Volume 5
Abstract: This paper presents a framework involving Federated Learning (FL) and Edge computing (EC) (FL-EC) to achieve privacy-preserving and real-time predictive maintenance in the Industrial Internet of Things (IIoT) systems. FL reduces the privacy risk by training machine learning models on local edge-devices without the transmission of sensitive data, whereas EC executes data nearer to the source decreasing latency and increasing responsiveness. The FL-EC framework demonstrates high accuracy, F1-score, and recall compared to traditional centralized models, establishing its utility for real-time fault detection. Scalability is also provided by the framework, making the system effective to support different and heterogeneous edge devices. FL-EC helps to solve the most important problems of traditional IIoT systems by maintaining data confidentiality and limiting latency. The research also contrasts FL-EC with the available literature and shows that it can be applied practically in the automotive and oil and gas industry as well as in industries where real-time monitoring and maintenance are essential. The results indicate that FL-EC has the potential to achieve significant cost-saving and efficiency improvement in industry. The future research must aim at improving federated learning algorithms, solving the non-IID data problem, and enabling the optimization of edge devices to be better performing and customized, which will guarantee a wider application in IIoT settings. Download this article: CPPJ - V5 N2 Page 104.pdf Recommended Citation: Vejendla, K., Chang, K., Aly, S., Maeng, B., (2026). Federated Learning and Edge Computing for Privacy-Preserving Real-Time Predictive Maintenance in Industrial IoT Systems. Cybersecurity Pedagogy and Practice Journal 5(2) pp 104-116. https://doi.org/10.62273/GPYS1669 | ||||||