Detection and Isolation of Covert Cyber Attacks Using Decentralized and Distributed Dynamic Observers

This paper proposes a detection and isolation scheme for covert cyber attacks in a generalized class of large-scale systems, referred to as a highly interconnected system. Specifically, this paper presents the modelling framework of the system subject to covert cyber attacks and disturbances as well as the design of dynamic observers for state estimation. The proposed scheme relies on two types of observers, namely, decentralized dynamic observer and distributed dynamic observer. Exploiting the combination of the dynamic observers, the paper demonstrates how detection and isolation of attacked subsystems are achieved. Moreover, extensive detectability analyses are provided, while also having multiple attackers and with using a reduced number of observers. The effectiveness and superiority of the proposed scheme are validated through detailed simulations.