False data injection attack resilient distributed exponential sliding mode consensus protocol for discrete...

This paper proposes a False Data Injection (FDI) attack resilient Distributed Exponential Sliding Mode Consensus (DESMC) protocol for Discrete Multi-Agent Systems (DMASs). By explicitly modeling FDI attacks on communication weights, this work extends resilient consensus theory beyond conventional sensor- and actuator-focused approaches, establishing a broader foundation for cyber-physical security to DMASs. A distributed Unknown Input Observer (UIO) is used to detect the FDI attack on the communication link in DMASs. The UIO residual triggers the switching between the sliding surface for normal condition (without FDI attack) to a sliding surface for abnormal condition (with FDI attack). Accordingly, a DESMC protocol is derived using the adaptive sliding surface to tolerate the effects of FDI attack in DMASs. This mechanism isolates malicious influence of FDI attack on communication weights and autonomously reconfigures consensus dynamics without controller redesign. The condition for global consensus stability of DMASs is derived using the Lyapunov function. The proposed FDI resilient DESMC protocol guarantees finite-time convergence, achieves ultra-tight O(T³) quasi-sliding bands, and reduces control effort while preserving global stability under the FDI attack. Simulation and experimental validation on a network of 2-DOF robotic manipulators confirm that the proposed protocol with switching surface strategy ensures reliable consensus, rapid recovery, and robustness against the cyber-physical attack.