Low-complexity Lattice Reduction Aided Detection for Generalised Spatial Modulation

Generalised spatial modulation (GSM) was first introduced with the maximum-likelihood (ML) optimum decoder. However, ML decoder may be infeasible for practical implementation due to its exponential complexity especially when the number of antennas or the constellation size is large. Lattice reduction (LR) aided linear decoders are known to have much lower complexity while achieving near-optimal bit-error-rate (BER) performance in MIMO V-BLAST systems. In this paper, LR-aided linear decoders are applied to GSM systems for the first time, but the simulation results demonstrate unsatisfactory BER performances. Thereby, two improved LR-aided linear decoders are proposed in this work. The proposed schemes achieve significant BER performance enhancement compared to that of conventional LR-aided linear decoders as well as linear decoders including zero forcing (ZF) detection and minimum mean square error (MMSE) detection. Compared to the ML decoder, the proposed schemes can provide fairly lower complexities with small BER performance degradation.