IoT Architecture and Machine Learning-Based Model for Network Traffic Classification

IoT Architecture and Machine Learning-Based Model for Network Traffic Classification

연구 분야: Networking

학회: International Conference on Advances and Applications of Artificial Intelligence and Machine Learning

The Internet of Things (IoT) refers to a comprehensive miscellany of embedded devices linked to the Internet, allowing them to transfer and intercommunicate data in smart domains. The periodic observation of network traffic spawned from IoT devices is essential for their correct working and adversarial act detection. It stimulates the administrator to observe the actions of IoT devices that can be beneficial for the appropriate implementation of detecting adversarial IoT devices. Various researchers have proposed techniques for IoT traffic classification by utilizing many machine learning (ML) classifiers. However, the accuracy of ML classifiers is based on the data spawned from miscellaneous IoT devices and feature extraction from network traffic. In this research, we present IoT architecture and show its role in IoT traffic classification. We employed the public dataset IoT23 for our experiment. The important features are extracted from network traces. Next, we create a machine-learning model by employing various classifiers. Further, we conduct a proximate performance analysis of those ML models that are based on classification, precision, recall, F-measures, false-positive rate, true positive rate, and ROC parameters. We finally show the best ML model that will be useful for the future direction of researchers in this field.