Stochasticity of Road Traffic Dynamics: Comprehensive Linear and Nonlinear Time Series Analysis on High Resolution Freeway Traffic Records

Abstract

The dynamical properties of road traffic time series from North-Rhine Westphalian motorways are investigated. The article shows that road traffic dynamics is well described as a persistent stochastic process with two fixed points representing the freeflow (non-congested) and the congested state regime. These traffic states have different statistical properties, with respect to waiting time distribution, velocity distribution and autocorrelation. Logdifferences of velocity records reveal non-normal, obviously leptocurtic distribution. Further, linear and nonlinear phase-plane based analysis methods yield no evidence for any determinism or deterministic chaos to be involved in traffic dynamics on shorter than diurnal time scales. Several Hurst-exponent estimators indicate long-range dependence for the free flow state. Finally, our results are not in accordance to the typical heuristic fingerprints of self-organized criticality. We suggest the more simplistic assumption of a non-critical phase transition between freeflow and congested traffic.