Improved Routing Model with Traffic Accident Prediction Using Machine Learning in Vehicular Communication

An innovative approach, IRTPML, aims to enhance both routing efficiency and road safety in vehicular networks. The proposed model integrates geometric particle swarm optimization for routing and convolutional neural networks (CNNs) for traffic accident prediction, leveraging vehicle-to-vehicle communication. By dynamically adapting routing paths based on network conditions and issuing early warnings for potential traffic accidents, the IRTPML model demonstrates significant improvements in energy efficiency, energy consumption, average delay, network throughput, and data delivery ratio compared to existing protocols. Extensive simulations and comparative performance analysis validate the efficacy of the proposed model, suggesting its promising potential for optimizing VANETs and enhancing transportation safety.