Cross ML for Io(H)T Network Traffic Classification: A New Approach Towards Standardization

The rapid proliferation of the Internet of Things (IoT) underscores the need for robust network security, especially in sectors like healthcare. While numerous datasets support cyber-attack detection in the IoT space, there remains a challenge due to the limited availability of publicly accessible data specific to certain sectors, like healthcare. This study delves into Cross-Machine Learning (Cross-ML), a novel approach to leveraging data from multiple sources to enhance Machine Learning (ML)-based Network Intrusion Detection Systems (NIDS). We introduce a well-defined framework for Cross-ML and demonstrate its application across three key datasets: ToN- IoT, X-IIoTID, and UNSWNB15. Preliminary results, when juxtaposed with analogous studies, reveal significant insights. The study also indicates that relying on a small number of features might be limiting for ML-based NIDS, suggesting an exploration of more comprehensive datasets, like those with NetFlow features, for more conclusive results. This research not only underscores the potential of CrossML in IoT network security but also emphasizes areas of further exploration to solidify its application.