(YIP) LEIOS COMPLEX NETWORK RESILIENCE THROUGH CONTROLLED ISLANDING AND RECONNECTION
Abstract
Interdependent failures, in which the loss of one node or edge causes neighboring nodes or edges to fail, can create cascades that lead to widespread network outages. High-profile examples include power system blackouts and malware-epidemic propagation in cyber, social, and biological networks. One approach to mitigating interdependent failures in defense, social, and infrastructure applications is islanding, in which a set of edges is deliberately removed after a disturbance in order to partition the network. This project will research and develop Leios, a framework for adaptive, agile, and resource-efficient controlled islanding and reconnection with minimal service disruption. Our framework will enable network nodes to compute and execute islanding and reconnection in a distributed manner. This distributed approach will improve scalability of the island computation while making islanding robust to changing network conditions and unexpected disruptions due to malicious attacks, and will improve resilience of applications including power grids, cyber systems, and social networks. The effort builds on the PI’s extensive track record in complex networks, optimization, game theory, and networked control systems. The project will consist of three research thrusts. The first thrust will formulate islanding models and metrics, and develop algorithms for computing the metrics. The second thrust will develop distributed algorithms for computing islanding strategies and deciding when to implement islanding after a disturbance. This thrust will leverage submodular (diminishing-returns) structures of the islanding metrics to develop near-optimal algorithms. The third thrust will research and develop methodologies for safe reconnection, and will develop models of selfish nodes who bias the island computation in order to improve their own utilities.
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Mar 07, 2023
- Source ID
- FA95502210054
Entities
People
- Andrew Clark
Organizations
- Air Force Office of Scientific Research
- United States Air Force
- Worcester Polytechnic Institute