Miguel Rio


Prof. Miguel Rio, University College London, UK
Title: Learning how to Internetwork
Abstract: Although the Internet has evolved to be the biggest engineering system developed in history, it still relies mainly in algorithms and protocols that are decades old. For example, routing still relies on the Dijkstra algorithm from the 1950s and congestion control still uses TCP, which has changed little since the 1980s. The ability of machine learning techniques to significantly upgrade the intelligence of these algorithms opens the possibility of a much more efficient Internet. This talk will look at the data workflow of the upper layers in the network (network, transport and application) and explore work on machine learning being currently done to improve all its components including network routing, congestion control, network management and service placement.
Short bio: Miguel Rio received the Ph.D. degree from the University of Kent, Canterbury, U.K., and the M.Sc. and M.Eng. degrees in informatics from the University of Minho, Braga, Portugal. He is Professor of Computer Networks, Department of Electronic and Electrical Engineering, University College London, London, U.K. He has authored more than 100 articles in top ranked conferences and journals and has been the principal investigator in numerous projects funded by industry, UK government, the European Union and the US Army. His research interests include network measurement, congestion control, IoT, network routing and the application of machine learning to a variety of network related optimisation problems.