Occupancy Prediction Algorithms for Thermostat Control Systems Using Mobile Devices

Several techniques have been proposed for the automatic control of just-in-time heating and cooling systems in indoor spaces that accommodate both the occupants' comfort and energy savings. Current techniques, however, do not provide an adequate solution for efficient thermostat control, because they require costly infrastructures to detect occupancy or because they inaccurately predict the occupancy due to irregular patterns. In this paper, we propose an automatic thermostat control system based on the mobility prediction of users, using contextual information obtained by mobile phones. We also present an arrival time prediction scheme that combines both historical pattern and route classification. The experimental results indicate that the proposed system can successfully predict at least 70% of the transit cases within 10 minutes' error and can decrease energy consumption by 26%.