Dissertation Defence: Covert Cyberattack Detection and Isolation in Interconnected Systems

Saeed Ansari Rad, supervised by Dr. Ahmad Al-Dabbagh, will defend their dissertation titled “Covert Cyberattack Detection and Isolation in Interconnected Systems” in partial fulfillment of the requirements for the degree of Doctor of Philosophy in Electrical Engineering.

An abstract for Saeed Ansari Rad’s dissertation is included below.

Examinations are open to all members of the campus community as well as the general public. Please email ahmad.aldabbagh@ubc.ca to receive the Zoom link for this exam.

Abstract

Interconnected systems consist of multiple subsystems coupled through physical and cyber interconnections to achieve coordinated objectives. While such interconnectivity enhances scalability and performance, it also increases vulnerability to cyber threats. Among these threats, a covert cyberattack is particularly challenging because it manipulates actuator commands to achieve a malicious objective while simultaneously manipulating sensor measurements to mimic nominal behaviour of the attacked system/subsystems and remain undetected. This thesis develops a set of schemes for detection and isolation of covert cyberattacks in interconnected systems, with emphasis on practical challenges arising from high interconnectivity, reliance on communicated information, and uncertainties on system matrices, including disturbances and perturbations.

First, a modelling framework is developed for highly interconnected systems subject to covert cyberattacks and disturbances, and a detection and isolation scheme is proposed based on a combination of decentralized and distributed dynamic observers for state estimation. Detectability analyses are provided, including scenarios involving multiple simultaneous cyberattacks, together with an approach for selecting a reduced number of observers to address computational costs. Next, the framework is extended to a data-driven setting by designing decentralized and distributed data-driven observers that enable detection and isolation when exact system matrices are difficult to obtain, along with suitable design conditions and mechanisms to adjust the number of deployed observers.

Then, detection with control objectives are integrated. A robust decentralized control scheme with a decentralized state-disturbance observer is incorporated to enable local detection within each subsystem in the presence of disturbances and perturbations, thereby reducing dependence on neighbouring subsystems and communicated information. Finally, an integrated control and monitoring scheme is developed using decentralized control with dynamic observers to support both local and neighbourhood-based detection while ensuring closed-loop stability and robust output regulation under uncertainties on the system matrices.