QR-decomposed generalized belief propagation with smart message reduction for low-complexity MIMO signal detection
Akihide David Shigyo, Koji Ishibashi · 2017
In this paper, we propose a QR-decomposed generalized belief propagation (QR-GBP) with smart message reduction (SMR) for low-complexity multi-input multi-output (MIMO) signal detection. QR-GBP requires exponentially high complexity to update every messages while it enables to calculate exact marginal probabilities on a valid region graph consisting of sets of multiple nodes in a factor graph defined by MIMO channels and provides the superior performance close to maximum-likelihood detection (MLD). In this paper, we propose a novel complexity reduction method called SMR, which exploits the structure of the region graph after QR-decomposition. Numerical results show QR-GBP with SMR can achieve near-optimal performance close to MLD as same as QR-GBP while its complexity is significantly low.