Event Details
A discrete queuing analysis for continuous self-similar traffic
Presenter: Tarek Nasser
Supervisor:
Date: Wed, November 27, 2002
Time: 14:00:00 - 15:00:00
Place: EOW 430
ABSTRACT
Abstract:
Self-similar traffic describes traffic on Ethernet LANs. A router accepting self-similar traffic will find that its buffer will be overwhelmed at certain times even at low traffic rates. The effect of traffic distribution on queuing systems and the management protocols was not adequately studied due to the difficulty of developing the proper queuing models. A previous analysis was done based on infinite queue size (i.e. M/M/1). That analysis under the infinite queue size would never be able to give any knowledge about the self-similar traffic effect on the packet loss probability. In this talk, we compare biased Poisson traffic with Pareto traffic. Then we develop a simple discrete queuing model with finite queue size and measure analytically the performance of the queue (i.e. M/M/1/B).
For more information contact Tarek Nasser at tnasser@engr.uvic.ca