Compressive-Sensing-Based Multiuser Detector for the Large-Scale SM-MIMO Uplink

Conventional spatial modulation (SM) is typically considered for transmission in the downlink of small-scale multiple-input- multiple-output (MIMO) systems, where a single antenna element (AE) of a set of, e.g., 2p AEs is activated for implicitly conveying p bits. By contrast, inspired by the compelling benefits of large-scale MIMO (LS-MIMO) systems, here, we propose an LS-SM-MIMO scheme for the uplink (UL), where each user having multiple AEs but only a single radio frequency (RF) chain invokes SM for increasing the UL throughput. At the same time, by relying on hundreds of AEs and a small number of RF chains, the base station (BS) can simultaneously serve multiple users while reducing the power consumption. Due to the large number of AEs of the UL users and the comparably small number of RF chains at the BS, the UL multiuser signal detection becomes a challenging large-scale underdetermined problem. To solve this problem, we propose a Joint SM transmission scheme and a carefully designed structured compressive sensing (SCS)-based multiuser detector (MUD) to be used at the users and the BS, respectively. Additionally, the cyclic-prefix single carrier (CPSC) is used to combat the multipath channels, and a simple receive AE selection is used for the improved performance over correlated Rayleigh-fading MIMO channels. We demonstrate that the aggregate SM signal consisting of multiple UL users' SM signals of a CPSC block exhibits distributed sparsity. Moreover, due to the Joint SM transmission scheme, aggregate SM signals in the same transmission group exhibit group sparsity. By exploiting these intrinsically sparse features, the proposed SCS-based MUD can reliably detect the resultant SM signals with low complexity. Simulation results demonstrate that the proposed SCS-based MUD achieves a better signal detection performance than its counterparts even with higher UL throughput.