Illumination gain estimation and tracking in a distributed lighting control system

We consider a distributed lighting control system with multiple networked luminaires, each equipped with a light and presence sensor, and a local controller. Using local sensor inputs and information exchange with limited neighboring controllers, each local controller determines the optimum dimming level of its luminaire so that the lighting power consumption is minimized. This is done under net illumination constraints specified in terms of light sensor set-points. The underlying optimization problem requires knowledge of the illumination gains, i.e. the contributions of luminaire light output to the light sensors. Reflectance changes, e.g. due to occupancy changes or object movements in the environment can change the illumination gains over time, thus affecting the achieved illumination. We address the problem of illumination gain estimation and tracking in a lighting system with distributed asynchronous controllers. A Euclidean projection of the best linear unbiased estimator on the non-negative orthant is used as an initial illumination gain estimate and bounds on mean-squared error performance are obtained. A recursive calibration mechanism to track gain changes is proposed. Illumination performance improvements are shown by numerical results using photometric data.