ScholarLens Back to search
2007•Unpublished venueRequires access

Inter-protocol fairness between TCP New Reno and TCP Westwood

Niels Moller, Chadi Barakat, Konstantin E. Avrachenkov, Eitan Altman

Open publisher page 5 citations

Abstract

In this paper we investigate the effect of introducing TCP Westwood+ on regular TCP New Reno. By means of analytical modeling and ns-2 simulations, we demonstrate that the two protocols get different shares of the available bandwidth in the network. Our main result is that the bandwidth sharing between the two protocols depends on one crucial parameter: the ratio between the bottleneck router buffer size and the bandwidth delay product. If the ratio is smaller than one, TCP Westwood+ takes more bandwidth. On the contrary, if the ratio is greater than one, it is TCP New Reno which gets the larger part. Inspired by our results, we propose a simple modification to the window decrease algorithm in TCP Westwood+ that solves the unfairness problem for large buffer sizes. For small buffers, the unfairness problem is still open.

About this research paper

What this paper is about

In this paper we investigate the effect of introducing TCP Westwood+ on regular TCP New Reno. By means of analytical modeling and ns-2 simulations, we demonstrate that the two protocols get different shares of the available bandwidth in the network. Our main result is that the bandwidth sharing between the two protocols depends on one crucial parameter: the ratio between the bottleneck router buffer size and the bandwidth delay product. If the ratio is smaller than one, TCP Westwood+ takes more bandwidth. On the contrary, if the ratio is greater than one, it is TCP New Reno which gets the larger part. Inspired by our results, we propose a simple modification to the window decrease algorithm in TCP Westwood+ that solves the unfairness problem for large buffer sizes. For small buffers, the unfairness problem is still open.

Why it matters

OpenAlex reports 5 citations for this work. Citation counts describe recorded attention and do not establish research quality.

Key contribution

A contribution statement is not available in the OpenAlex record.

Method / approach

Method details are not available in the OpenAlex metadata.

Main findings

Findings are not separately available in the OpenAlex metadata.

Limitations

Limitations are not available in the OpenAlex metadata.

Applications

Application details are not available in the OpenAlex metadata.

Available abstract

In this paper we investigate the effect of introducing TCP Westwood+ on regular TCP New Reno. By means of analytical modeling and ns-2 simulations, we demonstrate that the two protocols get different shares of the available bandwidth in the network. Our main result is that the bandwidth sharing between the two protocols depends on one crucial parameter: the ratio between the bottleneck router buffer size and the bandwidth delay product. If the ratio is smaller than one, TCP Westwood+ takes more bandwidth. On the contrary, if the ratio is greater than one, it is TCP New Reno which gets the larger part. Inspired by our results, we propose a simple modification to the window decrease algorithm in TCP Westwood+ that solves the unfairness problem for large buffer sizes. For small buffers, the unfairness problem is still open.

Key concepts: Zeta-TCP, TCP acceleration, TCP global synchronization, TCP Westwood, Computer science, Compound TCP, H-TCP, TCP Westwood plus

Related papers

Related paper 1Related paper 2Related paper 3
Back to paper searchBrowse research topicsOriginal source
Inter-protocol fairness between TCP New Reno and TCP Westwood — Research Paper | ScholarLens