东北大学学报:自然科学版 ›› 2019, Vol. 40 ›› Issue (1): 21-25.DOI: 10.12068/j.issn.1005-3026.2019.01.005

• 信息与控制 • 上一篇    下一篇

基于Adaboost学习的ICN自适应缓存算法

蔡凌1, 汪晋宽2, 王兴伟3, 胡曦4   

  1. (1. 东北大学秦皇岛分校 控制工程学院, 河北 秦皇岛066004; 2. 东北大学 信息科学与工程学院, 辽宁 沈阳110819; 3. 东北大学 软件学院 辽宁 沈阳110169; 4. 东北大学秦皇岛分校 计算中心, 河北 秦皇岛066004)
  • 收稿日期:2017-05-24 修回日期:2017-05-24 出版日期:2019-01-15 发布日期:2019-01-28
  • 通讯作者: 蔡凌
  • 作者简介:蔡凌(1980-),女,湖南武冈人,东北大学秦皇岛分校讲师,博士; 汪晋宽(1957-),男,辽宁营口人,东北大学教授,博士生导师; 王兴伟(1968-),男,辽宁盖州人,东北大学教授,博士生导师.
  • 基金资助:
    国家自然科学基金资助项目(61501102); 河北省高等学校科学技术研究项目 (QN2014327).

Adaptive Caching Algorithm Based on Adaboost Learning for Information Centric Networking(ICN)

CAI Ling1, WANG Jin-kuan2, WANG Xing-wei3, HU Xi4   

  1. 1. School of Control Engineering, Northeastern University at Qinhuangdao, Qinhuangdao 066004, China; 2. School of Information Science & Engineering, Northeastern University, Shenyang 110819,China; 3. School of Software, Northeastern University, Shenyang 110169, China; 4. Computing Center, Northeastern University at Qinhuangdao,Qinhuangdao 066004,China.
  • Received:2017-05-24 Revised:2017-05-24 Online:2019-01-15 Published:2019-01-28
  • Contact: CAI Ling
  • About author:-
  • Supported by:
    -

摘要: 针对信息中心网络(ICN)中缓存内容优化放置的问题,提出一种基于Adaboost学习的自适应缓存算法ACAL.该算法首先将提取的节点和内容数据流作为网络资源,然后利用集成学习算法Adaboost对数据流进行分析挖掘,利用挖掘出的状态属性与缓存匹配之间的函数映射关系对未来时间段内的节点与内容间的匹配关系进行预测,该预测结果用于指导缓存的部署.实验结果表明,ACAL在延时、缓存命中率和链路利用率等指标方面,与CEE策略、LCD策略、prob0.5策略和OPP策略相比有显著的优势.

关键词: 信息中心网络, 缓存网络, 缓存策略, 学习算法, Adaboost算法

Abstract: In order to optimize the cache placement in ICN(information centric networking), an ACAL(adaptive caching algorithm based on Adaboost learning) algorithm was proposed. According to the algorithm, first, the extracted data flow including node data and content data was employed as the network resources, then the ensemble learning algorithm Adaboost was used to analyze and mine the data flow, and the mapping relationship between the state attribution data and the matching relationship value was utilized to predict the matching relationship between the node and the content in next period. Finally, the matching relationship algorithm was used to guide the cache placement. The simulation experiments demonstrate that the proposed ACAL, compared with CEE, LCD, prob0.5 and OPP yields a significant performance improvement, such as delay, hit rate and average link utilization.

Key words: information centric networking(ICN), caching network, caching strategy, learning algorithm, Adaboost algorithm

中图分类号: