Incomplete Information-Based Resilient Strategy Design for Cyber-Physical Systems Under Stochastic Communication Protocol

Abstract

In this article, a novel resilient control strategy is proposed for a jammed cyber-physical system (CPS) with multiple transmission channels. In the presence of jamming attacks, the control performance of the time-varying system with nonlinear disturbances degrades due to the missing of the control command. To reduce the probability of being attacked, a novel multichannel transmission mechanism under a stochastic protocol is constructed, which can protect the information of which channel is selected from being known by the opponent. An incomplete information-based Bayesian game is introduced to describe the interaction of the control command transmitter and jamming attacker by means of taking both the control performance and energy consumption into consideration. Solutions for the dynamic Bayesian game are explored with the Q-learning method. The presented results for the jammed CPS are demonstrated via a helicopter system and a wheeled mobile robot platform.