Technical Report: Performance of the Expected Force on AS-level Inernet topologies

TL;DR

This paper evaluates the effectiveness of the Expected Force (ExF) metric in predicting epidemic outcomes on AS-level internet topologies, demonstrating its strong performance despite the network's real-world constraints.

Contribution

It provides empirical evidence that the ExF metric remains predictive on complex, real-world internet topologies, extending its applicability beyond theoretical models.

Findings

ExF performs well on AS-level internet topologies

Internet topology design facilitates information diffusion

Expected Force is a robust predictor in real-world networks

Abstract

Many concepts in network science are based on idealized network models. Infrastructure networks, however, are strongly constrained by specific engineering and economic constraints which do not appear in such models. One result is that concepts which are assumed to be fundamental by theoriticians severely underperform in real world situations. This motivated us to test if the Expected Force (ExF) metric, which was developed in the context of more theoretical network models, would continue to be predictive of epidemic outcomes on the highly structured topology of the internet at the autonomous systems (AS) level router connectivity. Results suggest that the ExF performs extremely well on these topologies, perhaps reflecting their design constraint of easing the diffusion of information.