[Paper] Efficient In-advance Shared Content Deployment Method Using Popularity Prediction in Hybrid Peer-to-Peer Network with Cloud Storage

Abstract

In content sharing with Peer-to-Peer (P2P) network systems, content has been smoothly acquired and prevented from being lost by replication of the content and deploying replicas on multiple peers. However, in some conventional methods, it takes a long time for the replicas of content to be spread over the network, which makes stable content sharing difficult. In this paper, we propose a method to predict the change in popularity of each content item over time and to deploy an appropriate number of replicas of popular content evenly on the network before the supply becomes insufficient. The method also keeps the ratio of each content replica's total stored capacity to the whole storage capacity in P2P in proportion to its demand, while considering each replica's size of capacity. Furthermore, we confirm the effectiveness of the proposed method by computer simulations.