Topology switching-based moving target defense against false data injection attacks on a power system
Qi Wang, Sifan Wu, Zhong Wu, Jianxiong Hu, Quanpeng He, Yujian Ye, Yi Tang
Abstract
• A topology switching (TS)-based moving target defense (MTD) method is proposed to defend the false data injection attacks. • The reinforcement learning-based optimal MTD method is utilized to realize sequential decision-making. • The detection effectiveness of the proposed method is verified by comparing with the heuristic algorithm. False data injection attacks (FDIAs), as strategically designed cyberattacks, can bypass bad data detection mechanisms and thus pose potential economic and stability risks to power systems. In addition to increasing the detection capability of the system, the proactive property transformation of the system can effectively utilize the information gap between the attacker and the system operator, increasing the detection rate against FDIAs. In this paper, a moving target defense (MTD) method based on topology switching (TS) actions is proposed to overcome FDIAs. Specifically, we investigated the feasibility of proactive defense via TS actions, which reconfigured the topology via busbar switching. To make sequential defense decisions on the basis of the perceived current state of the system, the game between the attacker and the defender was modeled as a Markov decision process (MDP). Finally, the deep reinforcement learning-based MTD optimal algorithm was designed to achieve a fast and efficient decision-making strategy. The simulation results demonstrated the effects of the proposed method against FDIAs.