Low-complexity nulling-canceling detection algorithm for coded MIMO systems with near-optimal performance

We propose a nulling/canceling algorithm for decoding coded multi-input multi-output (MIMO) transmissions over flat fading channels. The algorithm works with a soft-input soft-output error-control decoder to iteratively detect the information symbols. Our algorithm is based on the BLAST (Bell-Labs Layered Space-Time) nulling/canceling algorithm, with the following differences: i) prior information from the error-control decoder is incorporated, ii) the ordering is based on a posteriori probabilities instead of signal to noise ratio (SNR), and iii) soft (as opposed to hard) interferences computed from a posteriori probabilities are canceled. Based on a previous work on low-complexity nulling/canceling without prior information and with SNR ordering, we present a modified "square-root" algorithm for the proposed nulling canceling detection scheme, with cubic-order complexity. We compared the performance of an iterative detection and decoding scheme based on the proposed MIMO detector and that based on a maximum a posteriori (MAP) detector. In a 4/spl times/4 system, our proposed detector had the same performance as the MAP detector for binary phase-shift keying (BPSK) modulation. For 16-quadrature amplitude modulation, our algorithm came within 1 dB of the performance of an approximated MAP detector, list sphere decoder, at bit error rate of 10/sup -4/.