See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/335864589

Indoor Air Quality Monitoring and Controlling System based on IoT and Fuzzy
Logic

Conference Paper · July 2019

DOI: 10.1109/ICoICT.2019.8835246

CITATIONS READS

43 1,663

2 authors, including:

Nico Surantha
Tokyo City University
88 PUBLICATIONS 621 CITATIONS

SEE PROFILE

All content following this page was uploaded by Nico Surantha on 07 April 2020.

The user has requested enhancement of the downloaded file.

 Indoor Air Quality Monitoring and Controlling
System based on IoT and Fuzzy Logic
Fadli Pradityo Nico Surantha
Computer Science Department, Binus Graduate Program – Computer Science Department, Binus Graduate Program –
Master of Computer Science Master of Computer Science
Bina Nusantara University Bina Nusantara University
Jakarta, Indonesia,11480 Jakarta, Indonesia,11480
fadli.pradityo @binus.ac.id nico.surantha @binus.ac.id

Abstract — Air pollution is one of the biggest health make something smarter than before [6]. IoT system will
challenges in the world. Indoor air pollution is 2-5 times larger monitor the indoor quality of air in real-time [7]. With the IoT,
than outdoors, but people still do not care about it. Currently, problems related to one of the limitations of storage on
people open the window or use an exhaust fan to refresh the air embedded system devices can be solved. Embedded systems
condition inside the room. However, people sometimes are too do not need to use ample storage because all data is stored on
busy to pay attention to the air quality inside the room. cloud servers. Cloud servers can guarantee data integrity, data
Therefore, automation is required to bring the air quality into validity, and data values as a place to store and share data [8].
the required level. This study discussed indoor air quality
Raspberry Pi can also be used as a tool for IoT based projects
monitoring and controlling system that can monitor the air
because of its small form and requires low power. Raspberry
condition and control the air condition using an exhaust fan.
This system used the IoT concept in conducting real-time Pi 3 has built-in Wireless adapter to connect to Wi-Fi for
monitoring of carbon dioxide and PM10. The proposed system accessing the internet [9].
has fuzzy controls that could adjust the working interval of In this paper, the indoor air quality monitoring system
exhaust fan automatically depends on the concentration of each based on IoT that could monitor the concentration of CO2 and
pollutant. Experiment results show that the proposed system PM10 density and control the fan using fuzzy logic is
shows excellent performance in controlling indoor air quality in
proposed. The data of sensors push by raspberry pi 3 to cloud
terms of pollutant concentrator, AQI, and processing time to
server and present the data in the dashboard. This system used
remove the pollutants.
an exhaust fan to reduce the concentration of that materials
Keywords—Internet of Things, Fuzzy logic, Indoor air quality inside the room and bring it to a safe level for humans
[10][11]. This system has fuzzy logic to control interval of
I. INTRODUCTION exhaust fan depends on gases concentration. It makes power
usage of exhaust fan efficient when the concentration level of
Polluted air in an indoor environment can be contaminated
gases changes.
by harmful chemicals and other materials [1]. Air pollution
can lead to various diseases such as asthma, wet lung, even II. RELATED WORKS
coronary heart. Pollution can be done outside or indoors.
However, people spend around 90% of their activities indoor, There are several journals regarding indoor air quality
such as at the office, homes, and school. [2]. monitoring that has been studied previously. Mukesh and
Sakula made a system of indoor air quality monitoring using
CO2 is one of the highest elements in the indoor CO2 sensor, CO sensor, and raspberry pi 3 with the output
environment due to respiration and activities of a human system is DC fan, but the feature of DC fan only has ON/OFF
inside the room. High level of CO2 can make a variety of state[5]. On that Mukesh's system, the user can turn on the
irritants and decrease cognitive performance [3]. Another ventilation fan manually from application in smartphone,
material in the air that can effect for human health is PM10. when the air quality tolerable limit exceeds. Their suggestion
There is a standard for indoor air quality gases concentration for future research is that the system can be enhanced with
in a room from ASHRAE (American Society of Heating, various sensors and also increase the reliability and accuracy
Refrigerating, and Air-Conditioning Engineers). For PM10, of the system. Mukesh's system also does not have automation
Based on US Environmental Protection Agencies (EPA), the to power on the ventilation fan when the limit of gases
standards for PM10 concentration in 24-hour is 150 ug/m3. exceeds.
For carbon dioxide ppm maximum in an indoor room that still
make comfort for human odor in the room, ASHRAE has Another studied indoor air quality monitoring, Kumar
standard 1000 PPM maximum in the room [4]. Air quality has made a system of indoor quality monitoring that integrated
an index that represents the quality of air. That has a value with a server in the cloud. Data of the sensor showed as a
from 0 to 500 called AQI. graph in the dashboard cloud server. MQTT protocol is vital
to establishing communication between raspberry pi and the
Indoor air quality monitoring needs to be implemented to cloud server. A medium that used to connect from system
control air quality in the room. Indoor air quality monitoring indoor air quality monitoring to the internet is using a Wi-Fi
can ensure that the indoor environment in the room is safe for access point[9]. On Kumar's system used various parameter
a stay or do activities[5]. With the current technology, Indoor for monitoring, there are PM 2.5, carbon monoxide, carbon
air quality monitoring can also be integrated with the Internet. dioxide, temperature, humidity, and air pressure. However,
Internet of things (IoT) concept can be implemented on system the system that made by Kumar does not have the actuator to
indoor air quality monitoring. IoTis a technology that can pull out bad air in the environment of sensors.

XXX-X-XXXX-XXXX-X/XX/$XX.00 ©20XX IEEE

 Defuzzification
Actuator for Arduino
Cloud Ventilation Fan Raspberry Pi Uno
Server
Actuator
Ventilation
Fan

PM 10 Sensor PM 10 Value
Fuzzy
CO2 Sensor CO2 Value Logic
Rules CO2 Sensor
Indoor Air Observation Fuzzification

Fig. 1 General Architecture of the Proposed System

PM 10
Sensor

Fig. 3 System Implementation

Sensor PM 10 Arduino Uno Raspberry Pi
for CO2 and Sharp GP2Y1010AU0F for PM10 density. In
processing fuzzy logic, raspberry pi 3 is used in this system.
Raspberry Pi is using raspbian 9 operating system. Raspberry
Pi also has built-in Wi-Fi that can be used to connect to the
AWS cloud server and send the data to the cloud server.
Sensor MQ 135 DC Ventilation Fan Arduino UNO is also used for interfacing sensor and convert
the analog value of the sensor with feature ADC (Analog to
Fig. 2 Hardware Architecture of the System Digital Converter). All sensors are connected to Arduino
UNO and will send the data to raspberry pi via serial cable.
Dr. Vivek proposed a ducting system using fuzzy logic
This system is using exhaust fan as an actuator to pull out
control to distribute the air in the room. There are two inputs
the air from the room — the exhaust fan mounts on the wall.
of fuzzy logic temperature and humidity. The output of fuzzy
The specifications of exhaust fan have dimension 10 inches
is to control the speed of a fan, pump, and duct in a fan. The
with power usage 35 Watt. Moreover, it has 423 Cubic Feet
system is using Mamdani fuzzy and has 30 rules. These
Minutes (CFM) airflow. The exhaust fan will be working
system works appropriately to control the temperature and
based on the interval that is produced by fuzzy logic. Fig. 2
humidity of the room. However, Vivek's system does not have
shows hardware architecture and relation between each
a monitoring dashboard to monitor real-time of environment.
hardware. Figure 3 shows the implementation of all system
This system is using mat lab to compare the fuzzy in code with
installed in the room.
the simulation [12].
B. IoT Cloud Platform and Dashboard
III. PROPOSED SYSTEM DESIGN
Cloud server design includes designing all components in
The proposed system architecture has four parts of the the AWS cloud server. The modules used in the AWS cloud
system. There are room environment, Fuzzy Logic process, server are AWS IoT, Elastic Search, and Kibana. Cloud
Actuator system, and Cloud server. The room environment, server receives data, store, and present data in the form of
including PM10, CO2 sensors, and exhaust fan. The proposed dashboards to users. In the dashboard, the AQI values of
system indoor air quality monitoring has fuzzy logic control
CO2, PM10, and real-time values will be displayed for each
to set interval of exhaust fan automatically. There are fuzzy
rules to create the possibility of many different situations of sensor. The protocol used in sending data from raspberry pi
gases concentration. The result of defuzzification is a value of to AWS IoT is the MQTT (Message Queuing Telemetry
interval to tell how long exhaust fan will work. Data of sensors Transport) protocol. MQTT protocol uses the publish-
are sent to the AWS cloud server and provided as a dashboard subscribe method in sending and receiving data, which means
for monitoring purpose. The dashboard is using kibana in the sender does not need to wait for the recipient to send
AWS cloud. The dashboard consists of AQI PM10, AQI CO2, confirmation of receipt of data. Raspberry Pi will publish data
Real-time graph of PM10 value, and CO2 value in the room to brokers in the cloud server every 12 seconds. In the cloud
environment. User can access the dashboard from the browser server, there is an AWS IoT Rule service that will subscribe
of devices. Fig. 1 shows the general architecture of the to the topics sent by raspberry. AWS IoT rule will also query
proposed system. the incoming message and send it to service elastic search.
Elasticsearch services store data published by AWS IoT
A. Hardware Design
into the database and filter data. Elasticsearch search will
This system is using various hardware and sensors filter and query the topic of data contained in the AWS IoT.
connected. The sensors used in this system are MQ-135 sensor After configuration, service elasticsearch will generate a
URL to open the dashboard. The URL can be accessed on
 Fig. 4 Indoor Air Quality Monitoring Dashboard

Table 1 CO2 and PM10 Pre-evaluation Result

Parameters High To Low High To Medium To

(seconds) Medium Low
(seconds) (seconds)
CO2 155 38 117

PM10 17 5 5

Fig. 6 Membership Function of PM10

Fig. 5 Membership Function of CO2 Fig. 7 Membership of the Output

mobile devices such as smartphones and laptops. The stored dashboard is the graphic of the real-time concentration value
data can be displayed on the kibana dashboard. In the kibana of pollutants every 30 seconds
module configurations can be done, such as selecting the
chart type, color chart, and placement settings from the C. Fuzzy Logic for Controlling Exhaust Fan
dashboard. Kibana in real-time displays data received by Fuzzy logic is a cryptic logic that has a value of
elasticsearch. In addition to displaying data in realtime, membership between 0 and 1. The primary motivation in
Kibana also can see the history of data that has been entered fuzzy logic theory is to map input space into output space
so that it makes it easy to search for the data needed. using IF-THEN rules[13]. A complete fuzzy rule-based
Fig. 4 shows the dashboard of the monitoring system to system consists of three main components: Fuzzification,
present the data. In the dashboard, the data of PM10 and CO2 Inference, and defuzzification[14].
are shown. The upper part shows the indoor air quality (IAQ) This system proposed fuzzy logic to compute the interval
level of the pollutants. It displays the real-time value of the of ventilation fan with Mamdani type fuzzy inference system.
pollutants and informs the user whether it is at a safe level or There are two inputs of fuzzy logic in this system and one
not according to the IAQ standard. While the lower part of output. Two inputs are Carbon Dioxide (CO2), and PM10.
the The output of fuzzy will generates the fuzzy result as an
 Table 2 Fuzzy Rules of System Table 3 Processing Time Comparison in 1st Scenario

INTERVAL CO2 - Low CO2 - Medium CO2 - High No Proposed System Conventional System
Attempts
PM10 - Low Fan- Stop Fan-Medium Fan-High (minutes) (minutes)
PM10 - Medium Fan-Short Fan-Medium Fan-High
1 2.52 3:05
PM10 - High Fan-Short Fan-Medium Fan-High
2 2.29 3:15

3 2.53 3:19

4 2.40 3:05

5 2.27 3:19

Average 2:40 3:13

of exhaust fan only need low interval. The table 2 decscribes

Fig. 8 CO2 Concentration Level Comparison in 1st Scenario
rules of fuzzy logic
interval of exhaust fan in second. To determine the rule and IV. RESULTS AND DISCUSSION
membership value of the fuzzy output, we need to know the
specification and characteristics of sensors CO2 and PM10. The main purpose of the system is to keep the indoor air
MQ 135 as Carbon Dioxide sensor can reach a maximum quality index (AQI) at a safe level based on the ASHRAE
value of PPM is 2000. Also, Sensor Sharp GP2Y1010AU0F standard. If AQI level in a room exceeds safe level, the system
as PM10 that used in this journal has maximum read value is will work, and an exhaust fan will run for several times
500 ug/m3. To find the output membership range of the depending on the crisp value of the fuzzy result. After AQI
system, we need to figure out average value from exhaust fan level back to normal, the exhaust fan will stop. It will save
ability to reduce CO2 and PM10 substance from the high to power usage rather than the fan is always running. The system
low, high to medium, and medium to a low level. According testing is to see how effective the system that made compared
to the test of exhaust fan, the average time needed for the with a conventional ways exhaust fan that is always working
exhaust fan to pull out High level of CO2 from the room, is anytime, even the condition of the room is good. The testing
155 seconds. Moreover, for the average time needed to pull room used is a room that has a size of 3x3x3 meters, with
out a high level of PM10 are 17 Seconds. Table 1 present the ventilation that is only in the exhaust fan installed on the wall
result of average CO2 and PM10 for each level. of the room. The sensors of system, MQ-135, and PM10
sensors, are placed in the room around the exhaust fan. To
After all sensors have been specified, the next step is verify this system is running well, we test it with giving smoke
setting the membership to input fuzzy from each sensor. On to MQ-135 sensor. For PM10, we test it with giving aerosol
Fig. 5, there is 3 memberships function for CO2, which are particle until it reached 500 ug/m3. The test was carried out
low, medium, and high. Low membership of CO2 has a range with three scenarios. The first scenario when only CO2
from 0 - 1200 ppm with type trapmf. For Medium reached high concentrations. The second scenario when only
membership, it has a range 1000 - 1600 ppm with type trapmf. PM10 reached high concentration. The third scenario when
Also, For High membership it has a range from 1400 - 2000 both of them in high concentrations.
ppm also in trapmf.
A. Evaluation with 1 st Scenario (High CO2, low PM10)
For PM10, it also defines three membership functions. The
This test is carried out to see the system performance when
maximum value of this membership is 500 ug/m3. The low
the condition of the room is given smoke, which causes the
condition has range 0 to 200 ug/m3 with type trapmf. The
level of CO2 in the room to increase. Giving smoke to increase
medium condition has range 150 - 350 ug/m3 with type trapmf
CO2 levels is given at intervals of about 5 minutes in 30
and High Condition has range 300 - 500 ug/m3 also with type
minutes of testing. In this evaluation, we measure when the
trapmf — the details of PM10 membership present in Fig. 6.
CO2 achieves the high level (due to smoke) until it reaches
The interval of fuzzy output has four definitions. Fan-stop low value due to exhaust fan that is activated by the fuzzy rule.
will be generated if CO2-Low and PM10-Low. Fan-Short has This test compared the proposed system and the conventional
interval range from 10 - 30 seconds, Fan-Medium has interval method, which means it does not use the system and exhaust
100-140 seconds, and Fan-High has interval from 120 to 200 fan. The purpose of this test is to see changes in indoor CO2
seconds. All membership of fuzzy output present in Fig. 7 levels. Fig. 8. is the result of CO2 testing for conditions of high
below. CO2 levels and low PM10.
The proposed system has nine fuzzy rules. The rules are From table 3, The results of the average testing time
based on IF-THEN statement with AND operator. Due to the required above, then it appears that the average time needed
duration of wasting PM10 is fewer than CO2 in the same level to reduce CO2 concentration if using a fuzzy logic is faster
of membership, thus to reduce the high level PM10 only need than the conventional one. The average time needed to reduce
a few moments of fan interval. For example, if CO2 is in low CO2 levels to a safe level is 2:40 minutes, whereas if without
condition and PM10 in High Condition, then Interval output using the system, it takes 3:13 minutes.
 Fig. 9 PM10 Concentration Level Comparison in 2nd Scenario Fig. 10 CO2 AQI Level Comparison in 3rd Scenario

Table 4 Processing Time Comparison in 2nd Scenario

No Proposed System Conventional

Attempts
(seconds) System (seconds)

1 14 26

2 14 40

3 14 55 Fig. 11 PM10 AQI Level Comparison in 3rd Scenario

4 14 53
differences in CO2 AQI trends and PM10 AQI trends. Below
5 26 2:09 are the results of testing of AQI CO2 and PM10 for conditions
of high CO2 levels and high PM10.
6 28 40
According to Fig. 10, we can see that through the proposed
Average 18.33 57.16 system, the interval of safe CO2 AQI values that can be
obtained in 30 minutes test are 16.3 Minutes. Whereas the safe
AQI value interval that was obtained using the conventional
system is only 8.13 Minutes. From the comparison of CO2
B. Evaluation with 2nd Scenario (High PM10, Low CO2) AQI interval values, it can be concluded that the proposed
system can produce a safe AQI PM10 value longer than the
This test is carried out to see the system performance when
conventional system. From Fig. 11, the interval of safe PM10
the condition of the room is given PM10 particles which cause
AQI values obtained in 30 minutes test is 26.4 Minutes.
the level of PM10 in the room to increase. Giving aerosol
Whereas the safe AQI value interval that was obtained using
particle to increase PM10 levels is given at intervals of about
a conventional system is only 14.3 Minutes. From the
5 minutes in 35 minutes of testing. In this evaluation, we
comparison of the PM10 AQI interval values, it can be
measure when the PM10 achieves the high level (due to
concluded that the proposed system can produce a safe AQI
aerosol) until it reaches low value due to exhaust fan that is
PM10 value longer than the conventional system.
activated by the fuzzy rule. This test compared the proposed
system and the conventional method, which means it does not V. CONCLUSION
use the system and exhaust fan. The purpose of this test is to
see changes in indoor PM10 levels. Figure 9 is the result of In this paper, we have proposed an indoor air quality
PM10 testing for conditions of high PM10 levels and low monitoring based on IoT and Fuzzy Logic. Based on the
CO2. performance of the proposed system after implementation
and testing, it can be concluded that this proposed system can
From table 4, The results of the average testing time monitor the indoor air quality in real-time and effective to
required above, then it appears that the average time needed keep a good quality of air. Also, exhaust fan in the system is
to reduce PM10 concentration if using a proposed system is wasting PM10 faster than CO2, and so interval performance
faster than the conventional one. The average time needed to
of exhaust fan to waste CO2 is longer than PM10. Moreover,
reduce PM10 levels to a safe level is 18.33 Seconds, whereas
with the conventional system, it takes 57.16 Seconds. this made system could also generate AQI value in safe level
longer than the conventional system. Also, it can save more
C. Evaluation with 3rd Scenario (High PM10, High CO2) energy rather than exhaust fan always working like traditional
This test was conducted to see the system performance ways because on this system exhaust fan only work when
when the room conditions were given smoke, and PM10 AQI level is not normal.
particles simultaneously, which caused the levels of CO2 and
PM10 in the room increased. Giving aerosol particles and ACKNOWLEDGMENT
smoke is given at intervals of about 5 minutes in 30 minutes This publication is fully supported by Bina Nusantara
of testing. In this evaluation, we measure when the CO2 and University.
PM10 achieve a high level (due to smoke and aerosol) until it
reaches low value due to exhaust fan that is activated by the REFERENCES
fuzzy rule. This test compared the proposed system and the
conventional method, which means it does not use the system [1] Cincinelli, A., & Martellini, T., “Indoor Air Quality and Health,” In
and exhaust fan. The purpose of this test is to see the International Journal of Environmental Research and Public Health,
vol.14(11), pp. 1286, 2017.
[2] Kankaria, A., et al., “Indoor AirPollution in India: Implications on International Conference on Information Science and Control
Health and its Control,” In Indian Journal of Community Medicine, Engineering, 2017, pp.747–751.
vol. 39(4), pp 203-207, 2014. [9] Gupta, M. S. D., Patchava, V., & Menezes, V., “Healthcare based on
[3] Marcella, U., et al., “The Effect of Indoor Temperature and CO2 Levels IoT using Raspberry Pi,” In 2015 International Conference on Green
on Cognitive Performance of Adult Females in Universitas Building in Computing and Internet of Things, 2015, pp. 796–799.
Saudi Arabia,” In International Conference-Future Buildings & [10] Preethicandra, D. M. G,. “Design of a smart indoor air quality
Districts-Energy Efficiency from Nano to Urban Scale, vol.122, pp. monitoring wireless sensor network for assisted living,” In: IEEE
451-456, 2017. International Instrumentation and Measurement Technology
[4] ASHRAE/ANSI Standard 62.I- 2013 Ventilation for Acceptable Conference(I2MTC),2013.
Indoor Air Quality [Online]. Available: [11] Dobbin, A. et al., “The Benefit of kitchen exhaust fan use after cooking
http://www.myiaire.com/product-docs/ultraDRY/ASHRAE62.1.pdf – An experimental assessment, ” In Building and Environment,
[5] Mukesh, D. M., & Akula, S. K., “Automated Indoor Air Quality vol.135, pp. 286-296, 2018.
Monitor and Control,” In: International Journal of Computer [12] Sharma, V. et al., “Two Ways Ducting System Using Fuzzy Logic
Applications., vol. 159(6), pp. 0975 – 8887, 2017. Control System,” In 2013 International Journal of Electronics and
[6] Malche, T., & Maheshwary, P., “Internet of Things (IoT) for building Communication Engineering & Technology (IJECET).2013
smart home system,” In: 2017 International Conference on I-SMAC [13] Naba, A., “Belajar cepat fuzzy logic menggunakan Matlab,”
(IoT in Social, Mobile, Analytics and Cloud), pp.65–70, 2017. Yogyakarta, Andi, 2009.
[7] Kumar, S., & Jasuja, A., “Air quality monitoring system based on IoT [14] Suyanto., “Artificial Intelligence, Searching, Planning and Learning,”
using Raspberry Pi,” In 2017 International Conference on Computing, Bandung, Informatika Bandung, 2007.
Communication and Automation, pp.1341–1346, 2017.
[8] Shu, M., Tang, M., Yang, M., & Wei, N., “The Vital Signs Real-Time
Monitoring System Based on Internet of Things,” In 2017 4th
.

View publication stats