A game theoretical analysis of data confidentiality attacks on smart-grid AMI

The widespread deployment of smart meters in the advanced metering infrastructure (AMI) raises privacy concerns. Analyzing the data collected from smart meters can expose habits and can be potentially used to predict consumers' behaviors. In this paper, we analyze the confidentiality of information in the AMI consisting of nodes with interdependent correlated security assets. On each node, the defender can choose one of several security modes available. We try to answer the following questions: 1) What is the expected behavior of a rational attacker?; 2) What is the optimal strategy of the defender?; and 3) Can we configure the security modes on each node to discourage the attacker from launching any attacks? In this paper, we formulate the problem as a noncooperative game and analyze the behavior of the attacker and the defender at the Nash equilibrium. The attacker chooses his targets in order to collect the maximum amount of data on consumers, and the defender chooses the encryption level of outbound data on each device in the AMI. Using our model, we derive the minimum defense resources required and the optimal strategy of the defender. Finally, we show how our framework can be applied in a real-world scenario via a case study.