Research On Caching Models And Strategies In ICN For Content Consistency

With the rapid growth of the Internet,the traditional TCP/IP architecture has become difficult to cope with the growing user demand.The ubiquitous caching in the Information-Centric Network(ICN)brings content closer to users,effectively reducing user access latency,reducing link load,and improving network performance.In realworld applications,data updates can cause out-of-date content to appear in the cache,requiring the use of a strong cache consistency strategy to ensure that users get the latest content.Establishing an accurate analytical model of the cache is the basis for analyzing network performance and performing network planning.The few studies on cache strong consistency models have been conducted only based on LRU cache replacement policies or have only considered the cache performance of a single node.In addition,the current study of the content copy in the cache node decision strategy rarely considers the impact of content updates,which cannot ensure that users request the latest content while the caching system also has high performance,while the design of complex policies add a large number of fields,increasing the cost and difficulty of deployment.To address the above problems,this dissertation establishes a mathematical analysis model of cache strong consistency and gives an analysis method of cache performance at the theoretical level.Meanwhile,a cache decision strategy based on the attributes of cache nodes and content copies is designed based on the consideration of content consistency,and the specific research work is as follows.(1)Based on the cache strong consistency problem,an analytical model under a generic cache replacement policy is proposed for all cache replacement policies that can be approximated by characteristic time.The model provides the calculation of cache hit ratio and server load under three cache strong consistency policies by comparing the size of update interval and characteristic time.In the model,the time interval at which content update events occur can satisfy an arbitrary distribution.The model is analytically derived and experimentally validated by considering four common cache replacement policies in single-node cache and four types of mixed cache replacement policies in two-level cache,and the model can achieve high accuracy.(2)The Differentiated Caching Strategy Based on Content Consistency and Node Attributes(DCCNA)is designed to address the shortcomings of existing caching decision strategies.The strategy integrates the effects of popularity and update cycles from the perspective of content,and considers the proximity of cache nodes to centrality in the network topology graph from the perspective of cache nodes to select important hub nodes in the network topology.So that different contents are stored differently in different nodes to make the network high performance with lower resource consumption.The dissertation has conducted a large number of simulation experiments to prove the high accuracy of the cache strong consistency model by comparing the model computation results with the simulation results,and also to explore the variation law of network performance in different caching systems.The proposed caching decision strategy achieves higher total cache hit ratio,lower server load and fewer average hops than other commonly used caching decision strategies when the content is updated frequently.