Steps towards self-aware networks
We present the design of a packet network where paths are dynamically selected based on quality of service (QoS) metrics that can be specified by the users of the network or by the network's access points. The approach uses on-line measurement associated with the traffic itself, and reinforcement learning throughout the nodes of the network, to try to satisfy the users' QoS objectives. The talk will present numerous measurement studies of this approach on a large laboratory test-bed. The seminar is based on our recent paper in the Communications of the ACM (July 2009), which integrates different aspects of our research including network design, performance evaluation models and methods, and probability models of neuronal networks.
Erol is the Professor in the Dennis Gabor Chair in the Department of Electrical and Electronic Engineering at Imperial College. He is a Fellow of ACM and of IEEE, a member of the French National Academy of Engineering (http://www.academie-technologies.fr), of the Turkish Academy of Sciences (http://www.tuba.gov.tr) and of Academia Europaea(www.acadeuro.org). His awards include the ACM SIGMETRICS Life-Time Achievement Award, and several "honoris causa" doctorates and honors from Belgium, France, Italy and Turkey. Erol's recent papers have appeared in the Communications of the ACM, several ACM Transactions including those on Adaptive and Autonomous Systems, Sensor Networks, and Internet Technology, The Computer Journal, Neural Computation, Physical Review E, and the Proceedings of the Royal Society. His research is currently funded by EPSRC, the EU FP7 Program, and companies including BT and BAE Systems.