Optimal hidden SCADA attacks on power grid: A graph theoretic approach

Correct estimation of the state variables of a power grid is indispensable for its stable operation. The advent of smart grid has further enhanced the importance of secure real-time monitoring of the state variables. This paper studies hidden data attacks on power systems by an adversary trying to manipulate the state estimator. A novel graph theoretic formulation is developed for the problem of determining the optimum measurements to introduce spurious data for a hidden attack by the adversary. The paper discusses a polynomial-time solvable algorithm to optimally solve this problem under different settings. From the perspective of the system operator, techniques are proposed to help identify critical measurements for protection to prevent such hidden data attacks and to reduce their efficacy. The performance of the algorithms are demonstrated through simulations on IEEE test cases.