(英) |
Cascading failure, in which a large-scale catastrophic failure of nodes
in a network triggered by a failure of a small number of nodes, has
been extensively studied. Most of the existing studies on cascading
failure assume uniform traffic patterns in networks. In contrast, this
study assumes biased traffic patterns, and investigates the effects of
traffic patterns among nodes in a network on cascading failure. We
perform simulation of cascading failure using a biased traffic pattern
where each node generates traffic to a specific sink node. Our
simulation results show the biased traffic pattern in a network can
significantly affect cascading failure triggered by a failure of a
single node with the highest load, which increases the damage of the
cascading failure to the network. Moreover, we examine methods for
increasing robustness of networks against cascading failure through
tuning of tolerance parameter of each node, which determines the
capacity of the node. |