信息中心网络服务缓存方法研究

发布时间：2018-01-17 02:31

本文关键词：信息中心网络服务缓存方法研究　出处：《北京交通大学》2016年博士论文　论文类型：学位论文

【摘要】：传统以主机网络地址为中心的互联网通信模式在网络可扩展性、安全性、可控可管性和移动性等方面存在严重问题,无法适应经济和社会发展对信息网络的需求。因此,旨在满足用户对服务内容需求、提高网络传输效率的信息中心网络在近年来得到了业界的广泛关注。信息中心网络通过为信息命名,使得信息在网络中任意位置(节点)存储而不改变名字,从而便于用户获取内容,提升用户体验和网络效率。本文针对信息中心网络的服务缓存方法进行研究,旨在提高信息中心网络缓存效率。论文主要工作和创新点如下：1)为了确定信息中心网络中的内容命名粒度,分析了信息中心网络内容命名粒度对域内缓存性能的影响。本文首先给出了内容命名粒度的概念,分析了内容命名粒度与分块缓存的优点,及对网络可扩展性的影响。在此基础上,通过仿真分析了内容命名粒度在不同网络拓扑、用户请求模型和缓存空间大小等关键参数条件下,不同缓存策略对缓存命中率、用户请求时延和用户请求跳数的影响。结果表明,内容命名粒度在由大变小的过程中,缓存效率逐渐提高,但随着粒度变小效率的增加幅度也逐渐减小,同时会带来较大的网络可扩展性问题。基于文中所采用的网络参数,内容命名粒度在区间[1OMB-100MB]内可以获得较好的缓存效率,同时保证对网络可扩展性影响不大。2)为了更好的适应视频等大容量流媒体服务的分发和传输,本文提出了一种基于内容分块流行度的流媒体服务缓存方法。通过分析用户对大容量流媒体视频的请求模式,即同一内容的不同分块具有不同流行度的特性,进而得出用户对流媒体服务的请求模型。根据这一模型并结合同一内容不同分块可以在不同节点缓存的特性,提出了一种根据内容分块流行度分级并将缓存节点根据位置进行分级的缓存策略,通过在用户请求包和数据包中携带缓存标签的方式实现隐式协作缓存。仿真结果表明,本方法相比于其他域内缓存方法,可以更好的利用内容分块流行度的特性,具有更高的网络缓存命中率,更低的内容请求时延以及内容请求跳数,从而可以提升用户对网络流媒体服务的体验。3)现有信息中心网络域内缓存协作方案大多没有考虑到网络中缓存节点缓存失效或者不可达的情况。因此这些方案的缓存性能在缓存节点失效时将会受到较大影响。本文根据该问题提出了一种基于内容分块流行度并具有良好鲁棒性的域内缓存协作方法RINA。在RINA中,缓存节点被分为两类,备份缓存节点和普通缓存节点。为了提供良好的鲁棒性,RINA在备份缓存节点和普通缓存节点同时缓存了网络中最流行的内容分块,所以当网络中任何节点失效时,都不会对网络的缓存性能(如缓存命中率)造成较大的影响。本章通过大量仿真,将RINA与其他现有的缓存方案进行了比较。仿真结果表明,相比于其他缓存方案,不论是单个缓存节点失效还是多个缓存节点失效,RINA在提供较好缓存性能的同时保证缓存鲁棒性。
[Abstract]:The traditional mode of Internet communication to host network address as the center of the scalability in network security, there are serious problems in terms of controllable and mobility, unable to adapt to the demand of economic and social development of the information network. Therefore, in order to meet the user demand for the service content, improve the efficiency of the network information center network has been the wide attention in recent years. The information center through the network for information makes the information in the network name, any position (node) storage without changing the name, which is convenient for users to access content, enhance the user experience and network efficiency. This paper makes a research on the service cache method for network information center, information center network to improve the cache efficiency. The main work and innovation are as follows: 1) to determine the content of network information center named granularity, analyzes the network information in the heart Effect of particle size on the domain name containing the cache performance. This paper first gives the concept of content named particle size, particle size and content analysis of advantages of named block cache, and network scalability. On this basis, through the simulation and analysis of the content of different size in the name of network topology, user requests the key parameters of the model and the cache size under the conditions of different cache hit rate of the cache, the user request delay and user request influence hops. The results showed that the content in the process of naming the size changed from big to small, the cache efficiency gradually increased, but with smaller particle size efficiency increase is gradually reduced, at the same time will bring great the network scalability problem. The network parameters used in this paper is based on the content of naming can obtain better efficiency of the cache size in the interval [1OMB-100MB], and ensure the network scalability. Not influence.2) in order to better adapt to the large capacity video streaming service distribution and transmission, this paper proposes a content based block of the popularity of streaming media service cache method. Through the analysis of user request patterns of streaming media video on large capacity, that is the same content and different blocks with different characteristics of popularity and then request model of streaming media service users. Based on this model and combined with the characteristics of the same content and different blocks in different node cache, a content block according to the popularity and grading the cache node according to the location of the hierarchical caching strategy, implementation of implicit cooperative caching by carrying a cache tag the user request packets and data packets in the way. The simulation results show that this method compared to other domain caching method, can use better characteristics of popularity, with a higher The network cache hit rate, lower content request delay and content request hops, which can enhance the user experience of the streaming media service network.3) existing information center network domain cooperative caching scheme has not considered to the network cache node cache invalidation or unreachable. Therefore the performance of these schemes in the cache cache node failure will be severely affected. This paper proposes a method based on block content popularity and has good robustness in domain of cooperative caching method RINA. in RINA according to this problem, the cache node is divided into two categories, the backup cache node and cache node. In order to provide good robustness, RINA in the backup cache the cache nodes and ordinary nodes at the same time the cache is the most popular network content in the block, so when any node in the network fails, not on the network cache performance (e.g. The cache hit rate) has a greater impact. This chapter through a lot of simulation, RINA and other existing caching schemes are compared. The simulation results show that compared to the other cache scheme, both multiple cache node failure single cache node failure, ensure the robustness of the RINA cache in providing better cache performance at the same time.

【学位授予单位】：北京交通大学
【学位级别】：博士
【学位授予年份】：2016
【分类号】：TP393.02;TP333

【相似文献】