Low-Complex AI-Empowered Receiver for Spatial Media-Based Modulation MIMO Systems

Next-generation communication systems are expected to integrate artificial intelligence (AI) techniques into multi-antenna setups to improve system performance. One of the most important use cases of AI in wireless communications is multiple-input multiple-output (MIMO) systems, due to the superior AI capability of learning the best possible decision-making in complex transceiver structures. In this study, we propose a deep neural network (DNN) receiver structure for spatial media-based modulation (SMBM)-MIMO systems, which detects the symbols in an end-to-end manner. Instead of a conventional two-stage approach, which handles channel estimation and symbol detection separately, the proposed DNN receiver recovers the transmitted symbols directly, utilizing an offline training process. It is demonstrated that while the proposed DNN receiver structure and conventional maximum likelihood (ML) receiver utilizing linear minimum mean-square error (LMMSE)-based channel estimation, perform similarly for a single receive antenna case, DNN is superior for multiple receive antenna cases. We conclude that using DNNs in SMBM with multi-antenna receivers can provide higher performance and thus permits higher data transmission rates in addition to the reduced receiver complexity due to the removing channel estimation process.