Optimal packet scan against malicious attacks in smart grids

Abstract

With the integration of advanced computing and communication technologies, the Smart Grid is expected to greatly enhance efficiency and reliability of future power systems with renewable energy resources, as well as distributed intelligence and demand response. Along with the salient features of the Smart Grid, cyber security emerges to be a critical issue because millions of electronic devices are inter-connected via communication networks throughout critical power facilities, which has an immediate impact on reliability of such a widespread infrastructure. In this paper, we discuss the packet based attacks and study the Optimal Inspection Points (OIP) problem, which asks us to find a subset of nodes in a given network to perform the Deep Packet Inspection so as to maximize the number of scanned packets while satisfying the delay constraints. This problem finds many applications for malicious attack detection, especially for those cases where each single packet or the network traffic is required to be inspected. Accordingly, we prove OIP is NP-complete and provide an FPTAS in the case of single path routing. For the multiple path routings, we design an FPTAS when the routing graph takes a form of series-parallel graphs, which is commonly used to model electric networks. We also discuss the multi-scan scenario and design PIVOT algorithm to tackle the problem and evaluate the algorithms through experiments.