Speaker: Clémence Magnien (CNRS Paris)

Title: Modelling the Internet topology dynamics

Abstract: Understanding the structure of the Internet's infrastructure, i.e., the routers and the links between them, which can be considered as a graph, is a key question, for instance for ensuring network reliabality, designing efficient protocols and limiting the cost of the infrastructure. Obtaining a map of this topology is however a very difficult task, both for practical and fundamental reasons, and is currently an active research topic. In the same way, understanding the topology dynamics in an equally important, but harder, question. Indeed, we are faced with the same problems as for a one-time map measurement, and new problems, due to the periodical repetition of measurements, arise. To tackle this question, we focus on a small and well-defined part of this topology, called an ego-centered view, which can be measured periodically with a high frequency, so that it is possible to study its dynamics. Based on the insight gained from such measurements, we introduce a simple model, based on random graphs, for the internet topology, its dynamics, and the periodical measurement of ego-centered views. I will present this model and will show that the study of its behaviour raises new fundamental questions, in particular about the bias induced by the measurement frequency.