Enhancing QoS in Urban V2V Communications Through AI: A Machine Learning Approach in Vehicular Ad-Hoc Networks

This study introduces a machine learning-based Quality of Service (QoS) model that significantly enhances Vehicle-to-Vehicle (V2V) communications within urban vehicular networks. To validate our model, we compared its performance against the conventional Enhanced Distributed Channel Access (EDCA) mechanism and a scenario without QoS enhancements. Simulations were conducted in an urban area of Buenos Aires, Argentina. We utilized Open Street Maps (OSM) to provide a realistic geographic foundation, ensuring our simulation environment closely mirrors actual urban settings. SUMO (Simulation of Urban Mobility) was employed to accurately model vehicular movements and interactions, facilitating a dynamic and realistic traffic flow simulation. The ns-3 (Network Simulator 3) was used for the network communication aspects, simulating complex network protocols and interactions with high fidelity. Together, these tools enabled comprehensive testing of the QoS model, highlighting its capacity to improve packet delivery ratio (PDR), reduce end-to-end delays, and enhance received throughput. The results underscore the potential of machine learning to significantly optimize vehicular network communications, paving the way for more efficient and reliable urban transportation systems.