Detection and Prevention of Cyber-Attacks in Networked Control Systems

Abstract

Abstract This paper addresses the problem of detection and prevention of cyber attacks in discrete event systems where the supervisor communicates with the plant via network channels. Random control delays may occur in such networked systems, hence the control of the supervisor could be affected. Furthermore, there is an attacker targeting the vulnerable actuators. The attacker can corrupt the control input generated by the supervisor, and aims at driving the plant to unsafe states. We propose a new approach to model the closed-loop system subject to control delays and attacks. The notion of AE-safe controllability in the networked control system is defined: it describes the ability to prevent the plant from reaching unsafe states after attacks are detected. A method for testing AE-safe controllability is also presented.