On Data Scheduling Strategies Of Serving Peers In P2P VoD Systems

P2P VoD(Peer-to-Peer Video-on-Demand) is a hot topic in the network application at present. Data request task scheduling is one of the important issues for P2P VoD systems. Most existing data scheduling strategies only consider the requesting peers and ignore the serving peers in which the requesting peers request the absent data from partners according to the BM and obtain the data before the playback point. The serving peers respond the requests according to the FCFS and EDFR strategies. So these strategies are not well suitable for P2P VoD because of real-time data requirement and highly dynamic user behaviour (such as VCR operation). This dissertation makes a detailed study on serving peer data scheduling strategies.Firstly, this dissertation presents admission control strategy combining the incentive mechanism. The serving peers respond the requesting peers that contribute more data to systems with priority and refuse some requests which can not be processed in time. The refused peers send the requests in time, so it reduces the startup delay and hinder the selfish peers to degrade the system; Aiming at the problem of unbalanced requests received by peers in the P2P systems with mesh topology, this dissertation presents an algorithm of data request task migration in which light-loaded peer help the overloaded peer to upload the data; In order to make full use of upload bandwidth, the dissertation designs a hybrid PUSH/PULL scheduling algorithm which consider data popularity and urgency ;At last, we design a P2P VoD prototype system based serving peer data scheduling which takes request task migration and hybrid PUSH/PULL scheduling into consideration.The simulation results show that, the proposed algorithms can improve the system performance with higher quality of media playback and the efficiency of upload bandwidth utilization. Meanwhile, it also reduces the server stress and startup delay. So it can improve users'experience and be significant meaning for promoting the application of highly scalable P2P VoD systems.