Knowledge Distillation in Federated Edge Learning: A Survey

TL;DR

This survey addresses the challenge of enabling efficient, personalized, and robust Federated Edge Learning (FEL) in resource-constrained edge environments by compiling and classifying the ways Knowledge Distillation (KD) can be integrated into FEL. It delineates four KD roles—knowledge transfer, model representation exchange, backbone algorithm components, and dataset distillation—and three deployment modes (edge, end, and edge-end collaboration). The work analyzes limitations and open problems, including device dropout, asynchronous training, incentives, privacy threats, and reliance on public data, and offers practical deployment guidance such as resource-aware architectures and dataset-free, privacy-preserving KD strategies. Overall, KD is presented as a flexible toolkit to compress models, personalize learning, and enable heterogeneous device participation in FEL, with clear implications for scalable and privacy-conscious edge intelligence.

Abstract

The increasing demand for intelligent services and privacy protection of mobile and Internet of Things (IoT) devices motivates the wide application of Federated Edge Learning (FEL), in which devices collaboratively train on-device Machine Learning (ML) models without sharing their private data. Limited by device hardware, diverse user behaviors and network infrastructure, the algorithm design of FEL faces challenges related to resources, personalization and network environments. Fortunately, Knowledge Distillation (KD) has been leveraged as an important technique to tackle the above challenges in FEL. In this paper, we investigate the works that KD applies to FEL, discuss the limitations and open problems of existing KD-based FEL approaches, and provide guidance for their real deployment.

Figures (1)

• Figure 1: General framework of knowledge distillation.