IoT Malware Classification Through CNN and Naive Random Undersampling

Internet of Things (IoT) networks are rapidly expanding, increasing the vulnerability of these devices to sophisticated malware attacks. Effective and efficient malware detection mechanisms are crucial for maintaining the security of IoT ecosystems. While various methods have been proposed, accurate detection remains a significant challenge. This study presents a deep learning approach for malware detection in IoT networks using an advanced convolutional neural network (CNN) architecture. This study employs a three-step approach: First, preprocess the data using Naive Random Undersampling to address class imbalance across 34 malware categories. Second, apply Principal Component Analysis (PCA) to reduce dimensionality to 23 components. Finally, utilizing a sophisticated CNN model for classification. Trained using the Nadam optimizer, The proposed approach achieves results Accuracy 98%, precision of 99%, Recall of 99% and F1-Score of 99 %. With a false positive rate of 0.001 and a false negative rate of <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\mathbf{0. 0 1 0}$</tex>. These results demonstrate the efficacy of this approach in detecting diverse IoT malware with high precision.