FERA: A caching scheme in CCN using file-extension and regression analysis

Content-Centric Networking(CCN) is emerging technology for the future network infrastructure. In CCN, CCN routers support cache space called Content Store(CS), which can store passing contents. In this paper, we define two representative problems that can occur in CS of CCN routers, namely content search problem and cache replacement problem, and propose FERA(File Extension and Regression Analysis based Caching Scheme) which can solve these problems. In FERA, two approaches are proposed. Specifically, the first proposal is to divide CS into four types of CSs based on file extensions. Thereafter, when the user sends a request including the content name, the CCN router extracts and identifies the file extension from the content name, and transmits the user's request to the corresponding CS. After being transmitted, the user request is processed in the corresponding CS. The second proposal is to predict the content that is continuously decreasing in popularity based on regression analysis which is one of the machine learning algorithms and evicts that content. Finally, we verify the proposed scheme using CCN simulator. The simulation results show that the proposed scheme improves the cache hit ratio and the cache hit distance over the LRU.