Heuristic Approach to Determining Cache Node Locations in Content-Centric Networks

IEEE Access

Hiroki Tahara¹  Nattapong Kitsuwan¹  Eiji Oki² 

[1] Department of Computer and Network Engineering, University of Electro-Communications, Tokyo, Japan;Graduate School of Informatics, Kyoto University, Kyoto, Japan;

关键词: Computer networks;    content distribution networks;   

DOI  :  10.1109/ACCESS.2017.2779452

来源: DOAJ

【 摘 要 】

Internet users consider content information to be useful, but the current Internet approach treats location information as more important as so ties the former to the latter. A Content-Centric Network (CCN) allows the user to obtain content without regard to its location. CCN caches the contents information at its intermediate nodes. The content is searched along the shortest path between the user and the node that has the original content. If any cache node is located on the shortest path, the content can be obtained from the nearest cache node, so far fewer hops are needed compared to the network without any cache node. However, this efficiency is not achieved if no cache node is located on the shortest path. One proposal sets cache nodes that broadcast their contents to surrounding nodes; the user is able to obtain the content from the cache node, rather than the node that has the original content, if the cache node is closer to the user. The location of the cache node affects the number of hops. We formulate an optimization problem that determines the locations of cache nodes to minimize the hop count as an integer linear programming (ILP) problem. Since large ILPs cannot be solved in practical time, we introduce a betweenness centrality (BC) approach that determines the location of cache nodes by computing the BC value of each node and ranks the nodes in descending BC order. The BC value denotes the ratio of the number of shortest paths between source-receiver pairs passing through the node to the total number of shortest paths between source-receiver pairs. Simulations show that the BC approach offers drastically reduced computation time, while the average number of hops is just 5.8% higher than that determined with the ILP approach.

【 授权许可】

Unknown