An iterative space-time detector managing the performance-complexity trade-off

For space-time bit-interleaved coded modulation (ST-BICM) systems, applying a turbo-reception between the space-time (ST) detector and the error-correcting decoder leads to the best performance. However, this receiving technique requires the computation of high quality log-likelihood ratios (LLR). The computation complexity of optimal a posteriori probabilities (APP) by the ST detector is exponential in the ST system dimensions. We propose in this paper a detector that significantly reduces this complexity, while guaranteeing the good quality of LLR. Representing the multiple-input multiple-output (MIMO) system as a tree, the detection consists in searching for the best candidates through the tree by maximizing a metric. The LLR on decoded bits are computed over this list of best candidates. The use of a constant bias value in the cost function accelerates this search by favouring the deeper points of the tree.