Distributed Resilient Control of Cyber-Physical Multi-Agent Systems Under Attacks

This paper delves into the distributed resilient state estimation-based secure control in multi-agent systems under false-data injection attacks. Firstly, we propose a novel adaptive distributed output observer approach that is based on resilient outputs. The key feature of our proposed observer is its sole dependence on the coefficients of the minimal polynomial of the leader's system matrix, rather than on individual entries of the leader's system. Subsequently, to identify the target attacker, we assign both trust and confidence values to each agent and its neighbors, thereby enabling attack detection and localization. Furthermore, we propose a distributed finite-time resilient control strategy, grounded in trust and confidence, to safeguard against attackers while ensuring secure output estimation and maintaining a bounded upper limit for finite-time convergence. Additionally, we introduce a piecewise function-based approximation method to achieve a less conservative upper bound for the convergence time, compared to traditional methods. Finally, we provide a numerical example to demonstrate the effectiveness of our proposed theoretical analysis.