Towards Secure Consensus of Multi-Agent Systems via a Virtual Network and Heterogeneous Controller Gains

This paper addresses the problem of resilient consensus in a multi-agent system under cyber-attacks that may affect the shared state information among the neighboring agents. We solve this problem by considering that there exists a virtual network which is not exposed to the attacker and only the physical layer of the agents can be influenced by the attacker. In this direction, our control design methodology relies on integrating physical and virtual layers appropriately. The proposed controller uses the state information of both physical and virtual networks and operates with heterogeneous control gains, which provides a greater flexibility in achieving the consensus as close as possible to the desired value. Furthermore, the conditions on these controller gains are obtained explicitly and the stability of the attacked system is analyzed. Simulations are given to support our theoretical findings.