电子学报 ›› 2007, Vol. 35 ›› Issue (4): 640-646.

• 论文 • 上一篇    下一篇

基于VoiceXML的语音平台缓存一致性控制算法

王文林, 廖建新, 朱晓民   

  1. 北京邮电大学网络与交换技术国家重点实验室,北京 100876
  • 收稿日期:2005-10-24 修回日期:2006-09-06 出版日期:2007-04-25 发布日期:2007-04-25

Consistency Control Algorithm for the Cache of Voice Platform Based on VoiceXML

WANG Wen-lin, LIAO Jian-xin, ZHU Xiao-min   

  1. State Key Laboratory of Networking and Switching Technology,Beijing University of Posts and Telecommunications,Beijing 100876,China
  • Received:2005-10-24 Revised:2006-09-06 Online:2007-04-25 Published:2007-04-25

摘要: 在基于VoiceXML(Voice eXtensible Makeup Language)的语音平台上使用缓存有利于减少网络带宽的消耗,降低服务器的负荷,缩短用户的等待时间.通过分析语音平台的缓存模型,提出了拟合预测算法,对随机分布进行参数拟合,预测一定时间内VoiceXML文档修改的概率,从而确定缓存中的文档是否有效.仿真研究表明,该算法能得到低于1/10000的文档过期率以满足语音呼叫的需求,同时能有效提高系统性能,优于Alex算法.

关键词: 缓存一致性, 参数拟合, VoiceXML, 负指数分布, 对数正态分布, Pareto分布

Abstract: Using cache in the voice platform based on VoiceXML (Voice eXtensible Makeup Language) could reduce the cost of network bandwidth,server-load and delay of request.Through analyzing the cache model of voice platform,the Fitting & Predicting algorithm is proposed.It estimates the validity of the cached document by performing parameter fits to stochastic distribution and predicting the change probability of VoiceXML documents within a certain period.Simulated research indicates the algorithm surpasses the Alex protocol and can obtain a stale ratio lower than 1/10000 to meet the demand of the voice platform while effectively enhancing the system performance.

Key words: cache consistency, parameter fits, VoiceXML, exponential distribution, lognormal distribution, Pareto distribution

中图分类号: