论文部分内容阅读
对视频点播系统中用户行为进行建模和仿真,是研究系统使用状况、设计性能优化算法的重要手段.但在以往的研究中,对用户行为建模和仿真都是基于整体历史数据的统计进行的,而在很多情况下,对不同模式的行为采用不同的策略能够更好的提供视频传输服务.本文针对视频点播系统中用户点播行为的特性,以及系统优化策略的需要,提出用户行为时间序列模型和聚类方法,在中国科技大学视频点播系统实际数据基础上进行了仿真测试,结果表明了该方法的可行性.
Modeling and simulating user behavior in video-on-demand systems is an important means to study system usage and to design performance optimization algorithms, but in previous studies modeling and simulating user behavior were all based on statistics of overall historical data , And in many cases, using different strategies for different modes of behavior can provide better video delivery services.This paper aims at the characteristics of users’ on-demand behaviors in video-on-demand system and the need of system optimization strategy, and proposes the user behavior time series Model and clustering method, the simulation test is carried out based on the actual data of China University of Science and Technology video-on-demand system. The results show the feasibility of this method.