Research On In-Network Caching Machanism For Shareable Contents

Content-centric transmission characteristics in the Internet are becoming increasingly distinct,and the "end-to-end" connection based interconnection in current network makes it difficult to meet the increasingly demand of enhanced content transmission.Content Centric Networking(CCN),as a new type of network architecture with content naming,routing and in-network caching,has shown great potential in solving the problem of network expansion,security and mobility support,which is considered a probable solution to future network transmission requirement.As the core technology of CCN,researches on network cache mechanism have received wide attention from academia.However,from the perspective of supporting the future development of the network system,the existing network cache mechanism is in lack of researches on cache content selection,content popularity information acquisition and choice of content replacement in cache decision and cache replacement.This paper is based on the National Key Basic Research and Development Program(973 Program)-“Research on the Basic Network System of Reconfigurable Information and Communication” and “Fundamental Research on Intelligent Collaborative Network Theory”.By studying the definition of shareable content,the acquisition method and deployment mechanism of content popularity,on-path caching decision,intra-domain hybrid caching approach and cache replacement approach,content cache and replacement can be made more reasonable,thus improving the efficiency of cache system.The main research achievements in this paper are as follows:1.A popularity monitoring mechanism for shareable content is designed.First of all,according to the shortcomings of the research on the method to get the content popularity in CCN,the idea of using content classification for popular monitoring is proposed.Then the shareable content is defined,and the three-level monitoring mechanism based on Bloom filter and hash table is designed.Finally,in both theoretical analysis and experimental verification,the effectiveness of the three-level monitoring mechanism is proved.2.An on-path cooperative cache decision mechanism based on popular content recognition and tracking is proposed.Firstly,the evolution process of popularity is analyzed,and it is pointed out that the cache content popularity does not have global distribution characteristic,and the necessity of the popularity monitoring is analyzed.Then,the collaborative awareness scheme of on-path content popularity is proposed,where the relationship between the popular content threshold and the path length(hop count)is analyzed theoretically,and the cooperative caching decision mechanism based on the maximum betweenness node and the edge node is proposed.Finally,the effectiveness of the cache decision mechanism is verified by simulation experiments.3.A hybrid cache decision mechanism based on two-level popularity awareness is proposed.According to the topological features,the network is layered,and a virtual core area is constructed.After analyzing the demand of the content popularity awareness and the rationality of popularity monitoring node selection in the layered network,a two-level popularity awareness scheme is proposed,which theoretically analyzes the basis of popularity threshold in virtual core area.Finally,a hybrid cache decision mechanism is proposed to support content diversification and ordered content caching based on popularity awareness information.Simulation experiment is evaluated and the shortcomings of the cache decision mechanism are analyzed.4.A cache management mechanism based on dynamic node content popularity is proposed.First,the cache is divided into two parts.A popular content identification and trace method based on cache division and a cached popular content protection mechanism is devised.The relationships between the contents in the primary cache and the cache hit ratio are analyzed,and then to reduce the cache operation,cache management mechanism is improved.Finally,the basis of cache division and the accuracy,complexity and sensitivity of popular content identification are analyzed theoretically,and the corresponding simulation is carried out for verification.