Non-Poisson processes of email virus propagation

Abstract

Email viruses are one of the main security problems in the Internet. In order to stop a computer virus outbreak, we need to understand email interactions between individuals. Most of the spreading models assume that users interact uniformly in time following a Poisson process, but recent measurements have shown that the intercontact time follows heavy-tailed distribution. The non-Poisson nature of contact dynamics results in prevalence decay times significantly larger than predicted by standard Poisson process based models. Email viruses spread over a logical network defined by email address books. The topology of this network plays important role in the spreading dynamics. Recent observations suggest that node degrees in email networks are heavytailed distributed and can be modeled as power law network. We propose an email virus propagation model that considers both heavy-tailed intercontact time distribution, and heavy-tailed topology of email networks.