Attack Analysis and Resilient Control Design for Discrete-Time Distributed Multi-Agent Systems

摘要

This work presents a rigorous analysis of the adverse effects ofcyber-physical attacks on discrete-time distributed multi-agent systems, andpropose a mitigation approach for attacks on sensors and actuators. First, weshow how an attack on a compromised agent can propagate and affect intactagents that are reachable from it. This is, an attack on a single node snowballinto a network-wide attack and can even destabilize the entire system.Moreover, we show that the attacker can bypass the robust H-infinity controlprotocol and make it entirely ineffective in attenuating the effect of theadversarial input on the system performance. Finally, to overcome theadversarial effects of attacks on sensors and actuators, a distributed adaptiveattack compensator is designed based on a virtual estimator. The adaptiveattack compensator is augmented with the controller to achieve resilientcontrol. The developed controller achieves secure invulnerable consensus inpresence of the attacks. This controller provides remarkable benefits over aclass of the existing resilient controller, by making no restriction on thenumber of agents or agent's neighbor's under the effect of adversarial input,despite it redeems compromised agent from the effect of the attack. Theeffectiveness of proposed controller and analysis is validated throughsimulations.