Principle Analysis and Detection of Zero-dynamics Attacks in Cyber-physical Systems

With the development of cyber-physical systems in modern industry and manufacturing industry, the security of cyber-physical systems (CPSs) is becoming more and more important. Due to the specific structure of zero-dynamics attacks, they are difficult to recognized by the classical data class detector. Therefore, it is of great significance to study zero-dynamics attacks for the development of cyber-physical systems. In this paper, we utilize an autoregressive predictive model combined with Byrnes-Isidori normal form to detect zero-dynamics attacks. First, we discretize the underlying continuous time system by using a zero-order hold, and give a method to solve the relative order of the system. Then the attack model and the optimized attack model are constructed. For the optimized attack model, we utilize autoregressive prediction model to analyze the internal state of the system. Finally, the simulated results show that our model method is effective for detecting zero-dynamics attacks in CPSs.