Game-Guided Matching Theory-Based Resource Allocation for Secure Semantic Communications

Semantic communication (SemCom) has become one of the most promising techniques in breaking the performance bottleneck for sixth-generation wireless networks, but the SemCom performance is easily degraded under malicious jamming and eavesdropping attacks over open wireless links. Therefore, this paper designs a reliable and secure SemCom approach against a hybrid attacker based on resource allocation, with the objective to jointly optimize channel selection and the number of transmitted semantic symbols to maximize secrecy semantic transmission rate (SS-R) under different quality of service requirements. We adopt a hierarchical framework based on Stackelberg game to model the interactions between legitimate users and the hybrid attacker. Furthermore, we model the optimization problem as a many-to-one matching problem with externalities, and propose two swap-based resource allocation algorithms, aiming to maximize the overall SS-R and meet fairness awareness. The two proposed algorithms are able to guide the iteration properly based on the dynamic of attack behaviors, which improves the secure resource allocation efficiency. Simulation results show that the proposed approaches outperform the baselines and benchmarks under different scenarios.