Structured methods for underwater wideband signals localization using sensor networks

In this paper, the localization problem via a wireless sensor network for underwater wideband sources is mapped to a sparse approximation problem by exploiting inherent structure and geometry. In underwater applications, signaling is achieved via acoustics which experiences attenuation as a function of transmission distance, large propagation delays and scaling from Doppler effects. Furthermore, there is significant multipath. Herein, novel joint sparse, structured recovery approaches for source localization in underwater environments are developed. The proposed methods achieve 4 to 7 dB improvement in localization over methods that only exploit block-sparsity or element-wise sparsity.