Message passing-based single-carrier communications in deep-sea horizontal acoustic channels: Joint interference cancellation and symbol detection.

Deep-sea horizontal acoustic channels are characterized by widely separated clusters, with inter-cluster intervals spanning several hundred milliseconds. This channel type leads to severe inter-block interference (IBI) in zero-padded single-carrier block transmissions. In this paper, we propose a two-step equalizer based on joint channel estimation (CE), IBI cancellation (IBIC), and symbol detection (SD), termed JCE-IBIC-SD. The proposed equalizer effectively suppresses IBI through iterative CE and symbol detection, even with insufficient training sequences. In the first step, considering the large separation between multipath clusters, the equalizer decomposes deep-sea channels longer than one block into quasi-synchronous channels and retrieves the symbols for all blocks through IBIC and equalization. In the second step, based on the a priori knowledge of the symbols, the JCE-IBIC-SD is applied. Using the damped Gaussian generalized approximate message passing sparse Bayesian learning algorithm, iteratively computes the maximum a posteriori estimates of both the channels and symbols through within a joint factor graph model. Simulations and deep-sea experimental results demonstrate that, even when the receiving array is located in the acoustic shadow zone, the proposed equalizer outperforms traditional IBIC equalizers with reasonable computational complexity.