基于预测模型和缓存替换策略的网络资源访问研究  

Research on Network Resource Access Based on Prediction Model and Cache Replacement Policy

作　　者：程龙泉[1] Cheng Longquan(Sichuan Electromechanical Institute of Vocation and Technology,Panzhihua Sichuan 617000,China)

机构地区：[1]四川机电职业技术学院,四川攀枝花617000

出　　处：《科技通报》2017年第10期134-137,187,共5页Bulletin of Science and Technology

摘　　要：针对在网络制造资源共享中客户端对资源无法有效、快速访问的问题进行应用研究,提出基于预测的网络制造资源访问策略,解决网络延迟问题。首先,描述了在客户端如何进行局部Cache设计,然后描述了在传统替换算法的基础上,如何引入基于多项式回归的预取技术,如何将预取技术与语义缓存技术相结合。通过构建对用户查询频率进行实时监测和基于多项式回归预取技术的预测模型,实现动态在线预测,避免了用户兴趣漂移引起的预取不确定性,较好地解决了访问延迟问题。最后,设计相关实验对客户端访问资源的有效性、快速性进行验证。实验结果表明,该优化调度策略在命中率上相对于传统的替换算法大约提高了8%,在一定的缓存空间内较传统替换算法在性能上有一定的提高,提高了客户端访问资源的有效性、快速性。For the client in the network manufacturing resource sharing problem of resources cannot be effective and rapid access to application research,based on the prediction of network manufacturing resources access strategy is put forward,solving the problem of network latency.First of all,describes how to design the local Cache on the client,and then described on the basis of the traditional replacement algorithm,how to introduce the prefetch technology based on polynomial regression,how to prefetch technology combined with semantic caching technology.By building the user query frequency real-time monitoring and prediction model based on polynomial regression prefetching technique,the dynamic online prediction,avoid the user interest drift caused by prefetching uncertainty,better solve the problem of the access latency.Finally,design the related experiment on the effectiveness of the client to access resources and quickness.The experimental results show that the optimized scheduling policy in shooting replacement for traditional algorithms are improved about8%,within a certain cache space thantraditional replacement algorithm has certain increase in performance,improve the effectiveness of the client to access resources and quickness.

关 键 词：制造资源 预取技术 多项式回归 调度 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]