Abstract: In recent years, the peer to peer (P2P) has become the key support technology of the efficient distribution of big data. As the node selection algorithm in the BitTorrent Tracker server of the P2P file distribution system does not consider the problem of node liveness, this paper introduces a liveness-based Tracker server node selection algorithm. This algorithm can search for the high liveness nodes to build a more efficient distribution network that fits the demand of requesting nodes and boosts their download procedure. The experimental result verifies that the improved node selection algorithm can reduce file download time and improve the efficiency of the distribution network.

Key words: BitTorrent, node selection, distribution network

摘要： 近年来，对等网络（peer to peer，P2P）因其高效的分片和分发等机制，已成为大数据高效分发的关键支撑技术。针对P2P文件分发系统BitTorrent中Tracker服务器端节点选择算法没有考虑节点活跃度的问题，提出了一种基于活跃度的Tracker服务器端节点选择算法。该算法选择出活跃度高的节点来建立一个更高效的分发网络，使之更能符合请求节点的需求，帮助请求节点更加高效地完成下载任务。实验结果表明，改进后的节点选择算法可以缩短文件的下载时间，提高网络的分发效率，提升系统的性能。

关键词: BitTorrent, 节点选择, 分发网络