Successive over-relaxation based Markov chain Monte Carlo symbol detection for multiple-input multiple-output underwater acoustic communications

Markov chain Monte Carlo (MCMC) technique has been employed for symbol detection in underwater acoustic (UWA) communications. Existing MCMC detectors, however, may even be inferior to a conventional linear minimum mean square error detector in case of nonideal factors. Moreover, they suffer high complexity, limiting their practical applications. In this paper, we resort to the successive over-relaxation (SOR)-based MCMC algorithm and explore its feasibility for symbol detection in multiple-input multiple-output UWA communications. The proposed SOR-MCMC detector using Gibbs sampling, was verified by both simulated data and experimental data collected in the Acoustic Communications 2009 UWA communication experiment conducted in New Jersey, USA in 2009. All results showed it has faster convergence and better performance than a standard MCMC symbol detector. Moreover, it enjoys lower computational complexity.