🤖 AI Summary
To address single-point failures and network bottlenecks inherent in centralized federated learning (FL) for time-series anomaly detection in Industry 4.0, this paper proposes a decentralized/semi-decentralized FL (DFL/SDFL)-driven framework for industrial time-series anomaly detection. The method uniquely integrates DFL/SDFL with domain-specific time-series preprocessing—including stationarity transformation and domain-adaptive feature engineering—and implements a fully connected distributed training architecture built upon the Fedstellar framework. Evaluated on a chemical gas industry dataset, the approach achieves high detection accuracy while substantially reducing communication bandwidth (−42%), CPU usage (−38%), and memory consumption (−35%). It demonstrates strong robustness against node failures and exhibits low resource requirements—making it particularly suitable for resource-constrained industrial edge environments. This work establishes a novel paradigm for secure, trustworthy AI deployment in industrial settings.
📝 Abstract
Industry 4.0 has brought numerous advantages, such as increasing productivity through automation. However, it also presents major cybersecurity issues such as cyberattacks affecting industrial processes. Federated Learning (FL) combined with time-series analysis is a promising cyberattack detection mechanism proposed in the literature. However, the fact of having a single point of failure and network bottleneck are critical challenges that need to be tackled. Thus, this article explores the benefits of the Decentralized Federated Learning (DFL) in terms of cyberattack detection and resource consumption. The work presents TemporalFED, a software module for detecting anomalies in industrial environments using FL paradigms and time series. TemporalFED incorporates three components: Time Series Conversion, Feature Engineering, and Time Series Stationary Conversion. To evaluate TemporalFED, it was deployed on Fedstellar, a DFL framework. Then, a pool of experiments measured the detection performance and resource consumption in a chemical gas industrial environment with different time-series configurations, FL paradigms, and topologies. The results showcase the superiority of the configuration utilizing DFL and Semi-Decentralized Federated Learning (SDFL) paradigms, along with a fully connected topology, which achieved the best performance in anomaly detection. Regarding resource consumption, the configuration without feature engineering employed less bandwidth, CPU, and RAM than other configurations.
