Predictive-Deception-Attack-Detection-Based secure control for switched systems

Abstract

Deception attacks in switched systems can manipulate both the switching signal and system state, inducing severe asynchronous switching. While existing research predominantly employs passive mitigation strategies, this paper proposes a proactive secure control framework integrating prediction into attack detection. A model predictive control-based attack detection mechanism, augmented by a mode-state predictor, detects attacks by comparing predicted and received system modes and states. A detection-aware event-triggered mechanism and a mode-based try-once-discard protocol are implemented to reduce asynchrony between the subsystem and the controller. Additionally, an optimal subsystem switching rule, derived from the optimal states, ensures stability and security during switching. Asymptotic stability of the closed-loop system is analytically verified, and its feasibility is validated through an unmanned surface vehicle case study.