【摘要】 针对如何提高信息中心网络的网内缓存性能,提出了一种基于概念漂移学习(conceptdriftlearning,简称CDL)的自适应缓存策略.考虑到节点数据和内容数据的相互感知对缓存性能的影响,将节点和内容的状态数据流作为网络资源,对提取的多维状态属性数据和缓存匹配数据进行分析挖掘,利用学习到的状态属性与缓存匹配之间的函数映射关系,即概念,对未来时期内的节点与内容间的匹配关系进行预测.为提高匹配算法的准确度,在学习过程中,提出了一种基于信息熵的概念漂移识别算法,当根据状态属性的信息熵变识别出漂移后,利用提出的基于概念重现的缓存算法,重新定义函数映射关系.仿真实验结果表明,该策略与CEE,LCD,prob和OPP策略相比,降低了网络运行成本,提高了用户体验质量.更多还原

【Abstract】 In order to improve the caching performance in information centric networks, an adaptive caching strategy based on concept drifting learning(CDL) was proposed. Considering the supplementary action of the node data and content data on improving caching performance, firstly, the status data flow of nodes and content were used as network resources, and then the mapping relationship, namely concept, between the multidimensional state attribution data based on the status data flow and the matching relationship value was mined. Finally, utilizing this mapping function, a matching algorithm to predict the matching relationship between the node and the content in the next time period was proposed. In order to improve the accuracy of the matching algorithm, a concept drifting detection algorithm based on information entropy was proposed. When the concept drifting of the state attribution data by the information entropy was captured, a new mapping relationship was learning by the proposed recurring concept caching algorithm. Simulation results show that CDL outperforms CEE, LCD, Prob, and OPP when looking at cost reduction of network operation and enhancement in quality of user experience.更多还原