Generalized Spatial Modulation in Highly Correlated Channels

Abstract

Generalized spatial modulation (GSM) is a promising technique that can highly increase the spectral efficiency for ultra-high data rate systems. However, its performance degrades in highly correlated channels such as those in the millimeter wave (mmWave) and sub-Terahertz (sub-THz) bands. GSM conveys information by the index of the activated transmit antenna combination (TAC) and by the M-ary symbols. In conventional GSM, the legitimate TACs are randomly selected, where their number should be a power of 2. In this paper, a simplified EGSM (S-EGSM) based on TAC selection but without instantaneous channel side information (CSI) is proposed for highly correlated channels. Moreover, an efficient Index-to-Bit mapping for spatial bits based on Gray coding is proposed to reduce the spatial bit-error rate (BER) instead of using the normal binary mapping as in conventional GSM. Simulation results show that the proposed TAC selection without instantaneous CSI (S-EGSM) outperforms the existing method by 1.4 dB in highly correlated channels. In addition, the proposed S-EGSM when compared to TAC selection with instantaneous CSI can be considered as a good tradeoff between performance and complexity since it reduces the feedback and real-time computation overhead due to instantaneous TAC selection. Finally, simulation results of gray coding for spatial bits show a performance gain of the order of 1-2 dB in highly correlated channels, and which becomes less significant in case of low spatially correlated channels.