Research On Push-based Scheduling Algorithm For Peer-to-peer Live Streaming Media

The Internet streaming media service becomes more and more popular andattractive because of the development of the Internet. It can be divided into two types:Internet television and Video-on-Demand. The Internet television desires the real-timeplayback of media, and requires higher data transmission efficiency than theVideo-on-Demand. The P2P technology acts as the basic technical support for livestreaming media transmission. In P2P network, one children peer may have severalparent peers. Each children peer provides high efficiency and good performance forthe entire transmission network by scheduling the data from different parent peersreasonably. There are two data transmission mechanisms: Pull and Push. Theprocesses of periodic request and response in Pull mechanism may lead to long datatransmission delay and media playback delay. However, the Push mechanism needn’tthese processes. So that the Push-based scheduling algorithm for P2P live streamingmedia helps to improve the quality of live streaming media service and userexperience. It has great research significance and application value.It is difficult to come up with a global scheduling optimization scheme, because ofthe complex network environment and the various factors impacting on datatransmission. This thesis proposes a local optimum solution to achieve theapproximate global optimization goal. At the same time, a Push-based schedulingalgorithm for P2P live streaming media is studied and analyzed. And the bufferlimitation and network instability problems are solved by a simpler packet recoverystrategy. This thesis obeys a series of standards for quantizing and analyzing the performances of scheduling algorithms, and conducts an experiment with a networksimulation software NS-2. After comparing the improved Push-based algorithm withtwo Pull-based algorithms in different simulative network environment, we can getthe experiment results. The results show that the optimized Push-based algorithm canachieve lower delay, lower delay peak, less bandwidth utilization, better packetrecovery and less control overhead than the Pull-based algorithms.