This paper analyzes the stochastic security performance of a multiple-input multiple-output (MIMO) integrated sensing and communication (ISAC) system in a downlink scenario. A base station (BS) transmits a multi-functional signal to simultaneously communicate with a user, sense a target's angular location, and counteract eavesdropping threats. The attack model considers a passive single-antenna communication eavesdropper intercepting communication data, as well as a multi-antenna sensing eavesdropper attempting to infer the target's location. We also consider a malicious target scenario where the target plays the role of the communication eavesdropper. The BS-user and BS-eavesdroppers channels follow Rayleigh fading, while the target's azimuth angle is uniformly distributed. To evaluate the performance in this random network, we derive the ergodic secrecy rate (ESR) and the ergodic Cramer-Rao lower bound (CRB), for target localization, at both the BS and the sensing eavesdropper. This involves computing the probability density functions (PDFs) of the signal-to-noise ratio (SNR) and CRB, leveraging the central limit theorem for tractability. We characterize the boundary of the CRB-secrecy rate region, and interpret the performance tradeoffs between communication and sensing while guaranteeing a level of security and privacy in the random ISAC networks.
Soltani et al. (Thu,) studied this question.