Intrusion Detection and Analysis in IoT Devices Using Machine Learning Models

Abstract

This study proposes utilizing deep learning and machine learning techniques to identify network anomalies. The IoT-23 dataset serves as the basis for the analysis. The proposed approach models are designed to classify network flows as benign or assign them to one of the 11 labels in the dataset, as well as to differentiate between malicious and benign connections. Performance and time costs of various models are compared to determine the optimal algorithm for maximum performance in minimal time. This comparison identifies the better performing model with the least overhead cost for deployment on IoT devices, ensuring the security and privacy of users by blocking malicious connections. The experimental results show that decision tree offers maximum efficiency and the lowest overhead cost, making it suitable for use in IoT devices.