Alleviating road network congestion: Traffic pattern optimization using Markov chain traffic assignment

Exacerbated urban road congestion is a real concern for transportation authorities around the world. Although agent-based simulation and iterative design approaches are typically used to provide solutions that reduce congestion, they fall short of meeting planners' need for an intelligent network design system. Since Markov chains are remarkably capable of modeling complex, dynamic, and large-scale networks, this paper leverages their theory and proposes a mathematical model based on Markov chain traffic assignment (MCTA) to optimize traffic and alleviate congestion through targeted direction conversions, i.e. two-way to one-way flow conversions. The approach offers an intelligent traffic pattern design system, one which can analyze an existing complex network and suggest solutions taking into consideration network-wide interdependencies. Specifically, the paper presents a binary nonlinear mathematical model to optimize road network traffic patterns using maximum vehicle density. The model is then solved using Genetic Algorithm (GA) optimization methodology, and a fine-tuning search algorithm is proposed to improve upon GA results in terms of solution's practicality and fitness. The approach is applied to a city setting and experimental results are reported. Finally, an application in time-sensitive decision-making is discussed.