Proactive caching with mobility prediction under uncertainty in information-centric networks

Abstract

Proactive caching can be a key enabler for reducing the latency of retrieving predictable content requests, alleviating backhaul traffic and mitigating latency caused by handovers. In mobile networks, proactive caching relies on mobility prediction to locate the mobile device's next location and hence the node that must prefetch the content. Previously proposed proactive caching strategies use edge caching exclusively and cache redundant copies on multiple edge nodes to address prediction uncertainty. In this paper, we present a proactive caching strategy that leverages ICN's flexibility of caching data anywhere in the network, rather than just at the edge, like conventional content delivery networks. The main contribution of the paper is to use entropy to measure mobility prediction uncertainty and locate the best prefetching node, thus eliminating redundancy. While prefetching at levels higher in the network hierarchy incurs higher delays than at the edge, our evaluation results show that the increase in latency does not negate the performance gains of proactive caching. Moreover, the gains are amplified by the reduction in server load and cache redundancy achieved.