Framework of Air Pollution Assessment in Smart Cities using IoT with Machine Learning Approach

Abstract

Exhale and inhale filthy air has major health consequences. Air pollution's influence may be mitigated by conducting regular monitoring and keeping a record of it. Government organizations may also take proactive measures to protect the environment by accurately anticipating pollution levels in real time. In future smart cities, we propose using the Internet of Things and machine learning to track pollution levels in the air we breathe. The Pearson correlation test is performed to see whether pollutants and meteorological indicators have a high link. A cloud-centric IoT middleware architecture is used in this research instead of a standard sensor network to gather data from both air pollution and current weather sensors. This means that both reliability and cost have been greatly improved. Sulphur Dioxide (SO2) and Particulate Matter levels were predicted using an Artificial Neural Network (ANN) (PM2.5). The positive results show that ANNs may be used to monitor and forecast air pollution. RMSE values of 0.0128 and 0.0001 for SO2 and PM2.5 were found using our models.