A REAL-TIME AIR QUALITY MONITORING DEVICE WITH STATIONARY

AND MOBILE MODE CAPABILITY

AN UNDERGRADUATE THESIS

Presented to
the Computer Engineering Department of
School of Engineering and Architecture
Ateneo de Davao University
E. Jacinto St., Davao City

In Partial Fulfilment
of the Requirements for the Degree of
BACHELOR OF SCIENCE IN COMPUTER ENGINEERING

Agawin, Julian Marvin A.

2024
 ABSTRACT

Air pollution poses a significant threat to global health, ranking as the fifth

leading risk factor for mortality worldwide and correlating with increased

hospitalizations. Common sources of air pollution, including motorized vehicles and

fuel-consuming products, contribute to various health risks such as asthma and

Chronic Obstructive Pulmonary Disease (COPD). Among the major pollutants, ozone

(O3) and particulate matter (PM) are particularly concerning, with inhalation leading

to immediate breathing problems and long-term health effects. Particulate matter,

including aerosols, significantly contributes to air pollution-related mortality, estimated

to claim over 500,000 lives annually. Despite efforts by organizations like the

Department of Environment and Natural Resources (DENR), gaps persist in air

quality monitoring, especially in rapidly growing urban areas like Davao City. The

primary objective of this study is to develop a low-cost air quality monitoring device

capable of real-time data collection for pollutants such as O3, PM2.5, and PM10,

alongside GPS location data. This device aims to address the limitations of existing

monitoring systems, providing extensive measurements for both mobile and

stationary air pollution. Additionally, the study seeks to raise public awareness

through visual representations of air pollutant data, aiding decision-making and future

research efforts. By providing accurate and accessible air quality data, the study

Furthermore it aims to contribute to improved environmental management and public

health in Davao City.

The study details the calibration process of the sensor, utilizing the Linear

Regression Curve method for accuracy. Calibration is conducted in conjunction with

DENR EMB R-XI and AdDU air quality monitoring stations to ensure consistency.

Data gathering involves testing two deployment types: Mobile and Stationary

functionalities. The Mobile Route along Roxas Avenue provides running average

readings, while the Stationary location near the AdDU PM 2.5 Air Quality monitor
offers stable pollutant readings and corresponding Air Quality Index (AQI) values. A

comparison between data intervals reveals connectivity issues on April 1st due to

network limitations, resolved by using a different cellular network on April 12th.

Mobile monitoring generates a heat map indicating areas with varying air quality

along the route, aiding in informed decision-making. The study concludes with the

successful assembly of a low-cost air pollution device capable of real-time

monitoring, demonstrating versatility in both stationary and mobile applications. The

web application developed facilitates data visualization and management, with data

securely stored in a cloud database for accessibility and analysis. These findings

signify a significant advancement in air quality monitoring technology, with

implications for environmental research and mitigation efforts in Davao City.

 TABLE OF CONTENTS

TABLE OF CONTENTS iii

ACKNOWLEDGEMENT vii

LIST OF FIGURES viii

LIST OF TABLES x

LIST OF EQUATIONS xi

CHAPTER I 1

1.1 Background of the Study 1

1.2 Statement of the Problem 3

1.3 Objectives of the Study 4

1.4 Significance of the Study 4

1.5 Scope and Delimitations 5

1.6 Conceptual Framework 6

1.7 Definition of Terms 7

CHAPTER II 8

2.1 Air Quality 8

2.1.1 Air Pollution 8

2.1.2 Ozone(O3) 8

2.1.3 Particulate Matter 9

2.2 Detection Methods and Computations 9

2.2.1 Air Quality Index 9

2.2.2 Deduction Method 11

 5

2.3 Air Quality Monitoring System based on Arduino Microcontrollers 12

2.4 Real-Time Air Quality Monitoring Systems 13

2.5 Wireless Air Quality Monitoring Systems 15

2.6 Mobile Air Quality Monitoring Systems 16

2.7 Enclosures of Related Studies 18

2.8 Calibration and Evaluation of Air Quality Monitoring Device 19

CHAPTER III 20

3.1 Research Design 20

3.2 Research Locale 20

3.3 Research Procedure 21

3.3.1 System Design 22

3.3.2 Prototype Design 23

3.4 Software Design 27

3.5 Research Instruments and Methods 32

3.6 Research Evaluation and Calibration 35

3.6.1 Sensor Calibration 35

3.6.2 White box testing 43

3.6.2.1 Air Quality Monitor Testing 44

3.6.2.2 Web Application Testing 44

3.6.3 Black Box testing 44

3.6.3.1 Air Quality Monitor Testing 45

3.6.3.2 Web Application testing 45

3.7 Research Instrument 45

 6

CHAPTER IV 49

4.1 Development and Testing Results 49

4.1.1 White Box Testing for Air Quality Monitor 49

4.1.2 White Box Testing for Web Application 52

4.1.3 Black Box Testing for Smart Node Sensor 54

4.1.4 Black Box Testing for Web Application 55

4.2 Data Gathering and Analysis 56

4.2.1 Stationary Monitoring Data Gathering 56

4.2.2 Stationary Monitoring Data Analysis 59

4.2.3 Mobile Data Gathering 59

4.2.4 Mobile Monitoring Data Analysis 61

4.2.5 User Testing and Survey 61

CHAPTER V 63

5.1 Conclusions 63

REFERENCES 64

APPENDIX A 69

A.1 NodeMCU V3 70

A.2 Nova Particulate Matter Sensor 71

A.3 Spec Sensors Digital Ozone Sensor (DGS-O3 968-042) 72

APPENDIX B 73

B.1 Soil pH Test Report 73

B.2 Soil Moisture Test Report 74

APPENDIX C 75
 7

APPENDIX D 76

APPENDIX E 77

APPENDIX F 78
 8

ACKNOWLEDGEMENT
 9

LIST OF FIGURES

Figure 1.5.1: Stationary Location (2nd Floor Canisius Bldg) 5

Figure 1.5.2: Mobile Route along Roxas Avenue 6

Figure 1.6.1: Conceptual Framework 6

Figure 2.2.1 AQI with Respective Color and Health Messages 10

Figure 2.2.2: Ozone(O3) AQI 2012 Updated Breakpoints [23] 11

Figure 2.2.3: PM2.5 and PM10 AQI 2012 Updated Breakpoints [24] 11

Figure 2.2.4: Deduction Method of Research 11

Figure 2.3.1: Arduino Uno R3 microcontroller 12

Figure 2.4.1: Cloud Data Processing Model 13

Figure 2.4.2: MQTT Implementation: Air Quality Monitoring System Design 14

Figure 2.4.3: Electronic Nose (E-Nose) Proof of Concept 15

Figure 2.5.1: XBee XB24-Z7WIT-004 Zigbee Module 16

Figure 2.5.2: MQ-131 Ozone O3 Gas Sensor 16

Figure 2.6.1: Sample of a High Resolution Air Quality Map 17

Figure 2.6.2: Architectural Overview of Airify 17

Figure 2.7.1: Electrochemical Ozone Sensor Test(SPEC Ozone Sensor) 18

Figure 2.7.2: Enclosure Implementation for Fine Dust Sensor and Ozone Sensor 19

Figure 2.8.1: Linear Regression Calibration Graph 19

Figure 3.2.1: Stationary Location (2nd Floor Canisius Bldg) 21

Figure 3.2.2: Mobile Route along Roxas Avenue 21

Figure 3.3.1: Overview of the System Design 22

Figure 3.3.2: Prototype Design Preview 23

Figure 3.3.3: Prototype Area Type Designations 24

Figure 3.3.4: Actual Built Prototype 25

Figure 3.3.5: SPEC DGS-O3 968-042 Ozone(O3) Sensor 26

Figure 3.3.6: Nova SDS - 011 Fine Dust Sensor 26

 10

Figure 3.3.7: NodeMCU V3 ESP8266 ESP-12E Microcontroller 27

Figure 3.4.1: Main Page View of the Web Application 29

Figure 3.4.2: Table of Summary of Data for Stationary Monitoring 30

Figure 3.4.3: Google Maps API Selective Location Representation 30

Figure 3.4.4: Table of Summary of Data for Mobile Monitoring 31

Figure 3.4.5: Mobile AQI Monitoring Heat Map Representation 31

Figure 3.6.1: Linear Regression of Line of Best Fit (DENR PM 2.5) 37

Figure 3.6.2: Corrected PM 2.5 Curve vs DENR PM 2.5 37

Figure 3.6.3: Linear Regression of Line of Best Fit (AdDU PM 2.5) 40

Figure 3.6.4: Corrected PM 2.5 Curve vs AdDU PM 2.5 40

Figure 3.6.5: Linear Regression of Line of Best Fit (AdDU PM 10) 42

Figure 3.6.6: Corrected PM 10 Curve vs AdDU PM 10 42

Figure 3.6.7: Ozone Sensor Barcode 43

Figure 3.7.1 Survey Questionnaire first page 47

Figure 3.7.2 Survey Questionnaire second page 48

Figure 4.2.1: Stationary Monitoring April 1st Data Map 57

Figure 4.2.2: Stationary Monitoring April 12th Data Map 58

Figure 4.2.3: Mobile Monitoring April 13th Heat Map 61

Figure 4.2.4 Survey Respondents Identifier 62

 11

LIST OF TABLES

Table 3.5.1 Particulate Matter 2.5 Microns AQI Breakpoints Table 33

Table 3.5.2: Particulate Matter 10 Microns AQI Breakpoints Table 33

Table 3.5.3: Ozone AQI Breakpoints Table 33

Table 3.6.1: Linear Regression Data for Raw Data vs DENR PM2.5 Calibration 35

Table 3.6.2: Line of Best Fit DENR PM 2.5 Sensor with Corrected values 36

Table 3.6.3: Linear Regression Data for Raw Data vs AdDU PM2.5 Calibration 38

Table 3.6.4: Line of Best Fit AdDU PM 2.5 Sensor with Corrected values 39

Table 3.6.5: Linear Regression Data for Raw PM10 vs DENR PM10 Calibration 41

Table 3.6.6: Line of Best Fit AdDU PM 10 Sensor with Corrected values 41

Table 3.7.1: Four - Point Likert Scale 46

Table 4.1.1 Test Case 1 49

Table 4.1.2 Test Case 2 50

Table 4.1.3 Test Case 3 50

Table 4.1.4 Test Case 4 51

Table 4.1.5 Test Case 5 51

Table 4.1.6 Test Case 6 52

Table 4.1.7 Test Case 7 52

Table 4.1.8 Test Case 11 53

Table 4.1.9 Test Case 1 54

Table 4.1.10 Test Case 2 55

Table 4.2.1: Stationary Monitoring April 1st Data Collection 57

Table 4.2.2: Stationary Monitoring April 12th Data Collection 58

Table 4.2.3: Mobile Monitoring April 13th Data Collection 60

 12

LIST OF EQUATIONS

Equation 3.5.1: General Air Quality Index (AQI) Formula 32

Equation 3.5.2: Linear Regression Formula 34

Equation 3.5.3: Percent Error Formula 34

Equation 3.7.1: Mean Averaging Formula 46

 CHAPTER I

INTRODUCTION

This section presents the investigation by setting the viewpoint and

expressing the research problem. It likewise examines the study's objectives, the

scope and limitations, and the conceptual framework.

1 Hidden marker

1.1 Background of the Study

Air pollution is the fifth leading risk factor for mortality worldwide. There is also

a relationship between air pollution and increased hospitalizations [1]. In general, air

pollution is found in common emissions of motorized vehicles, appliances and other

fuel consuming products. From air pollution, comes tons of possible health risks like

asthma, Chronic Obstructive Pulmonary Disease (COPD), and more breathing

related problems. The health risks are enough reason to monitor the air quality so

that the proper standards are met [2].

According to the American Lung Association’s 2019 “State of the Air” report,

“the number of people exposed to unhealthy air increased to nearly 141.1 million.

That represents an increase from the past two reports: higher than the 133.9 million

in the 2018 report (covering 2014-2016) and the 125 million in the 2017 report

(covering 2013-2015).” Among the major pollutants that affect health are ozone (\

ce{O3} or smog) and particle matter (soot) pollution [1]. The effects of inhaling these

pollutants come in the form of health risks. Inhalation of Ozone and Particle pollution

pose immediate breathing problems, cardiovascular effects, and long term effects

show great bodily reactions to many air pollutants [3].

In air pollution, studies show that there is a consistent increase in cardiac and

respiratory mortality rates due to exposure to particulate matter. Particulate matter is

a major contributor in air pollutants and is estimated to kill more than 500,000 people
 2

each year [4]. Particulate matter are also identified as aerosols which are liquid

droplets and/or solid particles suspended in a gaseous medium. Aerosols adversely

affect health and also contributes to environmental pollution [5].

From the State of the Global Air report of 2019, “on-the-ground air quality

monitoring stations are few and far between in the rapidly growing urban areas of

countries at low and middle levels of development, as well as in rural and suburban

areas throughout the world” [2]. Methods of gathering data are now becoming a trend

and the technology for air quality monitors become readily available for public use.

The methods of measuring air quality are of international standards and allow for a

unified calibration setting [6].

In the Philippines, the Department of Environment and Natural Resources

(DENR), have already made efforts in attaining and monitoring ambient air quality of

a fixed area. As defined in RA 8749 as the general amount of pollution present in a

broad area, and refers to the atmosphere’s average purity as distinguished from

discharge measurements taken at the source of pollution. The DENR Environmental

Management Bureau (DENR EMB) regional monitoring stations are taking

measurements daily [7].

In Davao, DENR EMB of Region 11(DENR EMB R-XI), also has sensors for

air quality located on very specific areas, which are also unable to collect or provide

for most parts of the City that is highly congested with people and motor vehicles.

The DENR EMB R-XI also only has two monitor locations which are automated and

is able to provide data at any given time. Knowing that Davao is a very large city,

these monitors are not enough to provide quality data and no matter how far the

radius is covers is, it is unable and inefficient to cover where most of the population

experience the health risks of air pollution [8].

 3

1.2 Statement of the Problem

Davao City’s ambient air quality data is insufficient due to the lack of

monitoring stations. There is only six air quality monitoring stations deployed in the

city. These stations are situated in Ilang, J.P. Laurel/V. Mapa Street, Davao Memorial

Park, Toril Poblacion Open Park, Calinan National High School, and Davao

International Airport [8]. With the stations, DENR EMB XI used the deduction method

to cover the rest of the city. The deduction method is able to produce better results

when more data is available [9]. Expansion of monitoring stations for the DENR EMB

XI must be met to provide the public with higher quality data for further studies to

utilize.

Out of the six air quality monitoring stations deployed, only two of these are

automated. These are the Calinan District station and the Davao International Airport

station. With this, only the Calinan district station is registered in the Air Visual

database and website [10]. This station has recent data that range from the moderate

(81 air quality index) to a high (102 air quality index). For the both automated

stations, the data that these are able to produce are real-time averages of the

respective area. Most of the industry and population are not located in such far

places. There is higher concentration of people and air pollutants in urban areas,

where poorer air quality poses more detrimental health risks for its residents [7].

The DENR EMB R-XI air quality monitoring stations around the city are

commercial products and of an international standard. These are also difficult to

procure and expensive, thus an alternative solution must be explored. The alternative

must be able to produce similar quality of data, should be easily replicated, compact,

and mobile. The current implementation of the reference air quality monitoring

stations require the data retrieval team to collect the data after 6 days [11] [12].

However, there are studies that suggest a real-time implementation of gathering air

pollution data is much more convenient than having to conduct site visits often.
 4

1.3 Objectives of the Study

The primary objective of the study is to assemble an air quality monitoring

device that is able to display real-time data and save the data of deployment areas or

routes with GPS specified locations. To specify, the objectives of the study are the

following:

1. To assemble a low-cost air pollution device to monitor Ozone, Particle

Matter 2.5 microns, Particle Matter 10 microns(O 3, PM2.5, and PM10), and

GPS data, with the use of a NodeMCU V3 ESP8266 ESP-12E

microcontroller, a fine dust sensor for PM 2.5, and PM10 microns, and an

O3 sensor.

2. To include the capability of the device for real-time stationary and mobile

air quality monitoring of specific pollutants and Location Data.

3. To develop a web application for visualization of air pollutants in specific

locations of Davao City in the form of tabular data.

4. To store the data in the cloud database for database management.

1.4 Significance of the Study

The significance of the study is to deliver current and reliable data of air

quality

specifically Ozone(O3), PM2.5 and PM10 with GPS locations for mapping applications.

The study also stands to provide the DENR EMB R-XI with extensive measurements

of mobile and stationary air pollution measurements. The visual representations is

also for the benefit of the public by spreading awareness that certain locations are

not optimal for travel especially for commuters with breathing-related problems. The
 5

study will also open avenues for future research based on the air quality and pollution

data retrieved while also providing quality data as a reference.

The study aims to provide data for the DENR EMB R-XI, for a low cost,

accurate, and multi-point reading. The sensors to be used must be calibrated and

tested in accordance with the present and internationally calibrated sensors available

by the DENR EMB R-XI. The data produced by this study will also benefit the people

of Davao City by way of giving awareness and being able to map or show a

visualized proof of the status of air pollution in certain vicinity’s at given time slots.

1.5 Scope and Delimitations

Due to limitation in the deployment of the device and the calibration delay

from the DENR, the study will be limiting its area to one stationary and mobile

location respecitively. The stationary location is on the 2 nd floor of the Canisius Bldg.

in the College Campus, while the mobile location is a mobile route along Roxas

Avenue.

Figure 1.5.1: Stationary Location (2nd Floor Canisius Bldg)

 6

Figure 1.5.2: Mobile Route along Roxas Avenue

1.6 Conceptual Framework

Figure 1.6.1: Conceptual Framework

The conceptual framework as seen in Figure 1.3 has four levels. The first

level is the Air Pollution Monitor. This level comprises the hardware module of the

study and collects the data from the area. The data is composed of three major parts,

which are the value of the pollutants, the GPS location of the sensor and its time

stamp. These data are then sent to the next level.

The second level is the Data Delivery System. In this level, the data is

delivered from the sensors to the read only memory module of the system for offline
 7

data storage. Once connected to a Wi-Fi with internet service, the data will be

delivered to a cloud data storage for use with the next level of the system.

The third level is the Web Application. This level is one of the major software-

based level of the study as this will serve as the visualization of the data delivered to

the cloud storage. The data in the cloud storage will then be displayed in the web

application in the form of tabulated data and a constantly updated Air Quality Index

for the specific pollutant and location of the device. The web application level is

important for the succeeding level.

The fourth level is Air Quality Information. Having the ability to access the

previous level on any device with a browser installed, gives the users an avenue to

be aware of the current air pollution in the specific areas. The air quality information

also provides the user with quality visual representations of the data produced. This

also stands for the ability to cater future studies with the matter of air quality and the

data of the pollutants measured.

1.7 Definition of Terms

AQI A numerical value for measuring the quality of air in regards to air pollution and

health concerns.

DENR Government office of the Philippines on environmental protection.

DENR EMB Government office of the Philippines on environmental management.

DENR EMB R-XI Region 11 based Government office of the Region 11 on

environmental management.
 CHAPTER II

REVIEW OF RELATED LITERATURE

This chapter will discuss recent and past sources related on the study. These

sources include articles and research publications used to further understand the

topics and technology used in this study.

1 Hidden marker

1.1 Air Quality

1.1.1 Air Pollution

Air pollutants are generally classified in two categories based on their

physical form; gaseous and particulate (PM). Various gaseous and particulate

pollutants have an affect on the incoming or outgoing solar radiation which then

affects the temperature profile of Earth's atmosphere. The United States

Environmental Protection Agency(US EPA) have set global standards for Air Quality.

These standards lessened the widespread of visible air pollution ranging from the

1970's. However, air pollution is still harmful even when it is not visible. While the

health effects of air pollution differ depending on the gas, the common risks to health

are lung cancer, acute respiratory infections, and chronic obstructive pulmonary

disease (COPD) [5] [13].

1.1.2 Ozone(O3)

Ozone(O3) is a colorless gas that is derived from the reaction of nitrogen

oxides and volatile organic compounds emissions. Reduced ozone levels resulting

from ozone layer depletion implies that there is significantly decreased protection and

increased exposure from Ultraviolet(UV) radiation at ground level. High ambient

ozone concentrations are untied to heavily urbanized areas as ozone concentrations

are naturally present in the environment [14] [15].

 9

The health risks of atmospheric ozone exists in two scenarios. The first

scenario is when the ozone layer absorbs UV light, this reduces the possible

exposure to UV radiation. The UV radiation usually causes skin cancer and

cataracts. The second scenario is the adverse effects of inhalation of ozone which

can increase the frequency of asthma attacks, cause shortness of breath, aggravate

lung diseases, and other breathing related diseases through long-term exposure.

Thus, ozone gas is able to affect humans both internally (lungs) and externally (skin)

[13] [16] [17] [18].

1.1.3 Particulate Matter

Particulate Matter(PM), also known as aerosols, is the product of solid

particles and liquid droplets found in the air. Common particles like dirt, smoke, dust

or soot are large or dark enough to be visible. PM10 are inhalable particles that scale

up to 10 micrometers in diameter, while PM2.5 are inhalable particles that scale to

2.5 micrometers in diameter [4] [19].

The health risks of particulate matter defined depending on its size.

Supporting research indicates that PM2.5 exposure causes irritation of the airways,

heart or lung disease, and respiratory. While studies on PM 10 potentially indicates

that particles which are less than 10 micrometers in diameter have the tendency to

enter the bloodstream [13] [18] [20].

1.2 Detection Methods and Computations

1.2.1 Air Quality Index

The Air Quality Index(AQI) offers a visual idea of how polluted the air is. The

AQI is a numerical value which ranges from 0 - 500, and with respective colors and

levels of health concern. AQI computation varies on the specific pollutant

concentration over a specific averaging period because these have varying amounts

of potency. Pollutant concentrations are obtained from a gas sensor or forecasted

models. While the range of values and level of indices vary for some nations, the
 10

most common standard is the USA Environmental Protection Agency(USA EPA)

standards and it is being used globally [13] [21] [22].

The formula for Pollutant "X" and Air Quality Index as seen in (1). The value

for the AQI is then referred to the respective AQI levels with respective color coding,

and health messages shown in Figure 2.1.

Figure 2.2.1 AQI with Respective Color and Health Messages

The Air Quality Index breakpoints and respective health messages

varies depending on the pollutant. For Ozone (O3), the updated breakpoints

and health messages are shown in Figure 2.2. For PM2.5 and PM10, the

updated breakpoints and health messages are shown in Figure 2.3.

 11

Figure 2.2.2: Ozone(O3) AQI 2012 Updated Breakpoints [23]

Figure 2.2.3: PM2.5 and PM10 AQI 2012 Updated Breakpoints [24]

1.2.2 Deduction Method

Figure 2.2.4: Deduction Method of Research

 12

The deduction method, as seen in Figure 2.4, is a research approach that

builds a conclusion from a hypothesis and quality data. The definition of deductive is

to simply result to a general from given particulars. In this method, data must first be

assured of its quality so as not to build false conclusions. The amount of data must

also be able to suffice the necessary parameters of using this method. Insufficient

amounts of data collectively leads to inaccurate results or conclusions [9] [25].

1.3 Air Quality Monitoring System based on Arduino Microcontrollers

The implementation of the air quality monitoring systems was with the use of the

Arduino Uno R3 microcontroller, as seen in Figure 2.5. The environmental parameters

chosen were temperature, humidity, CO, CO 2, detection of leakage of any gas - smoke,

alcohol, or liquid petroleum gas (LPG). The gas sensors used by the study are all of the

"MQ" series line and these seem to be sufficient for the study as the objective is the

ability

to track the presence of the gases only. The study did fail to state the calibration

procedure

and conclusively states that the air quality monitor is to simply track the changes in air

quality [3].
 13

Figure 2.3.1: Arduino Uno R3 microcontroller

1.4 Real-Time Air Quality Monitoring Systems

The study of real-time implementation of air quality monitoring where the

sensor used by the study was also of the "MQ" series sensor and it also shows a

non-linear sensitivity graph for the sensor used. The data presented by the study was

pure raw data and this makes the AQI computed as impractical since the application

of AQI formula requires the sensor reading to have hourly averaging data. The study

tracks the data using an MQ135 gas sensor and sends raw sensor data to the cloud

using an android application. The MQ135 gas sensor was a generic air quality sensor

which was incapable of producing quality data as the sensor was able to gather

general gases in the environment. A better implementation would have been to

identify gases separately and compare air quality index levels using the different

gases recorded. Crowd sensing was the method for data collection and this was by

the use of multiple nodes having gas sensors. The cloud based implementation uses

the data processing model in Figure 2.6. The application uses Bluetooth capabilities

to retrieve data from the sensors and adds the necessary data such as location, time,

and device ID. The cloud service used was Google FireBase which contains the

database to sync and store the raw sensor readings [11].

 14

Figure 2.4.1: Cloud Data Processing Model

An alternative study on real-time air quality monitoring employ the use of

Message Queue Telemetry Transport (MQTT) protocol using an ESP8266

NodeMCU. The MQTT protocol allowed the communication of data from Machine-to-

Machine (M2M) and serves as an Internet of Things (IoT) tool as well. The system

design of the study in Figure \ref{fig:mqtt}. This study also used an air quality

monitoring dashboard using a responsive web application, which can be seen in

Figure \ref{fig:mqtt2}. The dashboard contains raw data values of the sensor and

display it in real-time for monitoring purposes. A better implementation of this would

be to produce an effective Air Quality Index value for the specific gas being

measured as this would make the value more comprehensible to the average person

[26].

Figure 2.4.2: MQTT Implementation: Air Quality Monitoring System

Design
 15

Figure 2.8: MQTT Implementation: Air Quality Monitoring Dashboard

As the Air Quality directly affects the health of the population, another

implementation for real-time monitoring focused on the breathing capabilities of the

human physique. The air quality monitor was built into a container that simulates the

inhalation and exhalation patterns using an improvised gas trap which functions

similarly as a "nose", hence the Electronic Nose (E-Nose) design. The

implementation seen in Figure \ref{fig:Enose} was the proof of concept of the study.

However, this study only uses the low cost "MQ" sensors to track the air pollutants

[27].

Figure 2.4.3: Electronic Nose (E-Nose) Proof of Concept

 16

1.5 Wireless Air Quality Monitoring Systems

The study implemented the air quality monitoring station using wireless sensor

networks. A Zigbee module, as seen in Figure 2.9, was used with the Arduino Uno R3

microcontroller to offer the wireless capabilities. The device which was setup as a client

node was able to send data to the server using a router node. The sensor node was

composed of two types. The type one sensor node was able to track Carbon Dioxide

(CO2), volatile organic compounds (VOCs), temperature, and humidity. While the type

two sensor node was able to track Carbon Monoxide (CO), Ozone (O 3), temperature and

humidity. The CO2 sensor was the MG811 chemical sensor which was able to detect

around 350 to 10000 ppm of CO2. The VOC sensor was the TGS2602 heating

semiconductor sensor, which gives results via changes in conductivity that produce an

output signal corresponding to a specific gas concentration. The CO sensor was simply

and MQ7 sensor and it has a detection range of 20-2000 ppm. The Ozone sensor was

the MQ131 sensor, as seen in Figure 2.10, which was the same MQ series that was

faulty in terms of calibration and would only be good for presence detection of the

specific gas [12].

Figure 2.5.1: XBee XB24-Z7WIT-004 Zigbee Module

 17

Figure 2.5.2: MQ-131 Ozone O3 Gas Sensor

1.6 Mobile Air Quality Monitoring Systems

The implementation of the Air Quality Monitor is by the use of mobile sensor

nodes which are placed on Buses. The concept of Bus as a Sensor (BaaS) is to use

the present implementation of public transportation as the option for deploying the

mobile sensor nodes. The study also uses location data to provide a mapping

visualization as an output. A sample of the air quality map is seen in Figure 2.11. The

implementation of public transport vehicles as a mobile sensing node proves to be

useful as these would be able to gather data on the whole route and show a visual

heat map of the air quality within the route [28].

Figure 2.6.1: Sample of a High Resolution Air Quality Map

Further implementations of mobile air quality monitoring adapt the use of

wearable technology. The wearable technology is able to send data to a smartphone

 18

application coined as "Airify". The architectural overview of the study is seen in

Figure 2.12. The wearable technology device purposely has integrated gas sensors

that detect the air pollutants and these are then sent to the Xamarin-based Android

application using Bluetooth. The data collected by the app is then processed using

the necessary USA EPA standard Air Quality index formula with the respective

breakpoints of the different pollutants. The air quality monitoring favors the use of the

wearable technology to prefer human traffic instead of vehicular transport as the

study is meant to cover urban monitoring while neglecting roadside congestion [29].

Figure 2.6.2: Architectural Overview of Airify

1.7 Enclosures of Related Studies

The implementations for using electrochemical sensors such as the SPEC

DGS-O3 968-042 Digital Ozone Sensor as seen for being tested in Figure 2.13 has

the ability to offer readings without the need of a sample space. The sample of air

being tested will be the ambient reading for the location.

 19

Figure 2.7.1: Electrochemical Ozone Sensor Test(SPEC Ozone Sensor)

The enclosure used by the study of wearable sensors for analyzing personal

exposure to air pollution, uses an open chassis. The open chassis, as seen in

Figure \ref{fig:enclosure1} will allow the sensors to read the values of an open air

environment and not of a sample space. The reading will then allow to measure the

natural environment and not a controlled environment. In measuring for air pollution,

studies insist to not use a controlled environment as these readings will only be for

the specific controlled environment [17] [30].

Figure 2.7.2: Enclosure Implementation for Fine Dust Sensor and Ozone Sensor
 20

1.8 Calibration and Evaluation of Air Quality Monitoring Device

The calibration method used in the study of using Low cost sensors for air

quality monitoring of Ozone and Nitrogen Dioxide, employs the use of Field

Calibration . In calibrating the sensor, the study used a reference air quality

monitoring station maintained by the European Reference Laboratory for Air Pollution

(ERLAP). The method of calibration used was Linear Regression and this was used

on data collected simultaneously within the same location of the reference sensor.

The calibration using Linear Regression seen in Figure 2.15 compares both

reference sensor and the built air quality monitoring device [31].

Figure 2.8.1: Linear Regression Calibration Graph

 CHAPTER III

METHODOLOGY

This chapter exhibits the determined approach in designing and developing

the Ozone(O3), PM2.5 and PM10 air pollution monitor prototype and web application

software. The chapter will also convey the research design, locale, and necessary

procedures to be used in the study.

1 Hidden marker

1.1 Research Design

The study will have a developmental research design since it is centered on

developing, calibrating, and validating the proposed air pollution monitor. The DENR

EMB R-XI have implemented stationary air quality stations which are all situated to

actively use the deduction method to determine the specific AQI of the city [8].

Having only six \gls{AQI} monitoring stations, two of which are automatic, the data is

prone to lapses in using the deduction method [25]. The current number of stations

are widely affected by the high cost of the commercially available sensors, the

immobility of the stations due to its size, and that most of the deployment areas are

private owned. The study seeks to improve this by being able to provide data for the

areas that are deficient in being monitored yet highly populated. The study improves

the current technology by using lower cost and power consumption rates. The study

also enhances the current technology by being able to produce quality real-time data

sent to a cloud storage and presented via the web application.

1.2 Research Locale

The stationary location of the study will be at The stationary location is on the

2nd floor of the Canisius Bldg. in the College Campus as seen in Figure 3.1, while the

mobile location is a mobile route along Roxas Avenue as seen in Figure 3.2. The
 22

locations were determined by the accessibility of the areas and also the presence of

an online air pollution monitor.

Figure 3.2.1: Stationary Location (2nd Floor Canisius Bldg)

Figure 3.2.2: Mobile Route along Roxas Avenue

1.3 Research Procedure

This section includes the selection of hardware components such as the

sensors, microcontroller, and the single-board computer and as well as the assembly
 23

of the hardware components into a single compact device and integrating them into

the overall system architecture.

1.3.1 System Design

Figure 3.3.1: Overview of the System Design

The overview of the system is shown in Figure 3.3. The system is composed

of three levels. The first level is the NodeMCU V3 ESP8266 ESP-12E microcontroller

module equipped with the fine dust sensor, ozone sensor, GPS module, memory

module, and a battery bank as a power supply. Air is freely flowed into a duct

directed to the fine dust and ozone sensors. The calibrated sensors will then take

running average value per minute with an interval time of fifteen minutes for Mobile

Air Quality monitoring and a running average value with an interval time of sixty

minutes for Stationary Air Quality monitoring. The averaged values will then be used

in the next level of the system.

The next level of the system is the data transmission to a local database on

the memory module or to the cloud data storage using a Wi-Fi connection. When
 24

there is no Wi-Fi connection available, the data is logged in the cache memory of the

NodeMCU. The offline data will then be uploaded to the cloud data storage once the

device is able to connect to Wi-Fi. Data contents will then be acquired via the

memory module or from the cloud data storage once it is uploaded.

The last level of the system is the Website Application software, which is

accessible via any browser with an internet connection. The web app will display the

data prepared and sent by the previous levels of the system. The data will be

tabulated showing the PM2.5, PM10, Ozone(O3) readings, their respective Air Quality

Index values, and their respective GPS locations per data entry.

1.3.2 Prototype Design

This subsection will discuss the design of the proposed air pollution monitor

including the preview of the prototype case and the selection of hardware to be used.

Preview of the Prototype

The visual plan of the prototype as seen in Figure 3.3.-2 is the proposed

design of the study and from previous studies, will be able to measure both mobile

and stationary values even without the use of a controlled environment. Air samples

will be in the form of ambient air as measuring samples in a controlled environment

will result in only data of the specific sample space of the environment [17].
 25

Figure 3.3.2: Prototype Design Preview

The chassis will be made up of two types of areas, open air areas and a

closed air areas as seen in Figure 3.3-3. The chassis material is mostly acrylic

plastic. The Particulate Matter sensor which has built-in inlet and outlet ports will be

in the closed area of the chassis but the sensor must be on a corner to accommodate

for the fan and ports. While for the Ozone sensor, it will need to be in the open area

of the chassis to allow it to measure ambient air instead of air samples of a controlled

environment. As the data is collected by the sensors, the GPS coordinates is

gathered via the Google Maps API embedded in the NodeMCU code which also

simultaneously tracks the location of the device tailored with a timestamp for every

data collected. Even with the sensors using mostly serial Rx/Tx connections, there is

only 1 microcontroller within the covered area of the chassis. After data collection,

the microcontroller will then apply the necessary computations using the calibration

method for quality control of data and delivers the data via the use of the Mobile Wi-

Fi device in the external part of the case or saves it offline using the NodeMCU cache

memory. To lessen the strain on computing for the NodeMCU the study opts to then

apply the necessary computations using the calibration method for quality control of
 26

data in the backend of the cloud data storage. The device also uses a single air

outlet to remove the previous sample.

Figure 3.3.3: Prototype Area Type Designations

Enclosure Design and Assembly

The actual built prototype as seen in Figure 3.3.-4, employs the designations

provided in the previous figure. With this, the casing used is a pre-built plastic casing

so modifications to the design was limited. The sensors and microcontroller are also

only attached using Velcro for easy debugging and removal in case of necessary

adjustments. The prototype also employs the use of an external battery source or

powerbrick and an external placement for the mobile wifi device to provide better

internet connection and reception.

 27

Figure 3.3.4: Actual Built Prototype

Selection of Sensors

In the selection of the sensors to be used for the specific air pollutants

mentioned, the comparison factor was brought down to the costing and quality of

data. The possible solution for a low cost monitor is the use of the "MQ" series

sensors that offer substandard quality data. The data offered by "MQ" sensors

produce non linear output of graphed data, which makes the calibration of the curve

to be difficult. While a possible solution for the preservation of quality of data was to

use sensors that are branded such as the SPEC Ozone and Nova SDS-011 fine dust

sensors.

For the preservation of the quality of data, the Ozone sensor selected is the

SPEC Digital Ozone Gas Sensor (DGS-O3 968-042), as seen in Figure 3.3-5, with a

measurement range of 0 to 5 ppm (the normal ozone level in the lower atmosphere is

generally 0.04ppm) [23]. The SPEC digital series line offers the ability to easily

integrate with most microcontrollers. The DGS-O3 968-042 consumes only 14mW

while in use and 100μW on standby [30].

 28

Figure 3.3.5: SPEC DGS-O3 968-042 Ozone(O3) Sensor

In line with the preservation of quality of data, the Fine Dust sensor selected

is the Nova SDS-011, as seen in Figure 3.3-6, which offers a serial input of data via a

Tx/Rx pin configuration which is also common for most microcontrollers. The SDS-

011 sensor offers the ability to output PM2.5 and PM10 which makes it a preferred

choice for a sensor. The measurement range of the SDS-011 sensor is from 0 – 999

μ/Ncm3 (good Particulate Matter levels range from 0 – 20 only) [32] [33].

Figure 3.3.6: Nova SDS - 011 Fine Dust Sensor

Selection of Microcontroller

For selecting the microcontroller, the factors to be considered are the

available I/O options and pin configurations as the computations are simple and

require only minimal processing capabilities. The NodeMCU V3 ESP8266 ESP-12E,

as seen in Figure 3.3-7, is the selected microcontroller for the study as it is of a small

form factor, and it uses only minimal power consumption. The NodeMCU V3 also has
 29

the necessary amount of digital input pins for the totality of the system and is able to

process the minimal computations needed by the study. Furthermore, there will be

three of the microcontrollers in order to accommodate all the serial Rx/Tx pins of the

sensors, which will still prove to be smaller space consumption than that of other

arduino modules [34].

Figure 3.3.7: NodeMCU V3 ESP8266 ESP-12E Microcontroller

1.4 Software Design

This subsection discusses the design of the software application to be

implemented in the study. This will also discuss the necessary programming tool to

be implemented in achieving the Web application software.

 30

Calibration Method

The calibration method of the different sensors will be applied respectively to

each. Simple Linear Regression is the calibration method used and will be computed

by side-by-side data gathering between the device and respective online reference

sensors. These online reference sensors are from the Canisius Bldg. PM2.5 sensor

in AdDU Jacinto Campus and the DENR PM2.5 and PM10 sensor located in Tagum

near the National Trade School. The calibration factors will be computed by

comparing the line of best fit with the raw data from each respective entry. This

comparison will return a difference which is also averaged for all instances to arrive

at a mean correction factor that adjusts the data based on the reference sensor being

used. For the Ozone sensor, the factory calibration is the main method used and will

be followed as the sensor has a correction factor coefficient set for each sensor

manufactured. The sensitivity factor printed on the sensor barcode is pre-configured

in the memory of the DGS2 module from the factory according to the datasheet. [35].

With this, the computation for the corrected values will no longer be attached to the

microcontroller as it is near the limits of its on-board memory cache. The computation

for the AQI and correction factors will now be done in the backed of the Web App

database since it is only minor computations.

Database Creation

The implementation of the database is done by using Google Sheets to

function as the main data source for the Web Application. This is accomplished using

Google Script which allows for a more efficient method of to send an ‘HTTP Post’

request and appending values to a Google Sheet for easier data manipulation.

Main Web Page

The main webpage is a paid domain setup by using Google Sites and

publishing it onto a custom domain that is bought from a legitimate domain provider.
 31

The Google Sites backend will contain all the data sources from the Google Sheets,

Google Scripts, and Looker Studio and this implementation already removes the

need for a running subscription for a database implementation. This also removes

the need for adjusting data by changing multiple data representations for the web

application since the previous implementation for Real time Database to immediately

show data as it is uploaded but also requires a cloud Database to store data for

further reference or representation. With the implementation of a Google Scripts

workaround the real time data is easily represented along with the store data which

allows for much faster navigation when sorting through data collections.

Figure 3.4.1: Main Page View of the Web Application

The desktop browser view of the Main Page of the website, “Tigair.online” is a

simple webpage that contains basic information about the study, such as the

objectives, AQI breakpoints for reference of the AQI values in the tables to follow as

well. The main page is also where you will be able to navigate through other pages

such as the Mobile Tracking Heatmap Page and other pages that have info regarding

the project.

The tabular data is the summary of information per entry that includes the

date and time of each entry, the corrected sensor values for each respective
 32

gas/pollutant monitored, the AQI values of each of the sensor and also the GPS

Coordinates of the entry. The Summary of Data for Stationary Monitoring in Figure

3.4.2 can be accessed and viewed on the main page of the website.

Figure 3.4.2: Table of Summary of Data for Stationary Monitoring

The tabular data is the summary of information per entry that includes the

date and time of each entry, the corrected sensor values for each respective

gas/pollutant monitored, the AQI values of each of the sensor and also the GPS

Coordinates of the entry. The Summary of Data for Stationary Monitoring in Figure

3.4.2 can be accessed and viewed on the main page of the website.
 33

Figure 3.4.3: Google Maps API Selective Location Representation

The Google Maps API is a mapped data representation that points to a

specific table value selected that can also select multiple table values at the same

time.

Figure 3.4.4: Table of Summary of Data for Mobile Monitoring

The next data is the same summary of information per entry that includes the

date and time of each entry, the corrected sensor values for each respective

gas/pollutant monitored, the AQI values of each of the sensor and also the GPS

Coordinates of the entry. The Summary of Data for Mobile Monitoring in Figure 3.4.4

can be accessed and viewed on the Mobile Tracking Heat Map page of the website.
 34

Figure 3.4.5: Mobile AQI Monitoring Heat Map Representation

The Mobile monitoring heat map as seen in Figure 3.4.5 is the Data

representation of the tabular data in Figure 3.4.4. The heat map is also configurable

by selected single or multiple data entries from the table above. This allows anyone

to manipulate the mapped data to view specific moments as the sensor travels one

location to another.

1.5 Research Instruments and Methods

Air Quality Index General Equation

The study employs the use of the General Air Quality Index Formula of the

Particulate Matter 2.5 Microns, 10 Microns, and Ozone(O 3) AQI. The general AQI

formula is as follows:
 35

I Hi−I Lo
I p= ( C −BP Lo ) + I Lo
BP Hi−BPLo p

Equation 3.5.1: General Air Quality Index (AQI) Formula

Where:

Ip = AQI value for the pollutant

Cp = Pollutant Concentration

BPHi = Breakpoint ≥ Cp

BPLo = Breakpoint ≤ Cp

IHi = AQI value corresponding to BPHi

ILo = AQI value corresponding to BPLo

 36

AQI Breakpoints for Specific Gases/Pollutants

Table 3.5.1 Particulate Matter 2.5 Microns AQI Breakpoints Table

Table 3.5.1 Particulate Matter 2.5 Microns AQI Breakpoints Table

Table 3.5.2: Particulate Matter 10 Microns AQI Breakpoints Table

Table 3.5.3: Ozone AQI Breakpoints Table

 37

Linear Regression Formula

The calibration method is Linear Regression formula as the study is able to

access reference data from the DENR EMB R-XI air quality monitoring stations. This

entails that the raw data produced by the prototype air pollution monitor will be able

to adjust to the line of best fit with regards to the specific correction factor derived

using the linear regression formula in Equation . The simple linear regression line

formula is as follows:

Y =mx+b

Equation 3.5.2: Linear Regression Formula

Where:

Y = the value along the vertical axis

x = the value along the horizontal axis

b = the y-intercept, when x = 0

m = the slope of the curve

Percent Error Formula

The percent error formula is to determine the error of the prototype data from

the reference data from the DENR EMB R-XI and AdDU air quality monitoring

sensors. The results should be shown as both with reference sensor/s and

prototype’s sensors with a comparison of which is to be more applicable for use. The

percent error formula is as follows:

Calculated −Accepted
percent error = x 100
Accepted

Equation 3.5.3: Percent Error Formula

 38

Where:

Percent error = the value in percentage of error

Calculated = the calculated value from sensor readings

Accepted = the reference value from reference sensors

1.6 Research Evaluation and Calibration

Upon completion of development procedures, the sensor is calibrated in

unison with the DENR EMB R-XI and AdDU air quality monitoring stations. The

prototype station and the reference station collects data simultaneously and located

in close proximity to produce precise values between each other. The accuracy on

the other hand, will be determined via the calibration method selected which uses

both reference and prototype values and is graphed using the Linear Regression

formula to account for the line of best fit that also provides the correction factor for

the prototype’s value calibration [36].

1.6.1 Sensor Calibration

The sensor calibration procedure has been conducted by side-by-side testing

so as to provide a linear correlation for the gas or pollutant readings with respect to

time. The linear regression charts are also shown for the Prototype, AdDU and

DENR monitoring stations and labelled respectively.

 39

Particulate Matter 2.5 Microns Sensor Calibration

Curve Calibration Using DENR PM2.5 Station

Table 3.6.1: Linear Regression Data for Raw Data vs DENR PM2.5 Calibration

% Error Raw PM
Raw DENR DATA
Date Time 2.5
PM2.5 PM2.5
vs DENR PM2.5
5:11:06
4/3/2024 PM 9.99 15.9 37.16981132
4:11:25
4/3/2024 PM 10.45 16 34.6875
3:11:44
4/3/2024 PM 3.81 12.3 69.02439024
Average % Error: 46.96056719

The tabular data gathering conducted in the Tagum Location of the DENR PM

2.5 sensor is seen in Table 3.6.1. The tabular data has been graphed and the curve

of the Raw Data has been determined to be Y = 3.090*X - 4.277, while the curve of

the DENR sensor is determined to be Y = 1.800*X + 7.533. Using these curves, we

are able to plot them and use Linear Regression to determine the Line of Best Fit,

with the curve of Y = 1.7922*X + 6.723. The Table 3.6.2 shows the values for the

Line of Best fit and solving for the correction factor by simply computing a value

difference against the Raw data readings. The average value difference is now

determined to be the correction factor of the Raw PM 2.5 data. The percent

difference of the corrected values and the DENR values are also shown in the last

column of Table 3.6.2. With this, the percentage error using the correction value of

approximately 5.80 μg/Nm3 is only approximately 6.94%.

 40

Table 3.6.2: Line of Best Fit DENR PM 2.5 Sensor with Corrected values

% Difference
Raw Line of Value Difference Corrected
Corrected
Date Time PM2. Best Fit Line of Best Fit vs Value
PM2.5
5 PM2.5 Raw Data PM 2.5 vs DENR
PM2.5
0.63857442
4/3/2024 5:11:06 PM 9.99 15.684 5.694 15.79846667
35
-
4/3/2024 4:11:25 PM 10.45 13.8918 3.4418 16.25846667 1.61541666
7
21.8010840
4/3/2024 3:11:44 PM 3.81 12.0996 8.2896 9.618466667
1
Correction Average % 6.94141392
5.808466667
Factor: Error: 3

Using the tabular data, below are the graphical figures during this calibration

process. Figure 3.6.1 shows the Curve of the Line of Best Fit and the PM 2.5

sensors. The Figure 3.6.2, includes the corrected values for the prototype’s PM2.5

sensor. A comparison of both AdDU and DENR sensor is also used due to the

number of data reference points.

Figure 3.6.1: Linear Regression of Line of Best Fit (DENR PM 2.5)

 41

Figure 3.6.2: Corrected PM 2.5 Curve vs DENR PM 2.5

Curve Calibration Using AdDU PM2.5 Station

Alongside the completed calibration for PM 2.5 using the DENR station, there

is also a calibration curve using the AdDU monitoring station. The tabular data in

Table 3.6.3 allows for more accurate calibration due to having more data entries as

the prototype station was able to run for a much longer duration. Although having run

for a longer duration, the raw percentage error is approximately 46.78%.

Table 3.6.3: Linear Regression Data for Raw Data vs AdDU PM2.5 Calibration

Date Time Raw PM2.5 ADDU PM2.5 % error

4/13/2024 1:41:14 AM 5.69 8 28.875
4/13/2024 12:41:33 AM 3.3 10 67
4/12/2024 11:41:52 PM 4.99 16 68.8125
4/12/2024 10:42:10 PM 9.04 19 52.42105263
4/12/2024 9:42:12 PM 10.92 32 65.875
4/12/2024 8:42:30 PM 22.7 31 26.77419355
4/12/2024 7:42:49 PM 16.49 22 25.04545455
4/12/2024 6:43:07 PM 11.7 14 16.42857143
4/12/2024 5:43:26 PM 7.04 13 45.84615385
4/12/2024 4:43:43 PM 5.43 9 39.66666667
4/12/2024 3:14:41 PM 3.26 7 53.42857143
4/12/2024 2:14:58 PM 2.92 7 58.28571429
 42

4/12/2024 1:15:16 PM 2.82 7 59.71428571

Average % Error: 46.78255108

The tabular data gathered with side-by-side testing using the AdDU PM2.5

sensor, located at 2nd Floor Canisius Bldg., and the prototype has the curve values of

Y = 0.3057*X + 6.037 and Y = 0.7802*X + 9.538 respectively. Using both curves,

Linear Regression is used again to determine the Line of Best Fit. The line of best fit

for this data is determined to be Y = 0.7810635*X + 9.037. Using this curve, we are

able to determine values for the correctional factor and apply it to the Raw Data for

that is seen in Table 3.6.4. With the calculations done, the correction factor for this

set of data is approximately 6.32 μg/Nm3 and the percentage error of the corrected

values vs the reference AdDU sensor values is only approximately 8.57%.

Table 3.6.4: Line of Best Fit AdDU PM 2.5 Sensor with Corrected values

Value % Difference
Line of Best Corrected
Date Time Difference vs Corrected Value
Fit Value PM2.5
Raw Data vs Addu PM2.5
4/13/2024 1:41:14 AM 19.1908255 13.5008255 12.01752142 50.21901779
4/13/2024 12:41:33 AM 18.409762 15.109762 9.627521423 -3.72478577
4/12/2024 11:41:52 PM 17.6286985 12.6386985 11.31752142 -29.26549111
4/12/2024 10:42:10 PM 16.847635 7.807635 15.36752142 -19.1183083
4/12/2024 9:42:12 PM 16.0665715 5.1465715 17.24752142 -46.10149555
4/12/2024 8:42:30 PM 15.285508 -7.414492 29.02752142 -6.362834119
4/12/2024 7:42:49 PM 14.5044445 -1.9855555 22.81752142 3.716006468
4/12/2024 6:43:07 PM 13.723381 2.023381 18.02752142 28.76801016
4/12/2024 5:43:26 PM 12.9423175 5.9023175 13.36752142 2.827087869
4/12/2024 4:43:43 PM 12.161254 6.731254 11.75752142 30.63912692
4/12/2024 3:14:41 PM 11.3801905 8.1201905 9.587521423 36.96459176
4/12/2024 2:14:58 PM 10.599127 7.679127 9.247521423 32.1074489
4/12/2024 1:15:16 PM 9.8180635 6.9980635 9.147521423 30.67887747
Correction Average %
Factor: 6.327521423 Error: 8.565173269

The curves plotted in Figure 3.6.3 include the Raw PM2.5, AdDU PM2.5 and

Line of Best fit lines. These indicate a correlation that proves the Particulate matter
 43

sensor used by the prototype has a linear curve with respect to the AdDU PM 2.5

station. With this, the Figure 3.6.4 shows that the corrected values are more closely

alike for the Line of Best fit curve and will give a more accurate reading when

deployed.

Figure 3.6.3: Linear Regression of Line of Best Fit (AdDU PM 2.5)

Figure 3.6.4: Corrected PM 2.5 Curve vs AdDU PM 2.5

 44

Particulate Matter 10 Microns Sensor Calibration

Using the same calibration method for PM 10 readings, there is only the

DENR PM 10 reference readings as this is the only available online reference sensor

in the within the scope the study. The Table 3.6.5 shows the Raw PM 10, DENR PM

10, and percent error of the data set. With this, the curve of the Raw PM 10 and

DENR PM 10 appear to be Y = 5.670*X - 2.473 and Y = 1.800*X + 7.533,

respectively. The percentage error of the Raw PM 10 data with respect to the DENR

PM 10 data is approximately 59.40%, and must have a correction factor to reduce

this.

% Error Raw PM 10
Date Time Raw PM10 DENR DATA PM10
vs DENR PM10
4/3/2024 5:11:06 PM 24.73 68.3 63.7920937
4/3/2024 4:11:25 PM 22.5 43.6 48.39449541
4/3/2024 3:11:44 PM 13.39 39.4 66.01522843
Average % Error: 59.40060585

Table 3.6.5: Linear Regression Data for Raw PM10 vs DENR PM10 Calibration

Using both curves, Linear Regression is still used to determine the Line of

Best Fit. The Table 3.6.6 employs this and shows that value for the correction factor

to be approximately 29.29 μg/Nm3. After applying the correction factor to the Raw

PM10 values, the percentage error is only approximately 2.07%. With this, the

prototype sensor for PM10 is now calibrated with a linear slope curve.

% Difference
Line of Best Value Difference vs Corrected
Date Time Corrected Value
Fit PM10 Raw Data PM 10 PM10
vs DENR PM10
4/3/2024 5:11:06 PM 63.675 38.945 54.02333333 20.90287945
4/3/2024 4:11:25 PM 49.5 27 51.79333333 -18.79204893
4/3/2024 3:11:44 PM 35.325 21.935 42.68333333 -8.333333333
Correction Average %
29.29333333 -2.074167603
Factor: Error:

Table 3.6.6: Line of Best Fit AdDU PM 10 Sensor with Corrected values
 45

The curves plotted in Figure include the Raw PM10, DENR PM10 and Line of

Best fit lines. These indicate a correlation that proves the Particulate matter sensor

used by the prototype has a linear curve with respect to the DENR PM 10 station.

With this, the Figure 3.6.5 shows that the corrected value curve is now more closely

related to the DENR reference curve and is now fit for data gathering.

Figure 3.6.5: Linear Regression of Line of Best Fit (AdDU PM 10)

Figure 3.6.6: Corrected PM 10 Curve vs AdDU PM 10

 46

Ozone Sensor Calibration

For the calibration of the Ozone sensor(DGS-O3 968-042), the datasheet

refers to each unit produced as being factory calibrated and includes a QR code that

contains a string text which includes a correction factor that is unique for the specific

unit sensor. This enabled the sensor to calculate ozone readings without

supplementary calibration which is ideal for research since it will remove the need to

do further testing on the sensor values [35]. The Figure 3.6.7 shows the actual

barcode of the sensor and scanning this will return the string “020420011529 110406

O3 2002 -41.41” which is a long form string that indicates the Serial Number, Part

Number, Type of Sensor, Date Manufactured(Day, Month), and sensitivity code. This

gives us the correcting factor indicated as the sensitivity adjustment value of -41.41.

Figure 3.6.7: Ozone Sensor Barcode

1.6.2 White box testing

The proponent used white box testing to verify the validity and reliability of the

uploaded code on to the prototype Air Quality Monitoring station.

 47

1.6.2.1 Air Quality Monitor Testing

The Air Quality Monitor Testing includes the validation of the methods

readPM25(), readPM10() and readO3() will collect the gas or pollutant data and add

the entry to the NodeMCU’s cache memory for a running average computation. After

the primary average raw data is collected the method requestGPSData() is now

called and will request GPS coordinates using the embedded Google Maps API

script to return and add the GPS data to a string builder for preparation. The

buildString() method uses concatenation to make a single line string separated by

commas in order to be sent as a HTTP Post request using the available internet

connection string. Google Script has implemented the use of an SSL handshake that

requires a ‘fingerprint’ that will authenticate the HTTP Post and allow the

sendHTTPPost() method to append the built string value in order to populate the

Google Sheet and function as the backend Database of the Website.

1.6.2.2 Web Application Testing

For the Web application testing, the Posted data from the microcontroller is

now processed by the backend, which is Google Scripts, and recalculated using

Looker Studio for the application of the correction factor and AQI computations.

Unloading these computations from the microcontroller to the backend allows for

more cache memory on the microcontroller for possible losses in internet

connectivity. From the Database in Google Sheets, the new data is now processed

with Looker Studio calculated fields in order to display the corrected data and AQI

values onto the web application which employs the use of Looker Studio to embed

the Tabular and Mapped data with ease.

1.6.3 Black Box testing

The study employs the use of black box testing to verify the functionality and

external behavior of the system.

 48

1.6.3.1 Air Quality Monitor Testing

The tests cases that will verify the functionality of the Air Quality Monitor,

along with the components such as the internet signal availability, cache memory

data retention, and the functionality to upload the data in succession.

1.6.3.2 Web Application testing

The test cases that will verify the usability of the website, accessibility of the

web application from other devices or locations and the functionality of the data

representation along with the manipulation to view specific data selections.

1.7 Research Instrument

The results of the real-time air pollution monitor are to be validated by the

DENR EMB R-XI. The validation must include the objectives of the study and provide

comments on the ability of the proposed air pollution monitor. The validation also

includes the real-time capabilities of the device in comparison to the reference air

quality monitoring stations.

The validation employs the use of the Four - Point Likert Scale in a conducted

survey. This will evaluate the system in a relevant and fair method to further validate

the system. The proponent conducted user surveys for the Air Quality Monitor and

the web application by allowing the users to access and use the website.

Four - Point Likert Scale

The Likert Scale will serve as the method for evaluating the functionality and

design of the prototype air pollution monitor. The four - point Likert scale will force the

respondent to make a choice when the respondent has no opinion. A sample four -

point Likert Scale is seen in Table 3.7.1.

Very Good 4

Good 3
 49

Fair 2

Needs Improvement 1

Table 3.7.1: Four - Point Likert Scale

Mean Averaging Formula

The Mean Averaging formula will be used to summarize the survey scores

from the Likert Scale. This will offer a numerical representation of the an rating that

the system and the different objectives it aims to resolve. The study employs an

hourly running average is the preferred data value [28]. The formula is as follows:

∑ ❑ Xi
μ= i=1
n

Equation 3.7.1: Mean Averaging Formula

Where:

n = the total number of data entries

X = the value for every entry

μ = mean value
 50

Figure 3.7.1 Survey Questionnaire first page

 51

Figure 3.7.2 Survey Questionnaire second page

 CHAPTER IV

RESULTS AND DISCUSSIONS

This chapter will provide all the results and discuss the implication of these

towards the functionality of the Hardware and Software side of the study. The study

also discuss the development and testing of the final prototype.

1 Hidden marker

1.1 Development and Testing Results

1.1.1 White Box Testing for Air Quality Monitor

Test Method Expected Actual

Test Case Test steps
Scenario name result results

Step 1:
Open
Arduino IDE

The data Step 2: Init() &

Microcontroller Connect the Prints
should be
initialization & NodeMCU WifiTest( “Connected”
displayed in PASS
Wifi availability V3 in serial
serial )
test monitor
monitor
Step 3:
Upload the
code and
open the
serial
monitor

Table 4.1.1 Test Case 1

 53

Test Method Expected Actual

Test Case Test steps
Scenario name/s result/s results

Step 1: Open
Arduino IDE
PM and Showing
Ozone The data Step 2: the gas or
sensor should be Connect the readPM25(), pollutant
Initialization displayed NodeMCU V3 readPM10() average PASS
and Data in serial & readO3() data in
Step 3: Upload
Acquisition monitor serial
the code and
test monitor
open the serial
monitor

Table 4.1.2 Test Case 2

Test Expected Actual

Test Case Test steps Method name
Scenario result results

Step 1:
Open
Arduino IDE
Receives
Tests GPS The data Step 2: and prints
Acquisition should be Connect the the value
Data from displayed NodeMCU requestGPSData() entry PASS
Google in serial V3 under
Maps API monitor Step 3: cache
Upload the memory
code and
open the
serial
monitor

Table 4.1.3 Test Case 3

 54

Test Method Expected Actual

Test Case Test steps
Scenario name result results

Step 1:
Open
Arduino IDE
Prints the
Step 2:
string in full
Building the The data Connect the
on serial
String of should be NodeMCU
monitor and
data for displayed V3 buildString() PASS
saves the
Cache in serial
Step 3: value entry
memory test monitor
Upload the under cache
code and memory
open the
serial
monitor

Table 4.1.4 Test Case 4

Test Test Expected Actual

Test steps Method name
Case Scenario result results

Step 1:
The code
Open
should be The serial
Arduino IDE
able to monitor will
connect to Step 2: print
the Google Connect the “Published”
sendHTTPPost() PASS
Script cloud NodeMCU when string
database V3 is appended
and store a to the
Step 3:
built string database
Upload the
HTTPS data into it
code
Post Test
Step 1:
The code is The serial
Open
unable to monitor will
Arduino IDE
connect to print
the Google Step 2: “Connection
Script cloud Connect the sendHTTPPost() Error” and PASS
database NodeMCU the Data
and store a V3 string is left
built string in the cache
data into it Step 3: memory for
Upload the the next
 55

code
turnaround
HTTP Post

Table 4.1.5 Test Case 5

1.1.2 White Box Testing for Web Application

Test Test Pre- Expected Actual

Test steps
Case Scenario condition result results

The Air Step 1: Open The

Quality the Google Google
Monitor is Sheet Script for
powered on the Sheet
and has should
Verify the collected Step 2: Verify allow the PASS
functionality hourly or 15 if the Data is data string
Air Quality of the Air minute Appended on to be
Monitor Quality running the first row appended
Data Monitor to average data on the
Posting & successfully readings database
Website Post data to
Step 3: Open
Display the database
Google
Test and display it The HTTP The data
Looker Studio
on the Post request should be
and check for
website. is successful, updated
data PASS
and the freshness and
website is displayed
online Step 4: at website
Access
Website

Table 4.1.6 Test Case 6

Test Pre- Actual

Test Case Test steps Expected result
Scenario condition results

Verify Data Confirm The The gas or

Displayed on the Data Google Step 1: pollutant average
Web App from the Sheet was Access the data is displayed PASS
Google able to website correctly on the
is
Sheet is store the Website UI
synchronized
displayed specified
with Google Step 2: Verify The GPS PASS
 56

data entry
The if the data
Google visible on the
Script and Coordinate data
website UI
Looker is displayed
corresponds
Sheet Data correctly Studio data correctly on the
with the Data
on the Web source is Website UI
from Google
Application active, and Sheets
website is
online

Table 4.1.7 Test Case 7

Expected Actual
Test Case Test Scenario Pre-condition Test steps
result results

Verify Raw Data Confirm the Correction The Data PASS

Displayed on Data from the Factor is Displayed
Step 1:
Google Sheet Google Sheet applied using on the
Open the
is to get the Google Looker Website
is recalculated Website UI
AQI values Studio and AQI Interface
by Looker Studio and Acess
and Corrected Computations is the
specific
Sensor have been Corrected
data point
Readings by assigned to sensor
applying the each sensor values
Correction reading and the
Step 2:
Factors AQI
Verify that
values are
the values
accurately
are already
computed
recalculate
d and AQI
value is
also
displayed
Step 3:
Confirm
the Data is
valid by
performing
calculation
manually
and cross-
referencing
using
Google
 57

Sheets raw
data

Table 4.1.8 Test Case 11

1.1.3 Black Box Testing for Smart Node Sensor

Test Pre- Expected Actual

Test Case Test steps
Scenario condition result results

Display Air Confirms the The Air Step 1: Plug The PASS
Quality successful Quality the Air Google
Monitoring execution of Monitor is Quality Sheet will
Functionality the read being Monitor to produce a
method, string powered any 5V new line of
builder continuously source via Data set
method, and with no the Micro that should
data cache power USB input be
memory source port complete
method issues with Raw
Step 2:
Sensor
Verify that
data, GPS
the LED on
data and
the
time
microcontroll
stamps
er is blinking
to indicate
Wifi
connectivity
 58

Step 3:
Verify that
the PM
sensor LED
is
continuously
blinking as
this is
averaging a
value for
every minute
of the hour

Table 4.1.9 Test Case 1

1.1.4 Black Box Testing for Web Application

Test Pre- Expected Actual

Test Case Test steps
Scenario condition result results

Assess the The web-

Usability of The Step 1: based
Web
the monitoring website is Access the application
Application
interface in online and monitoring monitoring
User
the web all the interface of interface PASS
Interface
interface to embedded the web- should
Usability
for Data systems based display the
Test
representatio are stable application data from the
n of calibrated smart node
 59

Step 2:
Interact with
the interface
to navigate
through
different
monitoring
views
sensor values
and AQI sensor
Step 3:
computations
Simulate
Relevant data
sensor changes to
data are observe the
integrated reflected
with the updates on
application the interface

Table 4.1.10 Test Case 2

1.2 Data Gathering and Analysis

The deployment of data gathering as determined in the research locale

section is in testing of two deployment types, Mobile and Stationary functionalities.

The Mobile Route is displayed as the Roxas Avenue mobile route which takes 15-

minute running average readings that also includes readings as the monitor is being

transported via walking from one GPS location to another. The Stationary location is

set to be the 2nd Floor Canisius Bldg just beside the PM 2.5 Air Quality monitor of the

AdDU in order to provide data for PM10 and Ozone and additionally compare more

data for PM2.5 after correction factors are applied.

 60

1.2.1 Stationary Monitoring Data Gathering

The stationary data gathering was conducted for multiple days as the initial

calibration data collection, done on April 1st, was unable to collect AdDU data from

their online monitoring website due to the history accessible only being 24hrs.

However, this data is still legitimate because on the 2 nd calibration data collection,

conducted on April 12th, the correctional factor for the PM2.5 sensor is now computed

properly and has a small percentage error with respect to the AdDU sensor readings.

The Table 4.2.1 shows the corrected data readings for the prototype air pollution

monitor data readings, the AQI value and GPS coordinates for April 1 st data

collection. The Figure 4.2.1 also supplies a visual mapped representation of the GPS

locations to help give clarity on locations of where each respective entry value is.

Hourly PM2.5 Hourly PM10 Ozone Ozone

Date Time GPS Coordinates
PM2.5 AQI PM10 AQI (Hour) AQI
Apr 2, 10:56:31
11.18 22 42.74 40 43.67 47 7.0722957,125.6132584
2024 PM
Apr 2, 9:56:50
11.81 24 42.31 39 43.88 47 7.0723009,125.613266
2024 PM
Apr 2, 8:57:08
14.75 29 46.37 43 43.68 47 7.0723009,125.613266
2024 PM
Apr 2, 7:57:26
17.16 34 48.37 45 43.53 47 7.0722561,125.6133041
2024 PM
Apr 2, 6:57:43
21.29 43 53.04 49 43.43 47 7.0723095,125.6132736
2024 PM
Apr 2, 5:58:00
12.76 26 43.47 40 43.6 47 7.0722179,125.6132584
2024 PM
Apr 2, 4:58:19
11.18 22 42.14 39 43.97 47 7.0722919,125.613266
2024 PM
Apr 2, 2:58:49
10.34 21 39.14 36 43.07 46 7.0723019,125.6132736
2024 PM
Apr 2, 1:56:40
10.20 20 37.41 35 43.83 47 7.0723047,125.613266
2024 PM
Apr 1, 11:16:47
10.70 21 41.51 38 43.3 46 7.0720248,125.6133041
2024 PM
Apr 1, 9:25:41
17.95 36 52.32 48 43.57 47 7.0718598,125.6133347
2024 PM
Apr 1, 4:10:31
10.04 20 40.04 37 43.6 47 7.0720038,125.613205
2024 PM

Table 4.2.1: Stationary Monitoring April 1st Data Collection

 61

Figure 4.2.1: Stationary Monitoring April 1st Data Map

The April 12th data gathering set is shown in Table 4.2.2, the table shows the

corrected data readings for the prototype air pollution monitor data readings, the AQI

value and GPS coordinates for April 12 th data collection. The Figure 4.2.2 also

supplies a visual mapped representation of the GPS locations to help give clarity on

locations of where each respective entry value is.

Hourly PM2.5 Hourly PM10 Ozone Ozone

Date Time GPS Coordinates
PM2.5 AQI PM10 AQI (Hour) AQI
Apr 13, 1:41:14
12.02 24 41.14 38 43.13 46 7.0722184,125.6133041
2024 AM
Apr 13, 12:41:3
9.63 19 38.05 35 43.93 47 7.0722303,125.6132584
2024 3 AM
Apr 12, 11:41:5
11.32 23 40.97 38 43.7 47 7.0722699,125.6132507
2024 2 PM
Apr 12, 10:42:1
15.37 31 47.61 44 43.6 47 7.0722694,125.6132813
2024 0 PM
Apr 12, 9:42:12
17.25 34 50.00 46 44.08 48 7.0722575,125.6132736
2024 PM
Apr 12, 8:42:30
29.03 58 64.11 59 43.63 47 7.0722518,125.6132736
2024 PM
Apr 12, 7:42:49
22.82 46 54.81 51 43.55 47 7.0722499,125.6132736
2024 PM
Apr 12, 6:43:07
18.03 36 48.15 45 43.32 46 7.0722799,125.6132736
2024 PM
Apr 12, 5:43:26
13.37 27 43.13 40 43.37 47 7.0722699,125.6132889
2024 PM
Apr 12, 4:43:43
11.76 24 41.45 38 43.58 47 7.0722504,125.6133118
2024 PM
Apr 12, 3:14:41
9.59 19 37.51 35 43.02 46 7.0722866,125.6132813
2024 PM
Apr 12, 2:14:58
9.25 18 37.98 35 42.82 46 7.072206,125.6132126
2024 PM
 62

Apr 12, 1:15:16

9.15 18 36.47 34 43.8 47 7.072268,125.6132813
2024 PM

Table 4.2.2: Stationary Monitoring April 12th Data Collection

Figure 4.2.2: Stationary Monitoring April 12th Data Map

 63

1.2.2 Stationary Monitoring Data Analysis

With the gathered data from the two days of data collection for stationary

monitoring, the study is able to provide the stable and informative Gas or Pollutant

readings that offer a bare view of the Air Quality for the location and the specific time

slots. Along with this data, the Stationary Air Quality Monitor also provides the Air

Quality Index value per entry, this offers a supplementary representation via the use

of colored breakpoints to interpret the values in a simple manner instead of unit

values from sensor readings. The Google Mapping solution also provides a visual

representation of the location of the readings and their respective AQI values. The

comparison of the interval between April 1st and 12th time stamps are also seen that

the April 1st data was unable to append some values due to an inability to connect to

the main Wifi network. The reason for this is the Cellular Data used on the April 1 st

was from a Globe network sim card while the April 12th is now using a Smart network

sim card. This is due to the cellular connectivity is better with a Smart network rather

than the Globe network for the specific downtown area location. Overall, the system

is still able to supply data that can be presented since this data is corrected by the

correction factor supplied from the calibration procedures in Chapter 3.

1.2.3 Mobile Data Gathering

The mobile data monitoring was conducted on April 13 th along Roxas Avenue

and the mobile route indicated in the research locale. The Table 4.2.3 shows the data

readings for 15 minute running averages of the Gas and Pollutants tracked along

with their respective AQI values and GPS coordinates. The Heat Map in Figure 4.2.3,

aims to provide a supplementary data representation and a heat map is opted rather

than a simple bubble map as the AQI value will provide a colored representation of

the breakpoints in the form of heat map zones per data entry.
 64

15 Min. PM2.5 15 Min. PM10 15 Min. Ozone

Date Time GPS Coordinates
PM 2.5 AQI PM 10 AQI Ozone AQI
Apr 13, 6:25:14
20.20 40 58.22 54 48.27 54 7.069088,125.614733
2024 PM
Apr 13, 6:10:40
20.42 41 53.69 50 48 54 7.0671206,125.6156921
2024 PM
Apr 13, 5:56:02
14.56 29 47.86 44 48.13 54 7.0688744,125.6144714
2024 PM
Apr 13, 5:20:01
13.02 26 50.90 47 47.07 52 7.0696259,125.6138687
2024 PM
Apr 13, 5:05:20
12.76 26 45.88 42 49.2 55 7.0705137,125.6134338
2024 PM
Apr 13, 4:50:43
13.18 26 51.77 48 48.47 54 7.072093,125.6123886
2024 PM
Apr 13, 4:36:05
14.39 29 49.47 46 48.07 54 7.0715642,125.6127319
2024 PM
Apr 13, 4:21:26
13.38 27 43.69 40 47.07 52 7.0708623,125.6133118
2024 PM
Apr 13, 4:06:49
12.91 26 48.60 45 47.4 53 7.0700965,125.6139679
2024 PM
Apr 13, 3:52:12
11.65 23 41.46 38 47.53 53 7.0689907,125.61483
2024 PM
Apr 13, 3:37:33
11.06 22 42.71 40 47.6 53 7.0683498,125.6153793
2024 PM
Apr 13, 3:22:50
11.26 23 43.82 41 48 54 7.0675755,125.6163025
2024 PM
Apr 13, 2:22:50
11.26 23 43.82 41 48 54 7.0675755,125.6163025
2024 PM
Apr 13, 1:22:50
26.33 53 79.29 73 55 64 7.0675755,125.6163025
2024 PM

Table 4.2.3: Mobile Monitoring April 13th Data Collection

 65

Figure 4.2.3: Mobile Monitoring April 13th Heat Map

[1.2.4] Mobile Monitoring Data Analysis

With the gathered data from Mobile monitoring along the route selected, the

device is able to map and gather accordingly which areas are more likely to have a

slight difference from Good air quality to Moderate or Unhealthy air quality. This

provides the study with information to indicate that the device performs mobile

tracking with the inclusion of a good cellular network in place. In terms of the heat

map in Figure 4.2.3, indicated also are the Air Quality values mapped per data entry

with GPS location. The representation allows the end user to be wary of the possible

locations along the route to avoid during the specific time slot.
 66

Figure 4.2.3: Mobile Monitoring April 13th Heat Map

1.2.4[1.2.5] User Testing and Survey

The study has gathered 9 respondents for the testing and survey. The range

of respondents will vary from DENR Engineers, Ateneo Students, Ateneo Teaching

and Non-Teaching Staff as these are the likely respondents that traverse along

Roxas Avenue, these are sorted in Figure 4.2.4

 67

Figure 4.2.4 Survey Respondents Identifier

 CHAPTER V

CONCLUSIONS AND RECOMMENDATIONS

This chapter will provide a conclusion based of the content provided by the

previous chapters. The chapter will also discuss the recommendations that will

further improve future studies related to Air Quality Monitoring.

1 Hidden marker

1.1 Conclusions

With the development process, calibration procedures, integration of the

system, objectives of the study, data gathering, and analysis of the data the

proponent was able to conclude the following:

● The assembly of a low-cost air pollution device capable of monitoring

Ozone, Particle Matter 2.5 microns, Particle Matter 10 microns (O3,

PM2.5, and PM10), and GPS data has been achieved. Utilizing

components such as the NodeMCU V3 ESP8266 ESP-12E

microcontroller, fine dust sensors for PM2.5 and PM10 microns, and

an O3 sensor, the prototype effectively captures real-time air quality

metrics alongside location information. This also proves that the

prototype is a compact, low power, cheap, and easily replicable

solution for the study.

● The device demonstrates versatility by enabling both stationary and

mobile air quality monitoring of specific pollutants and location data in

real-time. This capability enhances its utility for multiple monitoring

locations and timeslots. Selection of further high traffic areas are also

more plausible.

● The web application developed to visualize air pollutant data in

specific locations within Davao City is accessible via any browser with
 69

an internet connection. The tabular data representation that users can

easily access, select, and interpret air quality information, contributing

to informed decision-making regarding daily commuting and travel.

● Data collected by the air quality monitoring device is securely stored in

a cloud database, with the implementation being a Google Sheet

database. This provided an efficient management and accessibility

method of analysis and reference that can be accessed by the

website. This centralized approach enhances data accessibility and

allows data manipulation for the end user as well.

The culmination of these conclusions signifies a viable advancement in air

quality monitoring, particularly within the context of environmental research in Davao

City. By providing a cost-effective, real-time monitoring solution with robust data

visualization and management capabilities, the study sets a good foundation for

improved understanding and mitigation of air pollution challenges.

1.2 Recommendations

In line with the conclusion of the study, the proponent offer the list of

recommendations to amplify and explore ideas that could prove useful to the next

versions of the Air Quality Monitor. The following recommendations are provided:

● Expansion of the Monitoring Network: The proponent recommends to

increase the number of monitoring stations in densely populated areas

and high-traffic locations to ensure comprehensive coverage of air

quality data.

● Long-Term Monitoring: Establishing long-term monitoring programs

will enable tracking progress towards air quality goals and identifying

areas requiring further intervention this should be as the prototype is

 70

able to address concerns on Cellular Data inactivity and possible

losses in wifi connectivity.

● Integration with Existing Systems: The new monitoring device should

be integrated into existing systems operated by organizations such as

DENR and academic institutions like Ateneo de Davao University

(AdDU) to facilitate collaboration and data sharing.

● Continuous Calibration of Sensors: Regular calibration and validation

procedures should be established to maintain the accuracy and

reliability of the sensors over time.

● Policy Support and Community Engagement: Advocating for policies

such as emission controls and urban planning strategies is essential

to effectively reduce air pollution levels in Davao City. Likewise,

Involving local communities in monitoring programs will foster a sense

of ownership and responsibility for air quality improvement efforts.

 REFERENCES

[1 Health effects of ozone and particle pollution, 2017.

[2 State of Global Air 2019, Health Effects Institute 2019, 2019.

[3 N. Kaur, R. Mahajan, D. Bagai and P. G. Student, Air quality

] monitoring system based on Arduino microcontroller, vol. 5, 2016, pp.

9635-9646.

[4 S. Yadav and H. C. Phuleria, Oxidative Potential of Particulate

] Matter: A Prospective Measure to Assess PM Toxicity, Springer, 2020,

pp. 333-356.

[5 A. Chakraborty, Organic Air Pollutants: Measurement,

] Properties & Control, Springer, 2020, pp. 107-130.

[6 K. Kuklinska, L. Wolska and J. Namiesnik, Air quality policy in

] the U.S. and the EU – a review, vol. 6, 2015, pp. 129 - 137.

[7 W. P. Cuñado, National Air Quality Status Report, Department

] of Environment and Natural Resources, 2019.

[8 S. T. Manuel, Air Quality Management, Department of

] Environment and Natural Resources, 2017.

[9 K. Khalid, H. H. Abdullah, D. Kumar M and others, Get along

] with quantitative research process, International Journal of Research in

Management, 2012.

[1 Calinan District, Davao City Station - Davao City air quality

 72

0] index (AQI), 2020.

[1 S. R. Enigella and H. Shahnasser, Real Time Air Quality

1] Monitoring, pp. 182-185.

[1 S. Abraham and X. Li, A Cost-effective Wireless Sensor

2] Network System for Indoor Air Quality Monitoring Applications., 2014,

pp. 165-171.

[1 D. Mintz, Technical assistance document for the reporting of

3] daily air quality-the air quality index (AQI), 2018.

[1 C.-U. A. Q. Committee, I. J. Commission and others,

4] Canada/United States Air Quality Agreement: Progress report 2014,

International Joint Commission, 2017.

[1 N. E. P. United and E. A. P. Environmental, Environmental

5] effects of ozone depletion and its interactions with climate change:

progress report, 2015., vol. 15, 2016, p. 141.

[1 L. R. F. Henneman, C. Liu, J. A. Mulholland and A. G. Russell,

6] Evaluating the effectiveness of air quality regulations: A review of

accountability studies and frameworks, vol. 67, Taylor & Francis, 2017,

pp. 144-172.

[1 A. Ripoll, M. Viana, M. Padrosa, X. Querol, A. Minutolo, K. M.

7] Hou, J. M. Barcelo-Ordinas and J. Garc\'\ia-Vidal, Testing the

performance of sensors for ozone pollution monitoring in a citizen

science approach, vol. 651, Elsevier, 2019, pp. 1166-1179.

[1 R. W. Gore and D. S. Deshpande, An approach for

8] classification of health risks based on air quality levels, 2017, pp. 58-

61.
 73

[1 S. Fuzzi, U. Baltensperger, K. Carslaw, S. Decesari, H. van der

9] Gon, M. C. Facchini, D. Fowler, I. Koren, B. Langford, U. Lohmann and

others, Particulate matter, air quality and climate: lessons learned and

future needs, vol. 15, Copernicus, 2015, pp. 8217-8299.

[2 J. S. Apte, M. Brauer, A. J. Cohen, M. Ezzati and C. A. Pope III,

0] Ambient PM2. 5 reduces global and regional life expectancy, vol. 5,

ACS Publications, 2018, pp. 546-551.

[2 N. Castell Balaguer, F. R. Dauge, P. Schneider, M. Vogt, U.

1] Lerner, B. Fishbain, D. M. Broday and A. Bartonova, Can commercial

low-cost sensor platforms contribute to air quality monitoring and

exposure estimates?, 2017.

[2 O. Harrop, Air quality assessment and management: A

2] practical guide, CRC Press, 2018.

[2 A. W. Reitze Jr, The National Ambient Air Quality Standards for

3] Ozone, vol. 6, HeinOnline, 2015, p. 420.

[2 B. Bishoi, A. Prakash and V. K. Jain, A comparative study of air

4] quality index based on factor analysis and US-EPA methods for an

urban environment, vol. 9, 台灣氣膠研究學會, 2009, pp. 1-17.

[2 E. Babbie, The practice of social research: Cengage Learning,

5] 2010, pp. 73-85.

[2 S. Chanthakit and C. Rattanapoka, MQTT based air quality

6] monitoring system using node MCU and node-RED, 2018, pp. 1-5.

[2 A. Anyfantis and S. Blionas, Proof of concept apparatus for the

7] design of a simple, low cost, mobile e-nose for real-time victim

localization (human presence) based on indoor air quality monitoring

 74

sensors, vol. 27, Elsevier, 2020, p. 100312.

[2 S. M. Biondi, V. Catania, S. Monteleone and C. Polito, Bus as a

8] sensor: A mobile sensor nodes network for the air quality monitoring,

2017, pp. 272-277.

[2 I. Gruicin and M. Popa, Airify—A mobile solution for monitoring

9] air quality in urban areas, 2017, pp. 1-4.

[3 N. Dam, A. Ricketts, B. Catlett and J. Henriques, Wearable

0] sensors for analyzing personal exposure to air pollution, 2017, pp. 1-4.

[3 L. Spinelle, M. Gerboles, M. G. Villani, M. Aleixandre and F.

1] Bonavitacola, Field calibration of a cluster of low-cost available

sensors for air quality monitoring. Part A: Ozone and nitrogen dioxide,

vol. 215, Elsevier, 2015, pp. 249-257.

[3 S. den Elshout, K. Léger and H. Heich, CAQI common air

2] quality index—Update with PM2. 5 and sensitivity analysis, vol. 488,

Elsevier, 2014, pp. 461-468.

[3 M. Budde, T. Müller, B. Laquai, N. Streibl, A. Schwarz, G.

3] Schindler, T. Riedel, M. Beigl and A. Dittler, Suitability of the Low-Cost

SDS011 Particle Sensor for Urban PM-Monitoring, 2018.

[3 D. Thomas, B. Mistry, S. Snow and M. C. Schraefel, Indoor Air

4] Quality Monitoring (IAQ): A Low-Cost Alternative to CO Monitoring in

Comparison to an Industry Standard Device, 2018, pp. 1010-1027.

[3 DGS Ozone Sensor Datasheet, 2023.

5]

[3 J. Venkatraman Jagatha, A. Klausnitzer, M. Chacón-Mateos, B.

6] Laquai, E. Nieuwkoop, P. van der Mark, U. Vogt and C. Schneider,

 75

Calibration Method for Particulate Matter Low-Cost Sensors Used in

Ambient Air Quality Monitoring and Research, vol. 21, 2021.

[3 H. D. Pham, M. Drieberg and C. C. Nguyen, Development of

7] vehicle tracking system using GPS and GSM modem, 2013, pp. 89-94.

[3 M. Badura, P. Batog, A. Drzeniecka-Osiadacz and P. Modzel,

8] Regression methods in the calibration of low-cost sensors for ambient

particulate matter measurements, vol. 1, Springer, 2019, p. 622.

 77

APPENDIX A

DATASHEETS
 78

A.1 NodeMCU V3
 79

A.2 Nova Particulate Matter Sensor

 80

A.3 Spec Sensors Digital Ozone Sensor (DGS-O3 968-042)

 81

APPENDIX B

DOCUMENTATION

B.1 Letter of Intent to DENR EMB XI

B.2 AdDU Letter of Intent
83
 84

APPENDIX C

VALIDATION FOR SURVEY QUESTIONNAIRE

 APPENDIX D

SAMPLES FILLED UP SURVEY

APPENDIX E

GANTT CHART