Utilization of Artificial Intelligence (AI) techniques in intelligent transportation systems opens up new dimensions in choreographing sustainable urban mobility. However, one of the main issues concerns the appropriate context or situation where such techniques ought to be adopted. They have several alternatives, including the utilization of cloud computing, fog computing, edge computing, or even their mobile devices. A smart traffic management system based on the Internet of Things (IoT) concept is proposed in this paper. We optimize the use of evolutionary algorithms, starting with the Lightweight Random Early Detention (LRED) for Vehicles Dynamic (VD) mechanism. LREDfor VDs can be employed in controlled junctions to clear oncoming traffic and optimize the cycle and phases of the traffic lights. Then the authors explain that after LRED for VDs has been successfully optimized in a non-real-time environment, it is possible to deploy the approach to an unknown traffic situation without the need to involve AI in edge IoT devices. The versatility of this mechanism is extensively assessed using the traffic simulation package, SUMO. iREDVD outperforms all other competing designs since it minimizes the waiting time of vehicles, average travel time, fuel usage, and emission of solid and gaseous pollution, among other benefits.