A Comprehensive Review of Path Loss Prediction Methods and Models of Vehicular to Infrastructure (V2I) Communication

Abstract

Vehicle-to-Infrastructure communication is a key enabler of Intelligent Transportation Systems, supporting traffic efficiency and road safety. However, accurate path-loss prediction remains challenging due to dynamic urban propagation conditions, frequent non-line-of-sight situations, mobility, and heterogeneous roadside deployments. This paper presents a comprehensive review of V2I path-loss prediction methods, encompassing classical empirical and semi-deterministic models, as well as modern artificial intelligence approaches, including machine learning and deep learning. We summarize the main modeling assumptions, required inputs, strengths, and limitations across different environments and frequency bands, and we highlight recent trends such as hybrid physics-guided learning, data-driven feature engineering, and interpretability tools. Finally, we discuss open research challenges and future directions toward robust, generalizable, and practical V2I path-loss modeling.