中文摘要：

车载自组网已经在学术界和工业界受到广泛重视,新型应用和协议在实际工况条件下的测试需要很多人力和物力,并且不易控制,因此通常采用软件模拟方式进行实验验证.准确的移动模型是提高车载自组网模拟结果置信度的关键技术之一.文中从对交通特征的刻画粒度、移动模型的依赖与限制条件和移动模型的建立方法3个方面对已有的移动模型进行综合分析,提出若干面向宏观行为的统计特征量,并选取典型移动模型对其设计原理、微观特征和宏观特征等进行详细对比分析.微观方面,分析了各移动模型车辆的单次移动时间、单次移动距离、单次暂停时间和单次连接时间等特征量,结果发现不同模型在上述特征量的补累积分布下具有相似特征,而对应特征量的概率分布则具有一定差别.宏观方面,分析了各移动模型单个车辆的平均速度、总移动距离、总移动时间、平均邻居数量和平均加速度等特征量,结果表明不同模型在上述特征量的补累积分布下也体现出近似的相变行为,但各自的相变点不同,各模型对应特征量的概率分布具有明显的泊松分布特征,但均值差异较大.另外,也分析了各移动模型所有车辆的瞬时平均速度、瞬时移动距离、瞬时移动车辆数目、瞬时平均邻居节点数和瞬时平均加速度等特征量,结果表明各移动模型瞬时特性曲线的抖动强度和抖动趋势均不同,各模型特征量的平均值对车辆数量变化的敏感度也不尽相同.

英文摘要：

Vehicular ad hoc networks have attracted much attention both from academic and industrial fields.Considering the huge testing cost of human and material in the real scenario,the evaluation on the new proposed applications and protocols is commonly executed by using simulation methodology.An accurate mobility model of vehicular ad hoc networks is one of the key technologies to improve the confidence of simulation results.The available mobility models are analyzed from three aspects： the traffic characteristic,the dependence and restriction conditions,and the design methodologies,by which some novel macroscopic statistical metrics are proposed.Several typical mobility models are selected and implemented for detailed comparative analysis on design principles,microscopic characteristics,and macroscopic characteristics.At microscopic levels,the metrics such as single movement interval time,single movement interval distance,single pause time,and single connection time are analyzed in details.Simulations show that different mobility models exhibit similar characteristics under the complementary cumulative distribution,but there is plentiful of differences on the probability distribution of the corresponding metric.From the perspective of macroscopic,many metrics are also analyzed in details,e.g.the average velocity,the total movement distance,the total movement time,the average number of neighbors,and the average acceleration time.Results show that the mobility models display similar phase transition behaviors.However,the phase transition points are different.The corresponding metric follow apparently Poisson distribution,but the average values are completely different.Moreover,many other metrics such as the instantaneous average speed,the instantaneous travelling distance,and the instantaneous average neighbor nodes are also investigated.Results show that the jitter range and the jitter tendency of instantaneous metric curves of every mobility model are different from each other,and the average values of metrics have