Exploiting semantic locality to improve peer-to-peer search mechanisms

Show full item record

Title: Exploiting semantic locality to improve peer-to-peer search mechanisms
Author: Sharan, Ajitabh
Abstract: A Peer-to-Peer(P2P) network is the most popular technology in file sharing today. With the advent of various commercial and non-commercial applications like KaZaA, Gnutella, a P2P network has exercised its growth and popularity to the maximum. Every node (peer) in a P2P network acts as both a client and a server for other peers. A search in P2P network is performed as a query relayed between peers until the peer that contains the searched data is found. Huge data size, complex management requirements, dynamic network conditions and distributed systems are some of the difficult challenges a P2P system faces while performing a search. Moreover, a blind and uninformed search leads to performance degradation and wastage of resources. To address these weaknesses, techniques like Distributed Hash Table (DHT) has been proposed to place a tight constraint on the node placement. However, it does not considers semantic significance of the data. We propose a new peer to peer search protocol that identities locality in a P2P network to mitigate the complexity in data searching. Locality is a logical semantic categorization of a group of peers sharing common data. With the help of locality information, our search model offers more informed and intelligent search for different queries. To evaluate the effectiveness of our model we propose a new P2P search protocol - LocalChord. LocalChord relies on Chord and demonstrates potential of our proposed locality scheme by re-modelling Chord as a Chord of sub-chords.
Record URI: http://hdl.handle.net/1850/2891
Date: 2006

Files in this item

Files Size Format View
ASharanThesis11-01-2006.pdf 359.6Kb PDF View/Open

The following license files are associated with this item:

This item appears in the following Collection(s)

Show full item record

Search RIT DML


Advanced Search

Browse