Markov Chain Monte Carlo: Applications to MIMO detection and channel equalization

Abstract

In this paper, we present an overview of recent work on the applications of Markov Chain Monte Carlo (MCMC) techniques to both multiple-input and multiple-output (MIMO) detection and channel equalization. In the setting of MIMO detection, we have shown that, even for very large antenna systems with high spectral efficiencies of 24 bits/channel use (8 transmit and 8 receive antennas with 64 QAM modulation), the MCMC MIMO detector can bring us within 2 dB of the channel capacity with a greatly reduced complexity compared to several versions of sphere decoding based detectors. For frequency selective channels, we demonstrate that MCMC-based equalizers yield excellent performance even for severe inter-symbol-interference (ISI) channels. The MCMC equalizer achieves significant performance gain over minimum mean square error (MMSE) linear equalizer and performs closely to the optimal maximum a posteriori probability (MAP) equalizer. We will also discuss new approaches that effectively alleviate the well-known high SNR problems in existing MCMC detectors.