Optimal Allocation of False Data Injection Attacks for Networked Control Systems With Two Communication Channels

Abstract

Recently, the security issues in networked control systems (NCSs) have become a major challenge, since NCSs are vulnerable to malicious attacks. In this article, we consider the false data injection (FDI) attacks employed on two communication channels—namely: 1) sensor-to-controller and 2) controller-to-actuator channels. The FDI attacker aims to disturb the systems by injecting external control inputs and fake sensor measurements in the form of Gaussian noises. The problems of attack allocation with resource constraints are studied, i.e., the attacker decides instants of attack occurrence as well as the allocation of attack energy in order to obtain the optimal attack strategy. Also, the optimal attack strategies with respect to both time-invariant and time-varying attack covariances are proposed for two channels, where the time-varying attack covariances suffer a given budget. Finally, the feasibility of theoretical results is validated by a numerical example of an unmanned aerial vehicle.