Enabling a Metric Space for Content Search in Information-Centric Networks

The emergence of vivacious bandwidth-consuming services leveraged by the overwhelmingly content-centric usage on Internet has fostered over last years a new foundation for networking architecture design. Several Information-Centric Networking (ICN) architectures have then been proposed to tackle the challenging problems for content dissemination and distribution. Those architectures propose a forwarding plane based on a naming scheme that uniquely identifies content objects in the network regardless of their location. Routing in ICN is thus performed on content names by means of a name resolution service that can be operated independently on the current endpoint-centric networking infrastructure. In order to sketch out the efficient design of such information-centric networks, an important functionality need to be taken into account for resolving information usability which consists in finding appropriate content objects matching users' interest. We propose in this paper a novel forwarding scheme which integrates content search in the network path when resolving content names into locators (where content is actually stored). The content indexing and discovery function relies on distance sensitive hashing based DHT for similarity search between content objects. The proposed routing algorithm can be used for better performance than using classical DHTs for resolving names in ICN. We conceptually embed the high-dimension set of content objects into a low-dimension for reducing processing complexity, which allows us to design an efficient and scalable information-centric forwarding plane with in-network content search.