Passivity-Based Attack Detection and Mitigation with Switching Adaptive Controller and Quadratic Storage Function

This paper studies the consensus problem in a networked multi-agent system subject to actuator attacks. The concept of passivity is leveraged for the detection of destabilizing attacks, which relies on satisfying the dissipation-inequality with the quadratic storage functions. On identification of an attack, the controller switches to the defense mode, where the attack signal is mitigated via an estimator based on available measurements. We show that the state error, incurred due to an attack, remains bounded with bounded attack signals, and the system achieves consensus once the attack is mitigated. Simulations are provided to illustrate the theoretical findings.