A game-theoretic approach to optimal defense strategy against load redistribution attack

The wider deployment of advanced computer and communication technologies in the cyber monitoring and control layer of power system will inevitably make the power grid more vulnerable to various cyber attacks, such as false data injection attack and load redistribution (LR) attack. It is critical to develop methods to study the interaction between the attacker and defender for finding the optimal allocation of the limited defense resources. In this study, the LR attack considering the attack and defense is modeled by bilevel optimization. Game-theoretic approaches are developed to model the interaction of the attacker and defender for two scenarios for defending critical measurements and for defending critical substations. The attack and defense interaction is modeled by a zero-sum game if only the load curtailment is considered in the utility functions. And it can be modeled by a non-zero-sum game if both the load curtailment and the associated attack cost and defense cost are considered. The proposed approach is tested based on a representative IEEE 14-bus system, and optimal defense strategies are derived in different scenarios. This study can offer some meaningful insight on effectively preventing and mitigating the LR attack.