内容中心网络路由时延优化关键技术研究

发布时间：2018-10-18 16:05

【摘要】：互联网内容流量的飞速增长以及用户QoS需求的不断提高使得现有TCP/IP网络内容传输技术乃至架构面临全新挑战。以内容中心网络(Content Centric Networks,CCN)为代表的新型网络体系结构的研究发展迅速,越来越得到研究人员的关注。内容中心网络的核心思想为“节点的普遍缓存”以及直接面向数据内容的路由转发模式。但作为全新的内容路由模型,其命名机制、转发策略以及内容查询等方面的不完善也导致了其路由时延过大。在内容中心网络中,路由时延的主要原因有:(1)网络传输层面:原始的CCN内容路由传输方法仍采用传统网络的模式,将请求路由到内容提供商的源服务器获取内容,数据传输路径过长造成内容获取时延增加。(2)节点处理层面:CCN网络直接面向内容,名称具有唯一性且长度可变,数据规模过大;路由查询时,传统的匹配算法与名字长度线性相关,效率低下。本文研究聚焦在内容中心网络的路由时延优化关键技术。首先,以“最快获取内容”为目标,提出面向邻近缓存的引导式便捷路由机制,缩短内容请求所经历的“行程”,减小内容请求时延;其次,针对现有内容路由节点数据规模过大的问题,提出多径路由聚合算法压缩路由表项;最后,从“优化处理时间”的角度出发,提出基于哈希编码的内容查询机制,实现了路由节点的快速匹配。本文具体研究如下:1、提出一种面向邻近缓存的引导式便捷路由机制路由节点向其邻居通告部分活跃内容,建立快捷路由表,使节点知晓“内容存储”的位置,从而选取最优节点获取内容。便捷路由机制能够快速探知周围节点的缓存,缩短请求的传输路径,优化请求转发时间。理论分析与仿真结果表明:相较于其他机制,该机制能够有效利用邻近节点缓存,降低拓扑网络用户平均时延约5ms,服务器负载减少约30%。2、提出一种基于后缀摘要的多可选下一跳转发信息表聚合算法采用多径路由思想,针对每个聚合前缀都有对应可选多个下一跳,选择“最佳路径”使具有公共路径的前缀进一步聚合,在利用布鲁姆过滤解决后缀空洞的前提下压缩路由表项,减少节点路由查询数量。理论分析和仿真实验表明:将后缀摘要和可选下一跳转发信息表聚合相结合,可使路由表项缩减到原来的20%左右;随着布鲁姆过滤器哈希函数的增多,可使内容路由的成功率接近100%。3、提出一种基于哈希编码的内容路由查询机制提出一种有效的元素哈希编码机制,利用哈希函数压缩数据名字并采用改进的状态转换阵列实现名称最长前缀的快速匹配,设计一种快速的增量更新机制以满足内容网络转发过程中频繁的插入,修改与删除等操作。仿真结果表明,本文提出的方法使路由条目压缩率达到40%以上,匹配速率提高约10%。
[Abstract]:With the rapid growth of Internet content flow and the increasing demand of QoS, the existing TCP/IP network content transmission technology and architecture are facing new challenges. The research of new network architecture, represented by (Content Centric Networks,CCN), is developing rapidly and has been paid more and more attention by researchers. The core idea of content-centric network is "universal cache of nodes" and direct data content oriented routing and forwarding mode. However, as a new content routing model, its naming mechanism, forwarding strategy and content query are not perfect. In content-centric network, the main reasons of routing delay are as follows: (1) Network transmission: the original CCN content routing method still adopts the traditional network mode, routing the request to the source server of the content provider to obtain the content. The data transmission path is too long to cause the content acquisition delay to increase. (2) Node processing level: CCN network is directly content-oriented, the name is unique and the length is variable, the data scale is too large; when routing query, The traditional matching algorithm is linearly related to the length of the name and is inefficient. This paper studies the key technologies of routing delay optimization focused on content-centric networks. First of all, aiming at "the fastest access to content", a guided and convenient routing mechanism for neighboring cache is proposed to shorten the "journey" of content request and reduce the delay of content request. Aiming at the problem of excessive data size of existing content routing nodes, a multi-path routing aggregation algorithm is proposed to compress routing table items. Finally, from the point of view of "optimal processing time", a content query mechanism based on hash coding is proposed. The fast matching of routing nodes is realized. The specific research in this paper is as follows: 1. A guided and convenient routing mechanism for neighbor cache is proposed. The routing node notifies its neighbors of some active content, establishes a shortcut routing table, and makes the node know the location of "content storage". In order to select the optimal node to get the content. The convenient routing mechanism can quickly detect the cache of the surrounding nodes, shorten the transmission path of requests, and optimize the request forwarding time. Theoretical analysis and simulation results show that compared with other mechanisms, the proposed mechanism can effectively utilize the cache of adjacent nodes. The average delay of users in topology networks is reduced by about 5 msand the server load is reduced by about 30. 2. A multi-optional next-hop forwarding information table aggregation algorithm based on suffix summary is proposed using multipath routing. For each aggregation prefix has corresponding multiple hops, select the "best path" to further aggregate the prefix with common path, and compress the routing table items by using Bloom filter to solve the suffix hole. Reduce the number of node routing queries. Theoretical analysis and simulation experiments show that the routing table items can be reduced to about 20% by combining the suffix summary with the optional next hop forwarding information table aggregation, and with the increase of the hash function of the Bloom filter, The success rate of content routing is close to 100. 3. A content routing query mechanism based on hashing coding is proposed and an effective element hash encoding mechanism is proposed. Using the hash function to compress the data names and use the improved state transformation array to realize the fast matching of the longest prefix, a fast incremental updating mechanism is designed to satisfy the frequent inserts in the process of content network forwarding. Modify and delete operations. The simulation results show that the proposed method can reduce the compression rate of routing entries to more than 40% and increase the matching rate by about 10%.
【学位授予单位】：解放军信息工程大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TP393.02

【参考文献】