Significance of Feature Selection on IoT-based Botnet Attacks Identification Using Machine Learning
Nongthombam Joychandra Singh, Nazrul Hoque, Kh. Robindro Singh, D.K. Bhattacharyya · 2024
The rapid spread of Internet of Things (IoT) devices has introduced various issues in ensuring network security. One major issue with the integrity and availability of IoT network systems is botnet attacks. In this context, effective botnet detection strategies are crucial for safeguarding IoT networks. Feature selection (FS) emerges as a fundamental process in network traffic analysis, with a significant impact on identifying botnet activities amidst the vast and complex data streams generated by IoT devices. This paper presents a comprehensive overview of FS methods and explores how these techniques affect the efficiency of detecting botnets in IoT networks through machine learning (ML) methods. The significance of FS lies in its ability to improve the efficacy of machine learning algorithms employed in botnet detection systems. Furthermore, we conduct experimental analysis using N-BaIoT, a standard IoT network traffic dataset, to assess how FS affects the efficiency of botnet detection models. Through comparative analysis, we demonstrate how different FS strategies influence the ML methods’ computing efficiency and accuracy of detection. Our findings emphasize FS’s critical importance in enhancing botnet detection systems’ effectiveness within IoT infrastructure.