Object management policies in P2P-CDN networks

In this paper, we study the effectiveness of object management policies based on users' personal signals, available on YouTube-like content distribution systems, for improving content availability in P2P-CDN networks. Three policies are introduced and evaluated using real data collected from YouTube video collection, the most accessed multimedia distribution system in the Internet. We found those policies can improve content availability, over 60%, compared to the LFU policy, and over 25%, compare to the GDSP policy, in hybrid P2P-CDN networks.