一种缩短下载时间优先的自适应BitTorrent激励协议

摘要

The goal of current incentive mechanism for BitTorent is to guarantee the node's fair-ness between upload bandwidth and download bandwidth. However, the file download time is more preferable than such fairness for the users in real file sharing environments. Therefore, a new self-adaptive incentive protocol denoted as AIPS to minimize the file download time is provided in this paper for BitTorrent. This paper firstly analyzes the influence of incentive mechanism on file sharing, the file transfer structures under incentive mechanisms based on Markov model. The conditions to minimize the file download time are inferred further based on the transfer structures and then a game with file download time as its utility is defined in this paper. The AIPS is the strategy adopted by nodes under the Nash equilibrium for that game. Simulations show that the AIPS can improve the efficiency for file sharing.%BitTorrent激励机制的目标是保证节点上传和下载之间的公平性,但相比公平性而言,实际应用中的节点更优先考虑的是文件下载时间,据此文中提出了一种缩短文件下载时间优先的自适应BitTorrent激励协议AIPS.文中首先基于Markov模型对BitTorrent现有激励机制的效果给出了定量分析,分析了激励机制下的文件传输结构,并用概率分析方法给出了该传输结构下最小化文件下载时间的条件.应用分析结果文中定义了一个以缩短文件下载时间为效用的博弈,在该博弈达到Nash平衡时各节点采用的策略就是激励协议AIPS.模拟实验表明文中提出的AIPS较现有的BitTorrent激励协议能明显提高文件共享系统性能,提高文件下载效率.