A Big Data Security Framework for IoT Networks using Weighted Average Ensemble

Internet of Things technologies provide new challenges for network traffic security. As the types and volume of network attacks have increased in recent years, the concepts of artificial intelligence, imbalanced Data, and big data are not foreign in the field of network traffic classification. In this paper, we present a big data based security framework that solves the imbalanced classification problem in the Bot-IoT dataset. This framework combines two data sampling techniques to handle the problem of imbalanced traffic distribution. With our data sampling technique, the training dataset is divided into several resampled training subsets. In addition, the classification is carried out by employing Decision Tree algorithm in the Apache Spark environment. While the final predictions are made using a weighted average ensemble which can improve minority class classification performance. The experimental results exhibited high performance of our framework.