EXIT-Aided Scheduled Iterative MIMO Detection Under Non-Homogeneous Antenna Propagation Gain Scenarios

The non-homogeneous antenna propagation gain, potentially caused by the diverse large-scale fading effects in wireless communication channels, has a significant impact on the reliabilities of multiple-input multiple-out (MIMO) systems. We propose to utilize extrinsic information transfer (EXIT) to analyze the convergence characteristic of the factor graph (FG) based iterative MIMO detection mechanisms in an effective way. Based on the EXIT analysis, we propose a low-complexity scheduled algorithm for FG-based iterative MIMO detection, which speeds up the convergence of the mutual information exchange between the variable nodes and the observation nodes. To address the complexity issue in 5G new radio (NR) systems, where low-density parity check (LDPC) codes are adopted for data transmission, we also extend the proposed algorithm to concatenated detection and decoding in MIMO-LDPC systems to achieve low complexity. Simulation results show that the convergence speed of MIMO detection can be improved by at most 50%, while for the MIMO-LDPC system, the proposed algorithm can achieve 2 dB gain compared to the conventional minimum mean square error (MMSE) detection mechanism.