New Efficient Caching Strategy based on Clustering in Named Data Networking

Abstract

The Named Data Networking NDN is one of the most proposed architecture for the new model of Internet communications based on contents distribution, called Information-Centric Network ICN. It is widely accepted by the research community since it has become dominant in ICN design that resolves TCP-IP based Internet problems such as bandwidth, delay, location dependent and congestion. Based on location host IP addresses, TCP-IP designed for Peer-to-Peer communication P2P. NDN architecture is oriented Content Centric Networking CCN, where the data is stored on routers and distributed to users from the nearest router.  Cache capacities of routers are limited compared to forwarded contents. To move from TCP-IP model to CCN model, many papers propose several new contents distribution based architecture ICN. In this paper, we propose a novel strategy to optimize the use of network resources inspired from Network clustering and cluster head selection in MANETs. Specifically, the improved K-medoids cluster algorithm is used to divide the global network in clusters, where for each cluster; three routers are selected as content routers. The first is the main caching router as well as the second and the third are the secondary caching router. The caching router selection process relies on three relevant criteria consisting of the distance between a node and its cluster centroid, the number of neighbors, and the congestion level. Two Multi Attribute Decision–Making methods MADM are applied, namely TOPSIS and AHP. Performance analysis of our proposed strategy with the established criteria showed  � its effectiveness and strong potential.