Building upon this performance analysis, the thesis further investigates physical layer security in JCAS systems, accounting for both communication eavesdropping and sensing information leakage, including the scenario of a malicious target. To secure the considered random JCAS system, two novel closed-form beamforming schemes suitable for large antenna arrays are proposed. The first scheme combines a sensing-optimal precoder with a secure communication-optimal precoder. The second scheme integrates a secure sensing-optimal precoder—derived via a dedicated optimization framework—with the secure communication-optimal precoder. The proposed analytical designs avoid complex non-convex optimization and provide practical and scalable solutions for achieving joint communication security and sensing privacy.

Integrated Sensing and Communications (ISAC), which unifies radar sensing and wireless communication within a common framework, has emerged as a key enabling technology for next-generation wireless networks. By allowing the shared use of spectrum, hardware, and energy resources, ISAC enhances both sensing and communication functionalities. However, the intrinsic trade-off between sensing and communication objectives, together with emerging security and privacy challenges, introduces fundamental performance limitations. While ISAC represents a broad research paradigm, this thesis focuses on a joint communication and sensing (JCAS) framework in which a unified downlink MIMO transmit signal simultaneously supports communication and radar sensing.

The primary objective of this thesis is to characterize the fundamental S \& C trade-off in a downlink MIMO JCAS system with randomly located users and targets. This trade-off is quantified by deriving the complete Cramér–Rao bound (CRB)–rate region for target angle estimation and identifying its Pareto boundary. Exact analytical expressions are obtained for key performance metrics, including the ergodic communication rate, communication outage probability, sensing outage probability, and ergodic CRB, under both perfect and imperfect channel state information. Notably, the results demonstrate that the ergodic CRB provides a tighter performance characterization than the Bayesian CRB in the considered setting.