Stealthy false data injection attacks against distributed multi-agent systems

From the perspective of an attacker, this paper studies how to destroy the consensus of distributed multi-agent systems by employing False Data Injection (FDI) attacks. A stealthy FDI attack model is proposed to make the tracking errors diverge while allowing the consensus errors to remain as expected. The proposed model does not rely on real-time node information from the multi-agent systems. Furthermore, the minimum cost of attack edge sets is given, taking into account the limited energy available for the FDI attacks. The corresponding algorithm is further provided. Numerical simulations verify the effectiveness of the proposed FDI attack strategy.