Designing resilient and secure critical infrastructure networks against cascading failures is essential to mitigate natural disasters and unanticipated cyber attacks. This paper develops a dynamic and distributed resilient mechanism that enables efficient link recoveries and hence reduce the impact of cascading failures. The evolving system-wide link failure is captured by a continuous-time epidemic process over a network using mean-field approximation techniques. The optimal resilient policies are designed to maximize the global connectivity of the network in the worst-case failure state, while minimizing the cost of recovery.
展开▼