当前位置:首页>> >>


NS2平台的AdHoc网络路由协议仿真.rar

收藏

资源目录
    文档预览:
    编号:20180914220055298    类型:共享资源    大小:2.85MB    格式:RAR    上传时间:2018-09-15
    尺寸:148x200像素    分辨率:72dpi   颜色:RGB    工具:   
    45
    金币
    关 键 词:
    NS2 平台 AdHoc 网络 路由 协议 仿真
    资源描述:
    IEEE TRANSACTIONS ON MOBILE COMPUTING, VOL. 13, NO. 5, MAY 2014 1061Generation and Analysis of a Large-Scale UrbanVehicular Mobility DatasetSandesh Uppoor, Student Member, IEEE, Oscar Trullols-Cruces, Student Member, IEEE,Marco Fiore, Member, IEEE, and Jose M. Barcelo-Ordinas, Member, IEEEAbstract—The surge in vehicular network research has led, over the last few years, to the proposal of countless network solutionsspecifically designed for vehicular environments. A vast majority of such solutions has been evaluated by means of simulation, sinceexperimental and analytical approaches are often impractical and intractable, respectively. The reliability of the simulative evaluationis thus paramount to the performance analysis of vehicular networks, and the first distinctive feature that has to be properly accountedfor is the mobility of vehicles, i.e., network nodes. Notwithstanding the improvements that vehicular mobility modeling has undergoneover the last decade, no vehicular mobility dataset is publicly available today that captures both the macroscopic and microscopicdynamics of road traffic over a large urban region. In this paper, we present a realistic synthetic dataset, covering 24 hours of cartraffic in a 400-km2region around the city of Köln, in Germany. We describe the generation process and outline how the datasetimproves the traces currently employed for the simulative evaluation of vehicular networks. We also show the potential impact thatsuch a comprehensive mobility dataset has on the network protocol performance analysis, demonstrating how incompleterepresentations of vehicular mobility may result in over-optimistic network connectivity and protocol performance.Index Terms—Vehicular mobility, scenario generation, network connectivity, epidemic dissemination1INTRODUCTIONPRIVATELY owned cars and public transport vehi-cles are envisioned to become actual communicationhubs in the near future, as heterogeneous network inter-faces become integral part of the car equipment and theseamless Internet connection capabilities offered by tabletsand smartphones lure passengers’ attention. As a result,vehicular environments have recently emerged as a promis-ing area of research to the telecommunication networkingcommunity. The introduction of enhanced infrastructure-based systems, involving the likes of WiMAX and LTE-Atechnologies, and novel communication paradigms, such asad hoc and opportunistic networking, has paved the wayfor the proposal of a flurry protocols specifically designedfor the forthcoming communicating vehicles and coveringall the layers of the network stack.The performance evaluation of most of these vehicularnetworking solutions requires large-scale scenarios, mak-ing direct experimental assessments impractical due to theircost and complexity. Simulation becomes then the tool ofchoice in the validation of new network architectures andprotocols for vehicular environments.• S. Uppoor is with INSA Lyon and Inria, 69621 Lyon, France. E-mail:sandesh.uppoor@insa-lyon.fr.• O. Trullols-Cruces and J. M. Barcelo-Ordinas are with the UniversitatPolitècnica de Catalunya, 08034 Barcelona, Spain. E-mail: {trullols,joseb}@ac.upc.edu.• M. Fiore is with CNR-IEIIT and Inria, 10129 Torino, Italy. E-mail:marco.fiore@ieiit.cnr.it.Manuscript received 29 Oct. 2012; revised 1 Feb. 2013; accepted 8 Feb. 2013.Date of publication 20 Feb. 2013; date of current version 15 May 2014.For information on obtaining reprints of this article, please send e-mail to:reprints@ieee.org, and reference the Digital Object Identifier below.Digital Object Identifier 10.1109/TMC.2013.27Unfortunately, simulative performance evaluation ofvehicular networks are often biased by the underlyingmobility representation. As a matter of fact, as repeatedlyproven in the past [1]–[3], the movement of vehicles candramatically affect the behavior of network protocols, andan incorrect representation of car traffic can lead to mis-leading conclusions, even in presence of a flawless network-level simulation. As a result, it is today acknowledged that,for the results of a vehicular simulative campaign to becredible, mobility traces must be employed that capturethe unique macroscopic and microscopic dynamics of carmovement patterns.Such considerations have led to substantial progressin the quality of car movement traces for vehicular net-working research over the last few years. The simplisticstochastic models employed in early works [1] have beenreplaced by random mobility over realistic road topolo-gies [4] at first, and by microscopic vehicular modelsborrowed from transportation research [5] later on. Thesefeatures were then included in dedicated simulation envi-ronments, and integrated with road signalization [6], [7].Ever since, vehicular mobility simulators have been grow-ing their complexity and features, allowing to accuratelysimulate the individual movement of vehicles over realisticroad topologies [8]. Moreover, in parallel with the evolu-tion of synthetic traces of vehicular mobility, real-world cartraffic dataset have grown in number and scale.In this paper, we go a step further in assessing theimpact that realism in the vehicular mobility representationhas on the design and evaluation of networking solutions.To that end, we provide a threefold contribution. First,we present in Section 2 a concise yet comprehensive sur-vey of the state of the art in road traffic modeling and1536-1233 c© 2013 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.1062 IEEE TRANSACTIONS ON MOBILE COMPUTING, VOL. 13, NO. 5, MAY 2014TABLE 1Major Features of the Vehicular Mobility Traces Currently Available for Network Simulationtracking, discussing the strengths and weaknesses of thevehicular mobility traces currently available for networkingresearch. Second, we introduce an original synthetic datasetof urban vehicular mobility, that is characterized by anunprecedented combination of scale, detail and realism,and publicly available [9]. We detail the generation processof this new trace through a set of open-source state-of-art tools in Section 3, discuss the challenges it poses andhow to solve them in Section 4, and provide an analysisof its features in Section 5. Third, we show in Section 6the impact that such a comprehensive representation ofvehicular mobility can have on the evaluation of network-ing technologies. To that end, we compare our datasetagainst several traces commonly used in vehicular net-work simulation, in terms of pure connectivity propertiesas well as in a practical networking use case. Our analy-sis evidences the vehicular network performance bias thatmobility traces with limited microscopic and macroscopicdetail can induce, and leads to the future research directionsoutlined in Section 7.2RELATED WORKThe relevance of mobility modeling to the simulation ofvehicular networks has been long acknowledged, a fac-tor that has pushed the research community to seek foran ever-increasing realism in road traffic traces fed to net-work simulators. In this section, we overview the bodyof work on vehicular mobility traces for network simu-lation. We categorize the datasets based on the nature oftheir macroscopic traffic data, i.e., the sources employed todetermine the time and routes of trips traveled by individ-ual vehicles in the dataset. The most relevant features ofthe different mobility traces are summarized in Table 1.Formore details on the simulation environments mentioned inthe table, we refer the interested reader to [8], [10].2.1 Perception and Small-Scale MeasurementsA number of synthetic vehicular mobility traces were gener-ated by feeding real-world road topologies and perception-based macroscopic traffic information to microscopic-levelsimulators, such as SUMO [6] or VanetMobiSim [7].Notwithstanding the high level of detail granted by theuse of such simulators, these traces yield simplistic large-scale features. Indeed, the macroscopic traffic data theyemploy are based on the authors’ perception of the roadtraffic in the simulated area, as in the traces of Porto,Portugal [11], of several areas of Turin, Italy [12], and ofdowntown Karlsruhe, Germany [13]. As an alternative, sim-ple assumptions are made, as in the vehicular mobilitytrace of the city of Zurich, Switzerland [14], where largerroads attract more traffic. Finally, small-scale measurementsconducted by the authors themselves are also used to com-plement intuition, as in a trace of car traffic in the center ofBerlin, Germany [15]. However, all these approaches lackthe statistical rigor needed for a realistic representation ofthe macroscopic traffic distribution, and, as such, they canhardly capture the complexity of traffic flows in urban areasor their evolution over long time periods. Also for these rea-sons, these traces only cover modest geographical surfacesof few tens of square kilometers, or have a limited timeduration in the order of tens of minutes.2.2 Road Traffic ImageryAn original approach to the derivation of the macroscopictraffic information is adopted in [16], where stereoscopicaerial photography is leveraged to capture the vehicleUPPOOR ET AL.: GENERATION AND ANALYSIS OF A LARGE-SCALE URBAN VEHICULAR MOBILITY DATASET 1063distribution in the city of Porto, Portugal. A private aircraftwas flown over the city for two hours in the early afternoonof a weekday, and photographs were shot from the planeevery 5 seconds. The flight followed a parallel row patternso as to cover the whole geographical area of 41.3 km2cor-responding to the surface of Porto. By studying the aerialimagery, the authors were able to reconstruct a single snap-shot of the positions of 10566 vehicles in the urban area.Given that pictures of different city zones were taken atdifferent moments, car positions in the snapshot refer todifferent instants: the time error is of 23 minutes betweentwo cars, on average. Although this appears as an interest-ing way to derive static macroscopic data, its applicabilityto the generation of mobility traces is not immediate, due totime error above and to the the cost of running the aerialphotography campaign for a time sufficient to derive anactual mobility trace rather than a single snapshot.Another recent attempt at using imagery to estimate themacroscopic behavior of car traffic is presented in [17].There, the authors exploit the pervasiveness of road surveil-lance cameras to infer traffic densities in ten differenturban areas, including London, Sydney and Toronto. Thisapproach provides coarse information on the traffic flows,and could be used for the calibration of microscopic vehic-ular mobility, similar to the roadside detectors discussednext. However, no actual mobility trace leveraging suchdata is available to date.2.3 Roadside DetectorsInduction loops, infrared counters and roadside sensorsrepresent the traditional way to measure vehicular trafficflows in both freeways and urban road networks. In [18],two sets of empirical data are used, obtained from dual-loop and metal detectors from sections of the I-80 Freewayin Berkeley, CA, USA, and of the Gardiner Expressway inToronto, Canada. The detector information covers a spanof 24 hours and allows to determine the per-lane inter-vehicle arrival time and spacing. The data can be fed toa microscopic simulator to derive the position of individ-ual vehicles over time, however its validity is limited tohighway environments.Roadside detector are instead employed in a urban envi-ronment within the iTetris project [19]. Synthetic vehicularmobility traces of several areas of the city of Bologna weregenerated accounting for macroscopic traffic data acquiredthrough 636 induction loops spread over the road network,and complemented by user surveys on usual commutingtrips. The main trace covers 20.6 km2in the city center for aperiod of one hour, featuring the movement of 10333 vehi-cles. Thanks to the real-world nature of the macroscopicdata they are built upon, these traces reach an unprece-dented level of realism. Unfortunately, they do not coverlarge surfaces nor long time periods. Moreover, publicaccessibility to the mobility traces is not granted yet by theproject consortium.A similar approach has also been taken in [20], wherethe authors calibrate the microscopic mobility simulationof the city of Luxembourg through traffic counter informa-tion gathered by the local Ministry of Transport. As suchreal-world data only covers major traffic arteries, it is com-plemented by driver routes inferred from the differentnature of geographical zones in the area under study, andused to define traffic flows on medium- and small-sizedroads. The resulting mobility trace covers a very large areaof 1700 km2and features 150000 car trips. Although avery interesting dataset, the Luxembourg trace focuses onhighway and major road traffic, is limited to the morningperiod, and only accounts for in-bound flows, i.e., trafficmoving towards the city center.2.4 Socio-Demographic SurveysSocio-demographic surveys represent a significant sourceof information for the derivation of vehicular traffic data.The seminal work in [21] presents a synthetic mobilitytrace whose macroscopic model is derived by knowledgeof drivers’ activity in downtown Portland, OR, USA. Theresulting mobility dataset is acknowledged to be very real-istic, but only covers 15 minutes of car traffic in an areaof 21 km2, for a total of 16529 simulated cars. Similarissues affect the trace of Braunschweig, Germany, employedin [22], that features highly realistic road traffic description,but it is limited to a small region of 12 km2.Moreover,boththese datasets were obtained through commercial mobilitysimulators and are not publicly available.The largest vehicular mobility trace generated to datereproduces the car traffic in the whole Canton of Zurich,a 65000 km2region of Switzerland [23]. There, 24 hours ofcar traffic for the whole region are obtained from the SwissRegional Planning Authority and complemented using the1994 Swiss National Travel Survey. The resulting syntheticmobility dataset, as well as subsets of the same, are widelyemployed in the vehicular networking literature. However,the size of the road topology forces the authors to limitthe detail of the microscopic-level simulation. Thus, theyresort to a queue-based Multi-agent Microscopic TrafficSimulator (MMTS) [24], significantly less accurate thanstandard fine-grained vehicular mobility simulators basedon car-following models, employed in all the synthetictraces previously mentioned. Moreover, and again for scal-ability reasons, the road topology is pruned down to majortraffic arteries, and only the morning and afternoon trafficpeaks hours are modeled.2.5 Real-World TrackingMost of the previously discussed traces are syntheticallygenerated by injecting macroscopic data into a microscopicmobility simul
    展开阅读全文
    1
      金牌文库所有资源均是用户自行上传分享,仅供网友学习交流,未经上传用户书面授权,请勿作他用。
    0条评论

    还可以输入200字符

    暂无评论,赶快抢占沙发吧。

    关于本文
    本文标题:NS2平台的AdHoc网络路由协议仿真.rar
    链接地址:http://www.gold-doc.com/p-158431.html

    当前资源信息

    4.0
     
    (2人评价)
    浏览:25次
    bysj上传于2018-09-15
    1
    关于我们 - 网站声明 - 网站地图 - 资源地图 - 友情链接 - 网站客服客服 - 联系我们
    copyright@ 2014-2018 金牌文库网站版权所有
    经营许可证编号:浙ICP备15046084号-3
    收起
    展开