Iot based air quality index monitoring system using

ESP32
Mr. Karthikeyan S Sethu K Saran kumar N V
Dept. of Electrical And Communication Dept. of Electrical And Electronics Dept. of Electrical And Electronics
Engineering Engineering Engineering
Bannari Amman Institute of Technology Bannari Amman Institute of Technology Bannari Amman Institute of Technology
Sathy,Erode Sathy,Erode Sathy,Erode

spores. Emissions of methane and carbon monoxide are

factors in global warming. The Air Quality Index (AQI)
Abstract—It is predicted that the massive Internet of Things measures important pollutants such carbon monoxide
(IoT) would bring a plethora of applications for a completely (CO),
connected world. The deteriorating state of the air in many
biosphere zones can be attributed to human challenges such
as growing globalization and urbanization. Many cities have
air pollution levels that are higher than permitted by law nitrogen dioxide (NO2), sulfur dioxide (SO2), ozone (O3),
and the World Health Organization (WHO) for gaseous and particulate matter (PM2.5 and PM10), and sulfur dioxide
particle pollutants, which are present in amounts that are (PM10) to estimate ambient air quality. Lung cancer,
harmful to human health. Frequent exposure to high respiratory disorders, and cardiovascular illness have all
pollution intensities raises the death rate and increases the been related to PM2.5 and PM10 particles, which come
number of people suffering from respiratory conditions mostly from combustion and transportation. Burning fuel
including asthma and chronic obstructive lung disease. A
releases NO2, which can lead to asthma attacks and
parameter can be used to measure the air quality called the
Index of Air Quality. Particulate matter (PM2.5 and PM10),
bronchitis. Ozone, which is created when pollutants react
ozone (O3), nitrogen dioxide (NO2), sulfur dioxide (SO2), in the presence of sunshine, aggravates heart and
and carbon monoxide (CO) are the main pollutants used to respiratory conditions, triggers asthma attacks, and
construct the air quality index. In this article, the irritates the eyes. The average concentration of pollutants
deployment of a cloud-based Internet of Things (IoT) system recorded during predetermined time periods—24 hours for
for air quality monitoring is discussed. The sensors are most pollutants, 8 hours for CO and O3—is used to
utilized to determine the levels of CO, PM2.5 and PM10, O3, calculate the AQI.
SO2, and NOx pollution together with ambient parameters
like humidity and temperature. In this research, an Internet
of Things (IoT) air quality index (AQI) monitoring system
built on the ESP32 microcontroller platform is presented.
The system gathers real-time data on different air pollutants,
including particulate matter, nitrous oxide, carbon
monoxide, ozone, and sulfur dioxide, using a network of
sensors. In order to precisely anticipate air quality levels, the
gathered sensor data is processed and analyzed using deep
learning techniques, such as convolutional neural networks
(CNNs) and recurrent neural networks (RNNs). The ESP32
functions as a sturdy platform for data capture and
preprocessing, making it easier to gather and process sensor
data initially before incorporating it into deep learning
models. The accuracy of air quality predictions is increased The Air Quality Index (AQI) has five categories that
by the system's ability to extract spatial and temporal represent different pollution levels, ranging from 0 to 500.
patterns from the sensor data through the incorporation of While a satisfactory AQI (51–100) indicates that sensitive
deep learning. people should limit their outside activity, a good AQI (0–
50) indicates clean air. Sensitive populations, such as
youngsters, the elderly, and individuals with cardiac
Keywords—WHO,ESP32,air quality
problems, should avoid moderate pollution (101-200).
Older adults, children, and those with respiratory
I. INTRODUCTION conditions should avoid outdoor exertion when the air
quality index is poor (201-300). Everyone is advised to
avoid outside physical activity when the air quality index
A quarter of India's population is at significant risk of is very bad to severe (301–500), especially youngsters and
serious health problems due to air pollution, which is people with heart and lung conditions. The varied levels
brought on by a combination of solid particles and gases of pollution make it difficult to estimate the AQI when
released into the atmosphere by sources such as industry traveling across various cities. An IoT-enabled
chemicals, automobile emissions, dust, pollen, and mold environmental air pollution monitoring and rerouting

© 2024, IRJEdT Volume: 06 Issue: 04 |April-2024 Page 70

syst qual
em has been presented as a solution to this problem. ity monitoring system, utilizing ESP32 to enable data
transmission and sensor integration. When taken as a
whole, these observations provide insightful advances for
ESP32-based solutions in environmental monitoring
applications.[7],[8]

This study explores the details of an Internet of

Things (IoT)-based air pollution monitoring system, with
a focus on sensor curation, data processing, and
II. LITERATURE SURVEY visualization techniques. It clarifies the crucial part that
ESP32 plays in enabling these procedures—from sensor
An in-depth analysis of an Internet of Things selection to data transfer and visualization—through a
(IoT)-based air quality monitoring system that uses an thorough investigation. The solution guarantees smooth
ESP32. It clarifies the system architecture, sensor fusion, data management and accessibility by utilizing cloud
and data handling procedures of the system. It explores services, allowing for real-time monitoring and analysis.
the finer points of connecting sensors, analyzing data The study also emphasizes how important it is to choose
collected, and integrating them all through a thorough sensors wisely in order to accurately capture relevant
design and execution methodology. The insights offered environmental data. [9] ESP32 contributes to well-
highlight ESP32's critical role in enabling reliable air informed decision-making by converting unprocessed
quality monitoring solutions and shed light on its smooth sensor information into useful insights using sophisticated
integration inside the IoT framework.[1][2] data processing techniques. Integrating cloud services
improves accessibility and scalability while providing
An ESP32-based, machine learning-based, stakeholders with thorough information on air quality. All
reasonably priced air quality monitoring system was then things considered, this study opens the door for
developed. The study demonstrates the feasibility of improvements in environmental monitoring techniques by
utilizing IoT devices to attain accurate assessments of air offering insightful information about the planning and
quality by presenting this integration. By employing execution of ESP32-based IoT systems for efficient air
ESP32 in conjunction with machine learning pollution monitoring.[10]
methodologies, the system exhibits its capacity to furnish
precise evaluations of air quality metrics. This study
opens the door for a wider use of IoT solutions in
environmental monitoring projects by highlighting their
potential to make a substantial contribution to the field of III. METHODOLOGY
air quality assessment at a reasonable cost.[3]
Roadside air pollution is getting worse every day,
Then an ESP32-based IoT-based air pollution
which makes traveling uncomfortable for passengers and
monitoring system was done. It highlights how important
may even pose health risks depending on their individual
it is to integrate cloud services, use data transmission
circumstances. A low-cost wireless air pollution
techniques, and curate sensors. Focusing on these
monitoring system was created in this work. Particulate
components, the study clarifies the complexities
matter (PM2.5 and PM10), ozone (O3), nitrogen dioxide
associated in designing a strong monitoring system. The
(NO2), sulfur dioxide (SO2), and carbon monoxide (CO)
system's capacity to reliably acquire pertinent
emissions are measured to determine the quality of the air.
environmental data is ensured by the attention to sensor
Particulate matter is the phrase used to describe the
selection. Furthermore, information about cloud
various combinations of microscopic particles, smaller
integration and data transfer methods emphasizes how the
than 10µm in diameter, that pollute the air. The reason
system can provide remote access and real-time
this is so concerning is that the particles having a diameter
monitoring. All things considered, the article offers
of
insightful information about how to use ESP32-based IoT
solutions for efficient air pollution monitoring.[4]
A human being can inhale particles smaller than 10 µm,
which penetrates the lungs and can have major health
This paper presents a case study that was carried
implications. An optical dust sensor module, which uses
out in Portugal and demonstrates how edge computing and
light transmission to pass through airborne dust particles,
the Internet of Things are being used for air quality
can quantify this. Light is sent through the air in this
monitoring. It highlights the usefulness of ESP32-based
sensor, where it is reflected, refracted, and scattered,
monitoring systems by showing how they are
biting dust particles. It is possible to quantify the number
implemented in real-world scenarios.[6] The study also
of particles and determine their size based on the
describes the architecture and implementation of an
scattering property measure, reflective index, and light
Internet of Things-based air quality monitoring system,
absorption. Colorless and extremely unpleasant, ozone is
emphasizing the critical function of ESP32 in data
regarded as a secondary contaminant that is created when
collection, processing, and transmission. Additionally, it
pollutants released by automobiles, power plants,
suggests a blueprint for an Internet of Things-based air

71
© 2024, IRJEdT Volume: 06 Issue: 04 |April-2024
indu IV. D
strial boilers, and other sources chemically react with ATASET PREPARATION
sunshine. The MQ131 Ozone Sensor is perfect for
environmental and scientific research experiments, as well The LTE network is used by narrowband IoT
as for monitoring air quality. It can detect and quantify (NB-IoT) to transport data utilizing SCFDMA and
Ozone (03) Concentration between 10 ppb and 2 ppm. OFDMA waveforms. Because NB-IoT data is
Nitrogen oxides are combined with other airborne unstructured, analysis is challenging. The MQTT protocol
particles from fuel combustion and vehicle emissions is supported by WISE-4000 sensor nodes, allowing for
from buses, cars, and power plants. Breathing in nitrous easy setup and seamless cloud service interaction. Data
oxide at high concentrations can lead to respiratory from NB-IoT devices is provided to particular public
illnesses, which can aggravate asthma and produce cloud services for the purpose of monitoring air pollution.
wheezing. One dangerous gas is carbon monoxide which An estimation of the Air Quality Index (AQI) is made
is created when fuels like charcoal, propane, wood, and using information from multiple sensors. The AQI is a
gasoline are burned. Regularly breathing in carbon numerical scale that goes from 0 (good air) to 500
monoxide can lead to a number of health problems, and if (hazardous air). Based on pollutant concentration over
exposure to higher concentrations occurs, death may predetermined time periods, it is computed. The intention
result. The MQ7 is a reasonably priced sensor that is to swiftly distribute air quality data so that prompt
measures the amount of carbon monoxide in the air. action can be taken. Utilizing breakpoint concentrations
Sulfur dioxide is discharged into the atmosphere by power and matching AQI readings, the linear segmentation
plants burning coal and oil. Large ships, railroads, and approach is used in the AQI computation process. The Air
some diesel machinery that consumes fuel with a high Quality Index (AQI) framework reports pollutants that
sulfur content can also emit this. Humans' respiratory exceed regulations and their health effects. It also includes
systems may be impacted by this type of air pollution, internet dissemination and a website for national AQI
which can also irritate their eyes. The amount of gas display.
combined with air can be sent using 110-602 sulfur
dioxide electrochemical sensors, which also offer simple
integration into wireless solutions.

Start by connecting the ESP32 microcontroller to a variety

of environmental sensors. Particulate matter sensors, gas
sensors for pollutants such as sulfur dioxide, nitrogen
dioxide, carbon monoxide, and ozone, and temperature
and humidity sensors for environmental context are a few
examples of these sensors. To get raw sensor data from
the embedded sensors, use the ESP32. Analog to digital
converter (ADC) on the ESP32 makes it easier to convert
analog sensor readings into digital data that the
microcontroller can use.

V. PREPROCESSING

First, a variety of environmental sensors

connected with the ESP32 microcontroller provide raw
sensor data. These sensors gauge the quantities of sulfur
dioxide, nitrous oxide, carbon monoxide, particulate
matter, and ozone in the atmosphere. A number of crucial
processes are included in the preprocessing pipeline to
improve the raw sensor data. First, the dataset is cleaned
using techniques to eliminate noise, outliers, and
inaccurate readings. This guarantees that only data of the
highest caliber is saved for later examination.
Subsequently, the sensor data is combined and
synchronized to generate an extensive dataset that makes
it easier to comprehend the state of the air quality overall.
Then, transformation procedures are used to transform

72
© 2024, IRJEdT Volume: 06 Issue: 04 |April-2024
unp whe
rocessed sensor data into measurements that have re sensor data is collected from many sites. With this
meaning, such the Air Quality Index (AQI). The data is feature, the system may identify intricate links in the data
normalized across several sensors through the use of that conventional methods can miss. Without the need for
normalization techniques, allowing for uniform analysis manual feature engineering, CNNs can automatically
and fair comparisons. Furthermore, feature engineering extract pertinent features from raw sensor data. This is
techniques are used to extract pertinent characteristics especially useful in air quality monitoring systems
from the dataset, potentially improving the efficiency of because of the sometimes complicated and nonlinear
later analysis algorithms. Robust validation methods and interactions that exist between environmental parameters
error handling approaches are employed to properly and contaminants. CNNs can withstand noise in the data
address challenges such as environmental unpredictability, and spatial variations. Environmental factors might
data transmission faults, and sensor imperfections. fluctuate significantly between sites and times during air
The preprocessing phase greatly improves the quality monitoring, which makes data processing difficult.
gathered data's quality and usefulness, which raises the Because of their great efficiency and scalability, NNs can
system's overall efficacy in monitoring air quality. be implemented on devices with limited resources, like
Preprocessing makes data clean, consistent, and useful so ESP32 microcontrollers.Real-time data processing
that timely decisions to reduce environmental risks and capabilities of NNs allow for prompt detection and
safeguard public health may be made and reliable reaction to variations in air quality. This feature is crucial
assessments of air quality can be made. for situations where quick observation and action are
needed to reduce pollution and safeguard public
VI. DEEPLEARNIG TECHNIQUES health.The incorporation of CNNs into the ESP32-based
Internet of Things air quality monitoring system improves
Deep learning approaches provide a potent way its capacity to handle variability, learn from unprocessed
to improve data analysis and prediction skills in an sensor data, analyze spatial patterns, and conduct real-
ESP32-based IoT-based air quality index monitoring time analysis—all of which contribute to the air quality
system. Convolutional neural networks (CNNs) and monitoring system's increased efficacy and accuracy.
recurrent neural networks (RNNs) are two examples of
deep learning models that can be included into the system 2.VGG19
to interpret the complex and multidimensional sensor data
that ESP32 collects. Although ESP32 is primarily concerned with
CNNs are particularly good at removing spatial sensor data, there may be situations in which image
information from sensor data, which enables the system to processing is required, such as when examining photos of
identify complex relationships and patterns between pollution sources or using visual cues to track the quality
various environmental factors and contaminants. These of the air. The deep convolutional neural network
models can improve prediction accuracy by reducing the architecture VGG19 is well-known for its efficiency in
need for manual feature engineering by learning pertinent object recognition and picture categorization
features on their own, such as pollution concentrations applications.VGG19 can be used to process images taken
and their spatial distributions. by cameras that are part of the surveillance system. When
paired with sensor data, these photos can offer additional
Because RNNs are good with sequential data, they are a details about the surrounding environment or the sources
good choice for assessing time-series data on air quality of pollution, which enhances our knowledge of the
that is gathered over an extended period of time. RNNs dynamics of air quality. Moreover, anomaly identification
are able to detect trends, seasonality, and irregular patterns in photos using VGG19 enables the system to spot odd
in pollutant concentrations by capturing temporal occurrences or environmental alterations that might have
dependencies within the data. This allows for more an effect on air quality. This proactive strategy makes it
precise projections of future air quality levels. The ESP32 possible to take prompt action to reduce any risks.VGG19
is a reliable platform for gathering unprocessed sensor provides a level of sophistication to the analytical process
data and carrying out preliminary cleaning and when it is integrated into an ESP32-based Internet of
normalization operations. Then, for analysis and Things air quality monitoring system. This is especially
prediction, the preprocessed data is supplied into the deep useful in situations when visual data is used to supplement
learning models. Depending on the specifications and sensor readings, as it increases the overall efficacy and
available resources of the system, these models can be dependability of the monitoring system.
processed in batch or real-time on the ESP32 or directly
on remote servers.

1.CNN VII. RESULTS AND DISCUSSION

An important development in environmental sensing and
Spatial feature extraction from data is a specialty analysis is the combination of ESP32 microcontrollers
of CNNs. CNNs are able to capture spatial patterns and with deep learning models and techniques, such as
correlations between pollutant concentrations across Convolutional Neural Networks (CNN) and VGG19, for
different regions in the context of air quality monitoring, the purpose of monitoring the Air Quality Index (AQI).

73
© 2024, IRJEdT Volume: 06 Issue: 04 |April-2024
This
novel method improves the precision and efficacy of AQI
forecasts by fusing the ESP32's real-time data gathering
capabilities with deep learning models' complex pattern
recognition skills.
The system can give more precise and timely AQI
forecasts by using ESP32's sensor interface capabilities to
collect air quality data, such as pollutant concentrations REFERENCE
and ambient characteristics, and merging it with deep
learning models trained on historical AQI data. Relevant [1]. Li, Z. Jiang, X. Guo, and X. Yu, "Design and
characteristics can be extracted from sensor data to Implementation of an IoT-based Air Quality
forecast AQI levels with the help of per-trained models Monitoring System," in 2018 IEEE International
like VGG19, which are well-known for their strong Conference on Smart Internet of Things (SmartIoT),
Taormina, Italy, 2018
performance in image classification tasks, and CNNs,
which can automatically learn hierarchical features from [2]. Kumar and R. Singh, "Wireless Sensor Network
raw data. Promising results have been observed when for Air Quality Monitoring Using IoT," in 2019 5th
merging ESP32 for AQI monitoring with deep learning International Conference on Advanced Computing &
models. In comparison to conventional methods, the Communication Systems (ICACCS), Coimbatore,
system can accurately capture complicated interactions India, 2019
between different air contaminants and environmental
conditions, resulting in more accurate AQI estimates. [3]. Lee, S. Kim, and J. Lee, "A Low-Cost Air Quality
Furthermore, the deep learning models can adjust to Monitoring System Based on IoT and Machine
Learning," in 2018 IEEE International Conference on
shifting environmental conditions and increase forecast Consumer Electronics (ICCE), Las Vegas, NV, USA,
accuracy over time by continuously learning from fresh 2018
data.
[4]. Singh, N. Singh, and N. Kumar, "Design of IoT
Based Air Pollution Monitoring System Using
ESP32," in 2020 7th International Conference on
VIII. CONCLUSION Signal Processing and Integrated Networks (SPIN),
In conclusion, a reliable and effective method for Noida, India, 2020
assessing air quality in real time is provided by the
[5]. Pinto, F. J. M. Fernandes, and J. M. M. Gordo,
Internet of Things-based air quality index monitoring "Internet of Things and Edge Computing Applied to
system that uses ESP32. The system guarantees precise Air Quality Monitoring: A Case Study in Portugal," in
and dependable data collecting by utilizing sophisticated 2018 IEEE 17th Interna
preprocessing techniques and integrating ESP32 with
many environmental sensors. Deep learning algorithms [6]. Ranjan, V. Pandey, and S. Jain, "Design and
are also incorporated to further improve data analysis and Implementation of IoT Based Air Quality Monitoring
prediction skills, enabling preventative actions to reduce System," in 2019 3rd International Conference on
pollution and safeguard public health. Because of its low Trends in Electronics and Informatics (ICOEI),
Tirunelveli, India, 2019.
power consumption wireless connection, and adaptability
as a data gathering and processing platform, the ESP32 is [7]. Jing and L. Jing, "Design of Air Quality
a great choice for a wide range of environmental Monitoring System Based on Internet of Things," in
monitoring applications. Furthermore, data administration 2021 6th International Conference on Intelligent
and remote monitoring are made possible by the system's Computing and Signal Processing (ICSP), Hangzhou,
seamless integration with cloud services. All things China, 2021
considered, the Internet of Things (IoT)-based air quality
monitoring system provides a scalable and affordable way [8]. M. Swain, B. B. Satpathy, and A. B. Sahoo,
to track and manage air quality, enabling informed "Development of an IoT-Based Real-Time Air Quality
Monitoring System," in 2020 7th International
decision-making and promoting a more sustainable and
Conference on Power, Control and Computing
healthy environment. Technologies (ICPC2T), Vellore, India, 2020

[9]. K. Singh and A. Chauhan, "IoT Based Air

Pollution Monitoring System," in 2019 6th
International Conference on Signal Processing and
Integrated Networks (SPIN), Noida, India, 2019, pp.

[10]. D. Chauhan, R. Kumar, and V. Mishra, "IoT

Based Air Pollution Monitoring System," in 2020 6th
International Conference on Advanced Computing and
Communication Systems (ICACCS), Coimbatore,
India, 2020.

74
© 2024, IRJEdT Volume: 06 Issue: 04 |April-2024