Abstract

Optimistic unchoking plays an important role in BitTorrent Peer-to-Peer (P2P)[46, 45, 48, 4] file sharing networks. Peers use optimistic unchoking to find upload bandwidth information about their neighbors. However, free-riders can also take advantage of optimistic unchoking and download from the network without uploading anything. In this thesis, a novel optimistic unchoking algorithm for BitTorrent is proposed. The main purposes of our algorithm are to prevent free-riding and to improve the efficiency of optimistic unchoking. A stochastic model is then proposed to analyze the performance of my algorithm. We also verify the results by simulations. BitTorrent also have a built-in incentive mechanism called "Tit-for-Tat" [4] to prevent free-riding. Basically, a peer will upload to other peers (default is four) that give it the highest download rate. In this thesis, We will show that by adjusting the upload rate and the number of uploads, a selfish peer can take advantage of the "Tit-for-Tat" [4] to improve its download rate. However, this strategy of the selfish peer is harmful to the whole network. If many peers take the same strategy, the performance of the whole network will be significantly decreased. It is then theoretically proved that the "Tit-for-Tat" [4] is not an optimal incentive mechanism. To solve this problem, We propose a new incentive mechanism for BitTorrent. With this new mechanism, even if all peers are selfish, the performance of the whole network can still be maintained at a very high level