Transmit Antenna Subset Selection in Spatial Modulation Relying on a Realistic Error-Infested Feedback Channel

In this paper, we study the performance of spatial modulation (SM) employing Euclidean distance-based antenna selection (EDAS) operating in a realistic error-infested feedback channel, which has hitherto only been studied under ideal feedback channel conditions. Specifically, we model the feedback channel by a bit-flip probability δ and study its impact on the forward link employing EDAS. We show that the erroneous feedback channel severely degrades the performance of EDAS-aided SM (EDAS-SM) system by imposing an error floor in the forward link. Furthermore, we quantify the error floors associated both with the spatial and with the conventional symbols with the aid of asymptotic symbol error rate analysis. The expressions derived for the error floors in the forward link are utilized for optimizing the feedback signaling, which are shown to help reduce the error floor levels. Furthermore, a pilot-aided selection verification (PSV) algorithm is proposed for mitigating the effects of antenna-set mismatch between the transmitter and the receiver, which eliminates the error floor in the forward link. Simulations are conducted in order to validate the theoretical results presented in this paper. Furthermore, the bit-error ratio (BER) performance of the EDAS-SM is compared with that of the conventional antenna selection (C-AS) both in the PSV as well as in the no selection verification scenarios. It is observed that EDAS-SM outperforms C-AS in both the scenarios considered. Specifically, at a BER of 10
^-5
, EDAS-SM is observed to give a 3-dB signal-to-noise ratio gain compared with the C-AS, when operating at a spectral efficiency of 7 bits per channel use in the face of a feedback BER of δ = 0.05.