Reduced complexity maximum likelihood detection for V-BLAST systems

Proposed and studied in this work is a reduced complexity maximum likelihood (ML) detection scheme, which uses a pre-decoder guided local exhaustive search, for V-BLAST systems. A polygon searching algorithm and an ordered successive interference cancellation (O-SIC) sphere searching algorithm are major components of the proposed multistep ML detectors. The effects of pre-decoder's performance on the total decoding complexity was studied by simulation. The complexity of the proposed algorithm depends on the operating signal to noise ratio (SNR) of the system. At reasonable high SNRs, its complexity is comparable to that of O-SIC algorithm, a commonly used decoding algorithm for V-BLAST. For V-BLAST systems using a large number of transmit antennas, we also propose a new detection algorithm termed ordered group-wise interference cancellation (O-GIC) to further reduce decoding complexity. The O-GIC based detection scheme is not a ML detection scheme, however, its performance is shown to outperform that of the O-SIC.