TCP performance analysis over aggressive physical carrier sensing based wireless local area networks

Zexi Yang*, Yanfeng Zhu, Zhisheng Niu, Qian Zhang

*Corresponding author for this work

Research output: Chapter in Book/Conference Proceeding/ReportConference Paper published in a bookpeer-review

Abstract

1Aggressive physical carrier sensing (PCS), which improves the spatial reuse efficiency by shrinking the PCS range, is a promising technique to scale high density Wireless Local Area Networks (WLANs). Although the aggressive PCS can improve the throughput in MAC layer due to enabling more simultaneous transmissions, it results in serious frame loss rate caused by hidden terminal problems. TCP congestion control is loss sensitive, and thus high frame loss rate degrades TCP throughput seriously. In this paper, we propose an analytical model to investigate the TCP performance with aggressive PCS, and based on the proposed model the TCP throughput is expressed as a function with respect to the PCS threshold. Numerical results show that MAC throughput efficient PCS starves the TCP throughput, and thus we should not adjust the PCS threshold to maximize the MAC capacity in TCP based applications. In addition, extensive experimental results based on StarEast testbed are given to validate the analysis.

Original languageEnglish
Title of host publicationIEEE GLOBECOM 2007 - 2007 IEEE Global Telecommunications Conference, Proceedings
Pages5357-5361
Number of pages5
DOIs
Publication statusPublished - 2007
Event50th Annual IEEE Global Telecommunications Conference, GLOBECOM 2007 - Washington, DC, United States
Duration: 26 Nov 200730 Nov 2007

Publication series

NameGLOBECOM - IEEE Global Telecommunications Conference

Conference

Conference50th Annual IEEE Global Telecommunications Conference, GLOBECOM 2007
Country/TerritoryUnited States
CityWashington, DC
Period26/11/0730/11/07

Fingerprint

Dive into the research topics of 'TCP performance analysis over aggressive physical carrier sensing based wireless local area networks'. Together they form a unique fingerprint.

Cite this