Attack Detection on Internet of Things Devices using Machine Learning Techniques

Today, the exponential growth of technology and the boom in connection demand has resulted in significant growth of Internet of Things (IoT) devices. In light of the enormous increase in smart devices, secure communication between them has become a major challenge due to an increase in cyberattacks. Adding hardware-based security is challenging due to the complex architecture of IoT devices and limited computation ability. Since IoT devices produce enormous amount of data, machine learning (ML) can be utilized to detect attacks on IoT devices. This study proposes some Machine Learning models to detect and prevent Botnet IoT attacks. The Network-based Detection of Bait-and-Switch IoT Attacks (N-BaIoT) dataset is used to train the models. The performance of models is analyzed using precision, recall and F1-score by carrying out multiclass classification. Experimental result shows that some of the proposed models are efficient in detecting botnet attacks with 99% accuracy. The model can also extend to use complete dataset and new ML techniques.