Intelligent Semantic Communication Networks Empowered by Non-orthogonal Multiple Access

Intelligent semantic communication has appeared as an emerging paradigm for information exchange that improves communication efficacy across multiple fields. In this study, we consider the non-orthogonal multiple access technique in the transmission links from the base station to users to enhance the performance of the intelligent semantic communication system. Here, we examine a system delay minimization problem for extracting, transmitting, and reconstructing the transmitted data in an intelligent semantic communication system. To do this, we design a deep reinforcement learning framework to optimize the computation resource allocation and beamforming matrix at the base station. Accordingly, we demonstrate the convergence of the proposed framework and prove its efficiency through numerical results.