UC Santa Cruz

All Information-Centric Networking (ICN) architectures proposed to date aim at connecting users to content directly, rather than connecting clients to servers. Surprisingly, however, although content caching is an integral of any information-Centric Network, limited work has been reported on information-centric management of caches in the context of an ICN. Indeed, approaches to cache management in networks of caches have focused on network connectivity rather than proximity to content.

We introduce the Network-oriented Information-centric Centrality for Efficiency (NICE) as a new metric for cache management in information-centric networks. We propose a method to compute information-centric centrality that scales with the number of caches in a network rather than the num- ber of content objects, which is many orders of magnitude larger. Furthermore, it can be pre-processed off-line and ahead of time. We apply the NICE metric to a content replacement policy in caches, and show that a content replacement based on NICE exhibits better performances than LRU and other policies based on topology-oriented de nitions of centrality.