Review on Injection of False Data Attack into Power System State Estimation

Abstract

The process of state estimation is threatened to false data injection attacks and diagnosing such kind of bad-natured attacks has significant impacts on reliable operation of power system. Recent studies show that false data injection attack which is one kind of cyber-attack can bypass conventional methods of bad data detection (chi-square WLSE, maximum residual, Hypothesis testing), introducing dangerous effects on operation of power system state estimation. In this report detection-based methods are discussed to address the limitations of existing methods. One method is Kullback-Leibler Distance (KLD) which calculate the distance between two probability distributions p & q derived from variations of measurements (where, $q$ is the probability distribution of measurements variation from historical data. $p$ is the probability distribution of measurements variation between current measurements and previous measurements). Second method is absolute distance method which also used to calculate distance index. By calculating distance index between two distributions of measurements, injection of false data can be detected. If the distance index is larger compare to previous one then newly received measurements are likely to be manipulated.