Segment-wise Description of the Dynamics of Traffic Congestion

TL;DR

This paper demonstrates that segment-wise analysis of traffic data provides more detailed insights into congestion dynamics, revealing two types of congestion and linking the area enclosed by loopy trajectories to economic loss.

Contribution

It introduces a segment-wise approach to analyze traffic congestion, identifying two congestion types and correlating the area in fundamental diagrams with economic impact.

Findings

Segment-wise data is more informative than point-wise data.

Two congestion types are distinguished by the area enclosed in loopy trajectories.

The enclosed area correlates with economic loss due to congestion.

Abstract

We compare the point-wise and segment-wise descriptions of the traffic system. Using real data from the Taiwan highway system with a tremendous volume of segment-wise data, we find that the segment-wise description is much more informative of the evolution of the system during congestion. Congestion is characterized by a loopy trajectory in the fundamental diagram. By considering the area enclosed by the loop, we find that there are two types of congestion dynamics -- moderate flow and serious congestion. They are different in terms of whether the area enclosed vanishes. Data extracted from the time delays of individual vehicles show that the area enclosed is a measure of the economic loss due to congestion. The use of the loss area in helping to understand various road characteristics is also explored.