Real-time detection of false data injection attack using residual prewhitening in smart grid network

State estimation plays a critical role in the smart grid systems. However, according to recent researches, it is evident that a well-designed false data injection attack (FDIA) can bypass the security system and mislead the state estimator. Therefore, detecting such kind of attack efficiently is critical to ensuring the reliability of the smart grid systems. In this paper, we study the real-time FDIA detection mechanism. In the proposed mechanism, we use a residual prewhitening procedure to resolve the problem in the existing real-time FDIA detection mechanism, which is the covariance matrix of the residual is not full rank. Then, we construct a two-sided vector parameter test statistic, and propose a real-time FDIA detection method based on the cumulative sum (CUSUM) of the one-shot statistic. Moreover, the asymptotic closed-form expressions of the detection performance of the proposed method, in terms of the false-alarm period and the average detection delay, are theoretically derived. Numerical results validate that the proposed real-time detection method can detecte the FDIA efficiently.