IDS-based Machine Learning Model Specific for Secure Data Transmission within the Frame of IoT Network Infrastructure

Industry 4.0 represents a constant source of data originating directly from industrial plants. Data is transported via the communication network from the device that generates the data to the place of its permanent storage. In the process of transport, hacker attacks occur on the actors of communication – the sender, the recipient and the message itself. In the paper we propose a machine learning model aimed to analyze network traffic patterns and predict potential attacks, which is specialized for data generated within IoT devices network. For the development of the model, a publicly available dataset is used, which describes normal and problematic communication within the network of connected Internet of Things (IoT) devices. The developed and tested machine learning models are then compared and the algorithm with the best results is accepted as the final result. The comparative analysis of model performance is reflected in the comparison of the metrics of precision, accuracy, false positive rate, false negative rate, as well as the time required to execute different stages in the decision-making process. Summarizing all the results, the XGBoost model stood out as the model with the best performance.