Low-complexity iterative receiver based on variational Bayesian inference for multiple-input multiple-output underwater acoustic communication

The design of high-performance receivers with manageable complexity is crucial for underwater acoustic communication, especially in multiple-input multiple-output (MIMO) scenarios. In this paper, a low-complexity MIMO receiver, which is based on variational Bayesian inference (VBI), is proposed. First, an iterative channel estimation model is constructed, which is based on VBI, the high-dimensional channel vector in the MIMO system is decomposed into a parallel connection of multiple low-dimensional channel vectors with different sparsity, and the prior distribution of the low-dimensional channel vectors is jointly modeled using temporal correlation (TC) and sparsity. Next, the low-complexity vector approximate message passing (VAMP) technique is integrated into the VBI framework and a channel estimation method is derived, based on TC-VAMP-VBI. Finally, to reduce the complexity of MIMO channel equalization, a serial iterative equalization algorithm is proposed, which incorporates passive time reversal under the VBI framework. The proposed algorithm was validated using simulations and MIMO communication data collected from two field experiments. The results show that the proposed algorithm can significantly reduce the computational complexity of the MIMO system while maintaining robustness of channel estimation in short data block scenarios.