Wireless Sensor Network for Internet of Things Facility Management (IoT-FM) environment sensing

Abstract

This project presents an indoor air quality (IAQ) study to trend building underground space’s safe conditions using Wireless Sensor Network with Internet of Things (WSN-IoT). The improved WSN node houses 9 IAQ parameters namely, PM2.5, Temperature, Humidity, Carbon Monoxide, Methane, LPG, Smoke, Oxygen and Carbon Dioxide to monitor the indoor contaminants. The building underground location selected for data collection were Westgate (4days from 1pm to 5pm), Bedok Mall (12days from 12pm to 2pm), Tampines Mall (2days from 1230pm to 230pm) and Changi City Point (3days from 1pm to 3pm). For each run, 2 nodes were used at different locales. Polynomial regression and K-means clustering machine learning algorithms were used to model the surrounding air quality. Cross Validation Score Mean (CVSM) and Silhouette Coefficient was used to quantify the respective model’s goodness of fit, thereby characterizing monitored space’s safe condition. Temperatures achieve a better polynomial regression fitting and CVSM scores of 0.90. Also, PM2.5 had a better K-means clustering and silhouette coefficient of 0.627. These indicate that the parameters chosen are accurate in tending and well classified within its clusters. From observations, IAQ data is unique to its locale which suggests a building-wide coverage will be needed to monitor building underground spaces. The WSN-IoT solution prototyped in this work demonstrated the ability to continuously measure and model the building underground IAQ.