内容中心网络服务器选择和路由规划的研究

发布时间：2018-03-06 05:03

本文选题：发布订阅网络技术　切入点：内容中心网络　出处：《电子科技大学》2014年硕士论文　论文类型：学位论文

【摘要】：当前互联网有着根本性的缺陷,需要革命性的改变,于是内容中心网络这种新型的架构在近些年被提出。由于互联网中内容流量的飞速增长以及用户Qos需求的不断提高,内容中心网络被认为是下一代互联网发展趋向。本文介绍了PURSUIT这一网络体系,PURSUIT是一种发布订阅网络技术,并在网络平台实现的基础上对服务器选择算法和路由规划算法进行了研究。本文阐述了PURSUIT架构,以及BF转发机制,它实现了线速网际互联。本文利用OMNET++仿真软件对PURSUIT进行研究,服务器选择算法和路由规划算法本质上均为选路算法,它们都在系统的核心节点上实现。但目的不同,服务器选择算法是给定源点找宿点以及源点和宿点之间的路径,路由规划算法是给定源点和宿点找路径。这导致了它们的实现有很大区别。内容路由器与IP路由器的不同在于其真正具有缓存数据的功能,因而在PURSUIT中路由与缓存的结合显得尤为重要。目前关于内容中心网络中的关键策略研究仍很少。本文对于服务器选择提出了最近缓存点策略、最大重复度策略、最小链路利用率策略,这些策略均以最短路为基础。前两种方式要义都在于利用缓存,第三种策略要义在于负载均衡。在靠近客户机端附近内容的命中率高会使最近缓存点策略表现良好,然而总体来看,最大重复度策略命中率最佳。最小链路利用率策略在负载较重时最佳,在负载较轻时会较差。总的来说利用缓存更好的算法性能要更好。另外,本文对实现负载均衡的三种不同策略进行了对比,得到了有效的结论。本文对于路由规划提出了最近点策略、较多缓存次数策略、较多点策略以及缓存通告重路由策略。这些策略并不一定以最短路为基础,它们要找的是一条符合条件的简单路径,这是为了避开环路使得BF机制有效。最近点策略在命中率高时更有效,而较多缓存次数策略在命中率低时更好。缓存通告重路由策略在一跳甚至多跳的搜寻中表现良好。较多点策略的命中率在五种策略中最高。由于较多缓存次数策略与较多点策略都仅可求得近似解,因此本文对它们产生的近似解与最优解的性能进行了对比。另外,将路由规划中的算法应用于服务器选择,并且对相应的算法进行了对比,得到了有效的结论。
[Abstract]:Because of the fundamental defects of the Internet and the need for revolutionary change, the new architecture of content-centric network has been proposed in recent years. Because of the rapid growth of content traffic in the Internet and the increasing demand of users' Qos, The content center network is regarded as the development trend of the next generation Internet. This paper introduces PURSUIT, a network architecture, is a kind of publish / subscribe network technology. On the basis of the implementation of the network platform, the server selection algorithm and route planning algorithm are studied. The PURSUIT architecture and BF forwarding mechanism are described in this paper, which realizes the internetwork interconnection at line speed. In this paper, the PURSUIT is studied by using the OMNET simulation software. Both the server selection algorithm and the routing algorithm are routing algorithms, which are implemented on the core nodes of the system, but for different purposes, the server selection algorithm is a given source location point and the path between the source point and the location point. The routing algorithm is a given source point and a fixed location point to find a path. This leads to a great difference in their implementation. The difference between a content router and an IP router is that it really has the function of caching data. Therefore, the combination of routing and caching in PURSUIT is particularly important. At present, there are few researches on the key strategies in content-centric network. This paper proposes the nearest cache point policy and the maximum repetition policy for server selection. Minimum link utilization policies, which are based on the shortest path. The third strategy is about load balancing. A high hit ratio near the client side makes the nearest cache point policy perform well, however, in general, the. The maximum repeatability strategy is the best hit rate. The minimum link utilization strategy is optimal when the load is heavy and worse when the load is low. Overall, the algorithm with better cache is better. In addition, In this paper, three different strategies for load balancing are compared, and an effective conclusion is drawn. More poly-point policies and cache notification rerouting policies. These policies are not necessarily based on the shortest path, they are looking for a simple path that meets the criteria. This is to avoid the loop to make BF mechanism effective. The nearest point strategy is more effective when the hit rate is high. The cache notification rerouting strategy performs well in one-hop or even multi-hop search. The hit ratio of multi-point strategy is the highest among the five strategies. The approximate solution can only be obtained by using the simple and multi-point strategies. In addition, the algorithms in routing planning are applied to server selection, and the corresponding algorithms are compared, and an effective conclusion is obtained.
【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】：TP393.05

【参考文献】