Combating Transmit Antenna and Channel Correlations in Spatial Modulation Using Signature Constellations

Spatial modulation (SM) has a strong sensitivity to transmit antenna and channel correlations, because some of the information bits are assigned to active antenna selection, and the correlation limits the detection reliability of these bits. Recent approaches for the solution of this problem rely on either unequal error protection (UEP) of antenna and symbol bits with the addition of a channel encoder/decoder pair to the transceiver or precoding in the form of antenna-dependent rotation (or joint rotation and amplitude scaling) of the signal constellation. The UEP approaches have been shown to offer only limited efficiency in compensating for the adverse channel effects while increasing the latency and complexity due to the addition of the encoder/decoder. The precoding based approaches achieve good results for BPSK and QPSK signals, but the performance quickly degrades for higher-level QAM signal constellations. Also, the complexity of the precoder optimization problem increases with the number of transmit antennas and the modulation order, making this approach not very practical to use for large spectral efficiencies. This paper introduces a novel approach to this problem whose performance is independent of the modulation order. The key idea is to use signature constellations for different transmit antennas with an inter-constellation minimum Euclidean distance that is independent of the modulation order. The theoretical analysis and the simulation results show that compared to previous methods the new approach gives significant performance improvements in terms of robustness to transmit antenna correlation, particularly for Rician fading channels.