Contact: Jean-Louis Rougier (rougier@enst.fr)

Based on the work of a previous Ph.D. student of our group, we aim at developing and implementing some tools which would allow to analyse current Internet statistics (Ripe Atlas Streams and BGP data) in order to detect anomalies automatically. Using simple machine learning and data analytics on the wide number of RTT traces available word-wide, we have previously showed that we can extract very interesting results form our analysis method — even revealing potential congestion points and identifying their most likely location. The goal of this project is to design and implement these techniques measurement streams in order to have a on-line map of potential internet congestion. This publicly available tool could be very helpful for the research community to better understand the Internet behavior and for the network community at larger (for instance giving hints on potential reasons of network performance degradation for network practitioners).