Self similarity behavior in highway traffic arrival pattern

Abstract

This paper investigates the behavior of highway vehicle arrival patterns by using highways' traffic statistics provided by the Texas Department of Transportation. Inspired from Internet traffic self-similarity studies, this paper explores whether vehicle arrival patterns on highways exhibit the self-similarity property. This paper first statistically verify that vehicle arrival pattern on highways does not follow the Poisson process based on the above traffic data. It proceeds to investigate the self-similarity behavior of these vehicle arrival patterns by estimating the Hurst parameter that is used to quantify the existence of self similarity. Hypothesis Testing and Goodness of Fit test are then applied to examine the statistical confidence of the results obtained. Results show that traffic arrival pattern for highways under moderate to heavy flow conditions exhibit self similarity behavior. This conclusion is a new finding on the vehicle time headway distribution studies.