Comparative Analysis of ANN, SVR, and AdaBoost Models for V2V Path Loss Prediction in Urban and Suburban Environments

Advances in vehicle-to-vehicle (V2V) communication technologies have significant potential for improving road safety and traffic efficiency. This study investigated the use and performance of machine learning models for road loss estimation in V2V communication in urban and suburban environments. For this purpose, 34,000 datasets collected through extensive field measurements from Trabzon and Gümüşhane were analyzed to investigate and compare the performance of Artificial Neural Networks (ANN), Support Vector Regression (SVR), and AdaBoost models. Tree, vehicle, building density and distance were taken as common input parameters in all models and a 5-fold cross-validation technique was used to measure the fitness of the model and adjust the parameters. The results of our study show that the ANN model outperforms the other models, exhibiting good accuracy with the lowest Root Mean Square Error (RMSE) of 0.0966, Mean Square Error (MSE) of 0.0093, and Mean Absolute Error (MAE) of 0.0759. Additionally, it achieved the largest R-squared $\left(R^{2}\right)$ value of 0.9651. Consequently, this study aims to contribute to the development of intelligent transportation systems (ITS) by highlighting the importance of machine learning models that outperform traditional empirical models in path loss prediction in V2V and by performing performance comparisons.