Reduced complexity soft detection for BICM MIMO OFDM system

This paper presents a new approach to the soft detection of spatial data streams in MIMO channels in slow fading channels. We focus on high spectral efficiency bit interleaved coded modulation (BICM) MIMO OFDM system where, after serial to parallel conversion and per antenna coding, spatial data streams are simultaneously transmitted by using an antenna array. We propose a detection algorithm basing on the combination of linear and non linear detection techniques. The proposed algorithm orders the spatial streams as per V-BLAST ordering and then uses linear detectors as MMSE or zero forcing (ZF) to detect the streams which have seen good channel realizations and therefore enjoy higher signal-to-noise-ratios (SNRs). These streams after being detected are subsequently stripped off leading to the max log MAP detection of the streams which have seen comparatively poor realizations of the channel and consequently have lower SNRs. Unlike maximum likelihood (ML) detection, whose complexity is huge due to the need for enumerating all possible combinations of the transmitted constellation points, the proposed method has low complexity. The algorithm has a fully parallel structure, suitable for the implementation in parallel hardware. Numerical examples illustrate its performance on slow fading 3 times 3 and 4 times 4 complex MIMO channels.