Abstract: This paper analyzes the characteristics of time-shifting service for mobile multimedia broadcast and describes a strategy of resource scheduling for time-shifting proxy server based on predictable user behaviors.Then a time-shifting proxy cache replacement algorithm based on expectation prediction value is proposed.The simulation results show that the algorithm based on expectation prediction value achieves higher prediction hit rate of cache and lower average waiting delay than the algorithm based on FIFO.

Key words: time-shifting proxy, cache replacement algorithm, expectation prediction value, average waiting delay

摘要： 分析移动多媒体广播上时移业务的特点，描述了可预测用户行为的时移代理服务器资源调度策略，并提出一种基于期望预测价值的时移代理服务器缓存替换算法。仿真实验结果表明，该算法与传统的FIFO算法相比较，提高了缓存的预测命中率，减少了用户平均等待延迟。

关键词: 时移代理服务器, 缓存替换算法, 期望预测价值, 平均等待延迟