Nonasymptotic Performance Limits of Low-Latency Secure Integrated Sensing and Communication Systems

This paper considers an information theoretic model for secure integrated sensing and communication (ISAC) with the goal of establishing fundamental limits in low-latency scenarios. In this secure ISAC model, a message is transmitted through a state-dependent wiretap channel with decoder-side state availability. The model is studied under a strong secrecy constraint when only a part of the transmitted message should be kept secret. First, the secrecy-distortion rate region is established for a degraded channel by treating the model as a special case of a feed-backed secure ISAC model. Finite-length inner bounds are then proved by applying nonasymptotic random binning techniques. Bounds on the rates have a similar form to common finite-length bounds, and the distortion bound follows from a bound for letter-typical sequences.