User-search time analysis for hierarchical peer-to-peer overlay networks with time-dependent user-population process

  • Authors:
  • Anditto Heristyo;Hiroyuki Masuyama;Shoji Kasahara;Yutaka Takahashi

  • Affiliations:
  • Kyoto University, Sakyo-ku, Kyoto, Japan;Kyoto University, Sakyo-ku, Kyoto, Japan;Kyoto University, Sakyo-ku, Kyoto, Japan;Kyoto University, Sakyo-ku, Kyoto, Japan

  • Venue:
  • Proceedings of the 4th International Conference on Queueing Theory and Network Applications
  • Year:
  • 2009

Quantified Score

Hi-index 0.00

Visualization

Abstract

Traditional Internet telephony systems heavily rely upon central servers, especially during call initiations. The absence of these servers in Peer-to-Peer (P2P) implementations results in a significant increase in delay during the user search process. With this issue in mind, we focus on a hierarchical P2P overlay network with a tree structure where node groups form sub-network cells of a constant size bound. In our model, users are assumed to arrive into the system according to a non-stationary Poisson process. The sojourn time for a node is assumed to be exponentially distributed. Through this modeling, we analyze the time-dependent user population process with an Mt/G/∞ queue. The mean user search time is then calculated by examining the overall shape of the hierarchical overlay network structure at a given time. Numerical results indicate that fluctuations in the number of layers generated in the system produce a greater impact on the user search latency when compared with the effects of changes in the cell size bound.