基于区分服务网络的QoS队列算法研究
发布时间：2018-12-07 13:52
【摘要】：随着计算机网络技术的快速发展,网络的规模不断增大,网络中的应用与日俱增。如何满足不断增长的应用业务的QoS要求,一直是研究热点,具有较大的研究价值和现实意义。区分服务网络体系作为IETF提出的体系结构,具有简单有效、扩展性强的特点,它采取边缘分类、中心转发的原则,顺应了网络技术的发展趋势,是未来高速网络首选的方案。队列管理和队列调度技术作为区分服务体系中的核心技术,是网络中进行资源管理、带宽分配的关键。队列管理技术和调度技术对网络的QoS性能(吞吐率、丢包率、时延、时延抖动)有着巨大的影响。基于此,本文在区分服务队列算法的基础上,提出了改进的队列管理和队列调度算法。本文研究内容及创新点如下:1介绍了区分服务网路QoS队列算法的研究背景及其意义,分析了国内外研究现状,对网络QoS的基本概念、性能指标进行了介绍,对区分服务网络模型的基本概念、体系架构进行了研究,对网络仿真工具NS2进行了研究,剖析了其中的区分服务模型。2论文重点研究了区分服务模型下的队列管理算法。通过对比典型的队列管理算法,提出RIO-C算法适合于区分服务模型。针对RIO-C算法对分组丢弃过于直接、系统的利用率较低、对突发流量处理能力不足,本文提出了INRIO-C算法。该算法结合平均队列长度与丢包率之间的非线性关系,考虑到网络的实际拥塞情况,提出了分段的非线性的丢包方法,采取更加平滑的丢弃曲线来实现分组丢弃函数。通过在NS2.35仿真验证,INRIO-C算法有效地实现了AF队列区分服务的要求,能够保护较高优先级业务,同时在网络负载较轻时减少丢包,提高了系统的利用率,在网络负载较重时加大丢包,减缓拥塞。通过理论推导和仿真验证,算法有效地维持了队列的稳定性,高优先级业务的丢包率和吞吐率均有增加。3论文重点研究了区分服务模型下的队列调度算法。通过对比典型队列调度算法,得出基于轮询的队列调度算法更加适合于区分服务网络。本文在WRR算法的基础上,针对其不能解决因分组长度不同引发的公平性问题,提出IFWRR算法,该算法结合队列中的分组平均长度,合理改变权值,从而能够更加公平地进行带宽的分配。通过仿真实验和理论分析,新算法在队列的带宽分配、公平性能方面能够克服WRR算法的不足,满足了带宽实际分配的要求。
[Abstract]:With the rapid development of computer network technology, the scale of network is increasing, and the application of network is increasing day by day. How to meet the QoS requirements of the growing application business has been a hot research topic and has great research value and practical significance. As the architecture proposed by IETF, the differentiated service network has the characteristics of simple and effective, strong expansibility. It adopts the principle of edge classification and center forwarding, conforms to the development trend of network technology, and is the first choice of high-speed network in the future. Queue management and queue scheduling, as the core technologies of differentiated service system, are the key to resource management and bandwidth allocation in network. Queue management and scheduling technology have great influence on QoS performance (throughput, packet loss, delay jitter). Based on this, an improved queue management and queue scheduling algorithm is proposed on the basis of differentiated service queue algorithm. The research contents and innovations of this paper are as follows: 1 the research background and significance of QoS queue algorithm in DiffServ network are introduced, and the current research situation at home and abroad is analyzed. The basic concept and performance index of network QoS are introduced. In this paper, the basic concept and architecture of the differentiated service network model are studied, the network simulation tool NS2 is studied, and the differentiated service model is analyzed. 2 the queue management algorithm under the differentiated service model is mainly studied in this paper. By comparing the typical queue management algorithms, the paper proposes that the RIO-C algorithm is suitable for the differentiated service model. In view of the fact that the RIO-C algorithm is too direct for packet dropping, the utilization ratio of the system is low, and the ability to deal with burst traffic is insufficient, this paper proposes a INRIO-C algorithm. This algorithm combines the nonlinear relationship between the average queue length and the packet loss rate, considering the actual congestion of the network, a piecewise nonlinear packet loss method is proposed, and a smoother dropping curve is adopted to realize the packet dropping function. The simulation results in NS2.35 show that the INRIO-C algorithm can effectively meet the requirements of AF queue differentiated service, protect higher priority services, reduce packet loss when the network load is lighter, and improve the utilization rate of the system. Increase the loss of packets when the network load is heavy, reduce congestion. Through theoretical derivation and simulation, the algorithm effectively maintains the stability of the queue, and the packet loss rate and throughput rate of high-priority services are increased. 3 the queue scheduling algorithm based on the differentiated service model is studied in this paper. By comparing the typical queue scheduling algorithm, it is concluded that the polling based queue scheduling algorithm is more suitable for differentiated service networks. Based on the WRR algorithm, this paper proposes a IFWRR algorithm, which can not solve the fairness problem caused by the different packet length. The algorithm combines the average packet length in the queue to change the weights reasonably. Thus, the bandwidth can be allocated more fairly. Simulation experiments and theoretical analysis show that the new algorithm can overcome the shortcomings of WRR algorithm and meet the requirements of bandwidth allocation in terms of the fairness of queue bandwidth allocation.
【学位授予单位】：电子科技大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TP393.09
本文编号：2367304
[Abstract]:With the rapid development of computer network technology, the scale of network is increasing, and the application of network is increasing day by day. How to meet the QoS requirements of the growing application business has been a hot research topic and has great research value and practical significance. As the architecture proposed by IETF, the differentiated service network has the characteristics of simple and effective, strong expansibility. It adopts the principle of edge classification and center forwarding, conforms to the development trend of network technology, and is the first choice of high-speed network in the future. Queue management and queue scheduling, as the core technologies of differentiated service system, are the key to resource management and bandwidth allocation in network. Queue management and scheduling technology have great influence on QoS performance (throughput, packet loss, delay jitter). Based on this, an improved queue management and queue scheduling algorithm is proposed on the basis of differentiated service queue algorithm. The research contents and innovations of this paper are as follows: 1 the research background and significance of QoS queue algorithm in DiffServ network are introduced, and the current research situation at home and abroad is analyzed. The basic concept and performance index of network QoS are introduced. In this paper, the basic concept and architecture of the differentiated service network model are studied, the network simulation tool NS2 is studied, and the differentiated service model is analyzed. 2 the queue management algorithm under the differentiated service model is mainly studied in this paper. By comparing the typical queue management algorithms, the paper proposes that the RIO-C algorithm is suitable for the differentiated service model. In view of the fact that the RIO-C algorithm is too direct for packet dropping, the utilization ratio of the system is low, and the ability to deal with burst traffic is insufficient, this paper proposes a INRIO-C algorithm. This algorithm combines the nonlinear relationship between the average queue length and the packet loss rate, considering the actual congestion of the network, a piecewise nonlinear packet loss method is proposed, and a smoother dropping curve is adopted to realize the packet dropping function. The simulation results in NS2.35 show that the INRIO-C algorithm can effectively meet the requirements of AF queue differentiated service, protect higher priority services, reduce packet loss when the network load is lighter, and improve the utilization rate of the system. Increase the loss of packets when the network load is heavy, reduce congestion. Through theoretical derivation and simulation, the algorithm effectively maintains the stability of the queue, and the packet loss rate and throughput rate of high-priority services are increased. 3 the queue scheduling algorithm based on the differentiated service model is studied in this paper. By comparing the typical queue scheduling algorithm, it is concluded that the polling based queue scheduling algorithm is more suitable for differentiated service networks. Based on the WRR algorithm, this paper proposes a IFWRR algorithm, which can not solve the fairness problem caused by the different packet length. The algorithm combines the average packet length in the queue to change the weights reasonably. Thus, the bandwidth can be allocated more fairly. Simulation experiments and theoretical analysis show that the new algorithm can overcome the shortcomings of WRR algorithm and meet the requirements of bandwidth allocation in terms of the fairness of queue bandwidth allocation.
【学位授予单位】：电子科技大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TP393.09
【参考文献】
相关硕士学位论文 前1条
1 张军;QoS控制技术在UMTS环境下应用问题的研究[D];山东大学;2008年
，本文编号：2367304
本文链接：https://www.wllwen.com/guanlilunwen/ydhl/2367304.html
