Chaotic behaviour of noisy traffic data

Abstract

This paper describes the procedure of extracting information about the dynamics of highway traffic speed. The wavelet shrinkage is used to diminish the effect of the noise. Afterwards, the dynamical properties of the system are estimated through the 0-1 test for chaos, Lyapunov exponents and the notion of Shannon entropy. The results indicate the strong chaotic dynamics in the traffic speed data. In addition to that, the predictability of the system is related to the values of the maximal Lyapunov exponent and Shannon entropy. The higher those values are, the worse the predictability of the system is. Furthermore, it is shown that Shannon entropy can be used to detect changes in dynamics on different time scales.