Development of a federated learning-enabled IoT framework for indoor air quality and HVAC optimization in healthcare buildings

Abstract

Maintaining optimal indoor air quality (IAQ) in healthcare buildings is essential for occupant health, energy efficiency, and HVAC system performance. This paper presents a novel IoT-based air quality monitoring and ventilation control system powered by federated learning (FL) for real-time IAQ management. The system deploys multi-sensor IoT units to monitor PM2.5, PM10, CO₂, CH₂O, TVOC, temperature, and humidity in emergency rooms, doctors’ offices, and reception areas across three hospitals. A central hub dynamically adjusts HVAC settings based on real-time sensor data and predictive analytics, ensuring proactive air quality management. Addressable RGB indicators provide real-time IAQ displays and 30-min predictive warnings, enabling timely interventions. To enhance scalability, security, and computational efficiency, we introduce the Hierarchical Adaptive Federated Aggregation (HAFA) algorithm, which improves non-IID data processing and model accuracy in decentralized IAQ monitoring. HAFA achieves 90.8 % predictive accuracy (LSTM) and 88.0 % (CNN), outperforming conventional FL models. Additional performance metrics (R² = 0.87, RMSE = 0.09) validate its robustness. The system integrates LoRaWAN for low-power, long-range communication and HTTPS encryption for secure cloud-based data transmission. This paper demonstrates a scalable and intelligent IAQ control system for sustainable building management in healthcare facilities. By integrating IoT, federated learning, and HVAC optimization, it provides an energy-efficient, secure, and adaptive solution for indoor air pollution control in smart healthcare environments.