电子学报 ›› 2016, Vol. 44 ›› Issue (11): 2704-2712.DOI: 10.3969/j.issn.0372-2112.2016.11.020

• 学术论文 • 上一篇    下一篇

基于节点动态内容流行度的缓存管理策略

张果1, 汪斌强1, 张震1, 梁超毅2   

  1. 1. 国家数字交换系统工程技术研究中心, 河南郑州 450002;
    2. 解放军信息工程大学, 河南郑州 450001
  • 收稿日期:2015-04-16 修回日期:2016-01-07 出版日期:2016-11-25
    • 通讯作者:
    • 张果
    • 作者简介:
    • 汪斌强,男,1963年2月出生,安徽安庆人.国家数字交换系统工程技术研究中心教授、博士生导师,主要研究方向为宽带信息网络,网络安全.E-mail:wbq6301@163.com;张震,男,1985年出生,博士,讲师,研究方向为未来网络体系架构设计,网络测量;梁超毅,男,1978年出生,助教,研究方向为未来网络体系架构设计,内容中心网络.
    • 基金资助:
    • 国家自然科学基金创新研究群体项目 (No.61521003); 国家973重点基础研究发展计划 (No.2012CB315901,No.2013CB329104); 国家自然科学基金 (No.61372121,No.61309019,No.61309020,No.61572519); 国家863高技术研究发展计划 (No.2015AA016102,No.2013AA013505)

A Strategy Based on Dynamical Content Popularity for Cache Management

ZHANG Guo1, WANG Bin-qiang1, ZHANG Zhen1, LIANG Chao-yi2   

  1. 1. National Digital Switching System Engineering & Technological R & D Center, Zhengzhou, Henan 450002, China;
    2. PLA Information Engineering University, Zhengzhou, Henan 450001, China
  • Received:2015-04-16 Revised:2016-01-07 Online:2016-11-25 Published:2016-11-25
    • Supported by:
    • NSFC Innovation Research Group (No.61521003); National Key Basic Research Program of China  (973 Program) (No.2012CB315901, No.2013CB329104); National Natural Science Foundation of China (No.61372121, No.61309019, No.61309020, No.61572519); National High-tech R&D Program of China  (863 Program) (No.2015AA016102, No.2013AA013505)

摘要:

针对命名数据网络中节点无法感知内容流行度变化的缺陷,提出了基于缓存内容流行度动态变化的内容管理策略.将缓存分为主缓存(Primary Cache,PC)和副缓存(Secondary Cache,SC),分别用于识别和保护流行内容;采用标准布鲁姆过滤器(Standard Bloom Filter,SBF)过滤流行内容请求;引入滑动时间窗口算法和HASH表对副缓存内容进行细粒度的统计分析,进而管理缓存内容.仿真显示,与现有算法相比,该策略以增加少量复杂度为代价,延长高流行度内容的缓存驻留时间,提高了缓存命中率,减轻了服务器负载,并具有可扩展性,具备单线路40Gbit/s的报文处理能力.

关键词: 命名数据网络, 动态内容流行度, 线速, 内容管理

Abstract:

To overcome the drawback that nodes in Named Data Networking are insensitive to the change of the content popularity,a dynamic content popularity based cache management strategy is proposed.The strategy divides the cache into primary and secondary one.The former is used to identify popular content and the latter is used to protect it.Standard Bloom Filter is adopted by the strategy to filter popular content requests.The strategy also introduces sliding window and hash table to analyze the content of secondary cache in fine granularity and manage the cache content.Simulation results show that,compared with traditional strategies,our algorithm prolongs the cache residence time of high popularity content,increases cache hit ratio and reduces server loads.Our algorithm is also scalable and has the ability to process packets at 40Gbit/s.

Key words: named data networking, dynamical content popularity, line speed, content management

中图分类号: