Machine Learning-Based Intrusion Detection for Internet of Things Network Traffic

Abstract

The ad-hoc wireless sensor networks comprise numerous spatially distributed sensor nodes whose prime functionality is monitoring the environment continuously. With rapidly growing number of constraints connected devices in IoT and WSN, the scope for attack also increases exponentially. Therefore, an effective intrusion detection system is needed to efficiently detect the attack at faster rate in highly scalable and dynamic IoT environment. A model for classifying UNSW-NB15 data set samples was developed using various machine learning techniques like random forest classifier, decision tree, Naive Bayes (NB) and feed-forward neural network. The classification models like random forest classifier, decision tree and Naive Bayes (NB) classifies data to normal or attack-based data. The outputs of trained classification models are then used to train a neural network to further classify the attack data to different attack categories.