Peer Clustering System for Different Start Video Broadcasting
The different start video broadcasting is a new approach to manage video streaming applications. This approach allows unpunctual users to view broadcast programs from the beginning. This paper proposes a cluster model for different start video broadcasting. The model is composed of five processes: peer joining, super node selection, backup-node selection, download paths and leaving node process. The model is simulated and verified by NS-2. The results show that the video server load is diminished, the tracker load is reduced due to the existing of super nodes or clusters, the play out delay is only to 10 seconds and the bandwidth utilization is improved as the consequence of the reduced number of control messages.