Identification of "unobservable" cyber data attacks on power grids

This paper presents a new framework of identifying cyber data attacks on synchrophasor measurements. We focus on detecting “unobservable” cyber data attacks that cannot be detected by any existing detection method that purely relies on measurements received at one time instant. Leveraging the approximate low-rank property of phasor measurement unit (PMU) data, we formulate the unobservable cyber attack identification problem as a matrix decomposition problem where the observed data matrix is the sum of a low-rank matrix plus a linear projection of a column-sparse matrix. We propose a convex-optimization-based decomposition method and provide its theoretical guarantee in the attack identification. Numerical experiments on actual PMU data and synthetic data are conducted to verify the effectiveness of the proposed method.