Machine learning-enhanced IoT network security: a Black Hole Algorithm-based feature selection approach for intrusion detection
Neeraj Kumar, Jyoti Prakash Singh, Prabhat Kumar · Journal of Cyber Security Technology · 2025
The proliferation of Internet of Things (IoT) devices has substantially increased the vulnerability to cyber threats, worsened by limited computational capabilities and weak security mechanisms. To address these challenges, this study proposes a machine learning-based Intrusion Detection System (IDS) that leverages an optimized feature selection mechanism using the Black Hole Algorithm (BHA). The approach introduces a novel star encoding scheme and a hybrid fitness function to enhance solution convergence and classification performance. The methodology is evaluated using the AWID3 dataset, wherein feature preprocessing and selection are followed by classification using the XGBoost model. Experimental results demonstrate significant improvements over existing techniques, achieving an accuracy of 99.63%, precision of 99.57%, recall of 99.73%, F1-score of 99.56%, a false positive rate of 0.49%, and significant improvements in FNR, TPR, and TNR. These results outperform benchmark IDS models, including HHGS-ROA, GA-GWO, and BGWO. Furthermore, statistical analysis of the simulation results confirms the findings and demonstrates significant improvements in performance. This work establishes an efficient and scalable IDS framework for IoT wireless networks by combining metaheuristic optimization with machine learning classification, thereby offering a practical solution for real-time threat detection in resource-constrained environments.