An Application of IoT and Machine Learning to Air Pollution Monitoring in Smart Cities

Abstract

Exhale and breathe with polluted air causes serious health implications. The effect of air pollution can be minimized by continuous monitoring and track a record of it. Also, timely prediction of pollutants level can help government agencies to take proactive measures to protect the environment. In this paper we have proposed the application of Internet of Things and Machine learning so that air pollution can be monitored within future smart cities. A high correlation between pollutants and weather parameters is determined by using Pearson correlation. In contrary to traditional sensor network, this work utilizes cloud-centric IoT middleware architecture that not only receives data from air pollution sensors but also from existing weather sensors. Thus provides two-fold reliability and reduce the cost substantially. The Artificial Neural Network has been used to predict the level of Sulfur Dioxide (SO2) and Particular Matter (PM2.5). Promising results suggest that ANN is a reliable candidate that can be used in air pollution monitoring and prediction system. Our models have achieved Root Mean Squared Error of 0.0128 and 0.0001 for SO2 and PM2.5, respectively.