COD: caching on demand in information-centric networking

Abstract

By overcoming the well-known challenges of the current internet, great expectations are focused on information-centric networking (ICN). ICN extensively uses universal in-network caching. However, developing an efficient caching scheme remains an open question. To overcome the useless caching and duplication caching of previous caching schemes, we propose an adaptive caching scheme—caching on demand (COD). By following the change in potential demand from the consumer and the temporal patterns of content popularity, COD allows content to be cached only by some necessary nodes instead of all nodes on the path from content provider to content consumer. At the same time, in accordance with trading off bandwidth for cache (bandwidth-for-cache), content can be pushed to the adjacent node with more cache capacity. We present a theoretical model to evaluate cache usage for COD. Finally, we evaluate COD through extensive experiments and a wide range of performance metrics. The experimental results under diverse setting demonstrate that COD can yield a steady improvement of network performance and caching efficiency compared with CEE, EgoBetw and Probcache. Notably, COD improves performance with negligible overhead.