Using simple light sensors to achieve smart daylight harvesting

Lighting is the largest single energy consumer in commercial buildings. In this paper, we demonstrate how to improve the effectiveness of daylight harvesting with a single light sensor on each window. Our system automatically infers the window orientation and the cloudiness levels of the current sky to predict the incoming daylight and set window transparency accordingly. We evaluate our system with ten weeks of empirical data traces collected from windows around an office building and compare our approach with non-predictive feedback control. Experimental results show that our scheme can infer the orientation of a window to within ±7° of the actual orientation and improve energy savings by 10% over existing approaches without sacrificing user comfort.