IDEALS Home University of Illinois at Urbana-Champaign logo The Alma Mater The Main Quad

An Adaptive Content Distribution Protocol for Clustered Mobile Peer-to-Peer Networks

Show full item record

Bookmark or cite this item: http://hdl.handle.net/2142/13966

Files in this item

File Description Format
PDF onlinecodes993_cl15.pdf (170KB) PDF format PDF
Title: An Adaptive Content Distribution Protocol for Clustered Mobile Peer-to-Peer Networks
Author(s): Vu, Long; Malik, Rahul; Wang, Qiyan; Nahrstedt, Klara
Subject(s): Mobile Peer-to-Peer networks, clustered networks, content distribution, online codes
Abstract: In this paper, we show that the clustered mobile peer-to-peer (P2P) networks exist in numerous scenarios where mobile users collaborate to improve content distribution services. In order to understand the clustering behavior of nodes in clustered mobile P2P networks, we present a probabilistic model for path selection and use the Mobius Tool to study the clustering behavior of different movement areas. Our study shows that the cluster size distribution follows an exponential function. We then design an adaptive protocol, which blends cellular and P2P (i.e., wifi or Bluetooth) communications of the mobile devices and leverages the exponential-cluster-size function to improve content distribution services. With our protocol, mobile nodes periodically sample the current cluster size and predict the future cluster size using the exponential function. Then, nodes apply the predicted cluster size function to calculate the available data in P2P channel using Online Codes and tune the cellular download timer adaptively to meet the file download deadline. The simulation results show that our adaptive protocol achieves much higher performance than the non-adaptive protocol by reducing the downloading load on the cellular channel by 4% ~ 10%, and significantly reducing message overhead. Simulation results also confirm that our protocol adapts well to network dynamics since when the nodes get closer to the destination, the cluster size function is predicted more accurately.
Issue Date: 2009-09-28
Citation Info: PerCom 2010. The Eighth Annual IEEE International Conference on Pervasive Computing and Communications
Genre: Technical ReportOther
Type: TextOther
Language: English
URI: http://hdl.handle.net/2142/13966
Publication Status: published or submitted for publication
Peer Reviewed: is peer reviewed
Date Available in IDEALS: 2009-10-09
 

This item appears in the following Collection(s)

Show full item record

Item Statistics

  • Total Downloads: 430
  • Downloads this Month: 1
  • Downloads Today: 0

Browse

My Account

Information

Access Key