A Lightweight Few-Shot Learning-Based Traffic Classification System for Secure Internet of Vehicles

Abstract

The Internet of Vehicles (IoV), the latest generation of Vehicular Ad-hoc Networks (VANET) enables real-time, intelligent communication between vehicles and nearby transport infrastructures like roadside units, cloud servers, etc. It supports secure Vehicle-to-Vehicle (V2V), Vehicle-to-Infrastructure (V2I), and other communication forms, which are vital in intelligent transportation systems. However, the dependence on wireless channels introduces risks, such as Man-In-The-Middle (MITM) attacks, replay attacks, and impersonation attacks, that lead to potential data leakage. Furthermore, many existing schemes rely on a centralized Trusted Authority (TA) for mutual authentication, which creates scalability limitations and increases latency. To address these challenges, this paper proposes a privacy-preserving mutual authentication and key agreement protocol for IoV using blockchain and a multi-TA network, which enables decentralized V2V and V2I authentication. Additionally, a lightweight Few-Shot Learning (FSL)-based module is integrated at the Zone Manager (ZM) level to classify real-time traffic conditions based on limited labeled samples. This enhances traffic control without compromising security. The distributed ledger managed by multiple TAs ensures synchronized access to authenticated credentials, facilitating secure communication across different zones. The security and performance analyses demonstrate the proposed scheme’s security and efficiency over existing methods.