Modeling access network: Maximum node-degree determines the traffic of the internet

摘要

It has been demonstrated that both routers and domains in the Internet have a scale-invariant fractal structure. However, nodes in access network (AN) level are found demonstrate a random geographic distribution. In this paper, we investigate the AN by proposing a model in which the nodes are randomly posited at given regions. In particular, we design a new node-attachment strategy that can yield a connectivity pattern more consistent with real configuration. Taking into account of both the geographical distribution of the nodes and specific local connectivity, our model can therefore serve as more accurate description of real-world Internet. We furthermore use this model to study the traffic load of the Internet, and we found that the optimal betweenness centrality variance and average path length, which is associated with an optimal traffic load, can be obtained by setting a suitable maximum degree K_(max) for the AN. Our results indicates that the optimal value of K_(max) increases linearly with the number of nodes, which provide useful guidance to improve throughput and transmission efficiency when designing the access network.