Abstract—The effect of two different edge attacks on the robustness of random scale-free networks against cascading failure is investigated by establishing a cascading failure model for random scale-free networks. In this model, the initial load of an edge is defined as a nonlinear function of the product of the betweenness of its end nodes with an adjustable parameter, and the local preferential redistribution rule is applied to assign the broken edge’s load. An interesting conclusion is reached through theoretical analyses and numerical simulations: there is a threshold of the load parameter. When the value of the load parameter is larger than this threshold, attacking the edges with the higher load can result in larger cascading failures; while for the case of the parameter value smaller than the threshold, attacking the edges with the lower load will be more likely to lead to global collapse. Furthermore, the threshold value has a close relation with the degree exponent of the network. This work will be not only helpful to protect the key edges selected effectively to resist the cascading failure, but also useful in the design of high-robustness networks in order to stand against all kinds of attacks. 

 

Index Terms—Cascading failure, random scale-free network, robustness, edge attacks

Cite: Lin Ding and Minsheng Tan, "Robustness of Random Scale-Free Networks against Cascading Failure under Edge Attacks," Journal of Communications, vol. 11, no. 12, pp. 1088-1094, 2016. Doi: 10.12720/jcm.11.12.1088-1094