Title: Adaptive multi-level explicit congestion notification

Authors: A. Durresi, M. Sridharan, R. Jain

Addresses: Department of Computer and Information Science, Indiana University Purdue University Indianapolis, Indianapolis, IN 46202, USA. ' Department of Computer Science and Engineering, The Ohio State University, Columbus, OH 43021, USA. ' Department of Computer Science and Engineering, Washington University in St. Louis, St. Louis, MO 63130, USA

Abstract: We propose Adaptive Multi-level ECN (AMECN), a new TCP congestion scheme, as an extension to Multi-level Explicit Congestion Notification (MECN). AMECN allows network operators to achieve high throughput and low delays. However, AMECN performance depends its parameter settings and the level of congestion, hence, no guarantees can be given about delay. To achieve a predictable average delay with AMECN, constant tuning of the parameters to adjust to current traffic conditions is needed. We show how to tune AMECN parameters. The analysis of our simulations shows that Adaptive MECN performs better than Adaptive RED.

Keywords: adaptive congestion management; multilevel ECN; TCP congestion control; QoS; congestion notification; simulation; network congestion; internet; quality of service.

DOI: 10.1504/IJHPCN.2007.015759

International Journal of High Performance Computing and Networking, 2007 Vol.5 No.1/2, pp.3 - 11

Published online: 14 Nov 2007 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article