Research On Mobility Models Of Vehicular Ad Hoc Networks

Vehicular ad hoc networks are ad hoc networks built between vehicles and vehicles orbetween vehicles and roadside infrastructure. Its purpose is to provide communication amongvehicles so as to achieve information sharing. Vehicular ad hoc networks improve efficiencyand its security by the interaction among vehicle information (such as speed, acceleration,position, direction and, etc.) and traffic information (such as traffic congestion, traffic lights,some real-time road state and, etc.). Vehicular ad hoc networks are mainly used in followingfive aspects: intelligent transportation system of safety warning, assist driving, distributedtraffic information release, vehicle control and office entertaining based on vehiclecommunication.Vehicular ad hoc networks have attracted much attention both from academic andindustrial fields. At present, many related research based on vehicular ad hoc networks havebeen made. The node topology is changing frequently due to the fact that road restrictions andvehicles’ fast speed and many of existing protocols and applications cannot be suitable forvehicular ad hoc networks. Therefore, it is necessary for us to study out new protocols andapplications that meet the characteristics of vehicular ad hoc networks. Considering the hugetesting cost of human and material in the real scenario, evaluations on new proposedapplications and protocols are commonly executed by using simulation methodology.Since the fact that mobile nodes can meet the actual vehicle movement, characteristic isthe basis of the reliability software simulations, there is great need to design vehicle mobilitymodels to accurately simulate the vehicles in speed, acceleration, stop time, movementdirection, road restrictions, traffic rules and etc.An accurate mobility model of vehicular ad hoc networks is one of the key technologies toimprove the confidence level of simulation results playing a vital role in evaluate theprotocols and applications in vehicular ad hoc networks. Currently, research on vehiclemobility models is mainly devoted to add new inspection standards into the model or putforward to meet a certain scene. For the above reasons, in this article, we focus on theclassification and analysis of the existing vehicle mobility models. Typical models areselected and detailed descriptions of design principles are given, which generally reflect the classic mobility models’ development process. Some vehicle mobility characteristics areselected for statistics in macro and micro aspects, which shows the influence of differentdesign schemes of vehicle mobility model on the vehicle movement performance.In this paper, we analyze VANET mobility models based on the VanetMobisim platformin following aspects:(1) The available mobility model sanalyz the following three aspects: trafficcharacteristics, dependence and restriction conditions, and design methodologies, by whichsome novel macroscopic statistical metrics are proposed, furthermore, detailed classificationmethods and specific examples of each standard are given.(2) Several typical mobility models are selected and implemented for detailedcomparative analysis on design principles, some typical vehicle attribute are analyzed indetail. There are also comparative analysis of various microscopic characteristics andmacroscopic characteristics of each selected model.(3) At microscopic levels, metrics such as single movement interval time, singlemovement interval distance, single pause time, and single connection time are analyzed indetails. Simulations show that different mobility models exhibit similar characteristics underthe complementary cumulative distribution, but there is plentiful of differences on theprobability distribution of the corresponding metric.(4) From the perspective of macroscopic, many metrics are also analyzed in details suchas the average velocity, the total movement distance, the total movement time, the averagenumber of neighbors, and the average acceleration time. As shown from the result, themobility models display similar phase transition behaviors, however, the phase transitionpoints are different. Apparently, the corresponding metric follow the Poisson Distribution, butthe average values are completely different. Moreover, many other metrics such as theinstantaneous average speed, the instantaneous travelling distance, and the instantaneousaverage neighbor nodes are also investigated. Results indicate that the jitter range and thejitter tendency of instantaneous metric curves of every mobility model are different from eachother, and the average values of metrics have different sensitivity on the change of the totalnumber of vehicles.