Smart grid data injection attacks: To defend or not?

Data injection is a cyber-attack in which an attacker targets the state estimator of the smart grid with the aim to alter the estimation of the system's real-time state. Using data injection, an adversary can manipulate the calculation of the real-time locational marginal prices to reap financial benefit. Even though data injection attacks have attracted significant attention recently, remarkably, all of the existing works focus on cases in which a single adversary is present. In contrast, in this paper, multiple data injection attackers are considered. The problem is formulated as a noncooperative game between the attackers and the smart grid defender. In this game, each attacker chooses a data injection attack while the defender chooses measurements to protect in order to thwart the attacks. The Nash equilibrium of this game is characterized and the effect of the strategies taken by the attackers and defender on the system is analyzed. Our results show how, at the equilibrium, multiple attacks can eliminate the effect of one another thus requiring no defense. However, under different conditions, a defense mechanism can be beneficial in reducing the combined effect of the different attacks on the system. Numerical results using the WSCC 9-bus system are used to validate the derived analytical solution. These results also shed light on the effect of the cost of defense on the attackers' and defender's optimal strategies and utilities. Accordingly, we show that a higher cost of defense makes attackers more likely to attack and worsens the defender's expected utility.