A probability-bounded approach for resilient leader–follower consensus of nonlinear multi-agents under DoS attacks

Abstract

Networked engineering systems, including robotic collectives, autonomous vehicle platoons, sensor networks, and microgrids, must maintain coordinated operation in the presence of nonlinear system dynamics and communication disruptions arising from cyber-attacks. This paper addresses the leader–follower consensus problem for nonlinear multi-agent systems (MASs) subject to stochastic denial-of-service (DoS) attacks. A directed communication topology is examined, in which the interconnection graph alternates among a finite collection of configurations because of DoS attacks. Consensus is ensured under the mild condition that the union of these switching topologies contains a directed spanning tree. Instead of assuming exact knowledge of the switching probabilities, only lower and upper bounds are applied to consider a probability-bounded methodology for stochastic DoS attacks. For agents with locally Lipschitz nonlinear dynamics on a prescribed domain, sufficient conditions ensuring exponential consensus are derived and an explicit region of exponential convergence is characterized. A dynamic nonlinear feedback protocol is developed to enhance performance against DoS-induced switching, and a decoupling-based reformulation is provided for computationally tractable convex synthesis conditions to determine the controller gains. In contrast with the existing methods, the proposed approach deals with nonlinear MASs, considers stochastic DoS attacks with only probability bounds (rather than exact switching probabilities), incorporates locally Lipschitz dynamics (rather than global Lipschitz approach), employs nonlinear dynamic controller (rather than linear static control) and guarantees exponential leader–follower consensus. A consensus control simulation example on a network of UGVs under DoS attacks by considering nonlinear friction has been provided.