Federated learning with enhanced cryptographic security for vehicular cyber-physical systems

Area of Science:

• Cybersecurity
• Intelligent Transportation Systems
• Machine Learning

Background:

• Vehicular cyber-physical systems (VCPS) are crucial for intelligent transportation systems (ITS), enabling features like self-driving vehicles and real-time traffic control.
• The sensitive data handled by VCPS poses significant security risks, including threats to public safety and financial losses.
• Existing security measures often struggle to balance robust protection with the performance demands of modern transportation networks.

Purpose of the Study:

• To propose a novel federated learning framework designed to enhance data security within vehicular cyber-physical systems (VCPS).
• To ensure data privacy during decentralized model training and secure aggregation of local models into a global model.
• To address the critical need for robust security in evolving ITS infrastructure.

Main Methods:

• Implemented a federated learning approach enabling decentralized model training on vehicles and roadside units (RSUs).
• Integrated the Laplace method for adding noise to model updates, further protecting privacy during global model aggregation.
• Established a collaborative framework involving vehicles, RSUs, and a central server to prevent information leaks.

Main Results:

• The proposed federated learning design demonstrated superior performance compared to traditional cryptographic techniques on the CICIDS2017 dataset.
• The method effectively preserved high levels of data confidentiality and system security.
• Achieved these security enhancements without compromising computational efficiency or model accuracy.

Conclusions:

• The novel federated learning framework offers a promising solution for securing sensitive data in VCPS.
• This approach is essential for maintaining the integrity and reliability of intelligent transportation systems as they advance.
• Implementing such sophisticated security measures will ultimately contribute to improved transportation safety and efficiency.