Research On Peer-to-peer Streaming Service Quality Support Method

Mesh-based streaming media system is the most general solutions of P2P streaming mediasystem based on its advantages of simple structure, low management cost and good anti-churncharacteristics. But the mesh-based streaming media system inevitably exists some defects:1. Mesh-basedP2P streaming media system overlay structure is constructed with randomnode, its anti-churn ability is very strong, but its video content transmission delay is large.Transmission delay is related with the transmission count and peer’s bandwidth. Both of themare not controllable to the mesh-based structure. This makes the users suffer time intervalbetween entering the channel and playing the video, and different peers have different playingtime points.If users cannot play videoin a long time, they will very likely to leave the channel.Video playing time lag will seriously affect the user’s viewing experienceforlive streaming.2.In mesh-based P2P streaming system, the nodes’ relationships are based onthe list ofcandidate nodeswhich is received from Tracker server. But the process of generating candidatenode list is completely random, which makes the quality of neighbor nodes has greatrandomness.Neighbor nodes do not have data block node required, or lack of bandwidth willcause the node cannot start the video playing in a long time or jitter during playback, effectingthefluency of video playback.3. Bandwidth capability is an important factor in determining the quality of service of P2Pstreaming media system.Many studies have tried to introduce the third party nodes to increasethe bandwidth. Combining the social network to p2p networks is the main research trends. Mostof the third party nodes lack of willingness to share bandwidth resources.Meanwhile, the thirdparty nodes lack collaboration strategy, they cannot effectively balanced distribution ofbandwidth resources. Sotheimprovement of overall system bandwidth is not obvious4. With the commercial P2P streaming system, the malicious attacksto the P2P streamingsystem increase gradually. Content pollution is one of the most dangerous malicious attacksthatit will cause jitter in video playing, large amounts of data block retransmission and occupy extrabandwidth. To defend against content pollution, reputation management systems are appeared.The existed reputation management systems have good performance against single attack. But ithas the problems of slow identification speed and low identificationrate. Malicious nodesreputation value exists fluctuations, it makes the reputation system cannot rapidly and accuratelyidentify the malicious nodes. And there will be confusionbetween polluted nodes andmalicious nodes, the polluted nodescannotconnect other nodes after they are back to normal.In this paper, we carried out a series ofresearch to solvethe defects above. The maincontributions and innovations are as follows:1. We proposed a dynamic hierarchical model and concurrent transmission algorithm basedon the P2P Live Streaming Media System. Allocating the nodes to the different layer accordingto node heterogeneity of upload bandwidth capability. The distances between nodes andstreaming server arebased on the node upload bandwidthability.Based on the dynamichierarchical modelwe design the concurrent transmission algorithm. The nodes in the same layercomposite the concurrent transmission node group. Thenodes in the same group have thefather-son relationship with each other, and transmit the data in push form. Thenode ofconcurrent transmission group can get all video content in less transmission time. It caneffectively improve the transmission efficiency at different levels, decrease the system startupdelay and playback delay.2. Wedesigned the node evaluation framework of P2P Live Streaming Media System. Atfirst, we improved communication protocol between the nodes and the Tracker Server, it allowsthe Tracker server to collect more comprehensive information of the nodes status in the network.According to characteristic variables of requesting node, Tracker server selectsnodes from thecandidate node groups to meet current and future needs of the composition of the initialcandidate list. With reference to the calculated results of the evaluation function and the datablocks scarcity of node, the candidate node list will be repeated filtering based on the differentrequirements, thereby generating a final list of candidate nodes. The filtered candidate node listhas better stability and availability. It can effectively control the distribution of the startup delayand playback delay in mesh-based structure P2P streamingsystem.3. Combining the social network and P2P network, and based on the characteristics of socialnetwork nodes, we proposedsocial resource sharing framework. The social resource sharingframework decides social connection object based on the social relevance. In order to guaranteethe smooth running of existing peersand bandwidth utilization of social peers, we designeddifferent buffer zone split strategy and multi-node partitioning work strategies which aredepending on the type of streaming media systems. The buffer zone split strategy divided thenode buffer into different regions, modifiedthe corresponding scheduling algorithm toaccommodate the video data to obtain of different areas. Social resource sharing framework addsnew bandwidth source to theP2P network, which can increase the effective bandwidth ratio, improve the transmission quality of P2P streaming media system.4. We proposed the Fussy Combine Reputation Management System-FCRepS to solve themalicious attack in P2P streaming system, mainly by turning the local reputation value intovarious trust state, and combining the change trend of local trust state and network trust state toidentify pollution nodes. In order to avoid accidental injury phenomenon in node identify process,we designed time based isolation recovery mechanism. The number of isolation time decides thenodes isolated time length, it can reduce the back to normal time of restored node. Fuzzy logic inthe FCRepS solved reputation value slow change problem which is due to diversity of networkevidence value in traditional distributed reputation management system, to achieveP2P networkdistributed reputation system from a new point of view.This paper is committed to solve the problem of P2P streaming media system Quality ofservice. The research content is mainly about hybrid transmission structure, neighbor quality,social network combing, malicious attack defend. We verified the proposed method through aseries of simulation experiments which are based on the OMNET simulation platform, theexperimental results show the researches have good effect on improving the quality of the peerstreaming system service. The researches have noticeably improvement relative to the previousresearches.These methods are not only enhance the quality of the P2P streaming mediasystemservice, but alsoprovides the reference basis for the design and improvement of highquality streaming media service, it hassome certain theoretical significances and applicationvalue.