5G POWERED INTELLIGENT V2X DRIVING SYSTEM WITH AUTOMATED TRAFFIC LIGHT CONTROL FOR IMPROVED TRAFFIC FLOW AND SAFETY

Abstract

Traffic control is a major issue in smart urban cities, with existing systems failing to predict congestion and handle current traffic flow, leading to worsening jams. In this research, 5G-Traffic AI is presented, which is a novel framework utilising 5G and AI-based technologies for improved traffic movement and security. The integration of the V2X communication and 5G makes it possible for any information to be sent quickly and delivered almost instantly. The framework is designed in a multi-layered fashion where LSTM networks model congestion tendencies at the level of a city considering a mix of past and present time data. The GNN can represent the spatial structure among the nodes as a relation among the neighboring directional links and allows relating many traffic features interconnected. A Deep Reinforcement Learning (DRL) agent controls traffic at the fine granularity for better synchronization concerning global and local tendencies. This multi-level structure provides capability to adjust to real conditions constantly given the constant updating of data which makes LSTM forecasts, the spatial models based on GNN and DRL rules very much up to date. Experiments provide evidence that the traffic reduces significantly, the congestion is alleviated on a large scale and more traffic safety is introduced, which confirms the effectiveness of the mechanisms and their ability to real-time interaction with the environment.