On the Tradeoffs of On-Device Generative Models in Federated Predictive Maintenance Systems

TL;DR

This paper analyzes the performance and communication trade-offs of federated generative models like VAEs, GANs, and DMs for predictive maintenance in IoT, proposing a taxonomy for partial model sharing.

Contribution

It introduces a novel taxonomy for federated generative models emphasizing partial component sharing for personalization and evaluates their trade-offs in real-world scenarios.

Findings

GANs with full federation improve training stability but are less robust than VAEs and DMs.

Partial federation with decoder sharing can outperform full federation in bandwidth-limited, non-IID settings for DMs.

Distinct trade-offs exist in utility, stability, and scalability among the models under different federation setups.

Abstract

Federated Learning (FL) has emerged as a promising paradigm for preserving client data ownership and control over distributed Internet of Things (IoT) environments. While discriminative models dominate most FL use cases, recent advances in generative models -- such as Variational Autoencoders (VAE), Generative Adversarial Networks (GAN), and Diffusion Models (DM) -- offer new opportunities for unsupervised anomaly detection in time series analysis, with relevant applications in predictive maintenance (PdM) in critical industrial infrastructures. In this work, we present a comprehensive analysis of VAEs, GANs, and DMs in the context of federated PdM. We analyze their performance and communication overhead under both full and partial federation setups, where only subsets of model components are shared. Building on this analysis, the paper proposes a novel taxonomy for federated generative models that formalizes partial component sharing as a principled mechanism for model personalization. Our experiments over a real-world time series dataset reveal distinct trade-offs in model utility, stability, and scalability, especially in heterogeneous and bandwidth-constrained FL settings. For the evaluated GAN-based configurations, full federation improves training stability relative to independent local training, although the model remains less robust than the VAE- and DDPM-based alternatives. For DMs, however, partial federation -- especially decoder sharing -- can outperform full federation in bandwidth-constrained, non-IID settings.