Resilient Consensus of Nonlinear Multiagent Systems Under False Data Injection Attack on Communication Channels: An Attack Detection and Isolation-Based Approach

Abstract

A new resilient consensus control scheme based on the attack detection and isolation idea is proposed for nonlinear multiagent systems (MASs) experiencing false data injection attack (FDI) on communication links between the follower agents. At first, an improved dynamic surface control (DSC)-based consensus scheme is proposed for nonlinear MAS operating in safe conditions and its stability is analyzed. A local distributed diagnostic module based on the high-gain observer idea is proposed for each agent to monitor status of each agent and generate residual signal. Then, by evaluating residual signal, destination agent that receives false data is detected. After that, a bank of observers is activated for isolating under attack link from others ended to the detected agent by the attack detection unit. After identifying under attack link, validation index corresponding to the status of the identified under attack link is updated. Finally, by incorporating updated validation matrix into the control algorithm, the proposed resilient consensus algorithm is introduced. Simulation results on a group of single-link robot manipulators verify effective performance of the proposed scheme.