Energy-Aware Privacy-Preserving IoMT Framework for Consumer Electronics: A Federated Learning Approach to Remote Healthcare

The Internet of Medical Things (IoMT) relies on Internet of Things (IoT) devices to collect and share sensitive patient data, which raises significant privacy and security concerns. As IoMT systems become increasingly integrated into the Consumer Electronics (CE) and Consumer Technologies (CT) landscape, safeguarding patient privacy is critical. To address these challenges, this paper proposes a Two-Stage Federated Transfer Learning (FTL)-based IoMT framework tailored for CE/CT applications. The proposed framework enables IoT devices to train locally on sensitive data while sharing only encrypted model updates with a central server, ensuring secure aggregation and data confidentiality. Differential privacy (DP) is incorporated into the learning process to enhance privacy further, providing rigorous guarantees against data reconstruction attacks while maintaining model utility. To improve model convergence and mitigate accuracy degradation during FTL, this paper also introduces an Adaptive Learning Rate (ALR) mechanism that dynamically adjusts learning rates at different stages of training, optimizing performance across heterogeneous IoMT devices. Moreover, the framework employs a Cloud-Edge layer to address energy constraint issues to IoT devices to facilitate efficient model update exchanges, reducing communication overhead and energy consumption. The Edge node aggregates model updates securely before transmitting them to the central server using a Secure Aggregation (SA) protocol, ensuring privacy throughout the aggregation process. Experimental results demonstrate that the proposed framework achieves prediction accuracies of 94.65% and 98.25%, outperforming state-of-the-art approaches while maintaining strong privacy guarantees and reducing energy consumption.