Community detection and resilience in multi-source, multi-terminal networks

Abstract

Many networks, particularly infrastructure networks, have multiple source nodes and multiple terminal nodes. And many such networks exhibit community structures, wherein the network is partitioned into groups of densely connected nodes with sparse connections between groups, based on topology or spatial characteristics, among others. This article proposes an approach for evaluating the effects of disruptive events, or the disconnection of network components due to failures or attacks, to the community structures and to the total network. The approach enables the assessment of resilience, evaluating both the vulnerability of the network and the recoverability enabled by different network restoration sequences. Different predefined restoration sequences are compared from different perspectives, including cost and strategy characteristics as well as resilience objectives (partial or complete restoration). The approach is illustrated with the topology of an electric power network.