Observed time-headway distribution and its implication on traffic phases

Time-headway between two subsequent vehicles in traffic flow is determined both by traffic dynamics and individual fluctuations, showing a complex behavior. Analyzing real traffic data from multiple sections of urban freeway, we show that the time-headway distributions under different velocities are fitted well with a class of distributions which can be decomposed into an exponential distributed variable plus an independent Gaussian fluctuation, where the exponential component can describe traffic states in a statistical meaning and the random fluctuation is caused by complex traffic dynamics which relates to imperfect individual driving behavior. Based on analyzing the empirical time-headway distribution, traffic flow is recognized as different phases which were observed at urban freeway previously. Moreover, high-velocity highway flow and low-velocity urban freeway flow can be described in a unified framework by shifting distribution parameters.