Vulnerability of weighted networks

Luca Dall'Asta (1) , Alain Barrat (1) , Marc Barthelemy (2, 3) , Alessandro Vespignani (2)

1 LPT - Laboratoire de Physique Théorique d'Orsay [Orsay]
2 School of Informatics, Indiana University
3 DPTA - Département de Physique Théorique et Appliquée

In real networks complex topological features are often associated with a diversity of interactions as measured by the weights of the links. Moreover, spatial constraints may as well play an important role, resulting in a complex interplay between topology, weight, and geography. In order to study the vulnerability of such networks to intentional attacks, these attributes must be therefore considered along with the topological quantities. In order to tackle this issue, we consider the case of the world-wide airport network, which is a weighted heterogeneous network whose evolution and structure are influenced by traffic and geographical constraints. We first characterize relevant topological and weighted centrality measures and then use these quantities as selection criteria for the removal of vertices. We consider different attack strategies and different measures of the damage achieved in the network. The analysis of weighted properties shows that centrality driven attacks are capable to shatter the network's communication or transport properties even at very low level of damage in the connectivity pattern. The inclusion of weight and traffic therefore provides evidence for the extreme vulnerability of complex networks to any targeted strategy and need to be considered as key features in the finding and development of defensive strategies.

Domaines: Physique et Société [physics.soc-ph]; Mécanique statistique [cond-mat.stat-mech]

Luca Dall'Asta, Alain Barrat, Marc Barthelemy, Alessandro Vespignani. Vulnerability of weighted networks. Journal of Statistical Mechanics: Theory and Experiment, 2006, April 2006, pp.P04006. ⟨10.1088/1742-5468/2006/04/P04006⟩. ⟨hal-00021128⟩