中文摘要：

BitTorrent（BT）是一种基于P2P的文件共享软件,有着十分广泛的应用.在BT中,默认参与节点随机选择其它节点作为网络中的邻居构成覆盖网络,不能根据节点的位置优化覆盖网络,影响了BT中文件传输的性能.BT的有偏邻居选择,指的是BTTracker参照BTClient在互联网中的位置,向BT有针对性地提供相应的邻居,优化BT的覆盖网络,使得BT的文件传输效率得到提高.本文提出了一种基于邻近原则优化BT文件传输速率的机制.首先利用网络坐标对BT网络参与节点在互联网中的位置进行计算;之后,根据网络坐标,对于BT参与节点的邻居进行了有偏选择;BT参与节点经过分布式聚类得到基于邻近原则的覆盖网络.仿真实验显示,基于网络坐标的有偏邻居选择对于BT整体文件传输性能有较大的提升.

英文摘要：

BitTorrent （BT） is a widely used P2P file -sharing software. In basic BT, clients select their neighbors randomly to form the BT overlay network. As a result, they never consider the location of the participant nodes. This randomly constructed overlay will reduce the performance of the file transmission in BT. Biased neighbor selection in BT, means that BT tracker return the neighbor candidates to the BT clients according to their locations in Internet. This approach will improve the performance of the BT overlay network and increase the efficiency of the file transmission of BT. This paper proposes a scheme to construct the locality - aware BT overlay network to improve the performance of BT. Firstly, we use network coordinate to compute the location of the nodes. In addition, we help the BT clients to select their neighbors based on the locations of their network coordinates. Distributed node clustering will form a locality - aware BT overlay network. We use both simulation and experiments in Internet to prove that our network coordinate based locality - aware BT overlay will improve the efficiency of the file transmission a lot.