A Novel Grouping Strategy for Reducing Average Distribution Time in P2P File Sharing
Pui-Sze Tsang, Xiang Meng, Ki Lui · 2010
Peer-to-Peer (P2P) file distribution has been widely used for file sharing in recent years. When compared with the traditional client-server model, the P2P model is a lot more efficient as each user can act as both a client and a server. This enables the P2P file distribution to scale well with increasing number of users. Grouping strategy has been introduced to reduce the average distribution time among peers without prolonging the total time needed to obtain a file. In this paper, a novel grouping strategy which groups peers of similar bandwidth together is introduced. We mathematically illustrate that under certain circumstances, this new grouping strategy performs better than the Greedy Grouping mechanism. To understand the performance of our grouping mechanism more comprehensively, we conduct extensive simulations. The results show that our mechanism can enhance the performance significantly in different network settings.