A zonotopic set-invariance analysis of replay attacks affecting the supervisory layer

This paper presents a zonotopic set-invariance analysis of replay attacks affecting the communication network that serves the supervisory layer of complex control systems using an observer-based detection scheme. Depending on the attacker's access to the system's resources, two scenarios are considered: (I) Sensors and controller data are counterfeited; (II) Only sensor measurements are counterfeited. The effect of a physical attack against the plant during the data replay is also taken into consideration. The representation of invariant sets as zonotopes allows to derive analytical expressions for attack detectability under the presence of bounded uncertainties. The validity of the analysis is demonstrated through simulations using a quadruple-tank process