In the research field of streaming media service, server traffic is the main index of system performance. In the P2P VoD system, as both the user and provider of video resources, user behavior has great impact on server traffic. This characteristic does not exist in any other streaming media systems. By now analyses and modelling are lacking. In this paper, the relationship between user behavior and server traffic is studied on the basis of measured results of a real P2P VoD system, the timing model is established to characterize the user arriving pattern and the model of the relationship among server traffic, channel popularity and resource distribution is figured out. The model is verified with measured data. Finally, the model for evaluation of system performance is built in respect of the relationship among server traffic, population of users (i. e. , the cumulative number of system users all day long) and channel capacity (i. e. , the total number of channels of the system).%在流媒体服务研究领域,服务器流量是衡量系统性能的主要指标.P2P VoD系统中,用户作为视频资源的使用者和提供者,其行为模式对服务器流量产生较大影响.这一特点是以往其他流媒体系统所不具备的,而现有文献又缺乏对以上特点的研究分析及建模.本文以P2P VoD系统的实用网络测量数据为基础,研究用户行为模式与服务器流量的关系,建立用户行为的“时间性模式”模型,并由此推导出服务器流量与频道流行度及资源分布的关系模型,继而通过实测数据验证模型的合理性.最终确立了测评P2P VoD系统性能的模型,即服务器流量、用户规模(系统日用户数)、频道容量(系统总频道数)间的关系模型.此研究对于改善、提高P2P VoD系统性能具有理论和应用价值.
展开▼
