基于NICE的相对离开概率应用层组播覆盖模型及负载均衡机制的研究
发布时间：2018-10-15 09:18
【摘要】：近年来随着网民数量和互联网普及率的迅猛增长,新媒体产业也相应的快速发展,特别是其中的视频业务。视频业务由于数据传输量大,对网络带宽、时延、稳定性指标要求更高,这对网络通信质量及终端处理能力提出了较大的挑战。然而传统的单播技术是采用一对一数据传输来进行设计,显然无法满足多媒体数据的传输要求,进而推进了IP组播技术的出现。但初期的IP组播需要底层特殊路由器支持,制约了其可扩展性和实用范围,应用层组播应时而生。 应用层组播使用端主机来替代传统的路由进行数据包的转发、复制和管理等功能。但端主机性能的差异性较大,当较差的端主机作为非叶子节点进行数据转发时,可能影响系统的可靠性、延迟、可扩展性、传输效率,如何构建最优的应用层组播模型来满足实时多媒体传输一直是研究的热点。 本文主要从应用层组播的覆盖模型和负载均衡机制两个方面来对当前应用层组播系统进行了研究,并针对当前研究中的不足进行了改进,提出了基于NICE的节点相对离开概率覆盖模型(NRLP-NICE, Nodes' Relative Leaving Probability Based On NICE)和基于用户满意度评分自适应负载均衡机制(ALBS-SOCS, Adaptive Load Balancing Strategy Based On Score Of Customer Satisfaction)。 首先,本文针对应用层组播中节点在线时间的分布规律进行了分析,将其转换成了节点的相对离开概率,在此基础上提出了结合NICE协议分层分簇思想的NRLP-NICE,并采用Omnet++4.1中OverSim框架对NRLP-NICE进行仿真验证,将之与NICE协议进行了对比,仿真结果表明NRLP-NICE具有更低控制开销,同时降低了系统的延迟误差百分比。 其次,本文分析了实际生活中引起系统整体利用率不平衡的因素,把系统的服务质量、带宽利用率、信道容量等指标进行加权,提出系统平均组播用户满意度的概念,在此基础上设计了ALBS-SOCS.在与NRLP-NICE相同的仿真环境下,对ALBS-SOCS进行了仿真实验,通过仿真验证了ALBS-SOCS相对于NICE能够较好的减少平均传输时延和平均链路压力,增加系统的平均组播用户满意度,提高系统整体性能,改善用户的体验。
[Abstract]:In recent years, with the rapid growth of the number of Internet users and Internet penetration, the new media industry has also developed rapidly, especially the video business. Due to the large amount of data transmission, video services require higher network bandwidth, delay and stability, which poses a great challenge to the network communication quality and terminal processing capability. However, the traditional unicast technology is designed by one-to-one data transmission, which obviously can not meet the requirements of multimedia data transmission, and thus promotes the emergence of IP multicast technology. However, in the early stage, IP multicast needs the support of special routers at the bottom, which restricts its scalability and practical scope, so the application-layer multicast should come into being in time. Application layer multicast uses end host instead of traditional routing for packet forwarding, replication and management. However, the performance of the end host is quite different. When the poor end host is transmitted as a non-leaf node, it may affect the reliability, delay, expansibility and transmission efficiency of the system. How to construct an optimal application layer multicast model to satisfy real-time multimedia transmission has been a hot topic. This paper mainly studies the current application-layer multicast system from two aspects: the coverage model of application-layer multicast and the load balancing mechanism, and improves the shortcomings of the current research. In this paper, NICE based relative departure probability coverage model (NRLP-NICE, Nodes' Relative Leaving Probability Based On NICE) and adaptive load balancing mechanism (ALBS-SOCS, Adaptive Load Balancing Strategy Based On Score Of Customer Satisfaction).) based on user satisfaction rating are proposed. First of all, this paper analyzes the distribution of node online time in application-layer multicast, and converts it into the relative departure probability of nodes. On the basis of this, a new NRLP-NICE, based on the idea of hierarchical clustering of NICE protocol is proposed, and the OverSim framework in Omnet 4.1 is used to validate NRLP-NICE. The simulation results show that NRLP-NICE has lower control overhead than NICE protocol. At the same time, the delay error percentage of the system is reduced. Secondly, this paper analyzes the factors that lead to the imbalance of the overall utilization of the system in real life, weights the system's quality of service, bandwidth utilization, channel capacity, and puts forward the concept of system average multicast user satisfaction. On this basis, the ALBS-SOCS. is designed. In the same simulation environment as NRLP-NICE, ALBS-SOCS is simulated. The simulation results show that ALBS-SOCS can reduce the average transmission delay and average link pressure compared with NICE, and increase the average multicast user satisfaction. Improve the overall performance of the system, improve the user experience.
【学位授予单位】：华中师范大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TP393.03
本文编号：2272072
[Abstract]:In recent years, with the rapid growth of the number of Internet users and Internet penetration, the new media industry has also developed rapidly, especially the video business. Due to the large amount of data transmission, video services require higher network bandwidth, delay and stability, which poses a great challenge to the network communication quality and terminal processing capability. However, the traditional unicast technology is designed by one-to-one data transmission, which obviously can not meet the requirements of multimedia data transmission, and thus promotes the emergence of IP multicast technology. However, in the early stage, IP multicast needs the support of special routers at the bottom, which restricts its scalability and practical scope, so the application-layer multicast should come into being in time. Application layer multicast uses end host instead of traditional routing for packet forwarding, replication and management. However, the performance of the end host is quite different. When the poor end host is transmitted as a non-leaf node, it may affect the reliability, delay, expansibility and transmission efficiency of the system. How to construct an optimal application layer multicast model to satisfy real-time multimedia transmission has been a hot topic. This paper mainly studies the current application-layer multicast system from two aspects: the coverage model of application-layer multicast and the load balancing mechanism, and improves the shortcomings of the current research. In this paper, NICE based relative departure probability coverage model (NRLP-NICE, Nodes' Relative Leaving Probability Based On NICE) and adaptive load balancing mechanism (ALBS-SOCS, Adaptive Load Balancing Strategy Based On Score Of Customer Satisfaction).) based on user satisfaction rating are proposed. First of all, this paper analyzes the distribution of node online time in application-layer multicast, and converts it into the relative departure probability of nodes. On the basis of this, a new NRLP-NICE, based on the idea of hierarchical clustering of NICE protocol is proposed, and the OverSim framework in Omnet 4.1 is used to validate NRLP-NICE. The simulation results show that NRLP-NICE has lower control overhead than NICE protocol. At the same time, the delay error percentage of the system is reduced. Secondly, this paper analyzes the factors that lead to the imbalance of the overall utilization of the system in real life, weights the system's quality of service, bandwidth utilization, channel capacity, and puts forward the concept of system average multicast user satisfaction. On this basis, the ALBS-SOCS. is designed. In the same simulation environment as NRLP-NICE, ALBS-SOCS is simulated. The simulation results show that ALBS-SOCS can reduce the average transmission delay and average link pressure compared with NICE, and increase the average multicast user satisfaction. Improve the overall performance of the system, improve the user experience.
【学位授予单位】：华中师范大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TP393.03
【参考文献】
相关期刊论文 前10条
1 马宁;;动态反馈负载均衡算法的研究[J];华中科技大学学报(自然科学版);2010年02期
2 杨光;周敬利;;iSCSI系统中动态反馈负载均衡算法[J];华中科技大学学报(自然科学版);2010年10期
3 罗建光;赵黎;杨士强;;基于用户行为分析的应用层组播树生成算法[J];计算机研究与发展;2006年09期
4 柴云鹏;顾雷;李三立;;移动网格中负载自平衡的集群式边缘流媒体服务器[J];计算机研究与发展;2007年12期
5 林龙新;周杰;张凌;叶昭;;一种动态负载均衡的P2P应用层组播方案[J];计算机科学;2008年02期
6 王晋鹏,潘龙法,李降龙;LVS集群中的动态反馈调度算法[J];计算机工程;2005年19期
7 张昊;廖建新;朱晓民;;增强型动态反馈随机分发负载均衡算法[J];计算机工程;2007年04期
8 曾彬;张大方;黎文伟;吕磊;;基于节点性能估算的应用层组播算法[J];计算机工程;2009年08期
9 贺晓丽;;一种用于任务调度的广义遗传算法[J];计算机工程;2010年17期
10 闵帆,石兵,杨国纬,周明天;分布式系统中任务分配的一种结点自适应算法[J];计算机学报;2003年03期
，本文编号：2272072
本文链接：https://www.wllwen.com/guanlilunwen/ydhl/2272072.html
