Effective cache management and performance limits in information-centric networks

Abstract

Proliferation of content distribution traffic in the Internet is changing the requirements on the networking infrastructure to the extent that radical new architectures are envisioned, where information is addressed by semantic attributes rather than origin and destination identities. In that volatile environment, Information-Centric Networking (ICN) is a flexible communication model that meets the requirements of the new information distribution trends. Information is residing at various locations-nodes (surrogate servers) and users interested in a piece of information request it from the closest replica. In this paper, we propose a distributed, on-line gradient descent type cache management algorithm, designed for ICN, which re-assigns the information items between the nodes of the network with the aim to minimize the overall traffic cost of the information when the popularity and the locality of users' demand change. We also provide a method to calculate the overall network traffic cost lower bound for a certain category of network topologies, so that we can compare the performance of the proposed cache management algorithm.