Research On Peer Selection Strategy Based On User Behavior Capacity In P2P Streaming Media System

With the rapidly development of multimedia and Internet,the streaming media is used more and more broadly,especially in Internet. In recent years, the emergence and wide application of P2P and it's correlation techniques which promotes the new development of streaming media technology,nowadays, the P2P streaming media system becomes a research hotspot.In this paper,the existing P2P streaming media system structure and apply are analysed and summarized,based on which,the principal problem—node selection strategy is invesgated.It is well known that characteristics of streaming media are to download and play simultaneously, it is different from other file schedules, the real-time is very important. So when the algorithm is designed, it is needed to consider startup delay. To solve the above problems, a lot of research begin to consider the perspective of user behavior. User behavior, including the request arrival rate, the popularity of video objects, the number of user requests and users online time.At present,there has been some research for data scheduling. Such as random strategy, scarce priority strategy, etc.But, these strategies study on the perspective of bandwidth, and no research on customer's. In fact user behaviour has certain predictability, if the node chooses to consider these factors during selecting, can improve scheduling performance, to a certain extent ,reduce system dithering.The mainly considered consumer behavior ,in this paper, is users online time,users quit the system not just after reading all video,the study on users online time finds approximate to logarithmic normal distribution. Therefore,when choose service node, the request node can choose the node much more likely to continue to online,which can reduce the network jitter caused by service node leaving.The request sequence of data block should be determined first in algorithm design, when determine the request order,a comprehensive consideration of resource scarcity and data block priorities should be taken, calculate the node tolerance,then select those whose online time continue to be the biggest as provider,which can reduce the network jitter caused by the providers exit during transmitation lead to request requesting node.The simulation results prove that the method in this paper,to a certain extent,enhance the stability of the system,improve the efficiency of the system.The mainly innovation points of this paper :The data block torlerance is put forward,through which to decide block request order,and effectively reduce the delay. The selection strategy is put forward, which is according to online time,service capability,service probability to select node,which improve the system stable performance.