Air Quality Prediction and Monitoring using Machine Learning Algorithm based IoT sensor- A researcher's perspective

Abstract-Air Pollution (AP) is one of the serious and major
environmental problem worldwide. Many researchers have drawn
attention and have focused about these problems keeping in mind human health. Air quality prediction information is one of the better ways through which people can be informed to be more vigilant about serious health issues and protect human health caused by air pollution. In many metropolitan cities air pollution is a major challenging environmental issue. To analyze the present traffic condition of the city, local authorities can be enabled by real time monitoring of pollution data which makes appropriate decisions. Hence an early system is required for monitoring and calculating the level of AP using Air Quality (AQ) which is essential for predicting exactly the pollutant concentrations. The prediction of AQ can be improved by deploying Internet of Things (IoT) based sensor which are considerably changing the prediction of AQ dynamically. The prediction of AP discussed and estimated using many existing techniques are very expensive and have very low accuracy. The technological advancement of Machine Learning (ML) algorithm can be very fast increasing and searching almost all fields and applications whereas AP prediction has not prohibited from those fields. This paper describes about various studies of ML algorithm relating to AP prediction and monitoring based on the IoT sensor data in the context of various cities. This paper also summaries real time and historical data based on the AQ prediction models tools and techniques and describes about recent research methodologies merits and demerits of AQ prediction, along with the challenges based on real time monitoring and prediction of AQ. effective use of accessible resources by providing good health,
energy and transport facilities to their citizens and for the benefit of the people. At various points inside the city there are various types of data collection sensors are installed that are developed for management of city resources as a source of information. The major and basic aim of enhancing the smart city are controlling the pollution, energy conservation, good traffic control, waste management, enhanced public security and safety. The urban areas have growing population rapidly in recent years because of movement of people and industrialization from rural to urban areas. Approximately the world’s population of around 54 to 66% will migrate to urban areas by 2050 according to the report of UN [3]. Hence
increasing of population by adding additional industries and
automobiles to cites so the energy, transportation demand and assurance also increased in urban areas. This will in turn becoming a major concern by rising of pollution emission for local and national jurisdiction on the global stage of leaders. The government of national and local authorities provides best style of living through controlling pollution for their population like minimizing health issue among people. In many of the metropolitan cites, managing AP is the major and basic issues. The prediction problem of AP can be evaluated using statistical linear models. These techniques have variation in time-series data and provide poor estimation due to the complexity for AP [4] [5]. In order to overcome these challenges faced by prediction of AP over the last 60 years, the numbers of ML techniques are used to develop and help to address the problems.