Advanced Security for IoT Networks: A Novel Approach to Intrusion Detection and Feature Selection

The quick development of Internet of Things (IoT) has brought about concerns regarding the security of interconnected networks and devices. This requires the utilization of effective Intrusion Detection System (IDS) to mitigate cyber-attacks. The exiting methods struggle to capture the relevant features because they cannot capture the suitable features for classification. In this article, the Golden Strategy – Crystal Structure Algorithm (GS-CryStAl) and Gated Recurrent Unit (GRU) approach are developed to detect the intrusions in IoT. The GS-CryStAl-based feature selection algorithm is designed to choose the appropriate features from the whole feature subset by eliminating the irrelevant and inappropriate features. Subsequently, the selected features are classified by using the GRU classifier that effectively identifies intrusions with high accuracy. The proposed GS-CryStAl and GRU approach achieves 99.97% accuracy, 99.95% recall, 99.95% f1-score and 99.95% precision on the CICIoT 22 dataset, while on the CIC-IDS-2018 dataset, it achieves 99.21% accuracy, 98.69% recall, 98.71% f1-score and 99.95% precision. Hence, the proposed approach exhibits better performance compared to other conventional techniques.