Detecting IoT Botnets with Advanced Machine Learning Techniques
Anshika Sharma, Himanshi Babbar, Amit Kumar Vats, Bura Vijay Kumar · 2024
The widespread use of Internet of Things (IoT) devices has resulted in heightened susceptibilities and the rise of intricate botnet attacks, hence presenting noteworthy security obstacles. This work uses the IoT-POT dataset, a comprehensive source of IoT network traffic data, to investigate the effectiveness of machine learning (ML) techniques, namely Support Vector Machine (SVM), Gradient Boosting Machine (GBM), Extreme Gradient Boosting (XGBoost), and Light Gradient Boosting Machine (LightGBM), in detecting botnet attacks. To train and assess the previously described models using these features, the dataset must first be preprocessed to extract pertinent features. Performance metrics including accuracy, precision, recall, and F1-score are used to evaluate and compare the models' efficacy. Early findings demonstrate the potential of ensemble techniques like RF, XGBoost, and LightGBM for real-time detection of breaches in IoT networks since they beat conventional SVM for recognising botnet activity. In addition to offering insights into upcoming advancements in botnet detection techniques, this comparative analysis highlights the significance of sophisticated ML techniques in bolstering IoT security.