Research on the Application of Homomorphic Encryption and Federated Learning in the Internet of Vehicles Environment

Abstract

The rapid growth of the Internet of Things has significantly increased data volumes, leading to heightened concerns over security risks such as data theft and leakage. As machine learning becomes increasingly integral to various applications, data security in training processes has emerged as a critical issue. The Internet of Vehicles (IoV), as a crucial branch of the IoT, faces particular challenges in securely and efficiently training data. While current machine learning frameworks enable fast and efficient data training in IoV environments, security risks remain a pressing concern. This study explores the use of a federated learning framework enhanced with homomorphic encryption to address these issues. The research involves simulating real-world environments to test the basic performance and feasibility of the selected framework in IoV applications. Additionally, the impact of homomorphic encryption on the framework's effectiveness is assessed. Finally, a comparative analysis with traditional machine learning frameworks demonstrates that the chosen federated learning framework, when combined with homomorphic encryption, offers superior efficiency and security in IoV scenarios. This study underscores the potential of integrating advanced encryption techniques in machine learning frameworks to enhance data security in the IoV.