Data-driven Defense Strategies for an Infrastructure Network against Multiple Interdictions

Abstract

Critical infrastructures are significant for national security, economic development and social stability. With the development of economic globalization and information technology, the increasing network complexity and dynamic information have brought challenges to generate data-driven defense strategies for an infrastructure network against multiple interdictions. In order to optimize the defense strategy dynamically, we develop a data-driven finite Bayesian Stackelberg game model among a defender and multiple interdictors. In this model, the defender, with incomplete information on the interdictors’ risk attitude and objective weight, initiates to improve the network performance in the most cost-effective way; whereas the interdictor, with incomplete information on the valuation of components and effectiveness ratio, follows to destroy the network structure in the most cost-effective way. Over the dynamic interactions among the defender and multiple interdictors, the interdictor’s knowledge of the effectiveness ratio is updated by Bayesian updating rule. In order to solve the proposed model, smallest-depth binary-partition based hierarchical algorithm is developed to obtain the strong Stackelberg equilibrium. Finally, the practical applicability is demonstrated by a case study of Zhi Jiang network.