Actuator-Enabling Attacks in Discrete-event Systems with Unknown Supervisors
Ziyue Ma, Alessandro Giua, Carla Seatzu · 2025
In this work we study a property of resiliency in discrete event systems modeled by finite state automata. The closed-loop system consists of a plant and a supervisor which enforces a specification. There exists an external attacker who can eavesdrops the output of the system via a mask. The attacker has the knowledge of the exact plant model and knows the existence of the supervisor as well as the exact closed-loop language, but it does not have any knowledge of the supervisor nor the specification the supervisor is enforcing. The aim of the attacker is to let the system violate the specification by performing actuator-enabling attacks. We prove that the existence of such actuator-enabling harmful attacks can be verified by checking the supremal consistent supervisor.