TCP performance in wireless networks with delay spike and different initial congestion window sizes

Authors:
Fei Xin;Abbas Jamalipour
Affiliations:
School of Electrical and Information Engineering, The University of Sydney, Bldg J03, Sydney, NSW 2006, Australia;School of Electrical and Information Engineering, The University of Sydney, Bldg J03, Sydney, NSW 2006, Australia
Venue:
Computer Communications
Year:
2006

Citing 6
Cited 3

Congestion avoidance and control

SIGCOMM '88 Symposium proceedings on Communications architectures and protocols
An evaluation of TCP with larger initial windows

ACM SIGCOMM Computer Communication Review
On estimating end-to-end network path properties

Proceedings of the conference on Applications, technologies, architectures, and protocols for computer communication
The Eifel algorithm: making TCP robust against spurious retransmissions

ACM SIGCOMM Computer Communication Review
Parsec: A Parallel Simulation Environment for Complex Systems

Computer
Effect of Delays on TCP Performance

PWC '01 Proceedings of the IFIP TC6/WG6.8 Working Conference on Emerging Personal Wireless Communications

Architecture of a web accelerator for wireless networks

ACSC '07 Proceedings of the thirtieth Australasian conference on Computer science - Volume 62
Evaluation of VoIP quality over WiBro

PAM'08 Proceedings of the 9th international conference on Passive and active network measurement
Survey: Performance models for wireless channels

Computer Science Review

Quantified Score

Hi-index	0.24

Visualization

Abstract

Previous research indicates that increasing the initial congestion window size in transmission control protocol (TCP) from one segment to roughly 4KB can significantly improve its performance for file transfers that are small compared to the link bandwidth-delay product in wireless cellular networks. However, these networks are also likely to experience delay spikes exceeding the typical round-trip-time (RTT) figures, which can cause spurious timeouts that lead to unnecessary retransmissions and reduction of the TCP sender's transmission rate. Consequently, additional unnecessary retransmissions are needed for a large initial congestion window and thus the throughput of the TCP is degraded. This paper presents the research results on using large initial windows over wireless cellular networks. It has been shown that the throughput of TCP connection over a single bottleneck link with first-in-first-out (FIFO) transmission is degraded with enlarging initial window in the presence of delay spikes. Furthermore, TCP throughput performance is kept unchanged in the simple network topology when there is no burst delay. All these findings should be considered while setting up initial congestion window.