Towards secure IoT-enabled transportation: an explainable AI and deep learning-based approach for efficient threat detection

Towards secure IoT-enabled transportation: an explainable AI and deep learning-based approach for efficient threat detection

연구 분야: Safety

학회: Cluster Computing

The advent of Internet of Things technologies in intelligent transportation systems (ITS) brings not only advancements but also heightened security vulnerabilities. This paper presents an Intrusion Detection System (IDS) that employs a hybrid Deep Learning based model i.e., Convolutional Neural Network-Gated Recurrent Unit (CNN-GRU), augmented with Explainable Artificial Intelligence (XAI) to enhance security in ITS. The CNN-GRU performs efficient threat detection with high accuracy and a low false alarm rate while the SHapley Additive exPlanations mechanism renders the decision-making process transparent and interpretable to promote trust. We utilize a publicly available state-of-the-art dataset (CICIoV2024), a rich and diverse source of real-world traffic data tailored for ITS, to train and test our model. The efficacy of our approach is demonstrated through extensive evaluation metrics, where our model shows a very good performance in detecting and explaining cyber threats and achieved an accuracy of 100% with a recall of 99.99% for binary and 99.64% accuracy and 98.48% recall for multiclass classification problems. The application of XAI not only improves the accuracy of threat detection but also instills trust among stakeholders by making the underlying processes of the IDS comprehensible and justifiable. This work marks a significant stride towards developing more secure, reliable, and accountable intelligent transportation ecosystems.