Towards Resilient Federated Learning in CyberEdge Networks: Recent Advances and Future Trends

TL;DR

This survey addresses the resilience of federated learning in CyberEdge networks, focusing on non-IID data, unreliable devices, and sophisticated feature-oriented threats. It surveys two core approaches—joint training with agglomerative deduction and hierarchical aggregation, and defenses against poisoning, inference, and reconstruction attacks—alongside secure aggregation, anomaly detection, and privacy-preserving techniques. The paper also outlines future opportunities, including 6G integration, joint training with LLMs, and cross-domain/cross-silo ResFL with interoperable frameworks. The work highlights the practical significance of deploying ultra-low-latency, AI-driven, privacy-preserving learning for secure Metaverse-enabled applications across edge environments.

Abstract

In this survey, we investigate the most recent techniques of resilient federated learning (ResFL) in CyberEdge networks, focusing on joint training with agglomerative deduction and feature-oriented security mechanisms. We explore adaptive hierarchical learning strategies to tackle non-IID data challenges, improving scalability and reducing communication overhead. Fault tolerance techniques and agglomerative deduction mechanisms are studied to detect unreliable devices, refine model updates, and enhance convergence stability. Unlike existing FL security research, we comprehensively analyze feature-oriented threats, such as poisoning, inference, and reconstruction attacks that exploit model features. Moreover, we examine resilient aggregation techniques, anomaly detection, and cryptographic defenses, including differential privacy and secure multi-party computation, to strengthen FL security. In addition, we discuss the integration of 6G, large language models (LLMs), and interoperable learning frameworks to enhance privacy-preserving and decentralized cross-domain training. These advancements offer ultra-low latency, artificial intelligence (AI)-driven network management, and improved resilience against adversarial attacks, fostering the deployment of secure ResFL in CyberEdge networks.

Figures (8)

• Figure 1: FL in CyberEdge networks: The aggregation process
• Figure 2: Federated gradient scheduling for improving model convergence and stability in heterogeneous environments.
• Figure 3: Gradient-leak-resistant FL for detecting anomalous gradient patterns and preventing privacy breaches.
• Figure 4: Adversarial variational graph autoencoders for crafting malicious models from benign local updates alone, without direct access to training data.
• Figure 5: Detecting poisoning attacks on FL using GradCAM zheng2024detecting.