Auxiliary Innovation-Based Stealthy Attack Strategies Against Distributed State Estimation

This article investigates stealthy attack strategies that leverage auxiliary innovations in distributed state estimation. We propose a malicious attacker who exploits both local innovations and auxiliary innovations from neighboring nodes to construct the attack. Under such attacks, the dynamics of the compromised estimation error covariance are derived, and the attacker aims to maximize the trace while maintaining stealthiness. By leveraging singular value decomposition, we derive an analytical solution to the optimal stealthy attack, together with a degraded counterpart that does not exploit auxiliary innovations. Furthermore, to alleviate computational complexity for the attacker, a lower bound of the compromised estimation error covariance trace is established, based on which a numerical solution for a suboptimal stealthy attack is developed. Finally, the theoretical results are verified through simulations on a buoyant air turbine (BAT) model.