Approximation maximum likelihood detector for V-BLAST systems

An approximation maximum likelihood detector for vertical Bell Laboratories layered space-time (V-BLAST) systems is present in this paper, which is based on multistage processing. A classical sequential detector (OSIC), zero forcing, or low complexity square root algorithm is adopted to get the initial estimation of the transmitted data, after that, we utilize a novel iterative search algorithm to further refine the result. This scheme first determines the position of some important bits in the estimated data that significantly affect the mean square error (MSE), then a maximum likelihood search is performed at these finite positions. This procedure is iteratively performed for approximation maximum likelihood (ML) detection with a few bits changed in each cycle. Numerical results shows that the proposed scheme achieves either practically optimal performance or a performance only a fraction of a decibel away from the optimal maximum-likelihood detector with a significant reduction in the complexity, compared with the Viterbi algorithm based on the full trellis diagram.