Air Pollution Prediction Using Dual Graph Convolution LSTM Technique

In current scenario, Wireless Sensor Networks (WSNs) has been applied on variety of applications such as targets tracking, natural resources investigation, monitoring on unapproachable place and so on. Through the sensor nodes, the information for the applications is gathered and transferred. The physical coordination of these sensor nodes is determined, and it is called as localization. The WSN localization methods are studied widely for recent research with the study of small proportion of the sensor node called anchor nodes and their positions are determined through the GPS devices. Sometimes sensor nodes can be a IoT device in the network. With despite this, among the various applications, air pollution and air quality monitoring having many issues on how to place the sensor network in a wide area to monitor the air pollutants level such as carbon dioxide (CO2), nitrogen dioxides (NO2), particulate matter (PM), sulphur dioxide (SO2), ammonia (NH3) and other toxic gases involved in human and industrial activities. The responsibility of the WSN in air quality monitoring is to be positioning the sensor nodes in the large area with low cost and also gather the real time data and produce the monitoring system as an accurate one. In this proposed work, deep learning-based approach called dual graph convolution and LSTM (Long Short-Term Memory) network based (air quality index) AQI predictions were performed. This uses the infrared based technology to measure the CO2, temperature and humidity, Geo statistic method and low power wireless networking. Accuracy of the proposed system is maximum of 95% which is higher than existing techniques.