Power-law distribution in an urban traffic flow simulation

Abstract

We found power-law behavior in the distribution of traffic on road segments in urban traffic simulations using digitized map of Kobe city in Japan as an example of an actual road network. As a comparison, we performed simulations using an artificial random road network and Manhattan-type road network. Similar power-law behavior was confirmed in the former, but not the latter. The behavior appeared robustly with or without traffic congestion, which suggests that its origin is not the interaction between vehicles. The power-law exponent was fitted using least squares method and obtained as -1.1 for Kobe city and the random road network, with optimization to avoid traffic congestion. The result did not change with the use of a different origin and destination distribution. From these results, one of the reasons that caused the power-law behavior was considered to be the randomness of the road network connection and edge lengths, whose fluctuations are obvious both in Kobe city and the random road network, unlike the grid network.