P2P流媒体网络拓扑研究
发布时间：2019-02-09 08:31
【摘要】：随着互联网宽带技术、IPV6、3G等技术的飞速发展，流媒体已成为Internet承载的重要任务。对等网络流媒体系统的低消耗和高可扩展性很好的解决了传统流媒体系统中服务器和网络的高压力问题，使得为用户提供高质量网络视频服务成为可能。在随机的拓扑结构中，由于邻居选择的随机性和不确定性，使得网络系统中存在主干网络压力大、启动延迟和播放延迟长以及帧丢失率高等问题。 针对随机对等网络中主干网络压力大、启动延迟和播放延迟长以及帧丢失率高等问题，本文首先提出了一种基于RTT(Round-Trip Time,RTT)优先位图匹配结合的邻居选择算法。该算法利用Tracker服务器存储整个网络中所有节点之间的RTT信息和位图(bitmap)信息，当节点向请求邻居列表时，Tracker服务器通过节点的RTT信息和位图信息及邻居筛选函数为节点生成最佳邻居列表。通过OMNET++平台仿真实验显示，与随机邻居选择算法相比，基于RTT优先位图匹配结合的P2P邻居选择算法具有较小的启动延迟和播放延迟，同时很好地降低了帧丢失率和系统平均跳数。 同时，本文应用节点之间的RTT信息，，对CoolStreaming算法进行优化，提出一种基于RTT优化的CoolStreaming优化算法。该算法在节点选择新的父节点时加入RTT考虑，如果满足CoolStreaming算法原有的限制条件的伙伴节点个数大于1，则节点获取与这些伙伴节点之间的RTT，选择与请求节点之间RTT最小的伙伴节点作为新的父节点，请求服务。通过OMNET++平台仿真实验显示，与随机邻居选择算法和CoolStreaming算法相比，基于RTT优化的CoolStreaming算法虽然在启动延迟上没有很好的提高，但是在帧丢失率等方面，取得了很明显的改善，大大降低了帧丢失率和平均跳数。 最后，本文提出一种基于用户等级的P2P流媒体激励机制。通过用户等级将节点分类，当节点加入时，根据用户等级构建分层P2P拓扑结构，将用户等级高的用户放在离源服务器较近的层，使得其获得较好的服务质量，从而激励整个系统中的节点。通过OMNET++平台仿真实验显示，基于用户等级的P2P流媒体激励机制中用户等级高的节点的启动延迟和播放延迟较低；同时与随机邻居选择算法，具有较小的启动延迟和播放延迟。
[Abstract]:With the rapid development of Internet broadband technology and IPV6,3G technology, streaming media has become an important task for Internet. The low consumption and high scalability of peer-to-peer streaming media system solve the problem of high pressure of server and network in traditional streaming media system, which makes it possible to provide high quality network video service for users. In the random topology, due to the randomness and uncertainty of neighbor selection, there are some problems in the network system, such as high backbone network pressure, long startup delay and playback delay, and high frame loss rate. In this paper, a neighbor selection algorithm based on RTT (Round-Trip Time,RTT) priority bitmap matching is proposed to solve the problems of high backbone network pressure, long startup delay and playback delay and high frame loss rate in random peer-to-peer networks. The algorithm uses Tracker server to store RTT information and bitmap (bitmap) information between all nodes in the whole network. The Tracker server generates the best neighbor list for the node through the RTT information, bitmap information and neighbor filter function of the node. The simulation results of OMNET platform show that compared with the random neighbor selection algorithm, the P2P neighbor selection algorithm based on RTT priority bitmap matching has smaller startup delay and playback delay. At the same time, the frame loss rate and the average hops of the system are well reduced. At the same time, using the RTT information between nodes, this paper optimizes the CoolStreaming algorithm and proposes a CoolStreaming optimization algorithm based on RTT optimization. The algorithm adds RTT to the selection of new parent nodes. If the number of partner nodes satisfying the original restriction of the CoolStreaming algorithm is greater than 1, then the node acquires the RTT, between these partner nodes. Select the smallest partner node of RTT between the requesting node as the new parent node to request the service. The simulation results of OMNET platform show that compared with the random neighbor selection algorithm and the CoolStreaming algorithm, the CoolStreaming algorithm based on RTT has not been improved in the starting delay, but the frame loss rate has been improved obviously. The frame loss rate and average hops are greatly reduced. Finally, this paper proposes a P2P streaming media incentive mechanism based on user level. The nodes are classified by user level. When nodes are added, the hierarchical P2P topology is constructed according to the user level, and the users with high user level are placed in the nearest layer to the source server, so that they can get better quality of service. Thus motivating the nodes in the whole system. The simulation experiments on OMNET platform show that the node with high user level has lower startup delay and playback delay in P2P streaming media incentive mechanism based on user level, and with random neighbor selection algorithm, it has small startup delay and playback delay.
【学位授予单位】：吉林大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TP393.02
本文编号：2418888
[Abstract]:With the rapid development of Internet broadband technology and IPV6,3G technology, streaming media has become an important task for Internet. The low consumption and high scalability of peer-to-peer streaming media system solve the problem of high pressure of server and network in traditional streaming media system, which makes it possible to provide high quality network video service for users. In the random topology, due to the randomness and uncertainty of neighbor selection, there are some problems in the network system, such as high backbone network pressure, long startup delay and playback delay, and high frame loss rate. In this paper, a neighbor selection algorithm based on RTT (Round-Trip Time,RTT) priority bitmap matching is proposed to solve the problems of high backbone network pressure, long startup delay and playback delay and high frame loss rate in random peer-to-peer networks. The algorithm uses Tracker server to store RTT information and bitmap (bitmap) information between all nodes in the whole network. The Tracker server generates the best neighbor list for the node through the RTT information, bitmap information and neighbor filter function of the node. The simulation results of OMNET platform show that compared with the random neighbor selection algorithm, the P2P neighbor selection algorithm based on RTT priority bitmap matching has smaller startup delay and playback delay. At the same time, the frame loss rate and the average hops of the system are well reduced. At the same time, using the RTT information between nodes, this paper optimizes the CoolStreaming algorithm and proposes a CoolStreaming optimization algorithm based on RTT optimization. The algorithm adds RTT to the selection of new parent nodes. If the number of partner nodes satisfying the original restriction of the CoolStreaming algorithm is greater than 1, then the node acquires the RTT, between these partner nodes. Select the smallest partner node of RTT between the requesting node as the new parent node to request the service. The simulation results of OMNET platform show that compared with the random neighbor selection algorithm and the CoolStreaming algorithm, the CoolStreaming algorithm based on RTT has not been improved in the starting delay, but the frame loss rate has been improved obviously. The frame loss rate and average hops are greatly reduced. Finally, this paper proposes a P2P streaming media incentive mechanism based on user level. The nodes are classified by user level. When nodes are added, the hierarchical P2P topology is constructed according to the user level, and the users with high user level are placed in the nearest layer to the source server, so that they can get better quality of service. Thus motivating the nodes in the whole system. The simulation experiments on OMNET platform show that the node with high user level has lower startup delay and playback delay in P2P streaming media incentive mechanism based on user level, and with random neighbor selection algorithm, it has small startup delay and playback delay.
【学位授予单位】：吉林大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TP393.02
【参考文献】
相关期刊论文 前5条
1 胡进锋;洪春辉;郑纬民;;一种面向对象的Internet存储服务系统Granary[J];计算机研究与发展;2007年06期
2 郭朝阳,代亚非,韩华;燕星系统的设计及其实现中的技术问题[J];计算机工程与应用;2003年09期
3 王晓洁;;混合式P2P网络关键节点问题的研究[J];临沂师范学院学报;2008年03期
4 张茂伟;孙玉峰;;Skype协议分析[J];山东通信技术;2007年02期
5 张志明;周晋;陈震;李军;;基于网络编码的对等网流媒体传输模型和算法[J];软件学报;2012年03期
相关博士学位论文 前2条
1 韩志杰;基于P2P的流媒体若干关键技术研究[D];苏州大学;2009年
2 王嵩;P2P媒体服务系统研究[D];中国科学技术大学;2010年
本文编号：2418888
本文链接：https://www.wllwen.com/guanlilunwen/ydhl/2418888.html
