Vehicular Network Intrusion Detection Using a Cascaded Deep Learning Approach with Multi-Variant Metaheuristic

Area of Science:

• Cybersecurity
• Artificial Intelligence
• Intelligent Transport Systems

Background:

• Vehicular ad hoc networks (VANETs) are crucial for intelligent transport systems but face significant security challenges from attacks like Denial of Service (DoS) and Distributed Denial of Service (DDoS).
• Effective Network Intrusion Detection Systems (NIDS) are essential to mitigate these threats and ensure road safety.

Purpose of the Study:

• To develop an innovative Artificial Intelligence (AI)-based Network Intrusion Detection System (NIDS) specifically designed for the unique security demands of VANETs.
• To enhance the performance and accuracy of intrusion detection in VANETs by leveraging advanced Deep Learning techniques.

Main Methods:

• The proposed NIDS utilizes a combination of Deep Learning models: Cascaded Convolution Neural Network (CCNN) for high-level feature extraction and Self Attention-Based Bidirectional Long Short-Term Memory (SA-BiLSTM) for classification.
• The Multi-variant Gradient-Based Optimization (MV-GBO) algorithm is employed to optimize both CCNN and SA-BiLSTM models and for feature extraction, further improving detection capabilities.
• Model performance was rigorously evaluated on established datasets including KDD-CUP99, ToN-IoT, and VeReMi using the MATLAB platform.

Main Results:

• The AI-based NIDS demonstrated exceptional performance, achieving an accuracy rate of 99% across all evaluated datasets (KDD-CUP99, ToN-IoT, VeReMi).
• The integration of SA-BiLSTM and CCNN, optimized by MV-GBO, proved highly effective in identifying complex network intrusions within VANET environments.
• MV-GBO-based feature extraction significantly contributed to the enhanced learning and detection accuracy of the proposed model.

Conclusions:

• The developed AI-based NIDS, integrating CCNN, SA-BiLSTM, and MV-GBO, offers a robust and highly accurate solution for securing VANETs against sophisticated cyber threats.
• This research highlights the potential of advanced Deep Learning techniques in addressing critical security challenges in intelligent transportation systems, thereby improving overall road safety.
• The proposed model's 99% accuracy signifies a substantial advancement in intrusion detection capabilities for safety-critical vehicular networks.