面向参数化ε-隐匿性的远程状态估计最优隐匿攻击

In this article, we investigate the problem of designing stealthy attack against remote state estimation in cyber-physical systems, where the attacker intends to maximize the deterioration of the estimation performance while remaining stealthy. To enable the attack to bypass the detector with a lower alarm rate, a notion of parameterized $\epsilon$-stealthiness is introduced. Within the framework, a novel attack model is proposed, where all the available historical innovations are utilized and only one attack parameter needs to be designed at each step. Subsequently, the optimal attack strategy is presented. Compared to the existing results, the proposed attack strategy can be designed with fewer parameters while guaranteeing the desired attack performance. Simulations are given to verify the validity of the results.