Malware Detection in Network Traffic Data for Internet of Things

The expansion of Internet of Things (IoT) places an immediate need for strong security measures, particularly in malware detection. This research was done by using IoT-23 dataset to implement machine learning techniques for malware detection in an IoT network traffic. Random forest and gradient descent were selected and applied to pre-processed data to evaluate its effectiveness in differentiating benign and malicious traffic. Feature reduction was performed using principal component analysis (PCA) and both models were evaluated and tested using cross-validation. Random forest have scored an accuracy of 99.78%. with perfect precision Recall and F1 scores for both classes While gradient boosting achieved 99.45% accuracy and slightly lower precision and recall for non-hazardous traffic. The results indicate that random forest is more suitable for dataset which shows higher accuracy and consistency. The study highlights the efficacy of ensemble learning in malware detection within an IoT environments and emphasizes the importance of proper pre-processing and feature reduction to improve model performance.