ML-Driven Prediction of QoS in C-V2X Scenarios

This paper explores the efficacy of a Light Gradient Boosting Machine (LGBM) model in predicting downlink throughput within a Cellular Vehicle-to-Everything (C-V2X) environment. Utilizing the Berlin V2X dataset, the model demonstrates high accuracy, achieving an R2 score of 97% and a mean absolute error (MAE) of approximately 3 Mbps. The study underscores the model's utility in enhancing vehicular communication systems by facilitating reliable quality-of-service (QoS) predictions. The model ensures efficient and effective throughput predictions by focusing on a minimal set of impactful network features and employing a simple supervised regression approach.