A Markovian Framework for Modeling Dynamic Network Traffic

An urban transportation network is a complex and stochastic system with high degrees of unpredictability and uncertainty. While significant advances have been made in modeling dynamic network traffic, existing efforts often involve overly complex methods for modeling the physics of traffic and high cost in data acquisition. In this paper, we propose a novel Markovian framework for modeling dynamic network traffic by 1) using a full-system/subsystem scheme and data from Google Maps to estimate travel times for links, 2) relaxing the stationary assumption that previous work requires, and 3) allowing incoming and outgoing traffic streams to the network of interest. In the case study of downtown Baltimore, we simulate the dynamic network traffic with a transition matrix determined from real-world data. The simulation results show consistency with the estimates provided by Google Maps. Finally, we conclude the paper, summarize strengths and limitations of our work, and suggest future research directions.