General flow characteristics of P2P streaming considering impact to network load

Hiroyuki Kitada, Takumi Miyoshi, Akihiro Shiozu, Masayuki Tsujino, Motoi Iwashita, Hideaki Yoshino

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

This paper analyzes network traffic characteristics of peer-to-peer (P2P) video streaming services, which have been a recent source of annoyance for Internet service providers due to the large amount of data that they generate. We analyzed two popular P2P video streaming services, PPStream and PPLive, by capturing several hour-long packet streams using a personal computer. Through statistical analysis of the measured data, we identified flow-level characteristics of this P2P streaming. We observed that flow interarrival followed the Weibull distribution, and flow volume followed the Pareto distribution. Regarding network load, the interarrival among high-load flows followed an exponential distribution, and this distribution was valid as a general traffic model. Furthermore, flow volume almost followed a log-normal distribution, though the analysis failed to prove that this distribution can be used as a general model because the flow volume distribution greatly depends on P2P application.

Original languageEnglish
Title of host publicationComputer and Information Science 2010
EditorsRoger Lee, Tokuro Matsuo, Naohiro Ishii
Pages73-83
Number of pages11
DOIs
Publication statusPublished - 2010 Oct 7

Publication series

NameStudies in Computational Intelligence
Volume317
ISSN (Print)1860-949X

ASJC Scopus subject areas

  • Artificial Intelligence

Fingerprint Dive into the research topics of 'General flow characteristics of P2P streaming considering impact to network load'. Together they form a unique fingerprint.

  • Cite this

    Kitada, H., Miyoshi, T., Shiozu, A., Tsujino, M., Iwashita, M., & Yoshino, H. (2010). General flow characteristics of P2P streaming considering impact to network load. In R. Lee, T. Matsuo, & N. Ishii (Eds.), Computer and Information Science 2010 (pp. 73-83). (Studies in Computational Intelligence; Vol. 317). https://doi.org/10.1007/978-3-642-15405-8_7