Supervised detection of IoT botnet attacks

Abstract

Nowadays, more and more people start using IoT devices, which raise the threats of compromising these devices, since it's easily manipulated and hacked than desktop devices. This fact increased the number of cyberattacks that relay on IoT-based Botnet attacks. In this paper, we investigate the using of a supervised technique for detecting anomalies in IoT networks. The proposed model used a random forest classifier, the training data consider only 4 types of attacks while testing considers 10 types of attacks. The proposed model was effective in detection the new attacks and achieved 99% in terms of TPR, 100% in terms of TNR, and near-zero false alarms.