Detecting Generalized Replay Attacks via Time-Varying Dynamic Watermarking

Cyber-physical systems (CPS) often rely on external communication for supervisory control or sensing. Unfortunately, these communications render the system vulnerable to cyber attacks. Attacks that alter messages, such as replay attacks record measurement signals, and then play them back can cause devastating effects. Dynamic Watermarking methods, which inject a private excitation into inputs secure resulting have begun addressing challenges of detecting attacks, but been restricted linear time-invariant (LTI) systems. Though LTI models are sufficient some applications, other CPS, autonomous vehicles, require more complex models. This article develops time-varying (LTV) extension previous dynamic watermarking methods by designing matrix normalization factor accommodate temporal changes in system. Implementable tests provided with considerations real-world The proposed method is shown be able detect generalized both theory simulation using an LTV vehicle model.