The quick expansion of Internet of Things (IoT) devices has presented new cybersecurity challenges, with botnet attacks posing noteworthy threats to network stability and data integrity. This research presents a hybrid deep learning model that combines Convolutional Neural Networks (CNN) also Long Short-Term Memory (LSTM) networks to detect botnet attacks. This proposed model is trained and evaluated on the publicly available BoT-IoT dataset, which includes multiple attack types and realistic traffic patterns. To improve performance, Feature Engineering (FE) is used to reduce noise, and SMOTE is applied to address class imbalance in this proposed work. Experimental results are calculated with dataset which is balanced and unbalanced. Experimental results of the proposed model achieve high accuracy 99.77%, PR-AUC of 100%, and a ROC-AUC of 99.99% on the balanced dataset (D2). Compared to traditional deep learning models, the proposed hybrid model demonstrates better feature representation and temporal pattern recognition, which significantly improves attack classification in IoT scenarios. Compared to traditional deep learning models, the proposed hybrid model demonstrates better feature representation and temporal pattern recognition, which significantly improves attack classification in IoT scenarios. The model also performs on imbalanced data, demonstrating its generalizability. The proposed CNN-LSTM hybrid outperforms classical and deep learning baselines, offering a scalable and interpretable approach.
Duraibi et al. (Tue,) studied this question.