Special Section at CloudCom 2025
The proliferation of Internet of Things (IoT) devices and the increasing demand for privacy-aware intelligence at the edge have led to the widespread adoption of federated learning (FL) in resource-constrained environments. FL enables collaborative model training across decentralized clients—such as smart sensors, wearable devices, and autonomous vehicles—without transferring sensitive raw data to centralized servers. This paradigm holds strong potential for next-generation AI applications in healthcare, smart homes, transportation, and industrial automation. However, the intersection of FL and IoT introduces a complex and evolving landscape of security and privacy risks.
Unlike traditional cloud-centric architectures, IoT environments are inherently decentralized, heterogeneous, and bandwidth-limited. Devices participating in federated learning often vary widely in computational power, data quality, and connectivity, making them vulnerable to a wide range of attacks including inference attacks, poisoning, model inversion, and malicious aggregation. Furthermore, ensuring robust privacy protection—while preserving learning utility—remains a central challenge in real-world deployments.
This special session seeks to bring together researchers and practitioners to explore innovative solutions that enhance the security and privacy of federated learning in IoT ecosystems. We invite original research, system designs, and visionary perspectives that advance the theoretical understanding, practical implementation, and cross-domain integration of privacy-preserving techniques in federated IoT settings.
Topics of interest include, but are not limited to:
Authors are invited to submit original, unpublished research papers. Submissions should be in IEEE format and not exceed 6 pages. All submissions will be peer-reviewed. Accepted papers will be published in the conference proceedings and indexed by IEEE Xplore.