Detection of C-V2X Spoofing Attacks using Physical Layer Features and Graph Neural Networks
Danilo Greco, Muhammad Sohail, Mario Marchese
Abstract
Cellular Vehicle-to-Everything (C-V2X) communication underpins intelligent transportation systems but remains susceptible to spoofing attacks that inject false information, compromising safety and reliability. While traditional cryptographic security verifies sender credentials, it cannot assure the veracity of message content if credentials are compromised. Physical Layer Security (PLS) provides a complementary defense by analyzing inherent wireless signal characteristics. This paper details the implementation and simulation of an SDR-DL (Software-Defined Radio and Deep Learning) framework for C-V2X spoofing detection, inspired by recent research. The Pythonbased simulation, utilizing TensorFlow/Keras for the Position-Change Detector (PCD) and PyTorch Geometric for the Graph Neural Network (GNN), demonstrates the framework’s feasibility, achieving high detection accuracy for the modeled spoofing scenarios based on Received Signal Strength Indicator (RSSI) patterns. We discuss implementation details, highlight challenges including data requirements and computational complexity, and propose future research directions such as multi-feature fusion and explainable AI integration.