Bulk content distribution using peer -to -peer overlay: Design and analysis
Abstract
The research focus on Peer-to-Peer (P2P) networks had been on low-complexity content mapping and efficient search mechanisms, and many of the existing P2P applications provide an efficient substrate for sharing or distributing small files like MP3 or images because searching efficiency is essential to the overall performance. However, for bulk contents like multimedia documents or scientific data sets, the data transfer efficiency becomes critical to the overall distribution efficiency. This research explores the use of P2P overlays on distributing bulk contents, and the goal is to improve the distribution efficiency and make P2P overlay a cost-effective alternative to Content Distribution Networks. We consider a hybrid P2P model and propose a genetic-algorithm-based neighbor-selection strategy. The proposed strategy increases the content availability of peers from their immediate neighbors, and improves system performance without trading off users' satisfaction. Simulation results show that the proposed strategy does not provide incentives for peers to deviate from truthfully revealing their downloading progresses and is suitable for real-time deployment. To improve the efficiency of a P2P system; it is also important to provide incentives for the peers to participate and contribute their resources. We address the incentive provisioning problem and present a “seeing-is-believing” incentive-compatible mechanism in which a peer will decide how much resource to be assigned to which neighbor based on what it has experienced. The protocol applies a utility-based resource-trading concept where peers will maximize their contributions for a fair or better return. We show that by adopting this protocol, the system will achieve Cournot Equilibrium. The proposed protocol is light-weight, completely decentralized, and cheat-proof, and experimental results illustrate significant improvements on the distribution efficiency of our protocol over other adopted alternatives.
Degree
Ph.D.
Advisors
Kannan, Purdue University.
Subject Area
Electrical engineering|Computer science
Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server.