A misbehavior detection framework for cooperative intelligent transport systems.

Area of Science:

• Computer Science
• Network Security
• Intelligent Transportation Systems

Background:

• Vehicular networks require robust security due to the critical nature of human lives involved.
• Existing security measures in Cooperative Intelligent Transport Systems (C-ITS) need enhancement to counter malicious entities.
• The increasing complexity of Vehicle-to-Everything (V2X) communications necessitates advanced detection mechanisms.

Purpose of the Study:

• To introduce and detail a novel Misbehavior Detection Framework (MDF) for C-ITS.
• To evaluate the effectiveness of various Machine Learning (ML) models within the MDF for detecting malicious nodes.
• To identify the optimal ML model based on performance metrics and computational latency.

Main Methods:

• The proposed MDF employs consistency and local plausibility checks performed by Intelligent Transport System Stations.
• All Vehicle-to-Everything (V2X) messages are scrutinized through the MDF's detection mechanisms.
• Multiple ML-based models were evaluated using parameters including Recall, Precision, F1 Score, Accuracy, and others.

Main Results:

• The MDF effectively scrutinizes V2X messages for misbehavior using defined checks.
• Various ML models were compared based on their detection accuracy and computational efficiency.
• The study identified the best-performing ML model for the MDF, achieving optimal results across key performance indicators.

Conclusions:

• The developed Misbehavior Detection Framework significantly enhances security in vehicular networks.
• Machine learning offers a viable approach for real-time misbehavior detection in C-ITS.
• The chosen ML model provides a balance of high accuracy and low computational latency for practical deployment.