University of Houston
Department of Computer Science
In partial fulfillment of the Requirements for the Degree of
Master of Science
Archana Krishna
will defend her thesis
Performance Analysis Of Egress Admission Control Algorithm
Abstract
There has been an enormous growth of realtime multimedia applications in recent years. Ensuring minimum Quality of Service (QoS) to realtime multimedia applications is an important challenge for future packet networks since they are delay-sensitive and loss-tolerant. One possible way to meet this is by using admission control algorithms. Egress Admission Control (EAC) is one such admission control algorithm which is edge based. In a multi-hop network with many ingress-egress pairs, admission decisions for each flow are made based solely on aggregate measurements obtained at a flow's egress router without maintaining per flow state in either the network core nor the egress node and without the co-ordination of state with core nodes or other egress nodes. A frame work that provides such a property is called Egress Admission Control. In this thesis, we analyze the performance of EAC. First, we review the general EAC definition, design and key concepts. We then use simulations to experimentally evaluate the performance of EAC. For example we show that with variable traffic loads, multi-hop network nodes, unknown cross traffic, EAC still achieves efficient network resource utilization thus exploiting statistical multiplexing and utilization gains from interclass resource sharing. Next, we explore the accuracy of the theoretical results to that of the measured dynamic results and find that theoretical preformance predictions are quite close to measured values. Also, we show how EAC quarantees QoS for single and multi class flows and still achieves high utilization and preserves scalability. Finally, we compare the results obtained from network simulator (ns) to those obtained from a testbed of prototype routers and show that ns can thus emulate a realistic scenario.
Date: Thursday, June 20, 2002
Time: 11:00 AM
Place: 550-PGH
Faculty, students, and the general public are invited.
Thesis Advisor: Dr. Stephen Huang