Robustness of complex networks under cost-constrained cascaded attack strategies

Abstract

The robustness of complex networks to various kinds of attacks that could trigger cascading failures has attracted increasing attention. Most existing studies fail to consider that the cost to attack each network component (node or edge) may be unequal. Therefore, in this paper, we explore the network robustness to cascaded attacks based on heterogeneous costs. We introduce an attack cost model with both cost-sensitive and budget-constraint parameters. On this basis, three attack strategies are considered, including hub strategy, average degree strategy, and leaf strategy. Their cascaded attack effects are compared by considering the load local preferential redistribution rule. Both the fraction of failed nodes and the value of a new robustness metric, i.e. the budget-constraint threshold, are monitored in different complex networks. Numerical experiments indicate that as the attack cost changes from homogeneity to heterogeneity, the performance of the classic hub strategy decreases gradually. For the situation of weak heterogeneity of attack cost of each node, leaf strategy achieves the maximum attack performance gradually. Moreover, the budget, network structure, and robustness metrics may all affect the selection of the optimal attack strategy.