Optimal stealthy attack against stochastic event-based scheduling for remote state estimation in cyber-physical systems

This article concentrates on security issues of stochastic event-based (SEB) sensor scheduling in cyber-physical systems (CPSs). An SEB attack mechanism is proposed to compromise remote state estimation, in which the attacker can jam the feedback channel and inject false acknowledgments. Based on the Gaussian assumption, the minimum mean square error estimator under the SEB attack is obtained utilizing Bayes' theorem. In this article, the SEB attack stealthiness is composed of communication ratio invariance and anomaly detection invisibility with the forward channel. Furthermore, a double stealthy attack condition is given to complete persistent attacks from the perspective of the attacker. Aiming at maximizing average error covariances on the remote estimator, we design a generation algorithm of optimal SEB stealthy attack sequence and prove its optimality by rearrangement inequality and mathematical induction. Finally, the scalar and networked dc motor vector system simulations illustrate the corresponding theoretical results.