A Bayesian Network-Based Approach to the Critical Infrastructure Interdependencies Analysis

This paper presents a novel approach to the CI interdependencies analysis, based on the DBN formalism. An original modeling procedure is illustrated, which divides the DBN in three distinct levels: an atomic events level, a propagation level, and a services level. The first level models the adverse events that may impact on the analyzed CIs, the second one properly captures interdependencies among CIs' services and devices, and the last one allows to monitor the state of provided services. The resulting DBN permits to perform three kinds of analysis: a reliability study, which allows to find structural weaknesses of interconnected CIs, an adverse events propagation study, which highlights the role interdependency plays in the propagation of adverse events, and a failure prediction analysis, that can serve as an useful guide to the fault localization process (failures may have many different explanations due to interdependency). A specific case study provided by Israel Electric Corporation is considered, and explicative simulations are presented and discussed in detail.