Approximating maximum likelihood performance reduced dimension VBLAST detection algorithm

Abstract

SM (spatial multiplexing) can effectively increase the information rate in multiple input multiple output system. BLAST is the typical representation of SM, especially VBLAST, which has some simple detection algorithms such as ML, ZF-DFE and ML-DFE, etc. However, the existing algorithms cannot approach ML performance. This paper discusses the effect for detection performance by the correlation of channel matrix, proposes a new algorithm—HPML detection algorithm, which can approach ML performance with low complexity. In the new algorithm, we travel the first d layers, and use the DFE procedure for the remaining layers, then perform ML detection for all obtained signals. Simulation results show that HPML can approach ML performance when the traveling numbers are not less than half of the number of transmitting antennas, and the algorithm complexity is smaller than ML.