Fusion of Nonintrusive Environmental Sensors for Occupancy Detection in Smart Homes

We presented an approach to detect and count occupants using a fusion of environmental sensors from an indoor air quality measurement system. Environmental sensors, as opposed to motion detectors, are nonintrusive, easy to install, low cost, detect nonmoving occupants, do not have dead spots, and can even infer the number of occupants. For this paper, we conducted measurements of carbon dioxide, volatile organic compounds (VOCs), air temperature, and air relative humidity in four student apartments for a total of 49 days. We extracted features from the environmental sensors and selected subsets using correlation-based feature selection. Subsequently, we performed a comparison of the supervised learning models repeated incremental pruning to produce error reduction, naïve Bayes (NB), C4.5 decision tree, logistic regression, k-nearest neighbors, and random forest. We further proposed a method to greatly reduce time and effort of collecting training data in residential buildings. The results indicated that the predictive power of VOC sensing is comparable to that of carbon dioxide. With a simple NB classifier, our approach detected occupancy and estimated the number of occupants with an accuracy of 81.1 % and 64.7 %, respectively.