In-advance Deployment of Shared Content Replicas over Hybrid Peer-to-Peer Network Using Linear Popularity Prediction

Abstract

This paper proposes an efficient content deployment method over a Peer-to-Peer network in advance, before issuing content retrieval requests from the users, according to the users' location and whose content preferences. In that situation, the proposed method predicts the temporal alteration of each content items' popularity using simple linear approximation and deploys each item to its appropriate place. As a result, the method can be expected that the network load, content loss, and storage cost required for redundant replica deployment will be reduced. On top of that, the proposal is evaluated by computer simulations from the viewpoint of those three costs mentioned above, and from the results, the effectiveness of the proposed method is finally discussed.