On the Capacity and Performance of Generalized Spatial Modulation

Abstract

Generalized spatial modulation (GSM) uses N antenna elements but fewer radio frequency (RF) chains (R) at the transmitter. Spatial modulation and spatial multiplexing are special cases of GSM with R = 1 and R = N , respectively. In GSM, apart from conveying information bits through R modulation symbols, information bits are also conveyed through the indices of the R active transmit antennas. In this paper, we derive lower and upper bounds on the the capacity of a (N, M, R)-GSM MIMO system, where M is the number of receive antennas. Further, we propose a computationally efficient GSM encoding (i.e., bits-to-signal mapping) method and a message passing based low-complexity detection algorithm suited for large-scale GSM-MIMO systems.

Compressed-Sensing Assisted Spatial Multiplexing Aided Spatial Modulation

“…As observed from Fig. 8 (b), the SMx-SM system with the proposed MP detector outperforms the C-GSM system with Layer MP (LaMP) detector in [37] by 2.5 dB at BER=10 −5 . Fig.…”

“…Although, the authors of [37] proposed a simplified mapping method, it remains a challenge to estimate the TAC index at the receiver.…”

“…Additionally, the achievable rate and capacity of GSM system are analyzed in [36] [37]. However, for largescale GSM system having a large number of activated TAs, the cardinality of activated Transmit Antenna Combination (TAC) set is large.…”

“…In the conventional bit-to-symbol mapping of the GSM system, a specific TAC index is selected from 2 66 at the transmitter, which is extremely difficult. Although a low-complexity bit-to symbol mapping method is proposed in [37], it still a challenge to demap the estimated index into 66 bits at the receiver, which is impractical in simulations. Additionally, other SM variants have also been investigated in [39]- [43].…”

Neural Network Aided Computation of Mutual Information for Adaptation of Spatial Modulation

“…This calculation is needed, for example, in an adaptive SM system where the transmitter uses this mutual information, obtained and fed back by the receiver, to select the proper MCS. Results requiring numerical integration, for example by means of Monte Carlo simulations, can be found in the literature [10] and [11]. One value of this work is that it explores a radically new approach to solve an essential problem in the practical application of Information Theory: the mutual information of non-conventional modulations is computed by means of Machine Learning (ML) tools.…”