5G Narrow Band-IoT Based Air Contamination Prediction Using Recurrent Neural Network

Abstract

Air pollution plays a vital role among all other pollution; this is because it seriously has dreadful impact over the environment and the health of every individuals. Fine Particulate Matter (PM) 2.5 is the most significant ecological jeopardy feature contributing to global Cardio Vascular (CV) transience and respiratory syndrome. In India, the majority metropolitan spot has a deprived air quality with PM 2.5 levels exceeding the suggested limits. Apparently 60 % of Indian metropolitan areas has PM 2.5 levels greater than the index suggested by National Ambient Air Quality Standards. Hence, it is important to examine the level of PM 2.5 in air. In this paper, PM 2.5 level is monitored at various time stamp through Deep Learning (DL) technique called Long Short Term Memory (LSTM) which is a modification of Recurrent Neural Network (RNN). The proposed model will assist in alerting the dangerous level of PM 2.5 and taking action against it. Here, we use a 5 G protocol named 5 G Narrow Bands-Internet of Things (NB-IoT) communication protocol to alert officials quickly about the dangerous level of PM 2.5. This proposed techniques also helps to monitor the levels of various gaseous such as Carbon Monoxide (CO), Sulphur di Oxide (SO2), Oxide (O3), Hydrogen di Sulphide (H2S), Nitrogen di Oxide (NO2) along with Particulate Matter (PM 2.5). The proposed technology develops a cost effective and accurate stack based air pollution module that will provide a real-time, location based air pollutant statistics. The data is considered in division of one hour with different air quality sensors. The performance is measured using the loss function and the accuracy of predicting the PM 2.5 level.