Learning-Based Road Link Quality Estimation for Intelligent Alert-Message Dissemination

Abstract

Accurately assessing the quality of road links is essential for effectively sharing critical messages in dynamic vehicular network environments. Unfortunately, existing literature lacks models to estimate the quality of links between infrastructure and vehicles due to the complexity and variability of vehicular communication networks, including channel variations and interference patterns. To address this gap, we propose a prediction model based on supervised machine learning to estimate the Packet Reception Rate (PRR) on the road. Our model updates communication zones dynamically to align with traffic conditions. We train and evaluate our model using a dataset generated from a realistic mobility scenario simulated using NETSIM and SUMO. Our performance tests indicate promising results in terms of prediction accuracy. This work is an important step toward establishing an efficient and reliable scheme for disseminating alert messages, considering the fluctuations in traffic conditions and vehicular mobility.

Keywords

• vehicular communications
• quality of service assessment
• machine learning