A Resilient Intrusion Detection System for IoT Environment Based on a Modified Stacking Ensemble Classifier

The proliferation of the Internet-of-Things (IoT) paradigm has brought about transformative changes in various real-life applications and revolutionized how we interact with technology. However, the exponential growth of global IoT implementations has also intensified cybersecurity concerns. Breaches in IoT compromise not only the associated technology but also the information it handles. Therefore, developing IoT intrusion detection systems to handle and tackle these security issues has become imperative. Intrusion detection systems (IDS) have gained popularity due to their real-time intrusion detection capabilities, evolving into signature-based and anomaly-based detection technologies over the decades. In this article, the authors propose an intrusion detection framework to enhance the security of the IoT environment. The proposed model uses the modified stacking ensemble classifier to detect anomalies in a real-time framework. Cuckoo search optimization is used to reduce data dimensionality. The performance of the developed framework is tested over several datasets, including KDD Cup 99, CSE-CIC-IDS2018, and CICIoT2023; it excels in detecting both known and evolving cyber-attack patterns with an accuracy rate of 99.87%, 98.89%, and 99.58% in binary classification tasks and 99.43%, 98.17%, and 98.86% in multiclass classification tasks. The results validate the performance of the proposed framework in both scenarios, as it outperformed several existing state-of-the-art methods over numerous evaluation metrics.