Modeling indoor PM2.5 using Adaptive Dynamic Fuzzy Inference System Tree (ADFIST) on Internet of Things-based sensor network data

Abstract

Indoor air pollution is crucial for the building environment as it has a direct impact on human health, prosperity, comfort, and productivity levels. Repeated exposure to toxic materials and harmful chemicals leaves a considerable impact on occupant well-being. The degraded air quality in the indoor environment is closely connected to the rising cases of respiratory health issues, cardiovascular disease, lung cancer, and a wide spectrum of symptoms. Therefore, it is required to design a reliable system to assess indoor air quality (IAQ) in residential and commercial spaces to improve the health outcomes and productivity levels of the building occupants. The main contribution of this paper is to provide a rigorous design and development approach for the IAQ monitoring system using the Internet of Things-based sensor network. Furthermore, a forecasting model was designed using Adaptive Dynamic Fuzzy Inference System Tree to predict PM2.5 conditions in the target building environment. The performance of the proposed system is tested over multiple datasets obtained from different rural and urban locations; moreover, the findings are further validated against benchmark datasets. The updates about real-time IAQ conditions in the target building environment were made accessible to occupants via an online web portal, and the entire system was named “Vayuveda” – an approach for enhanced air quality in the living spaces. The authors tested multiple methods at every step of the development process and compared performance at each stage to come out with the best model for real-time IAQ assessment.