节点文献

无线自组织网络中的路由技术相关问题研究

The Research on Routing Issues in Wireless Ad Hoc Network

【作者】 吴磊

【导师】 曾家智;

【作者基本信息】 电子科技大学 , 计算机系统结构, 2010, 博士

【摘要】 未来的通信网络除了以低成本实现数据的高速传输外,还要求网络组网灵活,具有适应性和生存能力。无线自组织网络是一种无基础设施依托,可自组织重构的多跳无线网络。与其他传统无线网络不同的是,它具有临时组建、拓扑动态变化、节点能力有限等特点,可用于为地理上相邻但散布区域超过单个节点无线传输范围的一组节点之间提供信息传递服务。在无线自组网中,由于通信半径的限制,节点之间是通过多跳转发进行数据交互的,因而路由协议是其重要的组成部分。然而,由于无线自组织网络的上述特点,使得传统的路由协议不能有效地应用,故其路由传输技术成为了该领域的研究重点与难点。现今已取得了这方面的一些初步的研究成果,但现有的这些研究成果大多针对各自的应用场景,这就为无线自组网中的路由技术研究留下了非常广阔的空间。本文针对无线自组网中具有代表性的静态部署传感器网络(WSN)、移动车载自组织网络(VANET)以及近年来由移动自组织网络(MANET)发展演化出的机会型网络(Opportunistic Networks)进行了系统、全面地分析和总结,并在此基础上,结合无线自组网在上述三方面的不同特征,深入细致地研究了相关的路由传输策略,取得了若干创新和成果。本文的主要贡献包括:1.针对传感器网络的数据收集,提出了一种结合自组织映射(SOM)与免疫克隆选择算法的分簇路由策略SICR(SOM and Immune Clonal Selection Based Clustering Routing Scheme for Wireless Sensor Networks)。不同于以往的研究,SICR在分簇聚类时,充分考虑了网络节点密度、剩余能量以及与汇聚点间距离等因素,采用一种基于自组织映射原理的簇头竞争算法,构建了能量消耗均衡的分簇结构。该结构可以有效的应用于节点能量异构的网络场景。同时,为了减少簇重构次数,降低重构开销,在维护阶段引入了一种自适应调整机制,簇首可根据簇内各成员的剩余能量估算当前簇的稳定性,并据此调整簇规模,以维持簇的稳定。路由的组织则分为簇内通信和簇间通信两部分:簇内通信基于建立的簇内拓扑路径集进行;簇间通信则通过基于克隆选择算法建立的最小汇集树进行。模拟实验表明,与现有的几种算法相比,SICR能更好均衡节点的能量消耗和延长网络寿命。2.针对车载自组网的数据分发,提出了一种面向发布/订阅系统基于车辆移动分布感知的事件分发策略MDA (Mobile Distribution-Aware Data Dissemination)。基于车流的自组织性及自稳性的特点,建立VANET下的发布/订阅模型,通过计算车辆与移动订阅者的相遇概率,预测订阅者的移动分布,并以此为依据实时部署和调度广播令牌在网络中的转发,从而有效的控制事件代理的分布,保证了数据传递的有效性。与以前的相关研究相比,MDA采用的启发式算法,能够使事件代理的分布更好地适应网络环境的动态变化。此外,MDA采用了一种基于概率预测密度的令牌控制算法,通过实时的调整令牌的数量,进而控制事件代理的数量,降低了整个网络的负载。仿真实验表明,与现有的几种消息分发算法相比,MDA能以较低的网络负载和传输延迟获得较高的数据传输成功率。3.针对具有周期性社会移动轨迹的机会型网络,提出了一种基于周期性间歇连通的数据传输策略PICD (Periodic Intermittently Connected-Based Data Delivery in Opportunistic Networks)。通过有效利用节点间存在的周期间歇连通性来改善数据传输性能。节点传输概率的计算则充分考虑了其与汇聚点间存在的间歇多跳路径,并将其与消息容忍的传输延迟相结合,首先采用随机动态规划的方法来建立与延迟相关的传输概率模型,然后通过基于多跳的函数空间迭代法求出一个周期内的与延迟相关的传输概率分布矩阵,节点面向不同消息延迟的传输概率则基于分布矩阵计算获得,以此作为选择下一跳的依据。与延迟相关的概率转发机制提高了消息在容忍的延迟内被成功递交的可能。仿真实验表明,与现有的几种数据传输算法相比,在节点具有循环运动特征的环境下,PICD具有较高的数据传输成功率和较低的递交延迟。

【Abstract】 Next Generation telecommunication Network (NGN) requires not only high-speed data delivery at low cost, but also structural flexibility, adaptability and stability. Wireless Ad-hoc Network is an infrastructure-free, self-reconfigurable multi-hop wireless network. Different from other traditional wireless network, it can support temporarily-organizing, dynamic topology and computation capability of each node is limited. Therefore, Wireless Ad-hoc Network should be the right data delivery solution for a group of nodes scattered in a certain area, in which each node may reach some other nodes but not all nodes via wireless communication. Within Wireless Ad-hoc Network, considering limited wireless communication radius, normally data delivery between a pair of nodes may have to be done by means of multi-hop forwarding and as a result, routing protocol is the key player for data delivery. However, because of the characteristics of Wireless Ad-hoc Network, traditional routing protocol can not be applied to achieve expected performance. Therefore, routing protocols become hot topics. Although now there are some achievements, they have various limitations and only apply to specific scenarios. More research should be done for more generic and comprehensive routing solutions.This paper analyzes and summarizes Wireless Sensor Network (WSN), Vehicular Ad hoc Network (VANET) and Opportunistic Network, evolved from Mobile Ad Hoc Network (MANET) recent years systematically and comprehensively. Based on the result as well as the characteristics of Wireless Ad-hoc Network, dedicated research has been done on more generic and efficient routing protocols for Wireless Ad hoc Network and three novel solutions have been proposed:SICR for WSN data collection, MDA for VANET data delivery and PICD for Opportunistic Network data delivery. Here are details for these three routing protocols as follows:1. Based on both Self-Organised Mapping (SOM) and Immunity Clonal Selection algorithm, this dissertation proposes a novel clustering routing solution SOM and Immune Clonal Selection Based Clustering Routing Scheme for Wireless Sensor Networks (SICR) for data collection in WSN. There are two phases in SICR:clustering and maintenance. Unlike existing research, during clustering phase, SICR takes key factors such as node density, residual energy and distance to the sink into account. Based on self-mapping theorem, SICR adopts an innovative cluster head competition algorithm to establish energy consumption-balanced cluster structure. This structure can be effectively applied to heterogeneous energy scenario. Meanwhile, in order to reduce the reconstruction cost, self-adaptive mechanism has been introduced to maintenance phase:cluster head estimates current stability of the cluster based on each node’s residual energy and then adjusts cluster size accordingly to maintain the stability. Routing scheme can be divided into intra-cluster and inter-cluster communication: intra-cluster communication should be done based on topology inside the cluster while inter-cluster communication should be done based the minimum spanning tree, constructed by clone-selection algorithm. Simulation shows that, compared with several existing algorithms, SICR can lead to more balanced energy consumption among nodes and prolonged network life.2. The dissertation proposes a novel data delivery solution:Mobile Distribution-Aware Data Dissemination (MDA) based Publish/Subscribe mechanism for VANET. MDA takes advantage of VANET’self-organization and self-stabilization: firstly, establish appropriate Publish/Subscribe model for the VANET; then, predict the subscribers’distribution by calculating the delivery probabilities between vehicles and corresponding mobile subscribers; thirdly, based on the predicted subscribers’ distribution, adjust notification token’s real-time deployment and schedule the token forwarding to achieve effective distribution of notification brokers(notification-token holder) in the VANET and thus ensure efficient data delivery accordingly. Compared with existing solutions, a novel heuristic algorithm is introduced to alter notification broker distribution dynamically to adapt to real-time node topology in this VANET. Furthermore, MDA adopts an effective notification token control algorithm based on probability prediction density:dynamically adjust the number of notification tokens to control the number of notification broker. As a result, MDA can also reduce overall network load of VANET by restricting the occurrences of broadcasting. Simulation results based on a real city map and realistic traffic situations show that the MDA performs much better in terms of delivery ratio and delay with less network load over other solutions. 3. The dissertation proposes an innovative data delivery solution Periodic Intermittently Connected-Based Data Delivery in Opportunistic Networks (PICD) for Opportunistic Network. By effectively utilizing periodic intermittent connectivity between any two nodes, PICD can lead to better data delivery:Firstly, for a certain node, based on its periodic intermittently-connected paths to the sink and its data delay tolerance, PICD establishes its delay probability model by means of random dynamic programming. Then, PICD will calculate the node’s periodic delay probability distribution matrix within a certain cycle by multi-hop based function space iteration. Next, according to the node’s delay probability distribution matrix, current data delivery probability for a certain data within its tolerable delay can be calculated by specifying current cycle and the tolerable delay and this will become key-player of next-hop choosing. In brief, this probability forwarding mechanism is delay-relevant and can increase data delivery probability within tolerable delay. Simulation shows that compared with existing algorithms, which only take advantages of single-hop delivery probability, PICD can lead to better data delivery and lower delay.

节点文献中: 

本文链接的文献网络图示:

本文的引文网络