Compressive Self-Noise Cancellation in Underwater Acoustics

Abstract

The purpose of sonar is to detect the stealthy target in shallow water. The main barrier to locating the target is sonar's self-noise. Existing subspace-based noise suppression methods typically employ eigenanalysis-based methods involving high computational complexity. Recent approaches based on compressed sensing (CS) or sparse representations (SR) are computationally efficient. It is not straightforward to extend existing CS/SR-based methods for self-noise cancellation as, first, the energy of interference is much higher than the target, and second, it also exhibits similar sparsity properties. This work presents a novel method to combine the advantages of a subspace-based noise cancellation approach with low complexity of working with fewer CS measurements. Both target recovery and self-noise cancellation are done in the compressive domain only. Experimental results demonstrate the robustness of the proposed approach for both narrowband and broadband targets at very low signal-to-interference-noise (SINR). © 2022 IEEE.