The vehicles in the condition of car-following perform different features comparing their motion in free traffic, and the discrepancy reflect driving style, which serves as important part of uncertainly of traffic. Several phenomena and driving characteristics, such as hysteresis in velocity-spacing curve, were found and utilized to improve the understanding of driving and some other safety study. Here we show that the hysteresis phenomenon is discrepant between drivers in stop-and-go waves (a

more »

... pecial case of car-following which is relative stable), and the diversity which indicates driver style (such as aggressive and timid) is quantified using some low order car-following models. By calculating driving characteristics and examining the trajectory data in NGSim dataset, we demonstrated the safe distance -which was usually considered adiabatically on velocity -was significantly affected by driving characteristics. We further analyzed statistical properties of driving characteristics in NGSim data, and on eventual we provided a classification method to evaluate security of car-following behavior and to adjust the safe distance for different drivers.