基于大颗粒度的IP业务时延特性研究

发布时间：2018-07-05 17:12

本文选题：IP + over　；参考：《北京邮电大学》2014年硕士论文

【摘要】：IP业务时延作为一项衡量网络性能的指标,直接关系到整个网络的服务质量。此外,随着互联网业务的增长,光纤入户的不断普及,[P业务正朝着大颗粒度化转变。因此基于大颗粒度IP业务的时延特性分析已成为当前传送网性能研究的热点。经过调研国内外研究情况,目前基于大颗粒度IP业务时延特性主要从以下两个方面进行研究,一方面针对单层(IP层)网络或光层进行研究,另一方面从数据业务流进行讨论,只针对一种类型的单位业务数据流进行研究。这使得研究成果不能很好地反映出传送网中多层、多种类型业务数据流的时延特性。本文通过对IP over WDM网络结构的深入分析,重点研究承载IP业务的大颗粒度传输网多等级、多业务流种类的时延特性。本文一方面综合考虑光／电两层网络对IP业务的时延影响,以IP over WDM网络模型中的重叠模型和路由算法中的最短路径算法为基础,将IP层的排队时延,WDM层的O/E/O交换时延之和的值作为研究对象,提出一种新的时延测算算法。算法由以下三个过程组成：业务产生,业务调度和业务丢弃,其中业务产生采用自相似模型中的分形布朗运动,业务调度采用加权轮询调度算法,业务丢弃则根据节点的容量。另一方面,由于一条业务在IP over WDM网络中的传输会经历从IP层多个数据包到WDM层少量光突发再到IP层多个数据包的变化过程,因而本文在测算业务时延过程中考虑了业务的完整性因素。最后,本文利用多层网络规划系统对算法进行仿真实现并分析了业务分配带宽权重的标准差和业务数量对IP业务时延的影响。大量的实验仿真结果表明,提出的时延测算算法,不但能够动态地测量大颗粒度IP业务时延,而且可以通过控制业务的权重标准差来降低时延。
[Abstract]:As an indicator of network performance, IP service delay is directly related to the quality of service of the whole network. In addition, with the growth of the Internet service and the continuous popularization of the fiber into the household, the [P service is changing to a large granularity. Therefore, the time-delay analysis based on the large granularity IP service has become a hot issue in the performance research of the current transmission network. Point.
After investigating domestic and foreign research, the time delay characteristics of large particle IP service are mainly studied in the following two aspects. One is to study the single layer (IP layer) network or light layer, and on the other hand, the data traffic flow is discussed, only one type of unit traffic data flow is studied. It can well reflect the delay characteristics of multilayer and multiple types of traffic data in transport network.
Through in-depth analysis of the structure of IP over WDM network, this paper focuses on the multilevel transmission network with large granularity and the time delay characteristic of multi service flow type. On the one hand, this paper comprehensively considers the delay effect of the optical / electricity two layer network on the IP service, and the shortest path in the IP over WDM network model and the shortest path in the routing algorithm. Based on the path algorithm, the queuing delay of IP layer and the sum of the O/E/O exchange delay of the WDM layer are used as the research object, and a new time delay estimation algorithm is proposed. The algorithm is composed of three processes: business generation, business scheduling and service discarding, in which the service produces fractal Brown motion in the self phase like model and the business scheduling is added. The traffic discarding is based on the capacity of the node. On the other hand, because the transmission of a service in the IP over WDM network will undergo a change from a small number of packets in the IP layer to a small amount of light burst in the WDM layer to a number of data packets in the IP layer, this paper takes into account the integrity factors of the service in the process of calculating the service delay.
Finally, this paper simulates the algorithm by using the multi-layer network programming system and analyzes the influence of the standard deviation of the service allocation bandwidth weight and the number of services on the IP service delay. A large number of experimental simulation results show that the proposed time delay estimation algorithm can not only dynamically measure the time delay of large particle size IP, but also can be controlled by control. The weight standard deviation of the business is reduced to reduce the delay.
【学位授予单位】：北京邮电大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TP393.06

【参考文献】