Likelihood of Cyber Data Injection Attacks to Power Systems

Yingshuai Hao,Meng Wang,Joe Chow
DOI: https://doi.org/10.48550/arXiv.1512.05008
2015-12-15
Systems and Control
Abstract:Cyber data attacks are the worst-case interacting bad data to power system state estimation and cannot be detected by existing bad data detectors. In this paper, we for the first time analyze the likelihood of cyber data attacks by characterizing the actions of a malicious intruder. We propose to use Markov decision process to model an intruder's strategy, where the objective is to maximize the cumulative reward across time. Linear programming method is employed to find the optimal attack policy from the intruder's perspective. Numerical experiments are conducted to study the intruder's attack strategy in test power systems.
What problem does this paper attempt to address?