Toronto Networking Seminar




Distributed congestion control in networks: Solution concepts, distributed algorithms, and some recent results

Ravi R. Mazumdar
Department of Electrical and Computer Engineering
University of Waterloo

Date:  Friday, November  16, 2:10pm
Location: BA1220 (Bahen Center)


Abstract:

Distributed algorithms for congestion control in networks have their origins in the work of Arrow and Debreu in Economic Theory. Congestion control algorithms have two primary objectives: maintaining system stability while preventing link congestion and the appropriate allocation of flows in order to provide load balancing so that the system performance is improved. In the late 80's it was realized that these problems could be posed in the context of game theory and the solution concepts such as Nash equilibria, Nash bargaining, etc had important implications in terms of the performance of distributed algorithms and fairness. Following the seminal paper of Kelly 1n 1997, this approach had an even greater following due to its strong connections with TCP. This led to work on primal-dual schemes that are particularly attractive as user level and node level procedures with scalability and decentralizability as important engineering corollaries. More recently, investigations have returned to the issue of network stability.

In this talk I will present the historical overview, discuss the primal-dual schemes and their networking significance in the context of wireless networks, and then present some recent results on the performance of primal-dual algorithms when sessions or connections arrive and depart dynamically.

Bio:

The speaker was educated at the Indian  Institute of Technology, Bombay (B.Tech, 1977), Imperial College, London
(MSc, DIC, 1978) and  UCLA (PhD, 1983).  He is currently a University Research Chair  Professor in the Dept. of ECE at the University of Waterloo, Ont., Canada where he has been since September 2004.  Prior to this he was Professor of ECE at Purdue University, West Lafayette, USA where he continues to be an Adjunct Professor.  He is an editor of the IEEE/ACM Trans on Networking and has served as guest editor for a number of special issues of networking and queueing related journals.  He is a Fellow of the IEEE and the Royal Statistical Society. He shared the INFOCOM 2006 Best Paper Award and was runner-up for the Best Paper Award at INFOCOM 1998.  His research interests are in modeling, control, and performance analysis of both wireline and wireless networks, and in applied probability and stochastic analysis with applications to queueing, filtering, control, and mathematical finance.