Occupancy estimation for smart buildings by an auto-regressive hidden Markov model

One of the primary energy consumers in buildings are the Heating, Ventilation, and Air-Conditioning (HVAC) systems, which usually operate on a fixed schedule, i.e., running from early morning until late evening during the weekdays. This fixed operation schedule does not take the dynamics of occupancy level in the building into consideration, therefore may lead to waste of energy. An estimate of the number of occupants in the building with time can contribute to improving the control policy of the building's HVAC system by reducing energy consumption. In this paper, the auto-regressive hidden Markov model (ARHMM), is investigated to estimate the number of occupants in a research laboratory in a building using a wireless sensor network deployed. The network is composed of stand-alone sensing nodes with wireless data transmission capability, a base station that collects data from the sensing nodes, and a server to analyze the data from the base station. Experimental results and numerical simulation demonstrate that the ARHMM is more effective in estimating the number of occupants in the laboratory than the HMM algorithm, especially when the occupancy level fluctuates frequently.