RIS-Assisted MIMO Semantic Communication System for Speech Transmission

Semantic communication represents a paradigm shift from traditional communication systems by focusing on the transmission of meaningful information rather than raw data, leveraging advances in deep learning to interpret and convey semantics effectively. This paper considers point-to-point Multiple-Input Multiple-Output (MIMO) semantic communication systems assisted by reconfigurable intelligent surfaces (RIS), where a transmitter communicates with a receiver through an RIS. We introduce an end-to-end RIS-assisted MIMO communication architecture that jointly optimizes the encoder, precoder, decoder, and passive beamforming for RIS, avoiding the performance loss caused by alternately optimizing each function of the RIS-aided system. The simulation results demonstrate that our proposed system outperforms the traditional communications in both cases in terms of the speech signals metrics, such as Perceptual Evaluation of Speech Quality (PESQ) and Virtual Speech Quality Objective Listener (VISQOL), especially at low signal-to-noise ratio (SNR) regimes. Furthermore, our deep learning-based RIS-assisted MIMO semantic communication system maintains robust performance even when perfect CSI is unavailable.