Robust Stealthy Attacks Based on Uncertain Costs and Labeled Finite Automata With Inputs

This letter deals with the vulnerability analysis of cyber-physical systems subject to malicious actions. For this purpose, considered system is assumed be abstracted as a discrete event system. Labeled finite automata inputs are used model system's behavior along information that circulates in both input and output channels. In particular, we study here stealthy, i.e., undetectable, cyber-attacks aim drive from given normal state set forbidden states. We assume attacker has limited resources, credit, insert delete control sensors events. The proposed evaluates costs such attacks on controlled depending its structure, cost actions possible uncertainties may affect those costs. It provides systematic methods compute minimal robust weakly impacted by uncertainties. A case representing manufacturing plant illustrate results.