MDS) re-weighted scheme is also proposed by a linear combination of MUSIC magnitude weights and DS weights. The results show that, with the optimal number of sensors, the ARMSE and the resolution probability of the proposed scheme achieve the best state-of-the-art performance compared to other methods. © 2021 MTS.