On the Role of Information Sharing in the Security of Interconnected Systems

In this paper we consider a security problem for interconnected dynamical systems, where each subsystem aims to detect local attacks. Each subsystem has knowledge of only its local dynamics and, therefore, the subsystems share information among themselves to aid local attack detection. We develop a technique that processes the local and shared measurements and detects attacks with provable guarantees. Interestingly, we show that for some instances of the attack, the detection performance deteriorates if the subsystems share their measurements. We provide an explanation for this counter-intuitive behavior and illustrate our results through a numerical example.