TrellisBsP: Trellis-Based Belief-Selective Propagation for MIMO Detection

Abstract

Compared to its linear counterparts, the belief propagation (BP) MIMO detector achieves better performance in challenging MIMO scenarios. For practical applications, the recently emerged belief-selective propagation (BsP) detector mitigates the error floor issue and reduces the complexity. However, it still suffers from intractable complexity for high-order MIMO scenarios, and the serial message-updating mechanism limits its hardware feasibility. In this letter, a trellis-based BsP (TrellisBsP) detection algorithm is proposed. Two unexplored approaches: trellis representation and pathfinding strategy are utilized to enable parallel message-updating while achieving complexity reduction. Benchmarking with the BsP detection, the proposed TrellisBsP saves up to 72.76% complexity in an $8 \times 4$ , 256-QAM MIMO system without error performance penalty. In addition, the proposed TrellisBsP has great potential in hardware implementation due to its parallel message-updating mechanism.