Premium
An exploration of cache collaboration in information‐centric network
Author(s) -
Feng Bohao,
Zhou Huachun,
Gao Shuai,
You Ilsun
Publication year - 2014
Publication title -
international journal of communication systems
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.344
H-Index - 49
eISSN - 1099-1131
pISSN - 1074-5351
DOI - 10.1002/dac.2773
Subject(s) - computer science , cache , cache algorithms , cache invalidation , smart cache , computer network , page cache , cache coloring , cache pollution , distributed computing , cache oblivious algorithm , cpu cache
SUMMARY As one of key differences from traditional Internet, information‐centric networking introduces the network built‐in cache in network nodes. It is not proper to analyze a cache node isolated from cache neighbors, especially for the named data network in which all network nodes can cache the content. In this paper, we try to look insight into such an issue and propose a way of cache collaboration not only for the on‐path but also for the off‐path routers. The autonomous cache is not efficient and cache routers are required to guarantee the time of the cached content for the stability as well as the reliability. Therefore, we firstly introduce an interest counter, on one hand, to rank different contents within a period, in this way, we can use the advantage of the least frequently used (LFU) as well as avoid its cache pollution problem, on the other hand, to reduce the rate of content replacement and to maintain a relative stability in forwarding plane, thus, the cached content can be easily managed by the control plane. Secondly, we utilize the interest‐prune and the steady content advertisement mechanism in the control plane based on the interest counter in order to eliminate the duplicated cached content along en‐route routers and cache neighbors, as a result, more cache capacity for different contents can be saved and efficient cache collaboration can be obtained. We evaluate the performance of our scheme in line, tree and grid topology scenarios. The simulation results show that our scheme performs well compared with the perfect‐LFU, the LFU, the random, and least recently used replacement policies. Copyright © 2014 John Wiley & Sons, Ltd.