Malicious Vehicle Detection Using Layer-Based Paradigm and the Internet of Things

Area of Science:

• Artificial Intelligence and Machine Learning
• Cybersecurity in Transportation
• Blockchain Technology Applications

Background:

• Deep learning models are vital across many fields but vulnerable to malicious attacks.
• Protecting AI models, especially in critical applications like vehicle detection, is essential.
• Existing methods may lack robust security against parameter manipulation in vehicle networks.

Purpose of the Study:

• To propose a novel four-layered paradigm, the consortium blockchain-enabled conventional neural network (CBCNN), for detecting malicious vehicles.
• To ensure tamper-proof protection against parameter manipulation attacks within the Internet of Vehicles (IoVs).
• To enhance the security and reliability of vehicle management systems through multi-label classification and blockchain integration.

Main Methods:

• Developed a four-layered CBCNN model comprising convolutional neural network (CNN)-IoT, spatial pyramid pooling, fully connected layers, and a consortium blockchain.
• Implemented a proof-of-luck mechanism within the consortium blockchain to optimize energy consumption for vehicles.
• Utilized C++ for implementation and the ns-3.34 platform with the ns3-ai module for simulation and anomaly detection in IoVs.

Main Results:

• The CBCNN approach demonstrated superior performance in malicious label detection compared to state-of-the-art methods.
• Achieved higher average accuracy and a reduced loss ratio in identifying malicious vehicles.
• Showcased significant cost reduction benefits while ensuring tamper-proof protection of the AI model.

Conclusions:

• The proposed CBCNN model effectively identifies malicious vehicles and mitigates associated risks in IoVs.
• Consortium blockchain integration provides robust, tamper-proof security for deep learning models in vehicular systems.
• The CBCNN approach offers an energy-efficient and cost-effective solution for securing the Internet of Vehicles.