An Explainable Deep Learning Intrusion Detection in IoT-Enabled Transportation Networks

Cities are becoming sophisticated ecosystems that increase production and efficiency because to a notable increase in the Internet of Things (IoT). These technological advancements also increase scope of vulnerabilities to Cyber-attacks, making robust security measures a necessity.In this study, we explore IoT-enabled transportation networks’ need for intrusion detection systems. Adaptable and robust security solutions are essential as sophisticated cyber-attacks target critical infrastructure. The suggested framework applies deep learning and machine learning models to network intrusion detection through a thorough examination: Gradient Boosting Classifier which gave a 95.77% score, Gated Recurrent Unit (GRU) giving 96.54%, Autoencoder combined with k-Nearest Neighbors (kNN) Classifier gave 94.74%, and Feedforward Neural Network (FNN) with the highest score of 97.15%. But GRU excels with its rapid execution. This study focuses on evaluating and comparing these models. In addition, the framework can adapt to changing attack patterns and dynamic situations since it uses unsupervised learning techniques. This adaptability allows the framework to detect both known and unidentified attacks, ensuring its resistance to emerging dangers.