| Abstrakt: |
Traffic congestion increases day by day because of the number of vehicles, lack of traffic control, and the limitations of the technologies being deployed. People in big cities suffer from this issue and look for a permanent solution. This congestion causes air pollution, noise pollution, and may lead to a catastrophic situation. In big cities with large populations, traffic congestion happens daily and affects the quality of air. Various projects have been placed to resolve this issue; however, it remains unresolved, and drivers get totally annoyed. Numerous technologies, such as IoTs and deep learning are widely applied recently due to their results and costs. In this article, a novel Intelligent Adaptive Traffic Control and Management System (IATCMS) based on a deep-learning architecture and IoTs is proposed. This system utilizes Wireless sensor networks (WSNs), cameras, a Database, a Central Control Unit (CCU), and a newly developed Generative Adversarial Network (NDGAN). This approach aims to reduce congestion, provide air quality, reduce fuel consumption, control traffic, and save lives in case of accidents by letting ambulances reach destinations quickly. In addition, the method ensures smooth traffic with a reasonable speed. Various key factors, such as speed of vehicles, traffic density, length of roads, and allocated light signal times are considered to provide a feasible solution to remove or reduce congestion and minimize traffic waiting time at intersections. This algorithm determines which lane gets the highest priority to proceed first by computing the traffic density for each road and gives the lane with the highest density priority. The presented model is tested and analyzed in MATLAB. The approach reduces the average waiting time at intersections by nearly 34% to 48% and removes congestion in some scenarios. The achieved results indicate that the proposed system can be applied at any intersection around the world to provide a smooth trip and remove traffic congestion. [ABSTRACT FROM AUTHOR] |
Full text from SpringerLink