Abstract:

A cascading failure model for weighted networks is built by considering the influence of weight. In this model the networks’ initial loads are defined as the function of node strength and their distribution can be adjusted with the parameter α. When a node fails, the loads on it will be distributed to its neighbors through certain rules. The study shows that both factors of weight and topological structure should be considered in order to measure the cascading invulnerability, otherwise the cascading failure may be overrated. The following conclusions are validated under the framework of Barrat Barthélemy Vespignani (BBV) network: in the case of α＞1, attacking the larger load nodes is more prone to large scale cascading failures; while for α＜1, attacking the smaller load nodes is more easily leads to the whole network’s paralysis; and the modeling parameter δ negatively correlates with the networks’ invulnerability. Finally, theoretical analyses and simulations are made to explain these conclusions.