Distributed Adaptive Control Methods for Resilient Networked Multiagent Systems

Control algorithms of networked multiagent systems are generally computed without having a centralized entity monitoring agent activities. Therefore, adversaries such as attacks to the communication network and/or failure of agent-wise components can easily result in system instability and prohibit the accomplishment of system-level objectives. This invention studies resilient coordination of networked multiagent systems in the presence of these adversaries, i.e., agents that are subject to such challenges. In particular, we consider a class of adversaries consisting of external disturbances and internal uncertainties, and then present a new, adaptive control architecture to retrieve the nominal networked multiagent system behavior. Apart from the relevant existing literature that make specific assumptions on the graph topology and/or the fraction of misbehaving agents, we show that the considered class of adversaries can be mitigated through a novel adaptive control methodology that utilizes a local state emulator-even if all agents are misbehaving.