LPAttack: Leverage point attacks against state estimation in smart grid

Abstract

A novel class of malicious data attacks, LPAttack, are presented against the state estimation process in Smart Grid. Here LP represents leverage points, which are the outliers in the factor space of the regression model for Smart Grid state estimation. The attacker strategically manipulates the parameter data in Smart Grid to mislead the control center with incorrect system parameter information, such that leverage points are created within the factor space of the state estimation regression model. As a result, the attacker can freely inject arbitrary errors into the meter measurements corresponded with the leverage points, while bypassing the existing bad data detection mechanism. We first introduce the fundamental principles and strategies of launching LPAttack in Smart Grid. Then the potential countermeasure based on robust Schweppe-Huber Generalized-M estimator is proposed. Finally, we evaluate the LPAttack principles and its countermeasure through simulations in IEEE test system, and examine in particular the effect of the attacks on Locational Marginal Prices in real-time pricing power market.