一种基于逗留时间的新型内容中心网络缓存策略  

A New Cache Policy Based on Sojourn Time in Content-Centric Networking

作　　者：王国卿[1] 黄韬[1] 刘江[1] 陈建亚[2] 刘韵洁[1,2] 

机构地区：[1]北京邮电大学网络与交换技术国家重点实验室,北京100876 [2]北京邮电大学北京市网络体系构建与融合重点实验室,北京100876

出　　处：《计算机学报》2015年第3期472-482,共11页Chinese Journal of Computers

基　　金：国家"九七三"重点基础研究发展规划项目基金(2012CB315801);国家自然科学基金(61300184;61302089);中央高校基本科研业务费专项资金(2013RC0113)资助~~

摘　　要：内容中心网络(Content-Centric Networking,CCN)作为一种以内容为中心进行路由、缓存的新型未来网络体系架构受到了广泛的关注.在CCN中,关键技术问题之一是网络缓存问题,现有方案主要采用ALWAYSLRU缓存策略,然而该策略容易出现相邻节点重复缓存的问题,使得网络整体缓存效率较低.针对这一问题,文中提出了一种基于逗留时间的新型缓存决定策略,设计了一种适用于CCN的合作缓存机制.在请求泊松到达的假设下,通过在单个缓存器对LRU(Least Recently Used)替换策略使用马尔可夫链建模,该文得到了内容在各缓存器平均逗留时间的近似计算公式.数值仿真结果显示,该方案相比传统缓存策略,有效地提升了网络缓存的效率和缓存内容的多样性,进而减少了用户请求服务的总跳数,增加了内容访问的命中率.As a new Internet architecture,Content-Centric Networking（CCN）use the content name to route and cache,which has been attracted a lot of attention.In CCN,one of the most important issues is in-network caching problem.Although the ALWAYS-LRU cache policy is taken now,this policy leads to the neighbor nodes cache the same contents,which reduces the cache efficiency in the network.Therefore,the decision scheme based on Sojourn Time（ST）is proposed and a new cooperative cache policy is designed in this paper.By modeling the LRU replacement policy at a single node as Markov chain under the assumption of requests following Poisson distribution,the approximate expression of average Sojourn Time is derived.Finally,by comparing with the existing schemes,the numerical results are illustrated to show that the proposed scheme in this paper not only increases the cache efficiency and content diversity,but also decreases the total hops and increases the hit rate.

关 键 词：内容中心网络(CCN) 逗留时间 LRU 未命中率 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]