Block-Wise MIMO Detection with Near-Optimal Performance and Low Complexity

Abstract

A block-wise MIMO detection scheme which reduces the search space greatly comparing with the ML detection but still achieves the near-optimal performance is proposed. The specific feature of the proposed detection scheme is that, the search space is reduced as much as possible before performing the exhaustive search, in other words, the required search space that is large enough to guarantee the acceptable performance is obtained in a search space pre-determination stage. Based on the output of ZF equalization, the metrics used for determining the reduced search space, which are the posterior probabilities in this paper, can be computed. According to the simulation results, the near-optimal performance can be obtained while the detection complexity is far less than that of the ideal ML detection.