面向车联网的可靠路由方法研究

发布时间：2018-01-27 06:54

本文关键词： 车联网可靠路由 Q学习多跳分簇　出处：《天津理工大学》2017年硕士论文　论文类型：学位论文

【摘要】：随着通信技术的不断发展以及道路上车辆数目的不断增多,人们对安全出行提出了更高的要求。智能交通系统(ITS)作为构建智慧化城市的重要拼图,是近年来研究的热门话题。车载自组网(VANET)作为智能交通系统的主要通信技术,其有着网络拓扑变化频繁,通信链路不可靠以及车辆节点分布不均匀的特点。这些特性使得设计可靠性高,时延较低的路由算法成为一项具有挑战性的任务。有效的路由算法不仅能够保证节点之间可靠的通信,更重要的是其决定了应用部署的灵活性。因此,研究可靠的路由算法对实现智能交通系统有着重要的意义。本文首先就车联网中路由相关技术展开讨论,通过详细研究车联网中的通信架构以及路由算法原理,对车联网中影响路由性能的因素做了总结,并对车联网中路由设计方法做了详细的讨论,为下文设计路由算法做了充分准备工作。其次,通过详细的研究车辆运动与链路之间的关系,建立了节点间链路维持时间模型,并在此基础上证明了节点间的链路维持时间服从于对数正态分布,给出了链路可靠性计算方法。Q学习算法作为一种非监督型的智能算法,通过多次迭代能够找出源节点到目的节点的最短路径,本文通过修改原始Q学习算法,将其应用在路由算法设计中,设计了一种可靠的自适应路由算法RSAR。其利用分布式的方式将学习任务分散在每一个车辆节点中,节点通过周期性的与邻居节点交换信息来达到动态调整路由路径的功能,有效的解决了VANET中由于拓扑变化带来的路径不可靠的问题,极大的提高了路由算法的性能,并利用NS2网络仿真环境对RSAR算法的性能做了评估。最后,考虑到分簇算法作为一种分层架构,能够有效的提高路由算法的性能。但是,现有的分簇算法仅仅将车辆的相关移动性作为选择簇头的依据,而且并没有考虑所成簇中节点间链路的可靠性。簇头节点速度的轻微变化对簇结构影响很大,使得所成簇的稳定性以及可靠性很差。为了能够有效的解决这些问题,本文提出了一种基于多跳的反应式成簇机制PMC算法。在簇的形成过程中,提出了一种优先权车辆跟随策略,去保证所成簇的可靠性和稳定性,簇头作为簇中最稳定的节点被动的生成。在簇的维护阶段,通过将相邻簇头节点进行融合,提出了新的簇合并机制去进一步提高所成簇的稳定性以及扩展性。最后,在NS2仿真环境下对PMC算法的性能进行验证,并通过实验与其他相关算法进行了对比。
[Abstract]:With the continuous development of communication technology and the increasing number of vehicles on the road, people put forward higher requirements for safe travel. The Intelligent Transportation system (ITS) is an important puzzle to build intelligent cities. Vehicle Ad Hoc Network (VANET), as the main communication technology of Intelligent Transportation system, has frequent network topology changes. The characteristics of unreliable communication links and uneven distribution of vehicle nodes make the design reliable. Low delay routing algorithm has become a challenging task. Effective routing algorithm can not only ensure reliable communication between nodes, but also determine the flexibility of application deployment. It is very important to study the reliable routing algorithm for the realization of intelligent transportation system. Firstly, this paper discusses the related technologies in the vehicle networking, and through the detailed study of the communication architecture and routing algorithm principle in the vehicle networking. This paper summarizes the factors that affect the routing performance in the vehicle networking, and discusses the routing design method in the vehicle networking in detail, which makes sufficient preparation for the following design of routing algorithm. Secondly. Based on the detailed study of the relationship between the vehicle motion and the link, a link maintenance time model between nodes is established, and it is proved that the link maintenance time between nodes follows the logarithmic normal distribution. As an unsupervised intelligent algorithm, the link reliability calculation method .Q learning algorithm can find the shortest path from the source node to the destination node through multiple iterations. In this paper, the original Q-learning algorithm is modified. In this paper, a reliable adaptive routing algorithm RSAR is designed, which distributes the learning tasks in every vehicle node in a distributed way. Nodes exchange information periodically with neighbors to achieve the function of dynamically adjusting routing path, which effectively solves the problem of unreliable path caused by topology change in VANET. Greatly improve the performance of the routing algorithm, and use the NS2 network simulation environment to evaluate the performance of the RSAR algorithm. Finally, considering the clustering algorithm as a hierarchical architecture. It can effectively improve the performance of the routing algorithm. However, the existing clustering algorithms only take the vehicle mobility as the basis for the selection of cluster heads. Moreover, the reliability of the link between the nodes in the cluster is not considered. The slight change of the cluster head speed has a great influence on the cluster structure. The stability and reliability of the clusters are poor. In order to solve these problems effectively, a reactive clustering mechanism PMC algorithm based on multi-hop is proposed in this paper. A priority vehicle following strategy is proposed to ensure the reliability and stability of the cluster. The cluster head is the most stable node in the cluster to generate passively. In the maintenance phase of the cluster, the adjacent cluster head nodes are fused. A new cluster merging mechanism is proposed to further improve the stability and scalability of the clusters. Finally, the performance of the PMC algorithm is verified in the NS2 simulation environment. And through the experiment and other related algorithms are compared.
【学位授予单位】：天津理工大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：U495;TN929.5

【相似文献】