Skip to main content

Measuring the performance of user traffic in home wireless networks

Author(s): Sundaresan, S; Feamster, Nick; Teixeira, R

Download
To refer to this page use: http://arks.princeton.edu/ark:/88435/pr1vj8q
Abstract: This paper studies how home wireless performance characteristics affect the performance of user traffic in real homes. Previous studies have focused either on wireless metrics exclusively, without connection to the performance of user traffic; or on the performance of the home network at higher layers. In contrast, we deploy a passive measurement tool on commodity access points to correlate wireless performance metrics with TCP performance of user traffic. We implement our measurement tool, deploy it on commodity routers in 66 homes for one month, and study the relationship between wireless metrics and TCP performance of user traffic. We find that, most of the time, TCP flows from devices in the home achieve only a small fraction of available access link throughput; as the throughput of user traffic approaches the access link throughput, the characteristics of the home wireless network more directly affect performance.We also find that the 5 GHz band offers users better performance better than the 2.4GHz band, and although the performance of devices varies within the same home, many homes do not have multiple devices sending high traffic volumes, implying that certain types of wireless contention may be uncommon in practice.
Publication Date: 4-Mar-2015
Electronic Publication Date: 2015
Citation: Sundaresan, S, Feamster, N, Teixeira, R. (2015). Measuring the performance of user traffic in home wireless networks. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), 8995 (305 - 317. doi:10.1007/978-3-319-15509-8_23
DOI: doi:10.1007/978-3-319-15509-8_23
Pages: 305 - 317
Type of Material: Conference Article
Journal/Proceeding Title: Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Version: Author's manuscript



Items in OAR@Princeton are protected by copyright, with all rights reserved, unless otherwise indicated.