针对信息物理系统远程状态估计的隐蔽虚假数据注入攻击

The optimal strategy for stealthy false data injection (FDI) attacks in cyber-physical system (CPS) is explored from the attacker＇s perspective. The Kullback-Leibler (K-L) divergence is selected as the evaluation index of attack stealthiness, and the attack signal is designed to keep the attack stealthy and minimize the performance of CPS remote state estimation. First, the statistical characteristics of the residuals are used to calculate the error covariance of remote state estimation, which transforms the FDI optimal strategy problem into a quadratically constrained optimization problem. Second, under the constraint of attack stealthiness, the optimal policy is derived using Lagrange multiplier method and semi-positive definite programming. Finally, simulation experiments are conducted to verify that the method proposed in this paper has significant advantages in terms of stealthiness compared with existing methods.