Sum Rate Maximization for Probalistic Semantic Communication via Compression Ratio Optimization

In this letter, the sum rate maximization problem for a downlink probalistic semantic communication (PSC) system is investigated. In the model under consideration, the base station (BS) computes the small-size compressed information through knowledge base in the semantic information extraction process. At the user side, the received compressed semantic information is recovered with the help of shared knowledge base between the BS and the user. It is of importance to optimize the compression ratio in PSC, which is related to the communication efficiency during the information transmission process and also affects the computation efficiency during the semantic information extraction process. A sum rate maximization problem is formulated through optimizing the compression ratio, bandwidth allocation, and transmit power. In order to address this problem, an iterative algorithm is proposed where at each iteration the optimal solutions for compression ratio and bandwidth allocation are obtained in closed form. Simulation results confirm the effectiveness of the proposed algorithm, in comparison to the conventional scheme.