Clustering and congestion effects on cascading failures of scale-free networks

State Key Laboratory of Rail Traffic Control and Safety, Beijing Jiaotong University - Beijing 100044, PRC and Institute of System Science, School of Traffic and Transportation, Beijing Jiaotong University - Beijing 100044, PRC

Corresponding author: zygao@center.njtu.edu.cn

Received: 25 January 2007
Accepted: 5 July 2007

Abstract

In this work, we study cascading failures of scale-free networks by considering congestion effects, which should be very common in the transportation or communication systems. In terms of User Equilibrium condition, congestion effects can be described by cost functions or link performance functions, which map link flows to travel times. The cost on the link represents the travel time when passing through the link. Betweenness centrality of the link is introduced to define the “practical capacity ” of the link in the cost function. Different from betweeness centrality, by considering congestion effects, load dynamics can be determined in a more meaningful way. Simulation results indicate that, due to congestion effects, load distribution may become sensitive to the initial node removals for the scale-free networks. Moreover, based on congestion effects, scale-free networks with larger clustering coefficients are more sensitive to cascading failures.