Home Heating Using GPS-Based Arrival Prediction

Home heating is a major factor in worldwide energy use. We describe two experiments aimed at reducing the amount of time heating systems need to be on, without compromising occupants’ comfort. The first resulted in a machine learning algorithm based on GPS data to predict when an occupant will arrive at home. The second examined how long it takes to heat homes based on temperature measurements, telling us how far in advance arrival predictions are needed. Our findings suggest that GPS-based prediction has the potential to reduce home energy consumption compared to existing methods.

To appear in the Pervasive 2010 Workshop on Energy Awareness and Conservation.