Sample efficiency and sensor networks

The average, modern office building includes a complex combination of heterogeneous spaces that vary in dimensions, materials, placement, usage, ventilation and conditioning, and other factors that ultimately result in the existence of multiple environmental conditions within the same building. The target of building environmental monitoring is the detection of spaces and conditions under which occupant performance is likely to be impaired. In this paper we present the results of a study using carbon dioxide (CO2) measurements from a distributed sensor network to evaluate indoor air quality across 10 buildings and some 500 rooms. Carbon dioxide is commonly used as an indicator of adequate ventilation in occupied spaces. Evidence suggests humans face significant drops in their cognitive performance at levels commonly encountered in office buildings (Allen et al., 2016; Satish et al., 2012; Zhang et al., 2017). In this study, we explore the efficiency of CO2 sampling strategies across room types. We also explore the impact of room usage on CO2 levels. Findings show differences in CO2 distributions across rooms that are often missed by traditional approaches to commissioning buildings and building systems operations.