Semantic-bit coexistence transmission for securing ISASC system with RIS-assisted symbiotic radio

Abstract

The security of the integrated sensing and semantic communication (ISASC) system based on traditional encryption and decryption algorithms heavily relies on the absolute accuracy of transmitted ciphertext bits, which offsets the robustness and fault tolerance advantages of semantic communication. To tackle this issue, we propose a semantic-bit coexistence transmission framework with reconfigurable intelligent surface (RIS)-assisted symbiotic radio (SR) for physical layer security (PLS) enhancement. As the eavesdropper merely has the semantic-oriented communication structure, RIS can modulate its generated bit signal into the incident semantic signal to destroy intelligibility of the semantic data via reflect beamforming. Specifically, the superimposed signal produces a mutualistic symbiotic relationship for the intended legitimate users while inducing a parasitic interaction for the eavesdropper. Based on the constructed system model, the semantic secrecy rate (SSR) maximization problem is formulated while meeting minimum quality of service (QoS) requirements for both bit data decoding and target sensing. However, the problem faces significant challenges due to its inherent non-convexity and the strong coupling relationship between multiple variables. To further improve the PLS of the ISASC system, we develop a block coordinate descent (BCD) optimization structure by employing the introduction of slack variables, sequential convex approximation (SCA) and semidefinite relaxation (SDR) techniques, where the decomposed subproblems can be solved by deriving closed-form solutions or equivalent convex formulations. Monte Carlo simulation results show that the proposed approach achieves consistent SSR gains over the considered benchmark schemes.