Transmit Antenna Subset Selection for Single and Multiuser Spatial Modulation Systems Operating in Frequency Selective Channels

Transmit antenna (TA) subset selection (TAS) is a well known multiple-input multiple-output technique that exploits the channel state information (CSI) at the transmitter in order to improve the attainable bit error rate (BER) performance. The extensive study of TAS in the context of spatial modulation (SM) has recently revealed that a significant performance gains are attainable compared to SM systems without TAS. However, the existing TAS techniques conceived for SM were studied by considering a frequency-flat channel, which does not represent the practical channels which are frequencyselective. In this paper, we address this hitherto-not-addressed problem by studying the TAS schemes for zero-padded singlecarrier (ZP-SC) SM systems. Specifically, we employ the partial interference cancellation receiver with SIC in order to convert the frequency-selective channel into parallel sub-channels and invoke Euclidean distance based antenna subset selection (EDAS) over each of the sub-channels. This SIC aided TAS algorithm is termed as SIC-TAS. Furthermore, we show using theoretical analysis that the parallel sub-channels thus obtained are nearly identical, which enables us to employ a majority logic to obtain a single TA subset to be used in all the sub-channels. The majority logic based TAS scheme (MAJ-TAS) reduces the feedback overhead to that of frequency-flat scenario as it requires a single TA subset to be used over all the sub-channels. Furthermore, the computational burden of MAJ-TAS is further reduced by restricting the number of sub-channels over which the EDAS is invoked. This reduced complexity TAS scheme is termed as L-MAJ-TAS scheme, where L represents the number of sub-channels over which the EDAS is invoked. Furthermore, the proposed TAS schemes are extended to the multi-user scenario. All the theoretical insights are validated using simulation results. Furthermore, it is observed through numerical simulations that the proposed TAS schemes provide a significant BER performance improvement when compared to the systems without TAS. Specifically, a signal-to-noise ratio (SNR) gain as high as 3dB is observed in single user scenario and of about 1dB in case of two-user scenario while employing TAS.