SlideShare a Scribd company logo
Monitoring and Forecasting of Air Emissions with
IoT Measuring Stations and a SaaS Cloud
Application
Fernando Ar´evalo N.∗, M. Tahasanul Ibrahim∗, Rizky M. Diprasetya∗, Omar Otoniel Flores†, Andreas Schwung∗
∗Department of Automation Technology
South Westphalia University of Applied Sciences, Germany
Email: {arevalo.fernando,schwung.andreas}@fh-swf.de
†Applied Sciences School
Universidad Tecnol´ogica de El Salvador, El Salvador
Email: omar.flores@utec.edu.sv
Abstract—Environmental pollution is a major problem in
densely populated cities, due to the increase of respiratory
diseases in the population. The pollution level is used to inform
the public, in order to take special measures based on a pollution
scale. Particulate matter (PM), specifically PM10 and PM2.5, is
used to estimate the degree of pollution. The monitoring and
analysis of these two variables has attracted the attention of
the research community, particularly in the design of measuring
stations, forecasting, and software infrastructure to host user
applications. This paper proposes a modular and ready-to-use
architecture for a SaaS cloud application for air emissions fore-
casting using IoT measuring stations. The SaaS cloud application
is connected to the IoT measuring stations located in the densely
populated cities of El Salvador. The data is used to create deep
learning models, to make forecasts for the next day.
Index Terms—Air Emissions, IoT, LSTM, CNN-1D, Docker
I. INTRODUCTION
Pollution is a major problem in cities with large populations.
Direct consequences of high concentration of pollutants are
reflected in health problems in the population. Estimation of
the pollution level is used to assess the public, e.g. if the level
of pollution is within a certain range, outdoor sport should
be limited. Air pollution can be measured by the analysis
of particulate matter (PM), normally defined as PM10 and
PM2.5 corresponding to a diameter of 10 and 2.5 micrometers
respectively [1]. Karagulian et al. [1] presented a review of the
main contributors for the generation of PM10 and PM2.5 in
the air. The authors defined the following categories: traffic,
industry, domestic fuel combustion, natural sources such as
dust and sea salt, and unspecified sources. This research focus
on the use of PM10 and PM2.5 to estimate the air pollution.
Monitoring and Analysis of PM10 and PM2.5 are topics
of high interest in literature, specially for the design of
measuring stations, software for data analysis, and software
infrastructure for hosting user applications. Low cost IoT
measuring stations allow data to be collected in a massive
way, due to its fast implementation, access to communication
protocols, and its low investment. The proliferation of cloud
services have, on the other side, facilitated the implementation
of back-end infrastructures, so that the user application has all
the necessary services running on the background, such as
database, communication protocols, and user interface. These
days monitoring and analysis of air quality data is not enough,
because the population wants to know the forecast, similar
to the weather forecasting service, so in this manner it can
be aware of the pollution level of the next day and take the
necessary measures. This has motivated research on PM10
and PM2.5 forecasting, which have moved from classical sta-
tistical approaches such as Autoregressive Integrated Moving
Average (ARIMA) and regression, to more modern algorithms
based on deep learning.
This paper proposes a modular and ready-to-use architecture
for a SaaS cloud application for air emissions forecasting using
IoT measuring stations. Its docker-based architecture allows
the system to integrate micro-services such as M2M com-
munication, a database, forecasts and visualisation through a
web application. Its modularity allows new measuring stations
and additional sensors to be integrated into the SaaS cloud
application. This research presents a comparison of two time
series multivariate deep-learning-based approaches.
This paper is structured as follows: Section II portrays a
literature survey on IoT measuring stations for air emissions,
cloud-based applications, and air emissions forecasting. Sec-
tion III presents the proposed model. Section IV describes the
use case, including implementation and forecasting results.
The conclusions and prospective work are summarised in
section V.
II. RELATED WORK
Liu et al. [2] proposed an IoT microcontroller-based solution
for the measurement of PM2.5, PM10, NO, SO2. The real
time air quality monitoring system used Long Range (LoRa)
technology to communicate with the control centre. Venkata-
narayanan et al. [3] proposed a smart bicycle application for
monitoring air quality. The authors designed a microcontroller-
based station for measuring NO2 and SO2. The measurements
were sent to a cloud application. An Android application was
implemented for monitoring air quality. Kumar and Jasuja [4]
proposed an IoT system to monitor air quality monitoring
with Arduino, Raspberry and a cloud platform. The system
monitored PM2.5, CO, CO2, Temperature and Humidity.
Bao [5] proposed a mobile data acquisition node with a
microcontroller and a GPS-GPRS module. A cloud platform
received the particle concentration data PM2.5, PM10; as well
as the geographic location data. An Android application was
designed for the data monitoring.
Beeharry et al. [6] designed an IoT device that was based
on Arduino and Weather Shield. The device sent the weather
information to a local gateway. From this gateway the data
was sent to a cloud database. A server is assigned to retrieve
the data from the cloud database and processes it to make the
weather forecast. The prediction is performed using K-NN.
The result is sent to either a web page or a smartphone applica-
tion. Liu et al. [7] proposed an ensemble model for multivariate
time series forecasting based on LSTM. The authors predicted
PM2.5. Mahajan et al. [8] compared the PM2.5 forecasts
from an Autoregressive Integrated Moving Average (ARIMA)
model and a Neural Network Autoregression (NNAR). The
models predicted the next hour. Xu and Yoneda [9] proposed
a LSTM-Autoencoder Model for PM2.5 forecasting. They
tested their model using the Beijing dataset. They compared
their results with a RNN and ARIMA models. Du et al. [10]
proposed a hybrid model using 1D-CNN and a Bi-LSTM to
forecast PM2.5 using the multivariate time series from the
Beijing Dataset.
Huang et al. [11] proposed a mobile system architecture
consisting of an air quality measuring station, a smartphone
App and a cloud platform. The system measured PM2.5 from
inside a car, allowing samples to be taken from different
locations and at different times. The authors used a random
forest model to classify the air circulation on the vehicle
considering factors such as opening/closing windows, the use
of air conditioner in a in-loop and a out-loop modes. Song et
al. [12] proposed an IoT system using LoRa and LSTM for
PM2.5 forecasting. The Arduino-based sensing unit collected
PM2.5 samples, and sent them to a cloud platform via a
LoRa gateway. Raafat et al. [13] proposed an algorithm for
novelty detection in fog level with the measurements of light,
temperature, CO2 and humidity. The authors used Levene’s
test to identify novelties during feature extraction, and a Feed-
Forward Neural Network for classification on events detection.
Previous research focused either on the proposal of hard-
ware for air quality sensing units, time series multivariate
forecasting, or the integration of the monitoring systems to
commercial cloud platforms. This paper proposes a modular
and ready-to-use architecture for a SaaS cloud application
for monitoring and forecasting of air emissions using IoT
measuring stations.
III. MONITORIG AND FORECASTING OF AIR EMISSIONS
WITH IOT MEASURING STATIONS AND A SAAS CLOUD
APPLICATION
The proposed model provides the software infrastructure for
the connectivity of the IoT devices, as well as the features
of monitoring and analysis of the collected data. This is
illustrated in Fig. 1. The IoT device can be a microcontroller,
a sensor, a Programmable Logic Controller (PLC), or even
another computer. The forecasting feature is implemented in
the Dashboard and System Process module.
Fig. 1: Proposed Model Architecture
A. Air Emissions Data Acquisition with IoT Measuring Sta-
tions
The IoT measuring station is shown in Fig. 1 as an IoT
device. The station collects data from the sensors e.g. Tem-
perature, PM2.5, CO, CO2, humidity, etc. It can be based
on a low-cost implementation with microcontrollers or micro-
computers such as Raspberry Pi or Odroid. The IoT measuring
station can communicate directly with a cloud service, or with
a central control. It can have internet connectivity through the
use of communication protocols such as MQTT or through a
RESTful method. Implementations have been reported using
LoRa between stations and a gateway. The gateway finally
connects to the cloud or a central control system via MQTT
or a RESTful method.
B. SaaS Cloud Application
The software as a Service (SaaS) system is designed in a
cloud server where all the dynamics of the services are set up
according to the needs. The server has a modular and scaleable
architecture, since docker containers are used for each of the
services.
1) Docker Container System: The idea of a docker con-
tainer is to contain all the modular events and processing
within one or many virtual images, while keeping each ap-
plication isolated without direct environment dependencies.
There are several aspects of docker containment systems that
made this method suitable for virtualising services. Docker
containers run in self-isolated environments, which means that
no external system other than the docker daemon or the con-
tainer itself can access them. Docker container isolates its con-
tainers mainly in three different sectors: ”Process Isolation”,
”File System Isolation”, ”Network Isolation” [14]. Docker
containers has earned its reputation for speed, portability, and
scalability. The dockerfile contains instructions on what type
of operating system to install and what other applications
are necessary. Since each container is intended for specific
applications, not all components of the operative system are
needed. Therefore the whole process of container production is
faster and lighter for the container. As soon a container is built
or executed, the Docker Daemon wraps the entire container
together with its respective processes in a PID namespace. This
allows the host machine to allocate resources as needed, which
is much faster and better than traditional virtual machines. The
traditional ones assign the host resources completely to the
virtual machine regardless of its usage [15].
Fig. 2: SaaS Communication
2) Application Backbone Process: The Software as a Ser-
vice (SaaS) system is designed in a cloud server on which the
entire dynamic of services is set up according to requirements.
IoT devices and users around the globe can authenticate,
connect and use these services on demand through a com-
munication system hosted in the SaaS cloud application. The
services offered are divided into two groups: ”Pushing Data”
and ”Pulling Data”. The groups are determined by the type
of service, for example, publishing data from the server node
is categorised as ”pushing data” while viewing published
data from the server node is categorized as ”pulling data”.
Nevertheless, the principle of the SaaS cloud application is to
receive requests from either users or IoT devices and forward
them to the internal server, where multiple docker containers
are running to provide the requested service, see Fig. 2.
C. Air Emissions Forecasting using Deep Learning
Artificial Neural Network (ANN) tends to struggle with the
computational complexity of computer image data. Later, the
Convolutional Neural Network (CNN) was introduced to re-
duce the complexity of the problem. The CNN architecture are
based on three different types of layers: Convolutional layers,
pooling layers and fully-connected layers. There are different
types of convolutional layers such as 1D convolutional, 2D
convolutional and 3D convolutional. 1D Convolutional com-
prises a vector of neurons. 2D convolutional is based on a
2-dimensional array of neurons, which are defined as width
and height [16]. The 2D convolutional layer is a well-known
method for greyscale computer vision. The 3D convolutional
layer consists of a 3-dimensional array of neurons, which are
defined as width, height and depth. This type of convolutional
layer is normally used for a color image, e.g. RGB, which
has 3 different type of color [17]. For the forecasting, the air
emission data are treated as time series data that can be defined
in a vector form. For this purpose the 1D convolutional neural
network is used for forecasting.
Long short-term Memory (LSTM) is a well-known type
of recurrent neural network (RNN) for prediction based on
time series data. It consists of a cell, an input, an output
and a forget gate. The cell remembers all values over time
intervals. The input gate controls the signal from the network
to the cell. These prevent an irrelevant input from entering
the neuron. The output gate protects the memory cell from
disturbance originated from the neuron [18]. Both the input
and output gates are thus the flow control of constant error
carousel (CEC), which gives a constant error flow within the
cell. This solves the problem that the standard RNN has where
it cannot bridge more than 5 - 10 time steps, while LSTM can
bridge more than 1000 time steps.
IV. USE CASE: MONITORING AND FORECASTING OF
URBAN AIR EMISSIONS IN EL SALVADOR
The Universidad Tecnol´ogica de El Salvador (UTEC) pro-
posed a low-cost IoT measuring station to solve the problem
of missing online measurement of air emissions. The project
aimed at measuring air emissions in densely populated cities in
El Salvador, see Fig. 3. The IoT measuring stations are located
in the cities of Santa Ana (STNA), Chalatenango (CHLT), La
Libertad (ITCA) and San Salvador (UTEC). The SaaS cloud
application collects the data from all stations and visualises
it in the dashboard of the web application. The air emissions
forecasting focuses on the data of the city of San Salvador
(UTEC station), as it is the one with more data points.
A. Data Acquisition with an IoT Measuring Station
The IoT measuring station consists of four main compo-
nents: an ESP32 microcontroller, a PM sensor that measures
PM2.5 and PM10, a temperature sensor and a local display.
Fig. 3: Location of the IoT Stations in the Map
Fig. 4 shows the main components and the IoT measuring sta-
tion. The IoT measuring station communicates with the SaaS
Cloud application via the MQTT communication protocol. For
this purpose a MQTT topic is defined for each measurement
signal. The IoT measuring stations take a sample of PM10,
PM2.5 and the temperature every 15 minutes.
Fig. 4: Components of the IoT Measuring Station [19]
B. SaaS Cloud Application
In a previous paper, a first version of the current SaaS
Cloud Application was proposed in terms of services and
communication protocols [20]. The main server, which is the
backbone of the entire SaaS application, runs on a virtual
machine based on the Amazon Web Service (AWS). The
idea of the system is to run sets of docker containers for
different services within the server, which together will form
the overall air emissions service. For the air emission project
a total of 3 docker containers were used, see Fig. 5. Eclipse-
Fig. 5: SaaS Process
Mosquitto MQTT Broker is selected as the module responsible
for communications between the IoT measuring stations and
the server for pushing and pulling data. The MQTT broker is
lightweight and supports multiple ”Quality of Services” (QoS),
including device handshaking, making it a prime candidate
for the communication module. The MySQL Database runs
in a container, which is the primary data storage unit for
the system. All requests for data retrieval are processed via
this container. Dashboard and System Process is a custom
container build with an Ubuntu 18.04 image. The design of the
container enables dynamic data input on-the-fly. This is done
using an Excel spreadsheet that exists on the host machine
and is used together with the container to enter the node and
sensor names according to the variable changes. The Excel file
contains predefined formulas for the automatic generation of
the MQTT node addresses according to the user input. When
the container is spun, it traverses the Excel file and subscribes
to the respective MQTT nodes and MySQL database tables
that exist in the file. Fig. 6 and Fig. 7 show the dashboard
Home and the Chart of the UTEC IoT measuring station re-
spectively. Inside the container there is also a flask application
running that renders the HTML templates for the dashboard
using the data imported from the MySQL Databases. User
requests to view the dashboard panel are handled using this
flask application.
Data request flow process is similar to the generic SaaS
application, the air emissions system also works in three steps:
Request, Process, and Response. An IoT measuring station or
a user starts his journey with a simple request to the server.
Requests received from IoT measuring stations (TCP port
1883) take the form of a publication that the MQTT server
publishes to their respective nodes for further processing.
Other processes are performed by the user-defined container
(Excel file) , which scans for new MQTT publications and
in response stores the data in the corresponding tables in
the MySQL Database. On the other hand, there is only one
possible user request, namely to display the dashboard. This
way, requests from user devices (TCP port 80/443) are re-
ceived directly by the flask application of the custom container.
Once the request is captured, flask processes the requested
html template filled with the requested data from the MySQL
database. In response to the request, the edited template is
forwarded to the user’s web browser to be displayed with the
latest data from the MySQL database, see Fig. 5.
Fig. 6: Dashboard Home of IoT Stations
Fig. 7: Forecasting using data of IoT Station UTEC
C. Air Emissions Forecasting
1) Data Pre-processing: This paper proposes a forecasting
model in which 96 data points of the previous day are used
to predict the next 96 data points of the current day. The
IoT measuring station collects the air emission data and the
time stamp. Then the data is reconstructed into the following
structure:
• Previous Day
• Previous Hour
• Previous Minute
• Current Day
• Previous PM2.5
• Previous PM10
• Previous Temperature
The specific day, hour, and minute are taken into account,
as each different time has its own unique features. The data
is then fed into the forecasting model. The output will be
the predicted value of PM2.5, PM10, and temperature for the
whole day.
2) 1D-CNN: A 1D convolutional neural network is used,
since the forecasting problem is based on time series data.
This is followed by the max-pooling layer with a pool size
of 2, after which the result is flattened and fed into a fully-
connected layer with 50 neurons. Then again a fully-connected
layer with one neuron is used for the output.
3) LSTM: The LSTM model has 100 cells. After that there
is a 20% dropout chance. Then it is connected directly to the
output.
4) Implementation: The historical data of air emissions is
collected. Then the data is split into training data and testing
data. There are 9015 samples for the training data and 800
samples for the testing data. The samples are selected manu-
ally due to the unclean dataset. Adam is used as optimizer and
the mean squared error for the loss function. The training was
conducted for 100 epochs for each model. Fig. 8a shows that
LSTM and CNN models have the same performance. Both the
LSTM model and the CNN model have the same precision
of ±0.04. Fig. 8b and Fig. 8c show that the LSTM model
performs better than CNN model. For PM2.5, the precision of
the LSTM result is ±6.6, whereas the CNN result is ±6.7.
For PM10, the precision of the LSTM result is ±8, whereas
the CNN result is ±8.2. But in terms of losses, the LSTM
model still shows high losses.
D. Discussion
The IoT measuring station has a good performance in the
acquisition of air emissions signals. A drawback, however, is
that if a change to the firmware of the IoT measuring station
is required, the developer must travel to the IoT station and
make the changes on site. The SaaS Cloud is a scalable and
ready-to-use solution as it allows the addition of more signals
and more IoT measuring stations. However, the changes must
be manually updated in the git platform. As the results show,
the forecast model still does not perform well by checking
the high losses. There are many missing data points in the
historical data and some error measurements that result in the
dataset not being clean. In forecasting, there is still room for
improvement of the models to increase its performance.
V. CONCLUSION
This paper proposed a modular and ready-to-use architec-
ture for a SaaS cloud application for monitoring and forecast-
ing of air emissions using IoT stations. The performance of the
low-cost IoT station was satisfactory, and allowed to collect
(a) Temperature (b) PM2.5 (c) PM10
Fig. 8: Loss Test
data of PM2.5, PM10 and Temperature. The cloud application
proved to be scalable, since it allowed two additional IoT
stations to be added during the project development. The
changes were made simply by updating the user excel file
containing the MQTT topics. This updated the dashboard of
the web application as well as the tables in the database. This
research presented two deep learning approaches 1D-CNN and
LSTM. The performance of both algoritms have room for
improvement, the reason for this is based on missing data
points and error measurements.
The prospective work will focus on the development of a
smart IoT station that allows remote programming, the devel-
opment of a dynamic SaaS cloud application, and performance
improvement of the deep-learning-based forecasting.
REFERENCES
[1] F. Karagulian, C. A. Belis, C. F. C. Dora, A. M. Pr¨uss-Ust¨un, S. Bonjour,
H. Adair-Rohani, and M. Amann, “Contributions to cities’ ambient par-
ticulate matter (PM): A systematic review of local source contributions
at global level,” pp. 475–483, nov 2015.
[2] S. Liu, C. Xia, and Z. Zhao, “A low-power real-time air quality
monitoring system using LPWAN based on LoRa,” in 2016 13th
IEEE International Conference on Solid-State and Integrated Circuit
Technology, ICSICT 2016 - Proceedings. Institute of Electrical and
Electronics Engineers Inc., 2016, pp. 379–381.
[3] A. Venkatanarayanan, A. Vijayavel, A. Rajagopal, and P. Nagaradjane,
“Design of sensor system for air pollution and human vital monitoring
for connected cyclists,” IET Communications, vol. 13, no. 19, pp. 3181–
3186, nov 2019.
[4] S. Kumar and A. Jasuja, “Air quality monitoring system based on IoT
using Raspberry Pi,” in Proceeding - IEEE International Conference on
Computing, Communication and Automation, ICCCA 2017, vol. 2017-
Janua. Institute of Electrical and Electronics Engineers Inc., dec 2017,
pp. 1341–1346.
[5] L. Q. Bao, “Cloud connection oriented real-time monitoring system for
atmospheric particles,” IET Wireless Sensor Systems, vol. 10, no. 1, pp.
31–36, feb 2020.
[6] Y. Beeharry, T. P. Fowdur, and J. A. Sunglee, “A Cloud-Based Real-Time
Weather Forecasting Application,” in 2019 14th International Confer-
ence on Advanced Technologies, Systems and Services in Telecommu-
nications, TELSIKS 2019 - Proceedings. Institute of Electrical and
Electronics Engineers Inc., oct 2019, pp. 294–297.
[7] F. Liu, M. Cai, L. Wang, and Y. Lu, “An Ensemble Model Based
on Adaptive Noise Reducer and Over-Fitting Prevention LSTM for
Multivariate Time Series Forecasting,” IEEE Access, vol. 7, pp. 26 102–
26 115, 2019.
[8] S. Mahajan, L.-j. Chen, and T.-c. Tsai, “An Empirical Study of PM2
. 5 Forecasting Using Neural Network,” in 2017 IEEE SmartWorld,
Ubiquitous Intelligence & Computing, Advanced & Trusted Com-
puted, Scalable Computing & Communications, Cloud & Big Data
Computing, Internet of People and Smart City Innovation (Smart-
World/SCALCOM/UIC/ATC/CBDCom/IOP/SCI). IEEE, 2017, pp. 1–7.
[9] X. Xu and M. Yoneda, “Multitask Air-Quality Prediction Based on
LSTM-Autoencoder Model,” IEEE Transactions on Cybernetics, pp. 1–
10, oct 2019.
[10] S. Du, T. Li, Y. Yang, and S.-J. Horng, “Deep Air Quality Forecast-
ing Using Hybrid Deep Learning Framework,” IEEE Transactions on
Knowledge and Data Engineering, pp. 1–1, nov 2019.
[11] J. Huang, N. Duan, P. Ji, C. Ma, F. Hu, Y. Ding, Y. Yu, Q. Zhou,
and W. Sun, “A Crowdsource-Based Sensing System for Monitoring
Fine-Grained Air Quality in Urban Environments,” IEEE Internet of
Things Journal, vol. 6, no. 2, pp. 3240–3247, apr 2019. [Online].
Available: https://ieeexplore.ieee.org/document/8534351/
[12] W. Song, J. Han, J. Xie, Y. Gao, and L. Song, “System for detecting and
forecasting pm2.5 concentration levels using long short-Term memory
and lora,” in Proceedings - 2019 IEEE International Congress on
Cybermatics: 12th IEEE International Conference on Internet of Things,
15th IEEE International Conference on Green Computing and Commu-
nications, 12th IEEE International Conference on Cyber, Physical and
So. IEEE, 2019, pp. 834–841.
[13] H. M. Raafat, M. S. Hossain, E. Essa, S. Elmougy, A. S. Tolba,
G. Muhammad, and A. Ghoneim, “Fog Intelligence for Real-Time IoT
Sensor Data Analytics,” IEEE Access, vol. 5, pp. 24 062–24 069, sep
2017.
[14] T. Bui, “Analysis of Docker Security,” jan 2015. [Online]. Available:
http://arxiv.org/abs/1501.02967
[15] A. M. Joy, “Performance comparison between Linux containers and
virtual machines,” in International Conference on Advances in Computer
Engineering and Applications (ICACEA), 2015, {International Confer-
ence on Advances in Computer Engineering and Applications} and
ICACEA. IEEE, 2015, pp. 342–346.
[16] D. V. Sang, N. Van Dat, and D. P. Thuan, “Facial expression recognition
using deep convolutional neural networks,” in Proceedings - 2017
9th International Conference on Knowledge and Systems Engineering,
KSE 2017, vol. 2017-January. Institute of Electrical and Electronics
Engineers Inc., nov 2017, pp. 130–135.
[17] B. Chen and C. Jung, “Patch-Based Stereo Matching Using 3D Convo-
lutional Neural Networks,” in Proceedings - International Conference
on Image Processing, ICIP. IEEE Computer Society, aug 2018, pp.
3633–3637.
[18] Y. Heryadi and H. L. H. S. Warnars, “Learning temporal representation
of transaction amount for fraudulent transaction recognition using CNN,
Stacked LSTM, and CNN-LSTM,” in 2017 IEEE International Confer-
ence on Cybernetics and Computational Intelligence, CyberneticsCOM
2017 - Proceedings, vol. 2017-November. Institute of Electrical and
Electronics Engineers Inc., mar 2018, pp. 84–89.
[19] O. Flores-Cortez, R. Adalberto Cortez, and V. Rosa, “A Low-cost
IoT System for Environmental Pollution Monitoring in Developing
Countries,” in Proceedings of the 26th International Conference ”Mixed
Design of Integrated Circuits and Systems”, MIXDES 2019, 2019.
[20] F. Arevalo, M. R. Diprasetya, and A. Schwung, “A Cloud-based
Architecture for Condition Monitoring based on Machine Learning,”
in Proceedings - IEEE 16th International Conference on Industrial
Informatics, INDIN 2018. Institute of Electrical and Electronics
Engineers Inc., sep 2018, pp. 163–168.

More Related Content

What's hot

IRJET- Energy Saving in Governmental Educational Buildings: Case Study
IRJET- Energy Saving in Governmental Educational Buildings: Case StudyIRJET- Energy Saving in Governmental Educational Buildings: Case Study
IRJET- Energy Saving in Governmental Educational Buildings: Case StudyIRJET Journal
 
Text mining analysis of wind turbine accidents: An ontology-based framework
Text mining analysis of wind turbine accidents: An ontology-based frameworkText mining analysis of wind turbine accidents: An ontology-based framework
Text mining analysis of wind turbine accidents: An ontology-based frameworkGurdal Ertek
 
AN INVESTIGATION OF THE ENERGY CONSUMPTION BY INFORMATION TECHNOLOGY EQUIPMENTS
AN INVESTIGATION OF THE ENERGY CONSUMPTION BY INFORMATION TECHNOLOGY EQUIPMENTSAN INVESTIGATION OF THE ENERGY CONSUMPTION BY INFORMATION TECHNOLOGY EQUIPMENTS
AN INVESTIGATION OF THE ENERGY CONSUMPTION BY INFORMATION TECHNOLOGY EQUIPMENTSijcsit
 
Levelized economic method based paper
Levelized economic method based paperLevelized economic method based paper
Levelized economic method based paperkapilchopra2012
 
Insights into the Efficiencies of On-Shore Wind Turbines: A Data-Centric Anal...
Insights into the Efficiencies of On-Shore Wind Turbines: A Data-Centric Anal...Insights into the Efficiencies of On-Shore Wind Turbines: A Data-Centric Anal...
Insights into the Efficiencies of On-Shore Wind Turbines: A Data-Centric Anal...ertekg
 
Grid-Connected Photovoltaic System Requirement-Crimson Publishers
Grid-Connected Photovoltaic System Requirement-Crimson PublishersGrid-Connected Photovoltaic System Requirement-Crimson Publishers
Grid-Connected Photovoltaic System Requirement-Crimson PublishersCrimsonpublishersPRSP
 
IRJET- A Revieiw of Wind Energy Potential in Kano State, Nigeria for the ...
IRJET-  	  A Revieiw of Wind Energy Potential in Kano State, Nigeria for the ...IRJET-  	  A Revieiw of Wind Energy Potential in Kano State, Nigeria for the ...
IRJET- A Revieiw of Wind Energy Potential in Kano State, Nigeria for the ...IRJET Journal
 
Numerical simulation of Hybrid Generation System: a case study
Numerical simulation of Hybrid Generation System: a case studyNumerical simulation of Hybrid Generation System: a case study
Numerical simulation of Hybrid Generation System: a case studyIRJET Journal
 
OVERVIEW OF RENEWABLE ENERGY SOURCES IN THE PERSPECTIVE OF CARBON FOOT PRINTI...
OVERVIEW OF RENEWABLE ENERGY SOURCES IN THE PERSPECTIVE OF CARBON FOOT PRINTI...OVERVIEW OF RENEWABLE ENERGY SOURCES IN THE PERSPECTIVE OF CARBON FOOT PRINTI...
OVERVIEW OF RENEWABLE ENERGY SOURCES IN THE PERSPECTIVE OF CARBON FOOT PRINTI...ijiert bestjournal
 
Optimizing Size of Variable Renewable Energy Sources by Incorporating Energy ...
Optimizing Size of Variable Renewable Energy Sources by Incorporating Energy ...Optimizing Size of Variable Renewable Energy Sources by Incorporating Energy ...
Optimizing Size of Variable Renewable Energy Sources by Incorporating Energy ...Kashif Mehmood
 
Environmental asessement of switching to renewable energy sources
Environmental asessement of switching to renewable energy sourcesEnvironmental asessement of switching to renewable energy sources
Environmental asessement of switching to renewable energy sourcesSaad_Sarfraz
 
How costs affect deployment of low carbon technologies - analysis with JRC-EU...
How costs affect deployment of low carbon technologies - analysis with JRC-EU...How costs affect deployment of low carbon technologies - analysis with JRC-EU...
How costs affect deployment of low carbon technologies - analysis with JRC-EU...IEA-ETSAP
 
ESTIMATION AND COMPARISION OF COST OF COAL POWER GENERATION WITH EXTERNAL COS...
ESTIMATION AND COMPARISION OF COST OF COAL POWER GENERATION WITH EXTERNAL COS...ESTIMATION AND COMPARISION OF COST OF COAL POWER GENERATION WITH EXTERNAL COS...
ESTIMATION AND COMPARISION OF COST OF COAL POWER GENERATION WITH EXTERNAL COS...Ankur Tyagi
 
Optimal Operation of Wind-thermal generation using differential evolution
Optimal Operation of Wind-thermal generation using differential evolutionOptimal Operation of Wind-thermal generation using differential evolution
Optimal Operation of Wind-thermal generation using differential evolutionIOSR Journals
 

What's hot (17)

A42030110
A42030110A42030110
A42030110
 
IRJET- Energy Saving in Governmental Educational Buildings: Case Study
IRJET- Energy Saving in Governmental Educational Buildings: Case StudyIRJET- Energy Saving in Governmental Educational Buildings: Case Study
IRJET- Energy Saving in Governmental Educational Buildings: Case Study
 
GasAnalyzer
GasAnalyzerGasAnalyzer
GasAnalyzer
 
Text mining analysis of wind turbine accidents: An ontology-based framework
Text mining analysis of wind turbine accidents: An ontology-based frameworkText mining analysis of wind turbine accidents: An ontology-based framework
Text mining analysis of wind turbine accidents: An ontology-based framework
 
AN INVESTIGATION OF THE ENERGY CONSUMPTION BY INFORMATION TECHNOLOGY EQUIPMENTS
AN INVESTIGATION OF THE ENERGY CONSUMPTION BY INFORMATION TECHNOLOGY EQUIPMENTSAN INVESTIGATION OF THE ENERGY CONSUMPTION BY INFORMATION TECHNOLOGY EQUIPMENTS
AN INVESTIGATION OF THE ENERGY CONSUMPTION BY INFORMATION TECHNOLOGY EQUIPMENTS
 
Levelized economic method based paper
Levelized economic method based paperLevelized economic method based paper
Levelized economic method based paper
 
30120140505021
3012014050502130120140505021
30120140505021
 
Insights into the Efficiencies of On-Shore Wind Turbines: A Data-Centric Anal...
Insights into the Efficiencies of On-Shore Wind Turbines: A Data-Centric Anal...Insights into the Efficiencies of On-Shore Wind Turbines: A Data-Centric Anal...
Insights into the Efficiencies of On-Shore Wind Turbines: A Data-Centric Anal...
 
Grid-Connected Photovoltaic System Requirement-Crimson Publishers
Grid-Connected Photovoltaic System Requirement-Crimson PublishersGrid-Connected Photovoltaic System Requirement-Crimson Publishers
Grid-Connected Photovoltaic System Requirement-Crimson Publishers
 
IRJET- A Revieiw of Wind Energy Potential in Kano State, Nigeria for the ...
IRJET-  	  A Revieiw of Wind Energy Potential in Kano State, Nigeria for the ...IRJET-  	  A Revieiw of Wind Energy Potential in Kano State, Nigeria for the ...
IRJET- A Revieiw of Wind Energy Potential in Kano State, Nigeria for the ...
 
Numerical simulation of Hybrid Generation System: a case study
Numerical simulation of Hybrid Generation System: a case studyNumerical simulation of Hybrid Generation System: a case study
Numerical simulation of Hybrid Generation System: a case study
 
OVERVIEW OF RENEWABLE ENERGY SOURCES IN THE PERSPECTIVE OF CARBON FOOT PRINTI...
OVERVIEW OF RENEWABLE ENERGY SOURCES IN THE PERSPECTIVE OF CARBON FOOT PRINTI...OVERVIEW OF RENEWABLE ENERGY SOURCES IN THE PERSPECTIVE OF CARBON FOOT PRINTI...
OVERVIEW OF RENEWABLE ENERGY SOURCES IN THE PERSPECTIVE OF CARBON FOOT PRINTI...
 
Optimizing Size of Variable Renewable Energy Sources by Incorporating Energy ...
Optimizing Size of Variable Renewable Energy Sources by Incorporating Energy ...Optimizing Size of Variable Renewable Energy Sources by Incorporating Energy ...
Optimizing Size of Variable Renewable Energy Sources by Incorporating Energy ...
 
Environmental asessement of switching to renewable energy sources
Environmental asessement of switching to renewable energy sourcesEnvironmental asessement of switching to renewable energy sources
Environmental asessement of switching to renewable energy sources
 
How costs affect deployment of low carbon technologies - analysis with JRC-EU...
How costs affect deployment of low carbon technologies - analysis with JRC-EU...How costs affect deployment of low carbon technologies - analysis with JRC-EU...
How costs affect deployment of low carbon technologies - analysis with JRC-EU...
 
ESTIMATION AND COMPARISION OF COST OF COAL POWER GENERATION WITH EXTERNAL COS...
ESTIMATION AND COMPARISION OF COST OF COAL POWER GENERATION WITH EXTERNAL COS...ESTIMATION AND COMPARISION OF COST OF COAL POWER GENERATION WITH EXTERNAL COS...
ESTIMATION AND COMPARISION OF COST OF COAL POWER GENERATION WITH EXTERNAL COS...
 
Optimal Operation of Wind-thermal generation using differential evolution
Optimal Operation of Wind-thermal generation using differential evolutionOptimal Operation of Wind-thermal generation using differential evolution
Optimal Operation of Wind-thermal generation using differential evolution
 

Similar to Monitoring and Forecasting of Air Emissions with IoT Measuring Stations and a SaaS Cloud Application

IRJET-E-Waste Management using Robotics
IRJET-E-Waste Management using RoboticsIRJET-E-Waste Management using Robotics
IRJET-E-Waste Management using RoboticsIRJET Journal
 
Design and Implementation of Smart Air Pollution Monitoring System Based on I...
Design and Implementation of Smart Air Pollution Monitoring System Based on I...Design and Implementation of Smart Air Pollution Monitoring System Based on I...
Design and Implementation of Smart Air Pollution Monitoring System Based on I...BRNSSPublicationHubI
 
Implementation of Integration VaaMSN and SEMAR for Wide Coverage Air Quality ...
Implementation of Integration VaaMSN and SEMAR for Wide Coverage Air Quality ...Implementation of Integration VaaMSN and SEMAR for Wide Coverage Air Quality ...
Implementation of Integration VaaMSN and SEMAR for Wide Coverage Air Quality ...TELKOMNIKA JOURNAL
 
Design and Implementation of Smart Environmental Air Pollution Monitoring Sys...
Design and Implementation of Smart Environmental Air Pollution Monitoring Sys...Design and Implementation of Smart Environmental Air Pollution Monitoring Sys...
Design and Implementation of Smart Environmental Air Pollution Monitoring Sys...BRNSSPublicationHubI
 
An Efficient Tracking System for Air and Sound.pdf
An Efficient Tracking System for Air and Sound.pdfAn Efficient Tracking System for Air and Sound.pdf
An Efficient Tracking System for Air and Sound.pdfAakash Sheelvant
 
IRJET- Air Quality and Dust Level Monitoring using IoT
IRJET-  	  Air Quality and Dust Level Monitoring using IoTIRJET-  	  Air Quality and Dust Level Monitoring using IoT
IRJET- Air Quality and Dust Level Monitoring using IoTIRJET Journal
 
Implementation of environmental monitoring based on KAA IoT platform
Implementation of environmental monitoring based on KAA IoT platformImplementation of environmental monitoring based on KAA IoT platform
Implementation of environmental monitoring based on KAA IoT platformjournalBEEI
 
IRJET - Study on Smart Air Pollution Monitoring System based on IoT
IRJET - Study on Smart Air Pollution Monitoring System based on IoTIRJET - Study on Smart Air Pollution Monitoring System based on IoT
IRJET - Study on Smart Air Pollution Monitoring System based on IoTIRJET Journal
 
Final Synopsis -Bharathi(21-4-23).doc
Final Synopsis -Bharathi(21-4-23).docFinal Synopsis -Bharathi(21-4-23).doc
Final Synopsis -Bharathi(21-4-23).docmalli36
 
LORA BASED DATA ACQUISITION SYSTEM
LORA BASED DATA ACQUISITION SYSTEMLORA BASED DATA ACQUISITION SYSTEM
LORA BASED DATA ACQUISITION SYSTEMIRJET Journal
 
Adaptive Multi-Criteria-Based Load Balancing Technique for Resource Allocatio...
Adaptive Multi-Criteria-Based Load Balancing Technique for Resource Allocatio...Adaptive Multi-Criteria-Based Load Balancing Technique for Resource Allocatio...
Adaptive Multi-Criteria-Based Load Balancing Technique for Resource Allocatio...IJCNCJournal
 
Adaptive Multi-Criteria-Based Load Balancing Technique for Resource Allocatio...
Adaptive Multi-Criteria-Based Load Balancing Technique for Resource Allocatio...Adaptive Multi-Criteria-Based Load Balancing Technique for Resource Allocatio...
Adaptive Multi-Criteria-Based Load Balancing Technique for Resource Allocatio...IJCNCJournal
 
Design and Implementation of Portable Outdoor Air Quality Measurement System ...
Design and Implementation of Portable Outdoor Air Quality Measurement System ...Design and Implementation of Portable Outdoor Air Quality Measurement System ...
Design and Implementation of Portable Outdoor Air Quality Measurement System ...IJECEIAES
 
Real-time monitoring system for weather and air pollutant measurement with HT...
Real-time monitoring system for weather and air pollutant measurement with HT...Real-time monitoring system for weather and air pollutant measurement with HT...
Real-time monitoring system for weather and air pollutant measurement with HT...journalBEEI
 
Towards Adaptive Sensor-cloud for Internet of Things
Towards Adaptive Sensor-cloud for Internet of ThingsTowards Adaptive Sensor-cloud for Internet of Things
Towards Adaptive Sensor-cloud for Internet of ThingsTELKOMNIKA JOURNAL
 
CONTEXT-AWARE DECISION MAKING SYSTEM FOR MOBILE CLOUD OFFLOADING
CONTEXT-AWARE DECISION MAKING SYSTEM FOR MOBILE CLOUD OFFLOADINGCONTEXT-AWARE DECISION MAKING SYSTEM FOR MOBILE CLOUD OFFLOADING
CONTEXT-AWARE DECISION MAKING SYSTEM FOR MOBILE CLOUD OFFLOADINGIJCNCJournal
 
IRJET- Machine Learning for Weather Prediction and Forecasting for Local Weat...
IRJET- Machine Learning for Weather Prediction and Forecasting for Local Weat...IRJET- Machine Learning for Weather Prediction and Forecasting for Local Weat...
IRJET- Machine Learning for Weather Prediction and Forecasting for Local Weat...IRJET Journal
 

Similar to Monitoring and Forecasting of Air Emissions with IoT Measuring Stations and a SaaS Cloud Application (20)

IRJET-E-Waste Management using Robotics
IRJET-E-Waste Management using RoboticsIRJET-E-Waste Management using Robotics
IRJET-E-Waste Management using Robotics
 
Design and Implementation of Smart Air Pollution Monitoring System Based on I...
Design and Implementation of Smart Air Pollution Monitoring System Based on I...Design and Implementation of Smart Air Pollution Monitoring System Based on I...
Design and Implementation of Smart Air Pollution Monitoring System Based on I...
 
Implementation of Integration VaaMSN and SEMAR for Wide Coverage Air Quality ...
Implementation of Integration VaaMSN and SEMAR for Wide Coverage Air Quality ...Implementation of Integration VaaMSN and SEMAR for Wide Coverage Air Quality ...
Implementation of Integration VaaMSN and SEMAR for Wide Coverage Air Quality ...
 
Design and Implementation of Smart Environmental Air Pollution Monitoring Sys...
Design and Implementation of Smart Environmental Air Pollution Monitoring Sys...Design and Implementation of Smart Environmental Air Pollution Monitoring Sys...
Design and Implementation of Smart Environmental Air Pollution Monitoring Sys...
 
An Efficient Tracking System for Air and Sound.pdf
An Efficient Tracking System for Air and Sound.pdfAn Efficient Tracking System for Air and Sound.pdf
An Efficient Tracking System for Air and Sound.pdf
 
IRJET- Air Quality and Dust Level Monitoring using IoT
IRJET-  	  Air Quality and Dust Level Monitoring using IoTIRJET-  	  Air Quality and Dust Level Monitoring using IoT
IRJET- Air Quality and Dust Level Monitoring using IoT
 
Implementation of environmental monitoring based on KAA IoT platform
Implementation of environmental monitoring based on KAA IoT platformImplementation of environmental monitoring based on KAA IoT platform
Implementation of environmental monitoring based on KAA IoT platform
 
IRJET - Study on Smart Air Pollution Monitoring System based on IoT
IRJET - Study on Smart Air Pollution Monitoring System based on IoTIRJET - Study on Smart Air Pollution Monitoring System based on IoT
IRJET - Study on Smart Air Pollution Monitoring System based on IoT
 
Final Synopsis -Bharathi(21-4-23).doc
Final Synopsis -Bharathi(21-4-23).docFinal Synopsis -Bharathi(21-4-23).doc
Final Synopsis -Bharathi(21-4-23).doc
 
LORA BASED DATA ACQUISITION SYSTEM
LORA BASED DATA ACQUISITION SYSTEMLORA BASED DATA ACQUISITION SYSTEM
LORA BASED DATA ACQUISITION SYSTEM
 
Adaptive Multi-Criteria-Based Load Balancing Technique for Resource Allocatio...
Adaptive Multi-Criteria-Based Load Balancing Technique for Resource Allocatio...Adaptive Multi-Criteria-Based Load Balancing Technique for Resource Allocatio...
Adaptive Multi-Criteria-Based Load Balancing Technique for Resource Allocatio...
 
Adaptive Multi-Criteria-Based Load Balancing Technique for Resource Allocatio...
Adaptive Multi-Criteria-Based Load Balancing Technique for Resource Allocatio...Adaptive Multi-Criteria-Based Load Balancing Technique for Resource Allocatio...
Adaptive Multi-Criteria-Based Load Balancing Technique for Resource Allocatio...
 
Design and Implementation of Portable Outdoor Air Quality Measurement System ...
Design and Implementation of Portable Outdoor Air Quality Measurement System ...Design and Implementation of Portable Outdoor Air Quality Measurement System ...
Design and Implementation of Portable Outdoor Air Quality Measurement System ...
 
11. 23762.pdf
11. 23762.pdf11. 23762.pdf
11. 23762.pdf
 
40120140505008
4012014050500840120140505008
40120140505008
 
Real-time monitoring system for weather and air pollutant measurement with HT...
Real-time monitoring system for weather and air pollutant measurement with HT...Real-time monitoring system for weather and air pollutant measurement with HT...
Real-time monitoring system for weather and air pollutant measurement with HT...
 
Towards Adaptive Sensor-cloud for Internet of Things
Towards Adaptive Sensor-cloud for Internet of ThingsTowards Adaptive Sensor-cloud for Internet of Things
Towards Adaptive Sensor-cloud for Internet of Things
 
Seminar report
Seminar reportSeminar report
Seminar report
 
CONTEXT-AWARE DECISION MAKING SYSTEM FOR MOBILE CLOUD OFFLOADING
CONTEXT-AWARE DECISION MAKING SYSTEM FOR MOBILE CLOUD OFFLOADINGCONTEXT-AWARE DECISION MAKING SYSTEM FOR MOBILE CLOUD OFFLOADING
CONTEXT-AWARE DECISION MAKING SYSTEM FOR MOBILE CLOUD OFFLOADING
 
IRJET- Machine Learning for Weather Prediction and Forecasting for Local Weat...
IRJET- Machine Learning for Weather Prediction and Forecasting for Local Weat...IRJET- Machine Learning for Weather Prediction and Forecasting for Local Weat...
IRJET- Machine Learning for Weather Prediction and Forecasting for Local Weat...
 

Recently uploaded

İTÜ CAD and Reverse Engineering Workshop
İTÜ CAD and Reverse Engineering WorkshopİTÜ CAD and Reverse Engineering Workshop
İTÜ CAD and Reverse Engineering WorkshopEmre Günaydın
 
RESORT MANAGEMENT AND RESERVATION SYSTEM PROJECT REPORT.pdf
RESORT MANAGEMENT AND RESERVATION SYSTEM PROJECT REPORT.pdfRESORT MANAGEMENT AND RESERVATION SYSTEM PROJECT REPORT.pdf
RESORT MANAGEMENT AND RESERVATION SYSTEM PROJECT REPORT.pdfKamal Acharya
 
NO1 Pandit Black Magic Removal in Uk kala jadu Specialist kala jadu for Love ...
NO1 Pandit Black Magic Removal in Uk kala jadu Specialist kala jadu for Love ...NO1 Pandit Black Magic Removal in Uk kala jadu Specialist kala jadu for Love ...
NO1 Pandit Black Magic Removal in Uk kala jadu Specialist kala jadu for Love ...Amil baba
 
Fruit shop management system project report.pdf
Fruit shop management system project report.pdfFruit shop management system project report.pdf
Fruit shop management system project report.pdfKamal Acharya
 
A case study of cinema management system project report..pdf
A case study of cinema management system project report..pdfA case study of cinema management system project report..pdf
A case study of cinema management system project report..pdfKamal Acharya
 
Construction method of steel structure space frame .pptx
Construction method of steel structure space frame .pptxConstruction method of steel structure space frame .pptx
Construction method of steel structure space frame .pptxwendy cai
 
Online blood donation management system project.pdf
Online blood donation management system project.pdfOnline blood donation management system project.pdf
Online blood donation management system project.pdfKamal Acharya
 
Democratizing Fuzzing at Scale by Abhishek Arya
Democratizing Fuzzing at Scale by Abhishek AryaDemocratizing Fuzzing at Scale by Abhishek Arya
Democratizing Fuzzing at Scale by Abhishek Aryaabh.arya
 
Quality defects in TMT Bars, Possible causes and Potential Solutions.
Quality defects in TMT Bars, Possible causes and Potential Solutions.Quality defects in TMT Bars, Possible causes and Potential Solutions.
Quality defects in TMT Bars, Possible causes and Potential Solutions.PrashantGoswami42
 
Explosives Industry manufacturing process.pdf
Explosives Industry manufacturing process.pdfExplosives Industry manufacturing process.pdf
Explosives Industry manufacturing process.pdf884710SadaqatAli
 
Laundry management system project report.pdf
Laundry management system project report.pdfLaundry management system project report.pdf
Laundry management system project report.pdfKamal Acharya
 
Natalia Rutkowska - BIM School Course in Kraków
Natalia Rutkowska - BIM School Course in KrakówNatalia Rutkowska - BIM School Course in Kraków
Natalia Rutkowska - BIM School Course in Krakówbim.edu.pl
 
Introduction to Machine Learning Unit-4 Notes for II-II Mechanical Engineering
Introduction to Machine Learning Unit-4 Notes for II-II Mechanical EngineeringIntroduction to Machine Learning Unit-4 Notes for II-II Mechanical Engineering
Introduction to Machine Learning Unit-4 Notes for II-II Mechanical EngineeringC Sai Kiran
 
Pharmacy management system project report..pdf
Pharmacy management system project report..pdfPharmacy management system project report..pdf
Pharmacy management system project report..pdfKamal Acharya
 
Introduction to Casting Processes in Manufacturing
Introduction to Casting Processes in ManufacturingIntroduction to Casting Processes in Manufacturing
Introduction to Casting Processes in Manufacturingssuser0811ec
 
2024 DevOps Pro Europe - Growing at the edge
2024 DevOps Pro Europe - Growing at the edge2024 DevOps Pro Europe - Growing at the edge
2024 DevOps Pro Europe - Growing at the edgePaco Orozco
 
Courier management system project report.pdf
Courier management system project report.pdfCourier management system project report.pdf
Courier management system project report.pdfKamal Acharya
 
NO1 Pandit Amil Baba In Bahawalpur, Sargodha, Sialkot, Sheikhupura, Rahim Yar...
NO1 Pandit Amil Baba In Bahawalpur, Sargodha, Sialkot, Sheikhupura, Rahim Yar...NO1 Pandit Amil Baba In Bahawalpur, Sargodha, Sialkot, Sheikhupura, Rahim Yar...
NO1 Pandit Amil Baba In Bahawalpur, Sargodha, Sialkot, Sheikhupura, Rahim Yar...Amil baba
 
RS Khurmi Machine Design Clutch and Brake Exercise Numerical Solutions
RS Khurmi Machine Design Clutch and Brake Exercise Numerical SolutionsRS Khurmi Machine Design Clutch and Brake Exercise Numerical Solutions
RS Khurmi Machine Design Clutch and Brake Exercise Numerical SolutionsAtif Razi
 

Recently uploaded (20)

İTÜ CAD and Reverse Engineering Workshop
İTÜ CAD and Reverse Engineering WorkshopİTÜ CAD and Reverse Engineering Workshop
İTÜ CAD and Reverse Engineering Workshop
 
RESORT MANAGEMENT AND RESERVATION SYSTEM PROJECT REPORT.pdf
RESORT MANAGEMENT AND RESERVATION SYSTEM PROJECT REPORT.pdfRESORT MANAGEMENT AND RESERVATION SYSTEM PROJECT REPORT.pdf
RESORT MANAGEMENT AND RESERVATION SYSTEM PROJECT REPORT.pdf
 
NO1 Pandit Black Magic Removal in Uk kala jadu Specialist kala jadu for Love ...
NO1 Pandit Black Magic Removal in Uk kala jadu Specialist kala jadu for Love ...NO1 Pandit Black Magic Removal in Uk kala jadu Specialist kala jadu for Love ...
NO1 Pandit Black Magic Removal in Uk kala jadu Specialist kala jadu for Love ...
 
Fruit shop management system project report.pdf
Fruit shop management system project report.pdfFruit shop management system project report.pdf
Fruit shop management system project report.pdf
 
A case study of cinema management system project report..pdf
A case study of cinema management system project report..pdfA case study of cinema management system project report..pdf
A case study of cinema management system project report..pdf
 
Construction method of steel structure space frame .pptx
Construction method of steel structure space frame .pptxConstruction method of steel structure space frame .pptx
Construction method of steel structure space frame .pptx
 
Standard Reomte Control Interface - Neometrix
Standard Reomte Control Interface - NeometrixStandard Reomte Control Interface - Neometrix
Standard Reomte Control Interface - Neometrix
 
Online blood donation management system project.pdf
Online blood donation management system project.pdfOnline blood donation management system project.pdf
Online blood donation management system project.pdf
 
Democratizing Fuzzing at Scale by Abhishek Arya
Democratizing Fuzzing at Scale by Abhishek AryaDemocratizing Fuzzing at Scale by Abhishek Arya
Democratizing Fuzzing at Scale by Abhishek Arya
 
Quality defects in TMT Bars, Possible causes and Potential Solutions.
Quality defects in TMT Bars, Possible causes and Potential Solutions.Quality defects in TMT Bars, Possible causes and Potential Solutions.
Quality defects in TMT Bars, Possible causes and Potential Solutions.
 
Explosives Industry manufacturing process.pdf
Explosives Industry manufacturing process.pdfExplosives Industry manufacturing process.pdf
Explosives Industry manufacturing process.pdf
 
Laundry management system project report.pdf
Laundry management system project report.pdfLaundry management system project report.pdf
Laundry management system project report.pdf
 
Natalia Rutkowska - BIM School Course in Kraków
Natalia Rutkowska - BIM School Course in KrakówNatalia Rutkowska - BIM School Course in Kraków
Natalia Rutkowska - BIM School Course in Kraków
 
Introduction to Machine Learning Unit-4 Notes for II-II Mechanical Engineering
Introduction to Machine Learning Unit-4 Notes for II-II Mechanical EngineeringIntroduction to Machine Learning Unit-4 Notes for II-II Mechanical Engineering
Introduction to Machine Learning Unit-4 Notes for II-II Mechanical Engineering
 
Pharmacy management system project report..pdf
Pharmacy management system project report..pdfPharmacy management system project report..pdf
Pharmacy management system project report..pdf
 
Introduction to Casting Processes in Manufacturing
Introduction to Casting Processes in ManufacturingIntroduction to Casting Processes in Manufacturing
Introduction to Casting Processes in Manufacturing
 
2024 DevOps Pro Europe - Growing at the edge
2024 DevOps Pro Europe - Growing at the edge2024 DevOps Pro Europe - Growing at the edge
2024 DevOps Pro Europe - Growing at the edge
 
Courier management system project report.pdf
Courier management system project report.pdfCourier management system project report.pdf
Courier management system project report.pdf
 
NO1 Pandit Amil Baba In Bahawalpur, Sargodha, Sialkot, Sheikhupura, Rahim Yar...
NO1 Pandit Amil Baba In Bahawalpur, Sargodha, Sialkot, Sheikhupura, Rahim Yar...NO1 Pandit Amil Baba In Bahawalpur, Sargodha, Sialkot, Sheikhupura, Rahim Yar...
NO1 Pandit Amil Baba In Bahawalpur, Sargodha, Sialkot, Sheikhupura, Rahim Yar...
 
RS Khurmi Machine Design Clutch and Brake Exercise Numerical Solutions
RS Khurmi Machine Design Clutch and Brake Exercise Numerical SolutionsRS Khurmi Machine Design Clutch and Brake Exercise Numerical Solutions
RS Khurmi Machine Design Clutch and Brake Exercise Numerical Solutions
 

Monitoring and Forecasting of Air Emissions with IoT Measuring Stations and a SaaS Cloud Application

  • 1. Monitoring and Forecasting of Air Emissions with IoT Measuring Stations and a SaaS Cloud Application Fernando Ar´evalo N.∗, M. Tahasanul Ibrahim∗, Rizky M. Diprasetya∗, Omar Otoniel Flores†, Andreas Schwung∗ ∗Department of Automation Technology South Westphalia University of Applied Sciences, Germany Email: {arevalo.fernando,schwung.andreas}@fh-swf.de †Applied Sciences School Universidad Tecnol´ogica de El Salvador, El Salvador Email: omar.flores@utec.edu.sv Abstract—Environmental pollution is a major problem in densely populated cities, due to the increase of respiratory diseases in the population. The pollution level is used to inform the public, in order to take special measures based on a pollution scale. Particulate matter (PM), specifically PM10 and PM2.5, is used to estimate the degree of pollution. The monitoring and analysis of these two variables has attracted the attention of the research community, particularly in the design of measuring stations, forecasting, and software infrastructure to host user applications. This paper proposes a modular and ready-to-use architecture for a SaaS cloud application for air emissions fore- casting using IoT measuring stations. The SaaS cloud application is connected to the IoT measuring stations located in the densely populated cities of El Salvador. The data is used to create deep learning models, to make forecasts for the next day. Index Terms—Air Emissions, IoT, LSTM, CNN-1D, Docker I. INTRODUCTION Pollution is a major problem in cities with large populations. Direct consequences of high concentration of pollutants are reflected in health problems in the population. Estimation of the pollution level is used to assess the public, e.g. if the level of pollution is within a certain range, outdoor sport should be limited. Air pollution can be measured by the analysis of particulate matter (PM), normally defined as PM10 and PM2.5 corresponding to a diameter of 10 and 2.5 micrometers respectively [1]. Karagulian et al. [1] presented a review of the main contributors for the generation of PM10 and PM2.5 in the air. The authors defined the following categories: traffic, industry, domestic fuel combustion, natural sources such as dust and sea salt, and unspecified sources. This research focus on the use of PM10 and PM2.5 to estimate the air pollution. Monitoring and Analysis of PM10 and PM2.5 are topics of high interest in literature, specially for the design of measuring stations, software for data analysis, and software infrastructure for hosting user applications. Low cost IoT measuring stations allow data to be collected in a massive way, due to its fast implementation, access to communication protocols, and its low investment. The proliferation of cloud services have, on the other side, facilitated the implementation of back-end infrastructures, so that the user application has all the necessary services running on the background, such as database, communication protocols, and user interface. These days monitoring and analysis of air quality data is not enough, because the population wants to know the forecast, similar to the weather forecasting service, so in this manner it can be aware of the pollution level of the next day and take the necessary measures. This has motivated research on PM10 and PM2.5 forecasting, which have moved from classical sta- tistical approaches such as Autoregressive Integrated Moving Average (ARIMA) and regression, to more modern algorithms based on deep learning. This paper proposes a modular and ready-to-use architecture for a SaaS cloud application for air emissions forecasting using IoT measuring stations. Its docker-based architecture allows the system to integrate micro-services such as M2M com- munication, a database, forecasts and visualisation through a web application. Its modularity allows new measuring stations and additional sensors to be integrated into the SaaS cloud application. This research presents a comparison of two time series multivariate deep-learning-based approaches. This paper is structured as follows: Section II portrays a literature survey on IoT measuring stations for air emissions, cloud-based applications, and air emissions forecasting. Sec- tion III presents the proposed model. Section IV describes the use case, including implementation and forecasting results. The conclusions and prospective work are summarised in section V. II. RELATED WORK Liu et al. [2] proposed an IoT microcontroller-based solution for the measurement of PM2.5, PM10, NO, SO2. The real time air quality monitoring system used Long Range (LoRa) technology to communicate with the control centre. Venkata- narayanan et al. [3] proposed a smart bicycle application for
  • 2. monitoring air quality. The authors designed a microcontroller- based station for measuring NO2 and SO2. The measurements were sent to a cloud application. An Android application was implemented for monitoring air quality. Kumar and Jasuja [4] proposed an IoT system to monitor air quality monitoring with Arduino, Raspberry and a cloud platform. The system monitored PM2.5, CO, CO2, Temperature and Humidity. Bao [5] proposed a mobile data acquisition node with a microcontroller and a GPS-GPRS module. A cloud platform received the particle concentration data PM2.5, PM10; as well as the geographic location data. An Android application was designed for the data monitoring. Beeharry et al. [6] designed an IoT device that was based on Arduino and Weather Shield. The device sent the weather information to a local gateway. From this gateway the data was sent to a cloud database. A server is assigned to retrieve the data from the cloud database and processes it to make the weather forecast. The prediction is performed using K-NN. The result is sent to either a web page or a smartphone applica- tion. Liu et al. [7] proposed an ensemble model for multivariate time series forecasting based on LSTM. The authors predicted PM2.5. Mahajan et al. [8] compared the PM2.5 forecasts from an Autoregressive Integrated Moving Average (ARIMA) model and a Neural Network Autoregression (NNAR). The models predicted the next hour. Xu and Yoneda [9] proposed a LSTM-Autoencoder Model for PM2.5 forecasting. They tested their model using the Beijing dataset. They compared their results with a RNN and ARIMA models. Du et al. [10] proposed a hybrid model using 1D-CNN and a Bi-LSTM to forecast PM2.5 using the multivariate time series from the Beijing Dataset. Huang et al. [11] proposed a mobile system architecture consisting of an air quality measuring station, a smartphone App and a cloud platform. The system measured PM2.5 from inside a car, allowing samples to be taken from different locations and at different times. The authors used a random forest model to classify the air circulation on the vehicle considering factors such as opening/closing windows, the use of air conditioner in a in-loop and a out-loop modes. Song et al. [12] proposed an IoT system using LoRa and LSTM for PM2.5 forecasting. The Arduino-based sensing unit collected PM2.5 samples, and sent them to a cloud platform via a LoRa gateway. Raafat et al. [13] proposed an algorithm for novelty detection in fog level with the measurements of light, temperature, CO2 and humidity. The authors used Levene’s test to identify novelties during feature extraction, and a Feed- Forward Neural Network for classification on events detection. Previous research focused either on the proposal of hard- ware for air quality sensing units, time series multivariate forecasting, or the integration of the monitoring systems to commercial cloud platforms. This paper proposes a modular and ready-to-use architecture for a SaaS cloud application for monitoring and forecasting of air emissions using IoT measuring stations. III. MONITORIG AND FORECASTING OF AIR EMISSIONS WITH IOT MEASURING STATIONS AND A SAAS CLOUD APPLICATION The proposed model provides the software infrastructure for the connectivity of the IoT devices, as well as the features of monitoring and analysis of the collected data. This is illustrated in Fig. 1. The IoT device can be a microcontroller, a sensor, a Programmable Logic Controller (PLC), or even another computer. The forecasting feature is implemented in the Dashboard and System Process module. Fig. 1: Proposed Model Architecture A. Air Emissions Data Acquisition with IoT Measuring Sta- tions The IoT measuring station is shown in Fig. 1 as an IoT device. The station collects data from the sensors e.g. Tem- perature, PM2.5, CO, CO2, humidity, etc. It can be based on a low-cost implementation with microcontrollers or micro- computers such as Raspberry Pi or Odroid. The IoT measuring station can communicate directly with a cloud service, or with a central control. It can have internet connectivity through the use of communication protocols such as MQTT or through a RESTful method. Implementations have been reported using LoRa between stations and a gateway. The gateway finally connects to the cloud or a central control system via MQTT or a RESTful method. B. SaaS Cloud Application The software as a Service (SaaS) system is designed in a cloud server where all the dynamics of the services are set up according to the needs. The server has a modular and scaleable architecture, since docker containers are used for each of the services.
  • 3. 1) Docker Container System: The idea of a docker con- tainer is to contain all the modular events and processing within one or many virtual images, while keeping each ap- plication isolated without direct environment dependencies. There are several aspects of docker containment systems that made this method suitable for virtualising services. Docker containers run in self-isolated environments, which means that no external system other than the docker daemon or the con- tainer itself can access them. Docker container isolates its con- tainers mainly in three different sectors: ”Process Isolation”, ”File System Isolation”, ”Network Isolation” [14]. Docker containers has earned its reputation for speed, portability, and scalability. The dockerfile contains instructions on what type of operating system to install and what other applications are necessary. Since each container is intended for specific applications, not all components of the operative system are needed. Therefore the whole process of container production is faster and lighter for the container. As soon a container is built or executed, the Docker Daemon wraps the entire container together with its respective processes in a PID namespace. This allows the host machine to allocate resources as needed, which is much faster and better than traditional virtual machines. The traditional ones assign the host resources completely to the virtual machine regardless of its usage [15]. Fig. 2: SaaS Communication 2) Application Backbone Process: The Software as a Ser- vice (SaaS) system is designed in a cloud server on which the entire dynamic of services is set up according to requirements. IoT devices and users around the globe can authenticate, connect and use these services on demand through a com- munication system hosted in the SaaS cloud application. The services offered are divided into two groups: ”Pushing Data” and ”Pulling Data”. The groups are determined by the type of service, for example, publishing data from the server node is categorised as ”pushing data” while viewing published data from the server node is categorized as ”pulling data”. Nevertheless, the principle of the SaaS cloud application is to receive requests from either users or IoT devices and forward them to the internal server, where multiple docker containers are running to provide the requested service, see Fig. 2. C. Air Emissions Forecasting using Deep Learning Artificial Neural Network (ANN) tends to struggle with the computational complexity of computer image data. Later, the Convolutional Neural Network (CNN) was introduced to re- duce the complexity of the problem. The CNN architecture are based on three different types of layers: Convolutional layers, pooling layers and fully-connected layers. There are different types of convolutional layers such as 1D convolutional, 2D convolutional and 3D convolutional. 1D Convolutional com- prises a vector of neurons. 2D convolutional is based on a 2-dimensional array of neurons, which are defined as width and height [16]. The 2D convolutional layer is a well-known method for greyscale computer vision. The 3D convolutional layer consists of a 3-dimensional array of neurons, which are defined as width, height and depth. This type of convolutional layer is normally used for a color image, e.g. RGB, which has 3 different type of color [17]. For the forecasting, the air emission data are treated as time series data that can be defined in a vector form. For this purpose the 1D convolutional neural network is used for forecasting. Long short-term Memory (LSTM) is a well-known type of recurrent neural network (RNN) for prediction based on time series data. It consists of a cell, an input, an output and a forget gate. The cell remembers all values over time intervals. The input gate controls the signal from the network to the cell. These prevent an irrelevant input from entering the neuron. The output gate protects the memory cell from disturbance originated from the neuron [18]. Both the input and output gates are thus the flow control of constant error carousel (CEC), which gives a constant error flow within the cell. This solves the problem that the standard RNN has where it cannot bridge more than 5 - 10 time steps, while LSTM can bridge more than 1000 time steps. IV. USE CASE: MONITORING AND FORECASTING OF URBAN AIR EMISSIONS IN EL SALVADOR The Universidad Tecnol´ogica de El Salvador (UTEC) pro- posed a low-cost IoT measuring station to solve the problem of missing online measurement of air emissions. The project aimed at measuring air emissions in densely populated cities in El Salvador, see Fig. 3. The IoT measuring stations are located in the cities of Santa Ana (STNA), Chalatenango (CHLT), La Libertad (ITCA) and San Salvador (UTEC). The SaaS cloud application collects the data from all stations and visualises it in the dashboard of the web application. The air emissions forecasting focuses on the data of the city of San Salvador (UTEC station), as it is the one with more data points. A. Data Acquisition with an IoT Measuring Station The IoT measuring station consists of four main compo- nents: an ESP32 microcontroller, a PM sensor that measures PM2.5 and PM10, a temperature sensor and a local display.
  • 4. Fig. 3: Location of the IoT Stations in the Map Fig. 4 shows the main components and the IoT measuring sta- tion. The IoT measuring station communicates with the SaaS Cloud application via the MQTT communication protocol. For this purpose a MQTT topic is defined for each measurement signal. The IoT measuring stations take a sample of PM10, PM2.5 and the temperature every 15 minutes. Fig. 4: Components of the IoT Measuring Station [19] B. SaaS Cloud Application In a previous paper, a first version of the current SaaS Cloud Application was proposed in terms of services and communication protocols [20]. The main server, which is the backbone of the entire SaaS application, runs on a virtual machine based on the Amazon Web Service (AWS). The idea of the system is to run sets of docker containers for different services within the server, which together will form the overall air emissions service. For the air emission project a total of 3 docker containers were used, see Fig. 5. Eclipse- Fig. 5: SaaS Process Mosquitto MQTT Broker is selected as the module responsible for communications between the IoT measuring stations and the server for pushing and pulling data. The MQTT broker is lightweight and supports multiple ”Quality of Services” (QoS), including device handshaking, making it a prime candidate for the communication module. The MySQL Database runs in a container, which is the primary data storage unit for the system. All requests for data retrieval are processed via this container. Dashboard and System Process is a custom container build with an Ubuntu 18.04 image. The design of the container enables dynamic data input on-the-fly. This is done using an Excel spreadsheet that exists on the host machine and is used together with the container to enter the node and sensor names according to the variable changes. The Excel file contains predefined formulas for the automatic generation of the MQTT node addresses according to the user input. When the container is spun, it traverses the Excel file and subscribes to the respective MQTT nodes and MySQL database tables that exist in the file. Fig. 6 and Fig. 7 show the dashboard Home and the Chart of the UTEC IoT measuring station re- spectively. Inside the container there is also a flask application running that renders the HTML templates for the dashboard using the data imported from the MySQL Databases. User requests to view the dashboard panel are handled using this flask application. Data request flow process is similar to the generic SaaS application, the air emissions system also works in three steps:
  • 5. Request, Process, and Response. An IoT measuring station or a user starts his journey with a simple request to the server. Requests received from IoT measuring stations (TCP port 1883) take the form of a publication that the MQTT server publishes to their respective nodes for further processing. Other processes are performed by the user-defined container (Excel file) , which scans for new MQTT publications and in response stores the data in the corresponding tables in the MySQL Database. On the other hand, there is only one possible user request, namely to display the dashboard. This way, requests from user devices (TCP port 80/443) are re- ceived directly by the flask application of the custom container. Once the request is captured, flask processes the requested html template filled with the requested data from the MySQL database. In response to the request, the edited template is forwarded to the user’s web browser to be displayed with the latest data from the MySQL database, see Fig. 5. Fig. 6: Dashboard Home of IoT Stations Fig. 7: Forecasting using data of IoT Station UTEC C. Air Emissions Forecasting 1) Data Pre-processing: This paper proposes a forecasting model in which 96 data points of the previous day are used to predict the next 96 data points of the current day. The IoT measuring station collects the air emission data and the time stamp. Then the data is reconstructed into the following structure: • Previous Day • Previous Hour • Previous Minute • Current Day • Previous PM2.5 • Previous PM10 • Previous Temperature The specific day, hour, and minute are taken into account, as each different time has its own unique features. The data is then fed into the forecasting model. The output will be the predicted value of PM2.5, PM10, and temperature for the whole day. 2) 1D-CNN: A 1D convolutional neural network is used, since the forecasting problem is based on time series data. This is followed by the max-pooling layer with a pool size of 2, after which the result is flattened and fed into a fully- connected layer with 50 neurons. Then again a fully-connected layer with one neuron is used for the output. 3) LSTM: The LSTM model has 100 cells. After that there is a 20% dropout chance. Then it is connected directly to the output. 4) Implementation: The historical data of air emissions is collected. Then the data is split into training data and testing data. There are 9015 samples for the training data and 800 samples for the testing data. The samples are selected manu- ally due to the unclean dataset. Adam is used as optimizer and the mean squared error for the loss function. The training was conducted for 100 epochs for each model. Fig. 8a shows that LSTM and CNN models have the same performance. Both the LSTM model and the CNN model have the same precision of ±0.04. Fig. 8b and Fig. 8c show that the LSTM model performs better than CNN model. For PM2.5, the precision of the LSTM result is ±6.6, whereas the CNN result is ±6.7. For PM10, the precision of the LSTM result is ±8, whereas the CNN result is ±8.2. But in terms of losses, the LSTM model still shows high losses. D. Discussion The IoT measuring station has a good performance in the acquisition of air emissions signals. A drawback, however, is that if a change to the firmware of the IoT measuring station is required, the developer must travel to the IoT station and make the changes on site. The SaaS Cloud is a scalable and ready-to-use solution as it allows the addition of more signals and more IoT measuring stations. However, the changes must be manually updated in the git platform. As the results show, the forecast model still does not perform well by checking the high losses. There are many missing data points in the historical data and some error measurements that result in the dataset not being clean. In forecasting, there is still room for improvement of the models to increase its performance. V. CONCLUSION This paper proposed a modular and ready-to-use architec- ture for a SaaS cloud application for monitoring and forecast- ing of air emissions using IoT stations. The performance of the low-cost IoT station was satisfactory, and allowed to collect
  • 6. (a) Temperature (b) PM2.5 (c) PM10 Fig. 8: Loss Test data of PM2.5, PM10 and Temperature. The cloud application proved to be scalable, since it allowed two additional IoT stations to be added during the project development. The changes were made simply by updating the user excel file containing the MQTT topics. This updated the dashboard of the web application as well as the tables in the database. This research presented two deep learning approaches 1D-CNN and LSTM. The performance of both algoritms have room for improvement, the reason for this is based on missing data points and error measurements. The prospective work will focus on the development of a smart IoT station that allows remote programming, the devel- opment of a dynamic SaaS cloud application, and performance improvement of the deep-learning-based forecasting. REFERENCES [1] F. Karagulian, C. A. Belis, C. F. C. Dora, A. M. Pr¨uss-Ust¨un, S. Bonjour, H. Adair-Rohani, and M. Amann, “Contributions to cities’ ambient par- ticulate matter (PM): A systematic review of local source contributions at global level,” pp. 475–483, nov 2015. [2] S. Liu, C. Xia, and Z. Zhao, “A low-power real-time air quality monitoring system using LPWAN based on LoRa,” in 2016 13th IEEE International Conference on Solid-State and Integrated Circuit Technology, ICSICT 2016 - Proceedings. Institute of Electrical and Electronics Engineers Inc., 2016, pp. 379–381. [3] A. Venkatanarayanan, A. Vijayavel, A. Rajagopal, and P. Nagaradjane, “Design of sensor system for air pollution and human vital monitoring for connected cyclists,” IET Communications, vol. 13, no. 19, pp. 3181– 3186, nov 2019. [4] S. Kumar and A. Jasuja, “Air quality monitoring system based on IoT using Raspberry Pi,” in Proceeding - IEEE International Conference on Computing, Communication and Automation, ICCCA 2017, vol. 2017- Janua. Institute of Electrical and Electronics Engineers Inc., dec 2017, pp. 1341–1346. [5] L. Q. Bao, “Cloud connection oriented real-time monitoring system for atmospheric particles,” IET Wireless Sensor Systems, vol. 10, no. 1, pp. 31–36, feb 2020. [6] Y. Beeharry, T. P. Fowdur, and J. A. Sunglee, “A Cloud-Based Real-Time Weather Forecasting Application,” in 2019 14th International Confer- ence on Advanced Technologies, Systems and Services in Telecommu- nications, TELSIKS 2019 - Proceedings. Institute of Electrical and Electronics Engineers Inc., oct 2019, pp. 294–297. [7] F. Liu, M. Cai, L. Wang, and Y. Lu, “An Ensemble Model Based on Adaptive Noise Reducer and Over-Fitting Prevention LSTM for Multivariate Time Series Forecasting,” IEEE Access, vol. 7, pp. 26 102– 26 115, 2019. [8] S. Mahajan, L.-j. Chen, and T.-c. Tsai, “An Empirical Study of PM2 . 5 Forecasting Using Neural Network,” in 2017 IEEE SmartWorld, Ubiquitous Intelligence & Computing, Advanced & Trusted Com- puted, Scalable Computing & Communications, Cloud & Big Data Computing, Internet of People and Smart City Innovation (Smart- World/SCALCOM/UIC/ATC/CBDCom/IOP/SCI). IEEE, 2017, pp. 1–7. [9] X. Xu and M. Yoneda, “Multitask Air-Quality Prediction Based on LSTM-Autoencoder Model,” IEEE Transactions on Cybernetics, pp. 1– 10, oct 2019. [10] S. Du, T. Li, Y. Yang, and S.-J. Horng, “Deep Air Quality Forecast- ing Using Hybrid Deep Learning Framework,” IEEE Transactions on Knowledge and Data Engineering, pp. 1–1, nov 2019. [11] J. Huang, N. Duan, P. Ji, C. Ma, F. Hu, Y. Ding, Y. Yu, Q. Zhou, and W. Sun, “A Crowdsource-Based Sensing System for Monitoring Fine-Grained Air Quality in Urban Environments,” IEEE Internet of Things Journal, vol. 6, no. 2, pp. 3240–3247, apr 2019. [Online]. Available: https://ieeexplore.ieee.org/document/8534351/ [12] W. Song, J. Han, J. Xie, Y. Gao, and L. Song, “System for detecting and forecasting pm2.5 concentration levels using long short-Term memory and lora,” in Proceedings - 2019 IEEE International Congress on Cybermatics: 12th IEEE International Conference on Internet of Things, 15th IEEE International Conference on Green Computing and Commu- nications, 12th IEEE International Conference on Cyber, Physical and So. IEEE, 2019, pp. 834–841. [13] H. M. Raafat, M. S. Hossain, E. Essa, S. Elmougy, A. S. Tolba, G. Muhammad, and A. Ghoneim, “Fog Intelligence for Real-Time IoT Sensor Data Analytics,” IEEE Access, vol. 5, pp. 24 062–24 069, sep 2017. [14] T. Bui, “Analysis of Docker Security,” jan 2015. [Online]. Available: http://arxiv.org/abs/1501.02967 [15] A. M. Joy, “Performance comparison between Linux containers and virtual machines,” in International Conference on Advances in Computer Engineering and Applications (ICACEA), 2015, {International Confer- ence on Advances in Computer Engineering and Applications} and ICACEA. IEEE, 2015, pp. 342–346. [16] D. V. Sang, N. Van Dat, and D. P. Thuan, “Facial expression recognition using deep convolutional neural networks,” in Proceedings - 2017 9th International Conference on Knowledge and Systems Engineering, KSE 2017, vol. 2017-January. Institute of Electrical and Electronics Engineers Inc., nov 2017, pp. 130–135. [17] B. Chen and C. Jung, “Patch-Based Stereo Matching Using 3D Convo- lutional Neural Networks,” in Proceedings - International Conference on Image Processing, ICIP. IEEE Computer Society, aug 2018, pp. 3633–3637. [18] Y. Heryadi and H. L. H. S. Warnars, “Learning temporal representation of transaction amount for fraudulent transaction recognition using CNN, Stacked LSTM, and CNN-LSTM,” in 2017 IEEE International Confer- ence on Cybernetics and Computational Intelligence, CyberneticsCOM 2017 - Proceedings, vol. 2017-November. Institute of Electrical and Electronics Engineers Inc., mar 2018, pp. 84–89. [19] O. Flores-Cortez, R. Adalberto Cortez, and V. Rosa, “A Low-cost IoT System for Environmental Pollution Monitoring in Developing Countries,” in Proceedings of the 26th International Conference ”Mixed Design of Integrated Circuits and Systems”, MIXDES 2019, 2019. [20] F. Arevalo, M. R. Diprasetya, and A. Schwung, “A Cloud-based Architecture for Condition Monitoring based on Machine Learning,” in Proceedings - IEEE 16th International Conference on Industrial Informatics, INDIN 2018. Institute of Electrical and Electronics Engineers Inc., sep 2018, pp. 163–168.