Identifying multiple vulnerable areas of infrastructure network under global connectivity measure

Abstract

Infrastructure networks provide significant services for our society. Nevertheless, high dependence on physical infrastructures makes infrastructure networks vulnerable to disasters or intentional attacks which being considered as geographically related failures that happened in specific geographical locations and result in failures of neighboring network components. To provide comprehensive network protection against failures, vulnerability of infrastructure network needs to be assessed with various network performance measures. However, when considering about multiple vulnerable areas, available researches just employ measure of total number of affected edges while neglecting edges’ different topologies. In this paper, we focus on identifying multiple vulnerable areas under global connectivity measure: Size Ratio of the Giant Component (SRGC). Firstly, Deterministic Damage Circle Model and Multiple Barycenters Method are presented to determine damage impact and location of damage region. For solving the HP-hard problem of identifying multiple optimal attacks, we transform the problem into combinational optimization problem and propose a mixed heuristic strategy consisted of both Greedy Algorithm and Genetic Algorithm to attain the optimal solution. We obtain numerical results for real-world infrastructure network, thereby demonstrating the effectiveness and applicability of the presented strategy and algorithms. The distinctive results of SRGC indicate the necessity and significance of considering global connectivity measure in assessing vulnerability of infrastructure networks.