基于社会网络的车载自组网中数据转发机制研究与实现

发布时间：2018-03-14 02:09

本文选题：车载自组织网　切入点：社区检测　出处：《北京工业大学》2015年硕士论文　论文类型：学位论文

【摘要】：车载自组织网络通过使车辆自主的建立多跳无线网络,形成道路交通网络之上的信息分享平台,能广泛应用于现代交通管理、交通安全预防、城市车况监控等领域,已经成为当前的一个研究热点。数据转发是车载自组网数据交换的基础,而相关的路由协议则是影响网络性能至关重要的因素。在车载自组织网络中,由于车辆节点移动速度快,行驶受限于街道结构,以及交通规则的约束,路面通信环境恶劣,车辆分布不均,这些都导致网络中拓扑结构不稳定、网络频繁中断等不利于数据转发的因素。针对车载自组网的这一特性,本文研究基于社会网络的车载网数据转发机制,具体内容与贡献包括:第一基于社会属性的车载自组网数据转发机制。城市车辆移动具有较明显的社会特性,考虑到一般社会网络具有高度稳定的拓扑结构,本文提出基于社会网络设计车载自组网中的多跳数据转发机制,能通过降低大量无效的车辆间消息交换,提高车载自组网中网络资源的有效利用率。同时,本文还结合城市车辆移动特征提出使用社区性与中心度两项社会属性刻画节点社会特征,在此基础上设计节点效用计算方法,以及分布式的消息转发策略。第二分布式社区检测机制和节点中心度计算方法研究。同社区的节点间有相对较高的相遇概率,通常被作为消息路由的中继转发节点。在对比了代表性分布式社区检测算法k-clique、MODULARITY,发现k-clique由于具有更低的时间与计算复杂度,适合VANET数据转发场景。中心度反映了网络中移动节点遇见不同消息目标节点的能力。常见的社会网络中心度衡量方法有度中心性、介数中心性等。其中介数中心性方法无法分布式实现;而度中心性也由于车辆间的相遇次数较高,相对大小不易衡量,不适合车辆移动场景。提出使用移动熵值描述节点运动的规律性。第三提出基于分布式社区的路由算法DFDL,利用移动车辆间相遇时间间隔和相遇频率确定车辆的社区标签,并根据车辆运动的移动熵计算节点运动中心度。在转发过程中DFDL机制通过综合判断相遇车辆的社区标签以及运动中心度,为数据包消息选择合适的中继转发节点。第四DFDL路由算法性能仿真,使用ONE仿真平台,并针对大规模移动、分布式社会属性学习等进行了扩展,选择了Helsinki的地图作为仿真场景,确定了网络规模大小、传输率随时间的变化和数据包产生速率实验指标,对比DFDL与BubbleRap性能。
[Abstract]:The vehicle-borne ad hoc network can be widely used in modern traffic management, traffic safety prevention, urban vehicle condition monitoring and other fields, by making the vehicle establish multi-hop wireless network independently, and form the information sharing platform on the road traffic network, which can be widely used in modern traffic management, traffic safety prevention, urban vehicle condition monitoring and so on. Data forwarding is the basis of data exchange in vehicular ad hoc networks, and related routing protocols are the most important factors affecting network performance. Because of the fast moving speed of vehicle nodes, restricted by the structure of the street and the restriction of traffic rules, the communication environment on the road is bad, and the distribution of vehicles is uneven, all of these lead to the instability of topological structure in the network. In view of the characteristics of vehicular ad hoc network, this paper studies the data forwarding mechanism of vehicular network based on social network. The specific contents and contributions include: first, the data forwarding mechanism based on social attributes for vehicular ad hoc networks. Urban vehicle mobility has obvious social characteristics, considering that the general social network has a highly stable topological structure. In this paper, we propose a multi-hop data forwarding mechanism based on social network design for vehicular ad hoc networks, which can improve the effective utilization of network resources in vehicular ad hoc networks by reducing a large number of invalid message exchange between vehicles. This paper also proposes to use two social attributes of community and centrality to depict the social characteristics of nodes, and then design a method for calculating the utility of nodes based on the characteristics of urban vehicle movement. And distributed message forwarding strategy. Second distributed community detection mechanism and node centrality calculation method. There is a relatively high probability of meeting with the nodes in the community. Compared with the typical distributed community detection algorithm k-cliquen MODULARITY, it is found that k-clique has lower time and computational complexity. Suitable for VANET data forwarding scenario. Centrality reflects the ability of mobile nodes to meet different message target nodes in the network. The method of intermediation centricity can not be implemented in a distributed way, and the degree of centrality is difficult to measure because of the high number of encounters between vehicles. Mobile entropy is used to describe the regularity of node movement. Third, a distributed community based routing algorithm, DFDL, is proposed to determine the vehicle community label using the time interval and encounter frequency between mobile vehicles. According to the moving entropy of the vehicle, the motion center degree of the node is calculated. In the process of forwarding, the DFDL mechanism synthetically judges the community label and the motion center degree of the encounter vehicle. Selecting appropriate relay forwarding node for packet messages. 4th DFDL routing algorithm performance simulation, using ONE simulation platform, and aiming at large-scale mobile, distributed social attribute learning, etc., selected the map of Helsinki as the simulation scene. The size of the network, the change of transmission rate with time and the experimental index of data packet generation rate are determined, and the performance of DFDL and BubbleRap is compared.
【学位授予单位】：北京工业大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：U495;U463.67

【相似文献】