Active STAR-RIS-Enabled ISAC Networks Against Simultaneous Eavesdropping and Detection Attacks

Abstract

Reconfigurable intelligence surface (RIS) enabled integrated sensing and communication (ISAC) is vulnerable to different hostile attacks, especially the eavesdropping and detection attacks. Unlike the prior studies that tackle these attacks separately, this article investigates the use of an active simultaneously transmitting and reflecting RIS (STAR-RIS) to establish a unified security paradigm for an ISAC network against simultaneous eavesdropping and detection attacks. Thereinto, with the aid of active STAR-RIS, a multiantenna base station (BS) concurrently senses a point-like target and communicates with multiple secrecy and covert users (SUs and CUs) in the full-space. Particularly, target acts as an eavesdropper for wiretapping the transmitted data from BS to SUs, while Willie acts as an extra warden for detecting the existence of the wireless transmission from BS to CUs. To address these attacks, we first explore a joint physical layer secrecy and covert communication strategy. Concretely, BS continuously transmits the secrecy signals to SUs in all time slots, and opportunistically propagates the covert signals to CUs in the selected ones. On this basis, BS exploits the differences of fading channels to limit the amount of SUs’ information wiretapped by target, and uses the Gaussian signaling as an uncertainty to hide CUs’ communication behaviors from Willie. Obeying this, we utilize the Pinsker’s inequality and the large system analytical method to obtain lower bounds on the detection error probability (DEP) of Willie. Furthermore, to effectively balance radar sensing and secure communication, the transmit beamforming and receive filter at BS, and the transmission&reflection beamforming at active STAR-RIS are jointly designed to maximize the average sum of the minimum secrecy and covert rates, while meeting given constraints on the sensing signal-to-interference-plus-noise ratio at BS, the tolerable DEP against Willie, and the transmit power at both BS and active STAR-RIS. To solve this highly nonconvex problem, we develop an efficient overall iterative algorithm by invoking the successive convex approximation, semi-definite relaxation, and sequential rank-one constraint relaxation methods. Finally, numerical results verify the superiority of active STAR-RIS in alleviating the multiplicative fading, and manifest that driven by the proposed algorithm, the studied network greatly outperforms baseline schemes in defeating hybrid attacks.

Topics & Concepts