Toward user-classified P2P IPTV systems: A persona-based approach

Peer-to-Peer (P2P) approaches are a promising choice for media streaming in order to reach a global audience and to ease the deployment of such services. However, since P2P systems are in fact networks of users where the latter directly control the peers, user behavior is critical to the performance of these systems. Current literature propose global behavior models which do not help greatly to design adaptive control mechanism at peer level. In this paper, we propose a classification mechanism that associates a video streaming user to a behavioral class. This classification can help in building user-aware systems optimized for an improved streaming quality. We validate our model through simulations over a thousand users on a hundredday period. We also present a proof-of-concept application of our classifier for the design of churn-resilient topologies that can be used by service providers or network operators in P2P IPTV applications.