Reduced Complexity Sphere Decoder for MIMO Systems Using Reliability Discrimination Algorithms

The sphere decoding (SD) algorithm has been an important to solve complexity of the maximum likelihood detection (MLD). However, the complexity of the SD is higher than that of V-BLAST in low SNR region. But, the V-BLAST detection its performance suffers degradation because of the error propagation. In this paper, we propose two reliability discrimination (RD) schemes to reduce complexity of SD. Basically, estimate a symbol by V-BLAST in proposed schemes. One is RDED that discriminates the reliability of the estimated symbol by pre-calculated Euclidean distance. The other RDMD that discriminates the reliability by pre-calculated Manhattan distance. Then we can determine whether use SD algorithm or not by pre-calculated distances of RD. Therefore, the computational complexity of proposed schemes can be significantly reduced with acceptable performance degradation.