Cache-aware traffic engineering in Information-Centric Networks

The current Internet trends show that what is exchanged is becoming more important than who are exchanging it. As a result, the Internet is effectively moving from interconnecting machines to interconnecting information. Information-Centric Networking (ICN) has been proposed as a paradigm shift from the host-to-host Internet to a host-to-content one, or in other words from an end-to-end communication system to a native distribution network. The added value of the ICN architecture is that it allows Network Service Providers (NSPs) to participate in the actual content delivery, instead of simply being data-agnostic connectivity providers. This paper investigates an intra-domain traffic engineering (TE) approach for ICNs by leveraging the caching/replication of information items at the nodes of the network. Particularly, we present two variations of a distributed, on-line gradient descent type cache management algorithm, designed for ICNs, which re-assigns the information items between the nodes of the network with the aim, either to minimize the average link stress of the network, or to minimize the link stress (min-max) of the most congested link. Both approaches enhance the service capability of the network (high throughput and low latency) and minimizes congestion. Our extensive simulations, using realistic network topologies and synthetic workload generators, show that the proposed algorithms can significantly reduce information items' transfer time and link utilization.