A mean field-type game approach for resilient synchronization in multi-agent systems

Abstract

This paper addresses the problem of false data injection attacks in multi-agent systems, which the goal is to reach consensus or synchronization. In particular, we formulate the problem as an attacker-defender game, and we solve it using mean-field type game theory. Our methodology is able to mitigate the impact of false data injection attacks in the communication network under an intelligent and stealthy attacker. Our approach is illustrated through numerical simulations in applications such as average consensus, vehicle platooning, and oscillators synchronization.