Softgreen: Towards energy management of green office buildings with soft sensors

This paper describes an approach for saving energy in commercial buildings, based on the information gathered from pre-existing opportunistic context sources. Most energy management systems rely on a heavy instrumentation strategy to infer occupancies, and unfortunately ignore already available opportunistic context sources, that can provide significant information about occupancy. We present models to conduct a Context Profiling with available context sources, to infer spatial occupancy measures. Further, we model electrical loads of several types to infer potential energy savings. Through a pilot study of a building with 5 users for 30 days, we identify intra-building areas where additional instrumentation of occupancy sensors is not necessary and demonstrate potential for significant reduction in energy consumption. We believe such Context Profiling can provide insights to significantly reduce deployment and management costs for future occupancy detection and energy management systems.