Modeling vehicle indoor air quality using sensor data analytics

Abstract

A working person on an average spends 1.5–2 h every day traveling either to their places of work or for other daily activities, using metros, trams, buses, and cars, as common modes of travel. Most of such commuters regularly suffer from health conditions like headache, breathless condition, drowsiness, etc. Numerous accidents have been reported due to drowsiness while driving, which may occur due to the build-up of carbon-dioxide (CO2) build in the vehicle chamber. This paper attempts to monitor, analyze, and predict air quality inside the vehicle. This work proposes a sensing system using an off-the-shelf sensor Sensordrone which is connected to an Android Smartphone using Bluetooth Low Energy. The data obtained from the proposed sensing system are then utilized to perform predictive analysis of CO2 build-up inside the vehicular chamber using Auto Regressive Integrated Moving Average (ARIMA) and Support Vector Regression (SVR). Root-Mean-Square Error for SVR and ARIMA models is 47.91 ppm and 55.32 ppm CO2, respectively, indicating that SVR outperformed ARIMA in predicting the CO2 build-up inside the vehicle.