Cyber attacks, detection and protection in smart grid state estimation

Abstract

This paper reviews the types of cyber attacks in state estimation as well as detection and protection schemes. Recent studies show that adversaries can not only generate attack vectors, which can bypass the conventional detector, but can also optimize the attack vector to compromise least number of sensors. We examined four types of attack in state estimation process. Then, we examined least effort false data injection attack on how to find the optimal attack vector. Based on the analysis, we implement χ2 detector and Euclidean distance detector to detect attacks. We propose an effective way to protect power system sensors. The case studies are based on a 5-bus system and IEEE-14 bus system. It shows that least effort attack can make most significant deviation of state estimation by compromising least number of sensors. χ2 detector can detect random data injection, bad data injection and DoS attack. However, false data injection can bypass conventional statistical detector, such as χ2 detector. Euclidean distance detector can detect false data injection.