analysis_phd.xlsxMRP_PHD_DATA

Traffic congestion is a major challenge in urban transportation networks, leadingto increased travel time, fuel consumption, and emissions. This paper presents anovel approach for optimizing traffic signal control using deep Q-learning (DQL)algorithms. By leveraging real-world traffic data obtained from aerial footage ofan intersection, imported into the Simulation of Urban Mobility (SUMO) environmentusing DataFromSky, this study provides a practical implementation ofDQL in dynamic traffic signal optimization. The performance of the DQL-basedtraffic light control (TLC) is compared with traditional Q-learning (QL) andfixed-time control methods. Experimental results demonstrate that DQL significantlyreduces mean travel time and improves traffic flow efficiency compared toconventional methods.