Committee Chair

Liang, Yu

Committee Member

Wu, Dalei; Sartipi, Mina; Sun, Pengyuan

Department

Dept. of Computational Science

College

College of Engineering and Computer Science

Publisher

University of Tennessee at Chattanooga

Place of Publication

Chattanooga (Tenn.)

Abstract

Signalized intersections are persistent bottlenecks where inefficient operations contribute to congestion, delays, safety risks, and environmental impacts. Conventional control strategies provide stability under predictable demand but lack the adaptability required to manage stochastic and heterogeneous traffic conditions. This dissertation develops a decentralized graph-based multi-agent reinforcement learning (DGMARL) framework for adaptive traffic signal control. The framework advances the state of the art by (i) embedding operational constraints, including minimum/maximum green durations, pedestrian recalls, and clearance intervals, directly into the learning process; (ii) modeling intersections as decentralized agents that exchange direction-specific states through multi-head graph attention to capture asymmetric flows and upstream inflows, thereby enabling scalable coordination across large networks; and (iii) incorporating contextual pedestrian demand via point-of-interest weighting. Control policies are optimized within a constrained Markov decision process, where modular phase selection and fairness-aware rewards jointly balance vehicle efficiency and pedestrian accessibility. The framework is validated using a high-fidelity digital twin–based simulator with real-world traffic data and further demonstrated through preliminary on-street field testing on the MLK Smart Corridor. In simulation, the proposed approach reduced pedestrian waiting times by up to 24.7% and vehicle delays by 22.6%, while decreasing emissions (CO, CO_2, NO_x, and PM_10) by an average of 9.6% and increasing vehicle throughput by more than 22%. These improvements were achieved while ensuring compliance with safety-critical signal timing rules. Analysis of graph attention weights highlights interpretable coordination across intersections, confirming the robustness and scalability of the decentralized design under varied traffic conditions. In field operation, the decentralized agents correctly interpreted real-time traffic demand, switched signal phases adaptively, and responded to pedestrian push-button activations within approximately 10-15 s along with proper recall logic, maintaining Safety and Priority of Timing (SPaT) compliance and end-to-end processing latency between 80 and 120 ms. Together, these contributions establish a pathway toward deployment-ready, equitable, and sustainable traffic signal control across diverse network settings.

Acknowledgments

I would like to express my deepest gratitude to my advisors, Dr. Yu Liang and Dr. Dalei Wu, for their exceptional guidance, mentorship, and unwavering support throughout my doctoral journey. Their technical insight, patience, and continuous encouragement have profoundly shaped both my research direction and professional growth. I am sincerely thankful to Dr. Mina Sartipi for her visionary leadership, mentorship, and for providing me the opportunity to contribute through the Center for Urban Informatics and Progress (CUIP). Her guidance has been instrumental in aligning this research with real-world applications and community impact. I also extend my sincere appreciation to Dr. Penguan Sun for his valuable feedback and continued support during the later stages of this work. I am deeply grateful to my dissertation committee - Dr. Yu Liang, Dr. Dalei Wu, Dr. Mina Sartipi, and Dr. Penguan Sun - for their collective guidance, constructive feedback, and commitment that have greatly strengthened this dissertation. My sincere thanks go to my collaborators and field support team for their valuable contributions throughout this research. I would like to thank Dr. Abhilasha Saroj, Dr. Michael P. Hunter, and Dr. Angshuman Guin for their collaborative discussions, digital twin related insights, and guidance that strengthened the transportation research components of this work. I am grateful to Dr. Stevanovic Aleksandar, Dr. Ismet Goksad Erdagi, and Dr. Slavica Gavric for their technical collaboration in Eco\_PI development, field testing, and NTCIP-based integration that enabled the validation of the DGMARL framework under real-world conditions. Special thanks to Mr. Yasir Hassan and Mr. Austin Harris for their support with field data integration, computer vision inputs, SPaT and RTT communication, and real-field deployment coordination. I also acknowledge Mr. Toan Tran for his assistance in developing and integrating Eco\_PI function with the digital twin and Mr. Joseph Duhamel for his support with the economic impact analysis. I gratefully acknowledge the Department of Computer Science and Engineering and the Center for Urban Informatics and Progress (CUIP) at the University of Tennessee at Chattanooga (UTC) for providing the resources, facilities, and research environment that made this work possible. This research was supported in part by the U.S. Department of Energy (DOE) and the National Science Foundation (NSF) under associated research grants. Finally, I extend my deepest love and gratitude to my family and friends for their unconditional love, patience, and unwavering belief in me. Their constant encouragement and support have sustained me throughout this journey.

Degree

Ph. D.; A dissertation submitted to the faculty of the University of Tennessee at Chattanooga in partial fulfillment of the requirements of the degree of Doctor of Philosophy.

Date

12-2025

Subject

Intelligent transportation systems; Reinforcement learning; Traffic signs and signals--Control systems--Automation

Keyword

Traffic Signal Optimization; Multi-Agent Reinforcement Learning; Graph Neural Networks; Intelligent Transportation Systems; Digital Twin; Real-World Field Testing;

Discipline

Artificial Intelligence and Robotics | Graphics and Human Computer Interfaces

Document Type

Doctoral dissertations

DCMI Type

Text

Extent

xix, 225 leaves

Language

English

Rights

http://rightsstatements.org/vocab/InC/1.0/

License

http://creativecommons.org/licenses/by/4.0/

Date Available

1-1-2027

Download

Available for download on Friday, January 01, 2027

Share

COinS