MBP: a Max-Benefit Probability-based Caching Strategy in Information-Centric Networking

Nowadays, Information-Centric Networking (ICN) has attracted more and more attention, which allows named data to be cached within the network. Existing works mainly focus on decreasing the redundancy of replicas to enhance the cache hit ratio, while pay less attention to cache benefit maximization and often bring about frequent cache operations. In this paper, we first formulate the content placement problem and find two key factors, i.e., the content popularity and the content placement benefit. Then we propose a heuristic probability-based caching strategy, called MBP (Max-Benefit Probability-based Caching). In MBP, each cache node caches the passing content with certain probability, which is proportional to the content popularity and the content placement benefit. We evaluate MBP via extensive simulations by comparing it with state-of-art caching strategies under tree and graph topologies. The experimental results indicate that MBP can achieve great improvement compared with other caching strategies, in terms of average cache hit ratio, average access hop ratio, caching operation and link stress. Especially, when the cache size is small, MBP can also achieve dramatically performance improvement.