Safety-Mode Scheduling-Based Active Security Control Under DoS Attacks With Imperfect Detection

Abstract

This article investigates an active security $H_\infty$ control problem for networked control systems (NCSs) subject to denial-of-service (DoS) attacks with imperfect detection. The DoS process is modeled as a two-mode Markov chain to characterize its stochastic activation, persistence, and recovery without requiring deterministic bounds on the attack duration or occurrence frequency. To account for detection uncertainty, a hidden Markov model (HMM) is employed to describe the probabilistic mismatch between the true DoS mode and the detected mode. An active security control framework based on safety-mode scheduling is developed, under which the plant operating configuration and controller gains are reconfigured according to the detected mode. Under a common quadratic metric, the safety mode is designed to exhibit a strictly smaller one-step open-loop state-deviation growth factor than the nominal mode, thereby reducing the tendency of attack-induced state deviations. A mode-dependent Lyapunov analysis is then conducted to derive sufficient conditions guaranteeing asymptotic mean-square stability and a prescribed $H_\infty$ disturbance attenuation level for the resulting closed-loop system. Simulation results based on a networked permanent magnet synchronous motor drive system demonstrate the effectiveness of the proposed method.