The Effect Of AS's Geographic Locations On Internet's Stability

Recent studies show that Internet AS-level is a network with many macroscopical properties such as scale-free and high clustering. By studying the mechanisms that produce these macroscopical properties, scientists could design and develop efficient routing protocols, model Internet traffic and manage the organization relationship between ASes. Thus, the mechanism that produce these properties of AS network have been studied widely, which has become a popular problem in network science.In fact, the properties of AS network are influenced by economy, geography, politics and technology. Analyzing these mechanisms and modeling the AS network is the main method to study the AS network.This paper seeks to understand the effect of geographic factor on the Internet AS-level network. Furthermore, a relationship between geographic factor and network stability is shown. AS network is believed not an isolated system, but is in an opening environment where many systems interact together. A hypergraph structure that can describe the geographic locations of Internet resources is defined for the first time. The mechanisms that reflect the interactions between the AS network and the hypergraph structure on the geographic locations are presented, too.This paper models the Internet AS-level by simulating the interaction between the actual AS network and the geographic hypergraph. The effect of distance on the evolvement of AS topology is measured in two different conditions, one is that the AS expands randomly on the zones; the other is that the AS selects its neighboring zones to expand. The results show that if the AS selects its neighboring zones to expand all the time, then the average clustering coefficient of the AS network is influenced by the distance obviously. It presents that the average clustering coefficient decreases with the decrease of the distance. Finally, this paper discusses the relationship between the average clustering coefficient and the network toughness, our study indicates that the network toughness decays with the decrease of the distance which determines the connecting between ASes, and the network stability decays as well.