Relaxed Belief Propagation for MIMO Detection

TL;DR

This paper introduces a relaxed belief propagation (RBP) detector for MIMO systems that reduces computational complexity by focusing on partial edges in the factor graph, and enhances detection performance through Gaussian approximation and MMSE cascade.

Contribution

The paper proposes a novel RBP algorithm with a relax degree for MIMO detection, including a feedback-based Gaussian approximation and a cascaded MMSE-RBP approach, outperforming existing BP methods.

Findings

RBP reduces computational complexity compared to SBP.

Gaussian approximation with feedback improves detection accuracy.

MMSE cascaded RBP outperforms SBP in large MIMO systems.

Abstract

In this paper, relaxed belief propagation (RBP) based detectors are proposed for multiple-input multiple-out (MIMO) system. The factor graph is leveraged to represent the MIMO channels, and based on which our algorithms are developed. Unlike the existing complicated standard belief propagation (SBP) detector that considers all the edges of the factor graph when updating messages, the proposed RBP focuses on partial edges, which largely reduces computational complexity. In particular, relax degree is introduced in to determine how many edges to be selected, whereby RBP is a generalized edge selection based BP method and SBP is a special case of RBP having the smallest relax degree. Moreover, we propose a novel Gaussian approximation with feedback information mechanism to enable the proposed RBP detector. In order to further improve the detection performance, we also propose to cascade a minimum mean square error (MMSE) detector before the RBP detector, from which pseudo priori information is judiciously exploited. Convergence and complexity analyses, along with the numerical simulation results, verify that the proposed RBP outperform other BP methods having the similar complexity, and the MMSE cascaded RBP even outperform SBP at the largest relax degree in large MIMO systems.