Markov Model Research Based On Information Forecast Of P2P Node

With the rapid growing of Internet, P2P network technology is becoming more and more important. Comparing to traditional C/S pattern, personal computer (called peer) is client and server under P2P(peer to peer)technology, P2P greatly improve the serving ability of Web servers. All these facts make p2p repaid development in recent years. According to the the national internet report, P2P makes up 60 per cent of the network flowAs the internet research department ipoque in German say, P2P have absolutely held the internet nowadays. It is about 50 to 90 percent of the total flow from P2P program. The fact is that the application of VOD and live play a very import role in P2P flow. Users are often nagged by the low speed of download, unstability of the flow, and stumble of video in real live or VOD. The system performance is influenced by many factors restriction. Peer churn is one of the most importan elements. When mobile node (user) import to the live or VOD network, the other nodes are uncertainty on data transmission to the incomer.In order to solve the above problems, this paper tries to apply birth-death process in random process principle for the mothod that establish an information prediction of probabilistic parameters, then bulid markov model to mobile node (users), the father node (information providers) sets up a markov model, then comparative analysis of the two models, estimate the father node information to mobile node's probability, provide a reference for the mobile node in a certain extent, to ensure video watching is fluent and stable in mobile node.The main contribution of this paper:Applying birth-death process in random process principle to establish markov model on mobile node(the user) and father node (information providers), presenting detailed mathematical description, the important conclusion about the complete mathematical derivation on these two models, and the algorithm of achieveing P2P node information forcast.Combination of mobile nodes (users), the parent node (information provider) both subject and object, the probability at Positive and negative aspect between mobile node (users) and father node (information providers) on sending information, which makes the node probability evaluation more objective and comprehensive.Realized algorithm of P2P node information forcast by matlab and java.Obtained the maximum of k (markov model of the explore of informat-ion forecast based on mobile node) and the minimum of k (markov model of the explore of information forecast based on father node) under certain conditions to ensure the video is smooth and stable.