Modelling Road Traffic Congestion from Trajectories

Abstract

Traffic congestion has become a significant issue in almost all modern cities around the world. Existing research on congestion detection on roads takes a threshold-based approach that only detects the congestion as a binary notion, i.e., congestion or no-congestion. These approaches fail to address the intensity and residual effects of congestion over time, which is vital for real-time congestion detection and the propagation of congestion in adjacent road segments. To mitigate the above limitations, in this paper, we propose a mathematical model to quantify congestion on the road by incorporating the effect of change in average speed and the time decay of congestion. The parameters for calculating the congestion value is independent of road segments or varying structure of road network across cities. We have validated our proposed model of quantifying congestion through experiments on a real-world taxi trajectory dataset in an urban road network.