What question did this study set out to answer?

This survey aims to provide a comprehensive overview of Federated Learning in IoT systems, highlighting privacy-preserving techniques and challenges.

June 3, 2026Open Access

A Survey on Federated Learning for Privacy-Preserving Artificial Intelligence in Internet of Things Systems

Key Points

This survey aims to provide a comprehensive overview of Federated Learning in IoT systems, highlighting privacy-preserving techniques and challenges.
Analyzed advances in Federated Learning from 2020–2025, particularly studies post-2022.
Reviewed communication-efficient techniques, aggregation algorithms, and security threats.
Compared critical performance metrics such as model accuracy and privacy guarantees.
Identified key challenges like non-IID data and device heterogeneity affecting performance.
Established the importance of differential privacy and secure multi-party computation in future FL developments.
Highlighted the relevance of personalized Federated Learning and Blockchain integration in real-world deployments.

Abstract

🔹 Overview The explosive growth of the Internet of Things (IoT) has led to billions of connected devices generating vast amounts of sensitive data. While traditional machine learning relies on centralized data collection, this approach raises significant privacy, security, and regulatory concerns. Federated Learning (FL) has emerged as a transformative solution by enabling collaborative model training across distributed devices without transferring raw data to a central server. This makes FL one of the most promising technologies for privacy-preserving artificial intelligence in IoT ecosystems. What This Survey Covers This survey provides a comprehensive review of Federated Learning for IoT systems, focusing on research developments from 2020–2025, with particular emphasis on advances published since 2022. Key topics include: Communication-efficient federated learning techniques Aggregation algorithms and optimization strategies Privacy-preserving mechanisms and differential privacy Security threats such as poisoning, backdoor, and inference attacks Defense mechanisms and robust aggregation methods Comparative analysis of major FL frameworks and approaches Challenges arising from non-IID data, device heterogeneity, and resource constraints Comparative Evaluation The surveyed methods are analyzed across critical performance dimensions: Communication overhead Model accuracy Convergence behavior Privacy guarantees Suitability for real-world IoT deployments Future Research Directions The survey also explores emerging trends, including: Differential Privacy-enhanced Federated Learning Secure Multi-Party Computation (SMPC) Personalized Federated Learning Blockchain-integrated FL systems Large Language Model (LLM)-assisted federation TinyML and edge intelligence integration Target Audience This work serves as a structured reference for: Researchers in Federated Learning and Distributed AI IoT Security and Privacy practitioners Graduate students and academics Industry professionals developing privacy-preserving intelligent systems Keywords: Federated Learning, Internet of Things (IoT), Privacy-Preserving AI, Edge Computing, Distributed Machine Learning, Differential Privacy, Secure Aggregation, IoT Security.

Read Full Paperexternally

Mark Helpful

Bookmark

Relay

View Full Paper