Classifying Botnet Attack on Internet of Things Device Using Random Forest
Irfan Irfan, I M Wildani, Intan Nurma Yulita · IOP Conference Series Earth and Environmental Science · 2019
We live in Industry 4.0 where Internet of Things (IoT) is a new developing environment. A lot of researcher is trying to develop this new technology. As this technology is starting to become big, people try to attack the system of this technology. Luckily, a dataset contains of unattacked environment and attacked environment exist. The purpose of this research is to classify the incoming data in the IoT, contain a malware or not. In this research, we under sample the dataset because the datasets contain imbalance class. After that, we classify the sample using Random Forest. We use Naive Bayes, K-Nearest Neighbor and Decision Tree too as a comparison. The dataset that has been used in this research are from UCI Machine Learning Depository's Website. The dataset shows the data traffic from the IoT Device in a normal condition and attacked by Mirai or Bashlite. Random Forest gets greatest accuracy with 99.99% value with Precision, Recall, and F-Measure get 100% value. The score is followed by Decision Tree with 99.98% accuracy, KNN with 99.94% accuracy and Naive Bayes with 99.00% accuracy.