Resilient Secure Control of Networked Systems Over Unreliable Communication Networks

Abstract

This article considers the secure consensus of networked nonlinear multiagent systems (MASs) under denial-of-service attacks via unreliable communication networks in a directed graph. Different from existing consensus results in an undirected graph, the case that the communication topology of MASs may be frequently attacked by attackers in a directed graph is considered. The existence of attackers leads to the entire communication topology of MASs having no directed spanning tree that plays a key role in consensus analysis. Besides, to ease the communication load between agents, a resilient leader-following event-triggered control protocol and corresponding to the static triggering policy are first proposed, respectively. Then, an extra key variable is adopted to develop the dynamic triggering policy for the directed graph case to achieve secure consensus while reducing triggering instants, which subsumes several existing static ones as special cases. A unified hybrid framework is established to derive new sufficient conditions of achieving secure consensus without continuous communications among neighboring agents. Besides, parameters of sufficient conditions are designed by solving optimization problems. Finally, the theoretical analysis is verified by a robot manipulator system example.