Practical Sparse Channel Estimation for OTFS Underwater Acoustic Communications

Orthogonal Time Frequency Space (OTFS) modulation has demonstrated advantages in addressing the doubly selective underwater acoustic (UWA) channels. In this paper, we focus on channel estimation in OTFS-based underwater acoustic communications (UAC), where large Doppler effects in wideband systems increase estimation challenge. We first present an OTFS-UAC system based on the discrete Zak transform, incorporating fractional effects, and the scaling impact induced by the Doppler shifts in wideband systems. To address the superimposing of fractional effects and Doppler scaling, we analyze these characteristics in the time-domain and then map them into the delay-Doppler (DD) domain. Specifically, we formulate channel estimation as a compressed sensing problem and design a refined grid-based sensing matrix through the timedomain effective channel matrix. Utilizing the sparsity of the UWA channel impulse response, we propose a Sparse Bayesian Learning (SBL)-based algorithm that avoids DD domain channel spreading. Simulation and experimental results demonstrate that the proposed algorithm outperforms the comparison algorithms by at least 3dB signal-to-noise ratios (SNR) gain under bit error rate (BER) of 10⁻² on simulation channels, and by at least 5dB SNR gain under BER of 10⁻¹ on experiment channels.