Building occupant level estimation based on heterogeneous information fusion

It is of great practical interest to estimate the number of occupants at a zonal level in buildings, which is useful for energy-efficient control of air conditioning and lighting systems. We consider this important problem in this paper. First, the occupant level estimation problem is formulated as an information fusion problem with heterogeneous information sources with the criterion of the minimum mean square error (MMSE). Two fusion methods are developed. The first method assumes independent observation noises and the second method exploits the correlation among the multiple information sources to improve the estimation accuracy. The experimental results show that in comparison with individual RFID or video cameras, the two fusion methods improve the accuracy of occupant level estimation by 43% and 73%, respectively, and outperform the linear least mean square error (LLMSE) method [3]. Simulations and theoretical analysis are also conducted to analyze the performance of the two methods under different occupant levels and different correlated observations. It is shown that the second method is more effective for the cases where the multi-sensor measurements are highly correlated.