Federated Learning for Industrial Internet of Things in Future Industries

TL;DR

This paper reviews how federated learning can enable privacy-preserving, scalable AI applications in industrial IoT, discusses key applications, and presents a case study demonstrating its feasibility in future industries.

Contribution

It provides a comprehensive overview of federated learning applications in industrial IoT and discusses open research challenges for future industrial implementations.

Findings

Federated learning is feasible for industrial IoT applications.

FL enhances data privacy and scalability in IIoT.

Open research topics include communication efficiency and security.

Abstract

The Industrial Internet of Things (IIoT) offers promising opportunities to transform the operation of industrial systems and becomes a key enabler for future industries. Recently, artificial intelligence (AI) has been widely utilized for realizing intelligent IIoT applications where AI techniques require centralized data collection and processing. However, this is not always feasible in realistic scenarios due to the high scalability of modern IIoT networks and growing industrial data confidentiality. Federated Learning (FL), as an emerging collaborative AI approach, is particularly attractive for intelligent IIoT networks by coordinating multiple IIoT devices and machines to perform AI training at the network edge while helping protect user privacy. In this article, we provide a detailed overview and discussions of the emerging applications of FL in key IIoT services and applications. A case study is also provided to demonstrate the feasibility of FL in IIoT. Finally, we highlight a range of interesting open research topics that need to be addressed for the full realization of FL-IIoT in industries.