Research And Implementation Of Distributed Streaming Media System Based On Reputation Mechanism

Streaming media has become one of the most important ways of entertainment in people's daily life.There are many centralized servers in traditional C/S model,so it's vulnerable to be attacked.To solve this problem,researchers have tried to apply P2 P technology to streaming media.In this way,the pressure of central servers can be spread to the users in network.P2 P streaming media can use the bandwidth of nodes in the network to distribute resources.But there are some problems such as the instability of network,start delay and the difficulty to obtain the last block.Based on the analysis of existing P2 P streaming technology,we implement a distributed video on demand system and optimize the main algorithms in streaming media.Aiming at the dynamic problem of P2 P network,we propose a node selection strategy based on reputation mechanism.The algorithm evaluates the nodes by integrating global credibility and global credibility.It evaluates the credibility of nodes who giving feedback as well.The algorithm selects higher trusted nodes in the network to participate in data exchange.It can reduce the probability of choosing malicious nodes.For the nodes with similar reputation value,it counts the number of file segments to choose peers.For the problem of delay and the difficulty to obtain the last part of file,we add several backup servers to cache scarcity fragments for emergency.It can effectively solve the problem of uneven distribution of resources.We build a distributed video on demand system based on the Bit Torrent protocol and use the proposed node selection strategy.The system contains three modules,including video distribution module,tracking server module and seed management module.The video distribution module mainly distributes resources to remote backup servers and maintains video resources.The seed management module manages the video files and make seed files for the new videos.On the one hand,the tracking server statistics the peers' download information.On the other hand,it returns the list of peers to the request node.The experiments show the platform is of practical value and can meet the needs of users to watch video smoothly.We use Peer Sim simulation environment to do contrast experiments.In order to simulate the dynamics of real network,five nodes of every 10 s are randomly moved out from the network.The experiments show that the node selection strategy has greatly improved in download speed,start delay and other indicators compared to the original Bit Torrent algorithm.