The Research Of Node Selection Strategy In P2P Streaming System

With the development of computer and information technology, the user's demand for multimedia content is increasing. In the traditional C/S model, this model can not provide service quality assurance and meet the needs of customers.With the expansion of the user, P2P (Peer to Peer) technology appears to solve this problem, The applications of P2P streaming media that can take full advantage of the idle resources in the network nodes, each node is both client and server, which can provide data to other nodes and accept data. However, the heterogeneity and dynamic features of P2P streaming system also brings problems and challenges, In the P2P network, the joining and leaving nodes at any time may result in interruption and instability of service. How to guarantee the stable quality of service is the main problem in P2P streaming media systerm.Based on the above-mentioned problems, this paper analyzes the stability of node, and based on the original data statistics, we can see that offline probability of the node in the beginning of video is larger According to this characteristic,we design two characteristic parameters:online duration and broadcast fluency. We set up corresponding threshold value and specific algorithm to predict offline behavior of the node and broadcast fluency, and we choose nodes that off-line probability is smaller and broadcast fluency is better. Based on this kind of choice strategy, in this paper it introduces in detail the system design and algorithm procedures. It will select node that offline probability is smaller to request node for service. That can greatly improve the stability of nodes and guaranteed the quality of streaming media service.In order to evaluate the proposed node selection strategy, we build the system in P2P streaming media prototype system and it compares the random selection with offline prediction selection.At last, we analyze and compare two performance parameters (the play delivery and the playing fluency), that explains this node selection strategies has a certain superiority.