Resilient Design of Leader-Follower Consensus Against Cyber-Attacks

Abstract

This paper focuses on the development of a resilient cooperative control system for leader-follower consensus problems prone to external cyber-attacks. The attackers are assumed to adversely impact data integrity and privacy by (i) injecting unknown bounded exogenous signals to actuators and (ii) eavesdropping on the physical states of followers. To mitigate the adverse effects of such attacks on the consensus, privacy, and stability of leader-follower systems, we develop a resilient cooperative control system by introducing virtual states interconnected with the physical states in such a way that the leader-follower consensus is guaranteed under unknown false data injection cyber-attacks. The dynamics of the virtual states act as a dynamical output mask, which maps the physical states of followers to some virtual states that are exchanged via a communication network. A Lyapunov-based design framework is proposed to guarantee stability and the leader-follower consensus against cyber-attacks. The decentralized design of the control variables in the proposed resilient cooperative control approach facilitates creating a plug-and-play environment, where follower nodes can easily be plugged in/out. The effectiveness of the theoretical results is evaluated using several numerical examples and implementation on a planar robot experimental testbed.