节点文献

节点势能导向多下一跳路由协议研究与性能评价

Research on Node Potential Oriented Multi-NextHop Routing Protocol and Performance Evaluation

【作者】 张建辉

【导师】 汪斌强;

【作者基本信息】 解放军信息工程大学 , 通信与信息系统, 2009, 博士

【摘要】 随着信息技术的不断发展以及对人类生活各个方面影响力的日益加深,网络基础设施已经成为关系国计民生的战略资源,网络快速自愈的重要性也日益凸现。近年来,尽管大量的人力、物力被投入到网络的管理、维护环节,但是故障恢复技术仍是各个协议层次必不可少的组成部分。提高网络快速自愈能力仍然面临两个方面的挑战:首先,如何在军事打击、恐怖袭击以及自然灾害等极端环境下提高网络的抗毁自愈能力;其次,如何应付平时环境下针对网络设备的恶意攻击和侵害所造成的网络节点设备毁坏。网络快速自愈技术贯穿网络层、链路层以及物理层传输技术等研究开发的全过程。课题结合国家863计划信息技术领域目标导向类项目“快速自愈路由协议与试验系统”,着眼于在特定场景下增强网络路由层的快速自愈能力。在不改变现有网络基本路由架构以及路由协议的前提下,论文提出了一种适用于网络遭受重大破坏或普通网络故障在路由收敛期间能够保证报文正常传输的新型路由协议——节点势能导向的快速自愈路由协议,并对协议机制进行分析优化,提高协议的自适应能力和可扩展性,拓展协议应用环境;并提出了可高效实现新型路由协议的负载自适应路由节点体系结构。论文的主要内容如下:提出了多下一跳路由协议,该协议不改变IP逐跳转发的技术体系,通过路由度量约束报文的传输方向,利用多个下一跳进行网络报文并行转发,可最大限度利用网络中所有可行下一跳,使网络链路利用率整体均衡。与传统最短路径转发的单一下一跳路由协议相比,该协议可有效避免拥塞,保护切换时延较短;与多径路由相比,路径的建立维护通过相邻节点间信息通告完成,其协议控制开销较小,建立的可用下一跳数量更多,在并发多故障网络环境下,其故障修复概率更高。提出了基于节点统计信息的链路可用带宽估计算法,设计出跳数和可用带宽字典序结合的节点势能,并证明了势能具有等压性,保证了节点势能导向路由协议构造的多下一跳路由表的无环特性。研究并设计了一种网络拓扑信息和可达性信息相分离的路由信息传播机制,并分析评估了该机制降低路由协议控制开销的效果。通过该机制,可以减少网络中路由更新的频度,对提高网络性能具有重要价值。该机制具有较高实用价值,可推广应用于其他网络路由协议。提出了一种近邻同步的分布式广度优先遍历算法——NS-DBFS算法,以完成协议中节点势能信息扩散和网络势能图的分布式构造。提出一种基于带划分的路由协议可扩展方法,划分―带‖的依据是节点的转发时延。该方法首先定义基准的单跳转发时延,然后将节点按照转发时延的等级不同划分成若干环状带,每个环状带是一个路由子区域,基于节点势能路由度量运行节点势能导向路由协议;环状带之间基于环状带时延性能的路由度量进行路由选择。该体系解决节点势能导向多下一跳路由协议在异构异质网络环境的应用问题,并可推广应用于解决其他网络路由协议的可扩展性问题。为了进一步完善节点势能导向多下一跳路由协议,未来还需要从协议的安全机制、基于节点势能导向路由协议的网络规划等方面开展研究。

【Abstract】 With the continuous development of information technology and the increasing influence for every aspect of human life, network infrastructure has been strategy resource related the benefit of nations and people and the importance of the network fast recovery becomes increasing remarkably. Recently, though, people invest a large number of manpower resources and material resources into managing and maintaining network, recovery techniques are still indispensable. Nowadays, enhancing the fast self-recovery capability still faces two challenges: under extreme environments such as military attacks, terrorism attacks and natural disasters how to improve network self-recovery and under normal environments how to deal with node failures.Network fast self-recovery technique is used in several layers, including network layer, network layer,data link layer and physical layer. Combined with the national 863 objective orientation project in information technology field–―self- recovery routing protocol and test system‖, the subject is concerned on the fast self-recovery in network layer under certain scenarios. This dissertation proposes a new routing protocol—Node Potential Oriented Multi-NextHop Routing Protocol by the prerequistite not to alter the current network routing framework, which can guarantee packets to forward during the convergence time when the network suffers from huge destruction or normal failures. Through analyzing and optimizing protocol mechanism, self-adaptive ability and scalability of protocol is enhanced and application environment is extended.This dissertation is organized as follows:A Multi-NextHop routing forwarding architecture is proposed, which do not alter the current IP hop-by-hop forwarding mechanism. The propagation direction of packets is constrained by routing metrics. The packets can be distributed into multiple next hops in parallel, making maximum use of all feasible next hops, so that the link utility ratio is globally even. Compared with the single shortest-path routing, this architecture can effectively avoid congestion and shorten protection switching delay. The path establishment and maintenance can be accomplished by routing information advertisement mechanism between peers. The control overhead of the protocol is smaller than that of multi-path routing protocol. This mechanism can provide more feasible next hops and high recovery probability, when the network incurs multiple concurrent failures.This paper presents an algorithm for estimating available bandwidth of network nodes, based on statistical information, and designs node potential in combination with lexicographic order including hop count and available bandwidth. According to node potential, the multi-nexthop loop-free routing table can be established through proving the isopiestic pressure of the node potential.A routing information propagation mechanism is designed, which decouples the network topology information from the reachability information. This technique can cut down the control overhead of the protocol trough analysis and estimation and reduce the frequency of routing updated, which has the important value for enhancing the network performance. It also can be widely applied for other network routing protocols.A neighbor synchronize distributed breadth first search algorithm named NS-DBFS has been designed to complete node potential information propagation for NP-MNRP protocol and distributed construct a potential energy diagram in the network. This article proposes a multi-nexthop routing protocol based on node potential divided by“zone”which is determined by the node forwarding delay. This method defines the reference of the forwarding delay of the single next hop and then divides the forwarding delay into several cyclic zones according to different levels of forwarding delay. Every cyclic zone is a routing subregion, in which nodes implement node potential oriented Multi-NextHop routing protocol. Routing between two cyclic zones is selected on the basis of the cyclic zone delay. Using this hierarchical structure, node potential oriented multi-nexthop routing protocol can be applied in heterogeneous networks and solve the problem of scalability in other routing protocols.In order to improve node potential oriented multi-nexthop routing protocol further, security mechanisms, network plans and other potential definitions are needed to study.

节点文献中: 

本文链接的文献网络图示:

本文的引文网络