Summary: | Named Data Networking (NDN) has been recognized as a lever to the Internet of Things (IoT). One of the most founding features of NDN is in-network caching to improve data availability and reduce retrieval delays and network load. Despite the existence of several caching decision algorithms, the fetching and distribution of contents with minimum resource utilization remains a great challenge. In this article, we propose an efficient caching technique named PoolCache that augments the effective caching capacity of some defined conglomerates of nodes. This is accomplished by pooling the various caches and manage them in a way to insure zero content redundancy within any defined node conglomerate. The resulting high diversity of cached contents throughout the network tacitly amounts to much better overall performances. We conducted extensive simulations using the CCNsim simulator to evaluate the performance of PoolCache and compare it to that of some well known caching strategies. Simulations using a large Transit Stub topology show that PoolCache clearly outperforms the other caching strategies in terms of a much greater content diversity and consequently a limited number of content evictions, a much better cache hit ratio, and a much lower content retrieval delay. Simulations also showed that PoolCache benefits from any eventual content popularity.
|