复制策略DTN路由算法在真实车辆轨迹场景中的性能评估
发布时间：2018-12-17 06:01
【摘要】：随着网络技术的更新和进步，其应用也变得多样化。传统Internet结构在许多场景（如太空探测、战场搜救、野生动物研究等）中的应用已暴露出不可弥补的弊端。 这类领域具有相同的基本特征：即网络延迟大、链路拓扑变化频繁、传输速率不对称、差错率高以及节点容量受限等。延迟容忍网络（DTN）是为解决以上问题应运而生的一种新型网络。为使DTN网络能更有效地运行于各类场景之下，对其路由算法的评估则显得十分必要。 目前，有很多依据不同标准而进行的DTN路由算法分类。以复制策略为共同基础的分类中，包含了众多的路由算法。其中，比较有代表性的有：Epdemic算法、Spray AndWait算法，它们都基于洪泛思想；PROPHET算法基于概率，MAXPRO算法基于调度。它们都是理论上较为成熟的算法。 本文针对5种常见的复制策略DTN路由算法，采用机会网络模拟器ONE环境，进行系列实验对算法性能进行评估。与众不同的是，本文采用了ONE的开放模型ExternalMovement，以真实车辆运行轨迹数据为基础进行“模拟化”的真实场景重现，通过变换4类参数共进行155次具体实验。两组原始数据均来源于客观世界：一组是旧金山500余辆出租车运行轨迹数据；一组是国内南方某市的7000余辆出租车运行轨迹数据。由于出租车辆大多为小型车辆，活动方式与运行轨迹与普通车辆（路线固定的车辆如公交车、有轨电车等除外）并无太大差别，完全可以代表一般车辆运行规律。这种基于真实数据的采样模拟实验以及进行算法评估，很大程度上消除了软件模拟运动模型构造DTN网络的局限性，，更为客观精确地体现了路由算法在应用中的特性特征。 最后，基于投递成功率、平均延迟时间、资源开销比三个评估指标对5种算法进行了横向和纵向比较，得出了一般性结论。
[Abstract]:With the update and progress of network technology, its application has become diversified. The application of traditional Internet structure in many scenarios (such as space exploration, battlefield search and rescue, wildlife research, etc.) has exposed irreparable drawbacks. This kind of domain has the same basic characteristics: the network delay is large, the link topology changes frequently, the transmission rate is asymmetric, the error rate is high, and the node capacity is limited. Delay tolerance network (DTN) is a new network to solve the above problems. In order to make DTN network run more effectively under various scenarios, it is necessary to evaluate its routing algorithm. At present, there are many DTN routing algorithms classified according to different standards. Many routing algorithms are included in the classification based on replication strategy. Among them, the representative ones are: Epdemic algorithm, Spray AndWait algorithm, they are based on flood idea, PROPHET algorithm is based on probability, MAXPRO algorithm is based on scheduling. They are mature algorithms in theory. In this paper, a series of experiments are carried out to evaluate the performance of five common replication strategy DTN routing algorithms using the ONE environment of the opportunistic network simulator. What is different is that ONE's open model ExternalMovement, is used to reproduce the real scene based on the real vehicle track data, and 155 concrete experiments are carried out by transforming four kinds of parameters. The two groups of original data are from the objective world: one is the track data of more than 500 taxis in San Francisco; the other is the track data of more than 7000 taxis in a city in the south of China. Because most of the rental vehicles are small vehicles, there is no great difference between the moving mode and the running track of the ordinary vehicles (except the vehicles with fixed routes, such as buses, trams, etc.), so it can fully represent the running rules of the general vehicles. The sampling simulation experiment based on real data and the evaluation of algorithm to a great extent eliminate the limitation of software simulation motion model to construct DTN network and more objectively and accurately reflect the characteristics of routing algorithm in application. Finally, based on the delivery success rate, average delay time and resource overhead ratio, five algorithms are compared horizontally and longitudinally, and a general conclusion is drawn.
【学位授予单位】：吉林大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TP393.02
本文编号：2383773
[Abstract]:With the update and progress of network technology, its application has become diversified. The application of traditional Internet structure in many scenarios (such as space exploration, battlefield search and rescue, wildlife research, etc.) has exposed irreparable drawbacks. This kind of domain has the same basic characteristics: the network delay is large, the link topology changes frequently, the transmission rate is asymmetric, the error rate is high, and the node capacity is limited. Delay tolerance network (DTN) is a new network to solve the above problems. In order to make DTN network run more effectively under various scenarios, it is necessary to evaluate its routing algorithm. At present, there are many DTN routing algorithms classified according to different standards. Many routing algorithms are included in the classification based on replication strategy. Among them, the representative ones are: Epdemic algorithm, Spray AndWait algorithm, they are based on flood idea, PROPHET algorithm is based on probability, MAXPRO algorithm is based on scheduling. They are mature algorithms in theory. In this paper, a series of experiments are carried out to evaluate the performance of five common replication strategy DTN routing algorithms using the ONE environment of the opportunistic network simulator. What is different is that ONE's open model ExternalMovement, is used to reproduce the real scene based on the real vehicle track data, and 155 concrete experiments are carried out by transforming four kinds of parameters. The two groups of original data are from the objective world: one is the track data of more than 500 taxis in San Francisco; the other is the track data of more than 7000 taxis in a city in the south of China. Because most of the rental vehicles are small vehicles, there is no great difference between the moving mode and the running track of the ordinary vehicles (except the vehicles with fixed routes, such as buses, trams, etc.), so it can fully represent the running rules of the general vehicles. The sampling simulation experiment based on real data and the evaluation of algorithm to a great extent eliminate the limitation of software simulation motion model to construct DTN network and more objectively and accurately reflect the characteristics of routing algorithm in application. Finally, based on the delivery success rate, average delay time and resource overhead ratio, five algorithms are compared horizontally and longitudinally, and a general conclusion is drawn.
【学位授予单位】：吉林大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TP393.02
【参考文献】
相关期刊论文 前5条
1 樊秀梅;单志广;张宝贤;陈辉;;容迟网络体系结构及其关键技术研究[J];电子学报;2008年01期
2 陈垣毅;郑增威;;星际网络通信协议研究进展综述[J];计算机应用研究;2011年02期
3 王朕;王新华;隋敬麒;;机会网络模拟器ONE及其扩展研究[J];计算机应用研究;2012年01期
4 王战红;孙明明;姚瑶;;Dijkstra算法的分析与改进[J];湖北第二师范学院学报;2008年08期
5 张晓生;黄陶明;;车载自组网研究[J];现代计算机(专业版);2009年07期
本文编号：2383773
本文链接：https://www.wllwen.com/guanlilunwen/ydhl/2383773.html
