In the Peer-to-Peer file-sharing system, nodes at the edge of the network take part in the applications and share their resources. As the scale enlarged, search efficiency became the major problem. We researched on the organization and the search algorithm of the Peer-to-Peer file-sharing system to improve its efficiency. A group-based peer-to-peer network model is presented in this paper. In this model, peers with the same interest are organized into one group. All peers attempt to select the best neighbors by measuring other peers. Compared with Gnutella-like P2P systems, the group-based system can achieve high performance and reduce the loads of network as requests are transmitted in scope of the group instead of the whole system. Also we present the value-headed walk search algorithm and the measurement of downloading files among multiple answers. In value-headed walk, the peer forwards query to a subset of its neighbors based on the value of them. Results from theoretical analysis and simulations show that group-based P2P model and value-headed walk search algorithm could greatly reduce the amount of messages generated by searching files in P2P system.
[1]
Ian T. Foster,et al.
Mapping the Gnutella Network: Properties of Large-Scale Peer-to-Peer Systems and Implications for System Design
,
2002,
ArXiv.
[2]
Vana Kalogeraki,et al.
Finding good peers in peer-to-peer networks
,
2002,
Proceedings 16th International Parallel and Distributed Processing Symposium.
[3]
Ian T. Foster,et al.
Mapping the Gnutella Network: Macroscopic Properties of Large-Scale Peer-to-Peer Systems
,
2002,
IPTPS.
[4]
Stefan Saroiu,et al.
A Measurement Study of Peer-to-Peer File Sharing Systems
,
2001
.
[5]
Mary Baker,et al.
Nettimer: A Tool for Measuring Bottleneck Link Bandwidth
,
2001,
USITS.
[6]
Hector Garcia-Molina,et al.
Routing indices for peer-to-peer systems
,
2002,
Proceedings 22nd International Conference on Distributed Computing Systems.