Research On Cache Strategy In Named Data Networking

With the wide uses of Internet, the purpose of Internet usage has been changing from the original hardware resource sharing into the current resource acquiring. Because of this, the original networking architecture does not adapt to this new change, which lead to the urgent need of a new networking ar chitecture to meet people's access to the resource of networking. Information centric networking, as a new networking infrastructure that meet people's demand to acquire networking resource, also works out the TCP/IP networking issues of scalability, mobil ity and security. Named data networking(NDN), as a typical implementation of Information Centric Networking, has been unanimously looked up to by academia and industry.The main feature of Named Data Networking is that cached data package in router nodes can be requested by other nodes, which improve the utilization of networking resource by increasing the reusability of resource and avoiding duplicated transmission. In NDN, the efficiency of caching mechanisms directly determines the performance of networking, thus the research of caching mechanisms in NDN is of great significance. Since the default always caching strategy has the issues of high redundancy, and fierce cpmpetition of cache space in nodes etc, it is needed to study for more efficent caching strategies to reduce redundancy and increase hit rate of requesting.To address above problem, this research analyzes the factors that affect performance of networking, such as betweenness centrality of nodes, data package popularity and the distance between the current nodes and the requested nodes etc. By synthesizing these factors, this research build a probability caching model based on betweenness centrality and popularity(Prob-BP) which allows data package to decide whether to buffer on the return path by using this model; Then, by combining this probability caching model with consistent hashing algorithm, this research proposes a probability caching strategy based on consistent hashing(Prob-CH), which optimizes the caching of data packages from the aspects of neighboring nodes and requesting node path. In the aspect of neighboring nodes, the application of consistent hashing algorithm ensures that the maximum number of cached copy is only one, which greatly reduces the caching redundancy; In the aspec t of requesting node path, this research applies Prob-BP strategy. Finally, the research does simulation experiments on Prob-BP and Prob-CH strategies by using Icarus simulator, and analyzes these strategies by comparing them with those of Always, Prob etc. Experimental results show that Prob-BP strategy has better performance than Always and Prob strategies; meanwhile, Prob-CH strategy works better than Prob-BP and the Symmetric Hashing Strategy in terms of requesting delay and hit rate etc.