Low-Complexity Markov Chain Monte Carlo Detector for Single-Carrier Underwater Acoustic Communications

Markov chain Monte Carlo (MCMC) technique is an approximate Bayesian inference method and generally outperforms linear estimation. It has been adopted to realize channel estimation and symbol detection in underwater acoustic (UWA) communications. Despite its better performance, the high complexity limits its practical applications. To address this issue, we propose a low-complexity MCMC detector based on Gibbs sampling in this paper. The key idea is to utilize the reliability information in the Gibbs sampler. Specifically, when a bit is believed to be sufficiently reliable, it will be fixed, instead of being sampled and possibly modified in the remaining iterations. The proposed scheme is evaluated using both numerical simulations and experimental data collected in an at-sea UWA communication experiment conducted at Xiamen, China in 2019. The new method exhibits significant complexity reduction at the cost of slight performance degradation.