Tunnel-like road foliage path loss prediction for V2V communications for LoRa LPWAN 433 MHz and 920 MHz bands

An accurate path loss modeling is essential for reliable vehicle-to-vehicle (V2V) and infrastructure-to-vehicle communication. However, many existing propagation models and machine learning-based approaches primarily focus on environmental and distance-related parameters, while neglecting the impact of radio configuration and mobility factors. This study proposes a new path loss model which integrates both wireless channel characteristics and machine-related parameters, including frequency, bandwidth, spreading factor, transmit power, vehicle speed, and distance. In addition, a real-world measurement campaign was conducted in a tunnel-like foliage environment, representing a challenging and underexplored propagation scenario. The proposed model is evaluated against conventional models and other machine learning approaches using root mean squared error (RMSE) and MAE metrics. The results show that the ANFIS model achieves superior performance, with the lowest error values of 5.677 (RMSE) and 4.187 (MAE) across both 433MHz and 920 MHz frequency bands. These findings demonstrate that incorporating heterogeneous parameters with machine learning techniques significantly improves prediction accuracy and highlight the effectiveness of the proposed approach for complex V2V communication environments.