An Intelligent Wallpaper Based on Ambient Light for Human Activity Sensing

Using visible light for indoor human sensing has received a great deal of attention. Existing researches on visible light sensing have two limitations: (a) relying on a specific light (b) small sensing range. In this paper, using the light reflection model, we propose a human activity sensing system based on ambient light named Sensing-Wallpaper (SenWp). It could realize whole room human sensing using Photo-Diode (PD) hidden in the wallpaper without offline training or specific light. In the SenWp system, human activity sensing model is proposed to capture human activity semantic information and enhance signal characteristics. We have conducted a large number of experiments in three typical indoor environments. The accuracy of human activity sensing reaches 96%. Moreover, in the absence of artificial light, just using natural light, the activity sensing range can reach 6m. We also have conducted long-term research in real life to prove the potential of the system in practice.