Estimation of Underwater Acoustic Channel with Hybrid Sparsity via Static-Dynamic Discriminative Compressed Sensing

The technology of underwater acoustic communication plays an important role in linking the underwater sensors for data collection and networking. Due to simultaneous presence of direct path as well as reflections upon static or dynamic boundaries, underwater acoustic (UWA) channel exhibits hybrid sparse, i.e., consists of relatively static multipath arrivals and time varying ones. While the classic sparse channel estimation algorithms generally do not take this type of hybrid sparsity into account, dynamic compressed sensing (DCS) technique enables the exploitation of time varying sparsity by incorporating CS with Kalman filtering, namely, Kalman Filtered Compressed Sensing (KF-CS). However, by assuming the whole UWA channel as sparse set with time varying support, static sparse components will unavoidably lead to model mismatch and performance degradation. In this paper, by formulating the hybrid multipath UWA channels as sparse set consisting of static and time varying supports, a static-dynamic discriminative compressed sensing (SDD-CS) approach is proposed to explore the hybrid sparsity. Hybrid multipath arrivals are firstly decomposed into static and time varying components by discriminative orthogonal matching pursuit (DOMP), and then the magnitude of time varying and static multipath is estimated by KF-CS and SOMP respectively, after that estimated results of static and time varying component are summed up to obtain the whole channel response. Numerical simulations verify the superiority of the proposed SDD-CS algorithm. Finally experimental results obtained from a shallow water acoustic communication scenario are provided to demonstrate the effectiveness of the proposed algorithm, compared to conventional algorithms.