A computational approach to determine damage in infrastructure networks from outage reports

Abstract

We consider the problem of determining the set of damaged components in an infrastructure system solely from outage reports from customers. This problem is especially important in the context of cyber attacks on the information/control systems of the infrastructures since these attacks can present incorrect information about the status of components within the infrastructure. We develop an interdiction-based approach that iteratively determines a set of components to inspect whose damage would best replicate the outage report and thus provides decision-makers with an inspection priority list. Our approach is flexible in the sense that it can produce an updated set of damaged components even if a previous inspection reveals that a component is operational. We apply our approach to damage scenarios on realistic infrastructure systems.