Projet de recherche doctoral numero :4753

Description

Date depot: 1 janvier 1900
Titre: Improved algorithms for capturing the dynamics of Internet maps
Encadrant : Timur FRIEDMAN (LIP6)
Directeur de thèse: Prométhée SPATHIS (LIP6)
Domaine scientifique: Sciences et technologies de l'information et de la communication
Thématique CNRS : Non defini

Resumé: A network administrator in principle has a clear view of his or her own network: what are the routers that are deployed, where are they located, how are they interconnected, what is the content of the routing tables that govern how packets pass from one router to the other as they traverse the network. However, anyone outside of that network has a very limited view of what happens within, restricted to data from the BGP inter-domain routing protocol that is used between networks, and information that can be gleaned from sending probe packets into and through the network, guided by tools such as 'ping' and 'traceroute'. This limited view into the tens of thousands of networks that make up the Internet presents a challenge for efforts to construct a map of the Internet as a whole. Well known measurement systems such as Archipelago and RIPE Atlas have been constructed over the years to collect raw data, and algorithms have been devised to interpret this data into Internet maps. These systems consist of distributed probing agents that systematically explore the network. Current probing approaches arrive, at best, in refreshing the map of the Internet once every three days. However, the Internet is dynamic, and the map changes every hour and every minute, and current methods miss this dynamics. The value of better capturing the dynamics is inestimable, in terms of an improved scientific understanding of the Internet and also in terms of the augmented security of the Internet that would come from being able to track potentially disruptive changes in real time. This thesis subject is aimed at improving the capture of Internet map dynamics.

Doctorant.e: Vermeulen Kevin