Design analysis and prediction of an IoT based air pollution monitoring system

Abstract

Traditionally air samples are being collected using High Volume Samplers fixed at selected locations and the samples collected are tested at authorized laboratories using different techniques like Mass/Optical Spectroscopy, Gas Chromatography. The aforesaid equipment s are expensive and also involve lengthy procedures for determination of concentrations of air pollutants. The recent advent of nanotechnology based solid-state gas sensors are emerging as simple and compact instruments to measure concentration of pollutant levels with considerable ease. newlineThe objective of this research work is to design and develop a real time Air Pollution Monitoring System using gas sensor and to study the air pollution levels in order to suggest mitigation measures. Air Quality Monitoring Device will include five gas sensors MQ-2, DSM501A, ZE03-SO2, and NO2 2E N, Arduino UNO, LCD, Buzzer, GSM Module, WIFI, and LED Indicators. And for software part it includes Arduino IDE and Think Speak Cloud. newlineThe main sources of pollutant emissions in Chhattisgarh state are due to the industries, transportation (vehicle emissions), open-cast mining, and construction activities. On the basis of sources of pollutant in Chhattisgarh State we have targeted five cities: Raipur, Bilaspur, Durg-Bhilai, Korba, and raigarh. We have divided our work into two parts: Phase I and Phase II. newlineIn phase I, i.e., statistical analysis, we worked on the primary dataset, which was collected from the official site of the Chhattisgarh Environmental Conservation Board (CECB). In our statistical analysis, we further divided the data into three parts: before, during, and after the COVID period. First, we have shown the result city-wise; next, we have shown the result gas-wise. In addition, we have performed a statistical correlation-based feature selection to compute correlation and generated a correlation heatmap of the AQI. We have also shown the skewness present in the dataset. Additionally, we performed a statistical analysis of pollutants that included the Mean, Standa