Abstract
ULg Research Unit in Networking RUN
Abstract


Enhancement of TCP over wired/wireless networks with packet loss classifiers inferred by supervised learning

I. El Khayat2 , P. Geurts1 and G. Leduc2

1 University of Liège
2 Research unit in Networking, EECS department, University of Liège, Belgium

(2004)

Abstract

TCP is suboptimal in hybrid wired/wireless networks because it reacts in the same way to losses due to congestion and losses due to link errors. In this paper, we propose to improve TCP in wired/wireless networks by endowing it with an algorithm that can distinguish packet loss causes. To build this classification algorithm, a database of pre-classified losses is gathered by simulating random network conditions, and classification models are automatically built from this database by using supervised learning methods. Several learning algorithms are compared for this task. Our simulations of different scenarios show that improving TCP with such a classification algorithm can improve greatly the throughput of TCP in wired/wireless networks and at the same time maintain TCP-friendliness.

Keywords

loss cause classification, machine learning, TCP, wireless links

[ Home | People | Research Topics | Projects | Publications | IFIP WG 6.1 | Events and Conferences | CSS | How to Reach Us ]

Editor: - G. Leduc -
Webmaster: - C. Soldani -
Still running IPv4 at: 54.146.50.80... RUN | Montefiore | ULg
© 2000-2015.