Longest-path attacks on complex networks

TL;DR

This paper explores the vulnerability of complex networks to longest-path attacks, introducing new algorithms to identify critical paths and analyzing how network structure influences attack effectiveness.

Contribution

It proposes two novel algorithms for approximating the longest simple path in networks and analyzes their effectiveness across different network types.

Findings

Longest-path attack steps increase linearly with density in random networks.

Attack steps increase exponentially with density in scale-free networks.

Homogeneous degree distributions make networks more fragile.

Abstract

We investigate the longest-path attacks on complex networks. Specifically, we remove approximately the longest simple path from a network iteratively until there are no paths left in the network. We propose two algorithms, the random augmenting approach (RPA) and the Hamilton-path based approach (HPA), for finding the approximately longest simple path in a network. Results demonstrate that steps of longest-path attacks increase with network density linearly for random networks, while exponentially increasing for scale-free networks. The more homogeneous the degree distribution is, the more fragile the network, which is totally different from the previous results of node or edge attacks. HPA is generally more efficient than RPA in the longest-path attacks of complex networks. These findings further help us understand the vulnerability of complex systems, better protect complex systems, and design more tolerant complex systems.