Iterative path attacks on networks

Abstract

We investigate a path-attack process on model networks and real-world networks. Based on the local topological structure of a path, we propose an attack centrality measure with a control parameter α for quantifying the influence of a path. In the path-attack process, we iteratively remove the path with the largest attack centrality from a network. Results demonstrate that, for a specific network, there is an optimal α which results in maximum attack efficiency. The denser and more homogeneous the networks, the more robust the networks are against iterative path attacks. Our work helps to explain the vulnerability of networks and provides some clues about the protection and design of real complex systems.