In the p2p VoD system, the relationships between user watching behavior, distribution of chunks,download rates and the pre-fetch strategy are complicated.However, in previous research those elements are always studied independently without concerning the interactions between them.In this paper, measured data from a real p2p VoD system is studied firstly.Then,a mathematical model is proposed to analyze the relationship of those factors.In addition,a new pre-fetch strategy is designed, which aims to get high download rates while considering user watching fluency.This model provides a more accurate way to study user watching behavior.It also gives some help for user behavior study and new pre-fetch strategy design for other p2p applications.%peer-to-peer方式的VoD系统中,用户的观看行为、网络中资源块(chunk)的分布、下载速度和pre-fetch机制之间相互影响,相互制约.以往的研究通常专注于某个方面,而没有考虑各因素间的相互关系.本文通过对实际数据进行分析,建立模型描述p2p VoD系统中上述各因素间的关系,并基于该模型提出一种新的pre-fetch机制,能在保证用户流畅观看的前提下,尽可能提高下载速度.该模型提供一种更为精确的了解客户端行为的方法,为其他各类p2p应用中研究用户行为、设计下载机制、优化系统提供参考.
展开▼
