Social VoD: A Social Feature-Based P2P System

Abstract

Video-on-demand (VoD) service has been explosively growing since its first appearance. For maintaining an acceptable buffering delay, the bandwidth costs have become a huge burden for the service providers. Complementing the conventional client-server architecture with a peer-to-peer system(P2P) can significantly reduce the central server's bandwidth demands. However, the previous works focus on establishing a P2P overlay for each video, producing a high maintenance cost on users. Per-channel-based overlay construction was first introduced by Social Tube, which clusters the users subscribed to the same video channels into one P2P overlay. However, the current per-channel overlay structure is not suitable for users developing new watching preferences. Consider that a channel's subscribers tend to watch not only the videos from the channel, but also other videos from similar channels. In this paper, we propose a new overlay structure by exploring the existing social relations of users and the similarities of video channels. Our system creates a hierarchical overlay: subscribers of the same channel form the low-level overlay (also known as groups), and in high-level overlay, different groups are connected based on their similarities. The new structure has the small-world property, the existence of which has been found in most data-sharing patterns. Based on the new structure, we propose a routing algorithm for both channel subscribed and unsubscribed users. Extensive simulation results show the efficiency of our approach.