Learning-Based Infrastructure To Vehicle Link Quality Estimation

Timely and up-to-date information about road link quality is essential for the efficient and reliable dissemination of urgent messages in dynamic vehicular network environments. However, existing literature lacks models that provide infrastructure-to-vehicle link quality estimations. This is due to vehicular communication networks’ high variability and complexity (channel variations, complex interference patterns), making it difficult to integrate into a practical analytical model. To address this, we propose a supervised-machine-learning-based prediction model that estimates the Packet Reception Rate (PRR) on the road. Our model updates communication zones to adapt to changes in traffic conditions. It uses a dataset generated from a realistic mobility scenario using the NETSIM simulator and SUMO for training and evaluation. Our performance tests show promising results in terms of prediction accuracy. This work represents the first step toward developing an efficient and reliable scheme for disseminating urgent messages that considers traffic conditions and vehicular mobility changes.