Privacy-Preserving Human Activity Recognition in Smart Homes using Deep Learning and Edge Computing for Real-Time Processing

The rapid advancement of smart home technologies necessitates efficient human activity recognition (HAR) systems while ensuring user privacy. This research presents a novel architecture that integrates deep learning and edge computing for real-time HAR, emphasizing privacy protection. By employing differential privacy methods and federated learning, the approach minimizes personal data exposure risks. Using the UCI HAR dataset, the model achieves an accuracy of 96.50%, with precision, recall, and Fl-score metrics of 95.75%, 96.20%, and 95.97%, respectively. The results indicate high accuracy and adaptability to varying smart home environments, while edge computing ensures minimal latency. This work highlights the importance of balancing performance and privacy in HAR systems, paving the way for secure smart home applications. Additionally, it points to the potential of deep learning in medical diagnostics, suggesting future research could enhance diagnostic capabilities across various medical imaging modalities, including CT scans and MRIs, to improve healthcare outcomes.