Indoor health monitoring with VLC-based passive posture monitoring

With the growing emphasis on maintaining wellness without disrupting daily routines, the demand for natural indoor health monitoring solutions has never been more pertinent. Traditional health monitoring methods, such as camera-based systems, wearable devices, millimeter-wave, and Wi-Fi technologies, face challenges like privacy concerns, inconvenient wearability, and susceptibility to interference due to crowded frequency bands. Addressing these issues, we designed an integrated system that combines communication, sensing, lighting, and health applications using visible light communication (VLC) technology, achieving non-intrusive and passive human indoor posture and activity monitoring. Firstly, a low-cost VLC device is deployed for systematic basic functionality. A low-powered chip for data acquisition and processing is also designed to analyze modulated light signals reflected off or obstructed by individuals. Furthermore, an optical human posture and activity recognition model is developed to analyze VLC signals captured by receivers. The system employs a streamlined algorithm for signal analysis, incorporating preprocessing steps like filtering, normalization, and downsampling to enhance data quality. Three machine learning models are used for classification: Random forest, decision tree, and support vector machines (SVM), emphasizing prediction accuracy and computational efficiency in real-time monitoring. The monitoring signals and analyzing results are transmitted wirelessly to a user-friendly app on a smartphone. The experimental results demonstrate that our system achieves an accuracy rate exceeding 90% in the detection and monitoring of human postures and activities. The results suggest that our VLC-based system can achieve real-time, affordable, non-intrusive, scalable, and effective indoor posture and activity monitoring. This system has diverse applications for health monitoring in smart homes and healthcare facilities.