Partial MMSE-ML detection for coded MIMO systems

We propose computationally efficient MIMO detection scheme which can provide accurate soft decision information to the decoder. The detection process has two stages. The proposed scheme first detects MIMO symbols using MMSE method, and calculates noise variances of each detected symbol. Next, the detected symbols are demodulated in bits by maximum likelihood method respectively. The proposed scheme greatly reduces the complexity due to its computational algorithm compared to the maximum likelihood MIMO detection which exhaustively searches for the soft decision information.