Monitors under attack: Preliminary results on robustness analysis and synthesis

The problem of cyber attacks in the context of supervisory control layer is considered in this paper. Precisely monitor based supervisors, largely used to enforce generalized mutual exclusion constraints, are focused on. An attack model to capture a class of deception attacks, where the attacker has the ability to hijack a subset of sensor readings and mislead the supervisor, is proposed in this paper. The model reveals that a monitor supervised system under attack can become not live and can drive the plant to forbidden states. To make a monitor robust against attacks a restriction to its input and output transitions is considered.