Fast Time-Varying Sparsity Exploitation With Adaptive Observation Length Strategies in Underwater Acoustic Communication

Underwater acoustic (UWA) communication is essential for the information interaction between unmanned underwate vehicles (UUVs). However, the design of UWA communication systems is challenged by the inherent complexity of the UWA channel. While compressed sensing (CS) algorithms can enhance channel estimation performance by leveraging the sparsity of UWA channels, their effectiveness is limited in rapidly time-varying channel, especially in high-speed mobile scenarios. In this paper, we propose a variable observation length subspace pursuit (VOLSP) algorithm to exploit the fast time-varying sparsity in the UWA communication by dynamically adjusting the observation length. Since the residual error in the VOLSP indicates the unestimated portion of the target channel, the VOLSP adaptively searches for the optimal observation length along the negative gradient of the residual error during each iteration, so as to accommodate different degrees of channel variations. Additionally, we provide a convergence analysis of the VOLSP algorithm and compare its computational complexity with traditional methods. Finally, numerical simulation and field trial results verify that the proposed VOLSP algorithm attains superior reconstruction precision in fast time-varying sparse UWA channels compared to conventional algorithms.