Near-ML soft-MIMO detector with reduced complexity

This paper proposes a soft-detector with near-maximum likelihood (ML) performance in multiple-input multiple-output (MIMO) systems. The proposed detector performs an initial detection for candidate vectors by applying a low complexity channel matrix ordering. The detection order then is diversified to extend the list of candidates, from which high-quality soft-output can be generated. In addition, a method for reducing overhead from the diversification of the detection order is presented. Simulation results on a 4×4 system with a convolutional turbo code of 5/6 rate show that the proposed detector can approximate the performance of the soft-ML detector but its complexity is approximately 46% of a reference detector [5].