Two-stage constellation partition algorithm for reduced-complexity multiple-input multiple-output-maximum-likelihood detection systems

This study presents the analysis of a constellation partition (CP) algorithm for multiple-input multiple-output-maximum-likelihood detection (MIMO-MLD) systems. The authors consider an N_t by N_r MIMO system, where MLD algorithm is employed at the receiver side for MIMO signal detection. The authors show that for the case of orthogonal space-time block codes and MIMO beamforming (MIMO-BF) systems, the proposed CP algorithm achieves the same error-rate performance as the optimum MLD algorithm while cutting back significantly on the complexity of MIMO-MLD scheme. Thus complexity reduction is not achieved at the expense of the optimum MLD performance in these cases, unlike the existing simplified MLD schemes in the literature. For the case of MIMO-multiplexing system, however, it is observed that the complexity reduction is achieved at the expense of the optimum MLD performance, similar to the existing simplified MLD schemes. Thus the proposed algorithm is especially attractive in real-time implementation of space-time block code and MIMO-BF systems.