Near-optimal evaluation of network survivability under multi-stage attacks

Abstract

In this paper, a multi-stage attack-defense model is proposed. We consider cyber attackers and network defenders with complete understanding of the information about each other. In general, in the strategic interaction between cyber attackers and network defenders, both parties repeatedly interact with each other. These interactions should thus not be one-stage but multi-stage. From the network defenders' view, this model is used to support network operators and to predict all the likely strategies used by both cyber attacker and network defender. As a result, the Average Degree of Disconnectivity (Average DOD) is provided as a survivability metric for evaluating the residual network after malicious attacks. To solve the problem, a gradient method and game theory is adopted to find the optimal resource allocation strategies for both cyber attackers and network defenders. © 2012 Springer-Verlag.