Conference Proceedings of the 2005 IEEE International Performance, Computing and Communications Conference (2005)
Phoenix, AZ, USA
April 7, 2005 to April 9, 2005
In peer-to-peer (P2P) file sharing systems, the participating peers share the files with others. Two steps are needed for file sharing: first, a routing request is generated and sent to other peers by using a routing algorithm. The feedback received by the client contains the location information of the requested files; second, the client retrieves the file from one or more peers which have a copy of that file. Routing algorithms have great impact on the overall system performance, distributed hash table (DHT) based routing algorithms provide an elegant and efficient mechanism and become popular in recent years. However, two problems exist in DHT algorithms. First, to find out the location information, in some cases, the routing request may traverse distant peers around the world; second, in case of multiple copies of the requested file stored on different peers, there's no way to figure out which peer is the topologically closest to the client. Thus, the client may have to download the file from a remote peer and suffer from long retrieve latency. In this paper, we propose a hierarchical routing and retrieving algorithm to relieve these problems. The peers' topological information is utilized. Our algorithm is able to find out the closest copy for any routing requests. The simulation results show our strategy can significantly improve the system routing and retrieval performance.
Peer to peer computing, Routing, Helium, Feedback, Gears, Distributed control, Fault tolerant systems, Load management, System performance, Computational Intelligence Society
"Efficient file sharing strategy in DHT based P2P systems", Conference Proceedings of the 2005 IEEE International Performance, Computing and Communications Conference, vol. 00, no. , pp. 151-158, 2005, doi:10.1109/PCCC.2005.1460541