Auditing on Smart-Grid With Dynamic Traffic Flows: An Algorithmic Approach

Abstract

The introduction of Smart Grid systems has raised many new security concerns. If an attacker can compromise components of the Smart Grid communication network, they can fabricate malicious messages to interfere with the grid and ultimately cause outages. One method to address this concern is to conduct network audits by logging network traffic into dedicated servers in order to detect malicious messages. This may be done by using switches' or routers' ability to duplicate packets they receive with minimal overhead. The question of how many and which switches/routers to select for this task naturally follows. This paper considers the problem of finding minimal set of routers/switches in a Smart Grid communication network to use for auditing traffic. Accordingly, we devise three algorithms: the first one is highly effective with an approximation ratio of $(2+\theta)(\ln |V| + 1)$ . The second method is a highly scalable algorithm with a constant performance ratio. And the last one is a dynamic algorithm which can efficiently update its solution in response to changes to critical traffic. We experimentally evaluate our solutions and compare them to an optimal Integer Programming formulation, finding that they perform near-optimally and significantly outperform a simple heuristic in all cases.