Cascading Failures in Inter-Dependent Networks: Modeling, Vulnerability Analysis, and Epidemic Propagation
Project runs from 08/14/2015 to 05/20/2020
This project aims to develop building blocks towards a theoretical foundation of rapid mitigation of potentially catastrophic disturbances and control of inter-dependent dynamic networks. The nature of the failures and disturbances includes deliberate adversarial cyber-attacks on the infrastructures that are highly inter-dependent, such as tactical ad hoc networks, future power grids, and social networks. Our aim is to ensure large-scale network resilience against cascading failures so as to safeguard physical infrastructures such as the national power grid, transportation grid, and beyond these, the global information grid and defense strategic communication systems. The issues that are deemed fundamental are i) modeling approaches of inter-dependent networks in order to characterize the cascading effects among these networks, such as cascade failure evolution and to identify critical points and correlated events to guard against, ii) vulnerability analysis of cascading failures with respect to network topology, such as network partitions and blackholes, as well as capture the impacts of failures in the spatial-temporal domain, with uantitative and measurable limits and boundary properties; iii) epidemic propagation of failures due to cyber-attacks with and without countermeasure in mobile networks in that the increasing reliance on wireless communications, while offering great benefits of communications in highly dynamic environments, surrenders our information delivery to both active and passive malware attacks. The potential benefits are very promising: preemptive countermeasures can be designed by observing abnormal events, efficient design and planning of networking architectures and protocols for optimal system operation, and more importantly, rapid responses to failures by making the best use of islanding strategies to halt the cascade in progress, and, with minimal cost, damage and casualties, so as to achieve information assurance.