Botnet attack detection in IoT using machine learning models

Botnet is of great concern when dealing with security of computer networks globally. Bots readily attack network infrastructure through their malicious activities. It is pertinent to mitigate and control the level of threat posit by Bot and thus the need for advanced technologies in predicting their occurrences. Machine learning offered a greater support in this regard with ability to handle voluminous data from IoT devices and the robustness in predicting the potential attack from trained data. Both supervised (DT and RF) and unsupervised learning (Deep Learning) were used to investigate prediction of attack. Results of various machine learning models were compared along the performance metrics (Confusion Matrix, ROC Curves). The outcome showed that supervised learning did better with the study dataset than the unsupervised model.