Assessing Industrial Internet of Things Security at the Network Edge Using Trust-based Centralized and Federated Machine Learning
Open Access DepositedThe Internet of Things (IoT) in industrial and factory environments is rapidly growing toward enabling $3.3 trillion in economic value by 2030 (Chui et al., 2021). Increasingly, the deployment of Industrial Internet of Things (IIoT) is bringing information technology (IT) together with operational technology (OT) for the digitalization of industrial and safety-critical systems, ranging from factories to scientific laboratories to energy and public utility systems to transportation and national critical infrastructure. Many IoT devices provide little visibility into their services, configurations or operating states, which presents a security black box risk with poorly characterized behavior to the overall environment (CISA, 2020). Of deployed IoT devices, 57% are vulnerable to medium- or high-severity malware attacks (Wang & Lu, 2020). This praxis details research that improves an organization’s cyber resilience by providing a predictive trust model using federated learning at the network edge where IoT and IIoT devices are deployed to monitor and control the physical environment. Metadata commonly available in a network operating environment measure the trustworthiness of IoT/IIoT device behavior over time and form the attributes for an extensible machine learning model training and testing methodology. This research demonstrates that the methodology is effective as measured by F1-score and practical for detecting and classifying cyberattacks in a heterogenous IT/OT environment using conventional tree-based and boosted tree machine learning, multilayer perceptron and TabNet transformer deep learning, and horizontal federated learning architectures.
- All rights reserved
Notice to Authors
If you are the author of this work and you have any questions about the information on this page, please use the Contact form to get in touch with us.