IoT-botnet attack detection and mitigation using ensembled-deep-learning-model

In this study, we propose an ensembled-deep learning model for robust IoT-botnet attack detection. The model comprises six phases: data preprocessing, data augmentation using SMOTE, feature extraction (central tendency, dispersion, and information gain), feature selection using the clan updated grasshopper optimisation algorithm (CUGOA), and attack detection. The detection model integrates DCNN, attention-based bi-LSTM, and optimised RNN, with fine-tuning using CUGOA. When an attack is detected, our new botnet traffic filter (BTF) mitigates it, enhancing network reliability. Our model outperforms existing approaches in terms of accuracy, sensitivity, specificity, and precision.