AI-Enhanced Traffic Signal Optimization Using Microscopic Simulation Models for Congestion and Emissions Reduction in Mid-Sized U.S. Urban Corridors

Rapid urbanization, rising vehicle ownership, and limited roadway expansion have intensified traffic congestion and vehicular emissions in mid-sized U.S. cities. Conventional traffic signal control strategies, including fixed time and actuated systems, often lack the adaptability required to respond effectively to dynamic and stochastic traffic conditions. To address these limitations, this study proposes an AI-enhanced traffic signal optimization framework integrated with microscopic traffic simulation models to simultaneously improve traffic efficiency and reduce environmental impacts along urban corridors. The proposed framework employs reinforcement learning to dynamically adjust signal timing parameters based on real time traffic states such as queue lengths, vehicle delays, and flow variations. A microscopic simulation environment is utilized to model individual vehicle behaviors, including acceleration, deceleration, and idling, enabling precise estimation of performance indicators such as travel time, intersection delay, fuel consumption, and pollutant emissions. Comparative simulation experiments are conducted against traditional fixed time and actuated signal control methods. The results demonstrate that the AI-driven approach significantly reduces average travel time, intersection delay, and stop frequency across the corridor. Additionally, substantial reductions in carbon dioxide (CO₂) and nitrogen oxides (NOₓ) emissions are observed due to smoother traffic progression and reduced idle periods. The findings confirm that integrating artificial intelligence with microscopic simulation provides an effective and scalable solution for enhancing mobility and environmental sustainability in mid sized urban transportation networks, supporting the development of intelligent and sustainable smart city traffic management systems.