End-to-end single-rate multicast congestion detection using support vector machines
Xiaoming Liu · University of the Western Cape Electronic Theses and Dissertations Repository (University of the Western Cape) · 2008
IP multicast is an efficient mechanism for simultaneously transmitting bulk data to multiple receivers.Many applications can benefit from multicast, such as audio and videoconferencing, multi-player games, multimedia broadcasting, distance education, and data replication.For either technical or policy reasons, IP multicast still has not yet been deployed in today's Internet.Congestion is one of the most important issues impeding the development and deployment of IP multicast and multicast applications.Many congestion control schemes have been proposed to tackle multicast congestion problem.However, few of the schemes focus on using machine learning to detect multicast congestion in advance.Machine learning has already been successfully applied in a number of areas without much background information, and gives useful results.Because we tackle the multicast congestion problem with the end-to-end assumptions, we cannot obtain opportune and accurate congestion information directly from inside the network.Therefore, machine learning is particularly appropriate due to the absence of congestion information and the unpredictable variance of network congestion.To detect end-to-end multicast congestion, we propose an end-to-end multicast congestion detection scheme using support vector machines.Support vector machines are able to detect incipient congestion with great accuracy in an end-to-end multicast network after training by using structural information about the multicast network.To verify the performance of our scheme, we ran several ns-2 simulations and statistical experiments.Our simulations have shown that support vector machine is an appropriate mechanism for decision making in proactive multicast congestion detection.xi