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

net/publication/348164997

Agriculture Management Based on LoRa Edge Computing System

Chapter · January 2020

DOI: 10.1007/978-3-030-66471-8_10

CITATIONS READS
15 823

6 authors, including:

Mohammed Saleh Ali Muthanna Van Dai Pham

Southern Federal University Swinburne Vietnam
70 PUBLICATIONS 780 CITATIONS 36 PUBLICATIONS 197 CITATIONS

SEE PROFILE SEE PROFILE

Abdukodir Khakimov Ammar Muthanna

Peoples' Friendship University of Russia (RUDN University) Saint-Petersburg State University of Telecommunications
49 PUBLICATIONS 555 CITATIONS 216 PUBLICATIONS 2,981 CITATIONS

SEE PROFILE SEE PROFILE

All content following this page was uploaded by Abdukodir Khakimov on 16 January 2021.

The user has requested enhancement of the downloaded file.

 1

1
 Agriculture management based on LoRa Edge
Computing System

Fatkhullokhodzha Sharofidinov1[0000−0001−7913−3196] , Mohammed Saleh Ali

Muthanna2,4[0000−0002−1165−7812] , Van Dai Pham1[0000−0003−1363−0784] ,
Abdukodir Khakimov3[0000−0003−2362−3270] , and Ammar
Muthanna1[0000−0003−0213−8145] and Konstantin
Samouylov3[0000−0002−6368−9680]
1
The Bonch-Bruevich Saint-Petersburg State University of
Telecommunications,Prospek Bolshevikov 22, St. Petersburg, Russia
2
Department of Automation and Control Processes Saint Petersburg
ElectrotechnicalUniversity “LETI” Saint Petersburg, Russia
3
Peoples’ Friendship University of Russia (RUDN University),
6 Miklukho-MaklayaSt, Moscow, 117198, Russian
4
School of Computer Science and Technology, Chongqing University of Posts and
Telecommunications, Chongqing, China
fatkhullo0998@gmail,com,[email protected],[email protected],
[email protected],[email protected],[email protected]

Abstract. Internet of Things (IoT) technologies represent the future

challenges of computing and communications. They can also be use-
ful to improve traditional farming practices worldwide. Since the areas
where agricultural land is located in remote places, there is a need for
new technologies. These technologies must be suitable and reliable for
communication over long distances and, at the same time,consume little
energy. In particular, one of these relatively new technologies is the LoRa
communication protocol, which uses long waves to work over long dis-
tances.This is extremely useful in agriculture, where the communicating
areas are broad fields of crops and greenhouses. This study developed
a greenhouse monitoring system based on LoRa technology, designed to
work over long distances. The edge computing paradigms with a machine
learning mechanism are proposed to analyze and control the state of the
greenhouse, and in particular, to reduce thea mount of data transmitted
to the server.

Keywords: LoRa, Edge Computing, Precision Agriculture, Machine

Learning.

1 Introduction
Agriculture has always been one of the most critical sectors in the life of ev-
ery state. The foundation of human survival depends on agriculture in many
?
The publication has been prepared with the support of the “RUDN University Pro-
gram 5-100” (recipients A. Khakimov). For the research, infrastructure of the 5G
Lab RUDN (Russia) was used.
 Agriculture management based on LoRa Edge Computing System 3

respects,as the health of the population depends precisely on the quantity and
quality of manufactured products for everyday consumption in the agricultural
industry [1,2].Today, there is a rather serious threat to the food security of the
Earth’s population due to the exponential growth of its population, urbanization,
which suggests that the share of the population that is engaged in agriculture is
significantly falling[3].This problem is added to the process of global warming,
which is overgrowing. This means that to prevent the global crisis associated
with a lack of nutrition and the loss of a significant part of the crop, the intro-
duction of modern technologies is necessary. The development of modern tech-
nologies can help us face these challenges.Agriculture has been free from the use
of digital technologies [4][17]. Almost all jobs involved in people. All yield data
were collected manually, and analytic was not relevant for agriculture since the
amount of data on which conclusions and forecasts can be drawn was small. The
work was planned based on weather conditions, data on soil quality, and environ-
mental conditions, which were collected manually. Also, the standard processing
schedule, which includes demoralization, fertilizer, continuous irrigation, led to
cost overruns and undetected problems since local characteristic sand natural
variability was not taken into account. With the growing interest and demand
for intelligent agriculture, many attempts have been made to introduce Biotech-
nology on the farm[5]. IoT technology [6,7] allows farmers to control field plots
using connected sensors via the Internet from anywhere in the world. Farmers
and agronomists have already begun to use technology to improve their produc-
tivity. For example, sensors located in greenhouses allow you to receive detailed
data in real time in the form of variables, such as the temperature of the soil and
the environment,irrigation water and soil conductivity, the acidity level (PH) of
the soil and irrigation water, the properties of irrigation water, data on nutrient
composition of the soil, etc.These data can be transmitted and analyzed using
communication technologies and artificial intelligence (AI) paradigms. Farmers
will be able to remotely monitor crop sand equipment using their smartphones,
as well as analyze some statistics. Therefore,our proposed management system
based on modern communication technologies can be a solution for the intro-
duction of smart (precision) agriculture.
The organization of the paper will be as follows: Section 2 presents the related
work. Section 3 provides the precision agriculture system architecture based on
edge computing. Section 4 proposed a monitoring system for precision agricul-
ture. Section5 experiment and results And the conclusion will be given in section
6.

2 Related works

Many works have been done regarding the use of IoT in agriculture. Authors
in[8] analyzed the functionality of the most innovative and trending technologies
in the agricultural sector, as well as the possibilities of the Internet of things
in agriculture,this work also arguments are given, why the introduction of such
technologies is not only useful but also cost effective. Since LoRa technology is
4 F.Sharofidinov et al.

one of the most attractive in terms of its application in agriculture, many studies
have been cited in this area. For example, in [9,15], a study was carried out on the
implementation of LoRa technology as a transmission protocol between nodes of
a sensor network in a visual monitoring system for agricultural fields, the idea is
to transfer only the changing parts of the image frame and transfer data using
LoRa, when bandwidth is from 50 bits to 50 kbit / s. Authors in the [10] are
developing the LoRaWAN gateway mobile device,which can be used to increase
the productivity and accuracy of greenhouses. The presented development in-
cludes the use of the Halter LoRa mini gateway, managed by the Raspberry Pi
3 B +. There are cases where machine learning has been used to optimize costs
as well as the production process. For example, in [11] is a machine based, accu-
rate, and intelligent irrigation system with LoRa P2P networks for automatically
and seamlessly learning the irrigation experience from experienced farmers for
organic greenhouse crops. The proposed system will first calculate the amount
of water for each irrigation based on a trained irrigation model in combina-
tion with environmental data such as air temperature/humidity, moderate soil
moisture,light intensity, etc., and then automatically water crops through long
distance and lowpower wireless LoRa P2P network. Authors in the [12] present
an analysis of the influence of the parameters of the variant physical layer on the
performance of LoRa networks in a tree farm. The main goal of our study is to
develop a system for monitoring and controlling the process of growing crops in
year round greenhouses,where vegetables are grown for everyday consumption.
Our system is based on the principles of data transmission using LoRa technol-
ogy using edge computing paradigms to reduce the amount of traffic that will
be further transmitted to the cloud. Cloud computing will be used to calculate,
analyze, predict the state of culture and soil. Forecasting will be based on data
received from soil moisture sensors in the greenhouses, room temperature and
humidity sensors, crop status sensors, as well as a sensor that measures soil
composition.

3 Precision Agriculture System Architecture Based on

Edge Computing
This section provides details on the architecture system for four layers 1) sensor
layer 2) edge layer 3) network/cloud layer and 4) application layer . and results
of the architecture system using edge computing are discussed in terms of delay
time.

3.1 Layer-based architecture.

The architecture of our system consists of 4 layers, as shown in Figure 1. Sensor
layer. This level of our system consists directly of sensors: temperature, soil,
air humidity, CO2, and illumination. IoT devices are connected to the gateway
using energy efficient wireless technology LoRa. At this level, data is being read
from our greenhouse objects.
 Agriculture management based on LoRa Edge Computing System 5

3.2 Edge Layer

This level solves the following tasks: data exchange with sensors, preliminary
analysis of received data, diagnostics of sensor status, data compression. The
edge calculation paradigm is used. The advantages of using edge computing
for initial processing of data collected from sensors are in reducing the volume
of data uploaded to cloud servers, and in rapid response in critical situations,
which involves any action to resolve cases without waiting for a response from
the server [14]. Also at this level is the averaging of data collected from sensors.
Besides, the artificial intelligence paradigm is used at this level, which is to apply
a machine learning mechanism based on data collected from sensors to predict
the state of the greenhouse. This will significantly increase farm efficiency by
helping to take action to improve the quality of the crop being grown

3.3 Network / Cloud layer.

This level involves the transfer of data collected at the border through public
communication networks to cloud servers. Cloud computing is used for in depth
data analysis, as well as for their storage. At this level, all the advantages of
cloud computing are used.

3.4 Application layer.

The collected data must be analyzed and converted into a readable form so
that users can monitor and control the condition of the greenhouse in a way
convenient for them. At this level, a policy of work is defined, and the process
of control over the production of crops in the greenhouse takes place directly.

Fig. 1. Monitor and management system

Edge computation and its types.

Today, edge computing determines the future of IoT. It ensures the stability of
IoT devices and eliminates latency issues by providing data processing closer to
the source.
6 F.Sharofidinov et al.

IoT Edge Computing is a network of small data centers where missioncritical

data is stored and processed locally. These data centers are connected in a grid,
and they transmit the received data to a centralized data warehouse. This usu-
ally occurs within 10 square meters or less.
IoT Edge Computing is used to analyze and process data closer to the data
source. The smart devices used in IoT Edge Computing are capable of process-
ing critical data fragments and providing a fast realtime response. These devices
prevent delays caused by sending data over the Internet to the cloud and delay
response from the cloud.
These devices are designed to act as small data centers that provide nearzero
latency. With this advanced capability, data processing is decentralized, and net-
work traffic is significantly reduced. This data can then be collected by the cloud
for further evaluation and processing.
Types of IoT Edge Computing
Generally, we can define three types of edge computations of IoT [13]:
• Local devices to accommodate a specific and welldefined target. They can be
easily deployed and maintained.
• Local data centers to provide significant data processing and storage capabili-
ties. They are usually preengineered and configured. They are assembled locally
and provide good capital cost savings[16].
• Regional data centers, with a clear advantage in that they are closer to the
data source. Although they have more processing and storage power than local
data centers, they are expensive and require more maintenance. Such edge de-
vices are designed either in prefabricated or custom versions.
The Benefits of Edge Computing
As IoT Edge Computing is implemented and widely distributed, a large number
of industries are gaining potential benefits. In particular, edge computing offers
seven major advantages in intelligent manufacturing:
1) Fast response time: Computing and storage power is local and distributed.
Avoiding sending data back and forth in the cloud is key to reducing latency
and getting more rapid responses. This helps prevent vital machine operations
from being disrupted, or dangerous incidents occur
2) Sequential operations with sporadic communication: For many remote as-
sets, controlling unpredictable areas of Internet connectivity, such as agricultural
pumps, oil wells, windmills, or solar farms, can be challenging. The local stor-
age and processing capabilities of edge devices ensure no data loss and prevent
outages in the event of limited Internet connectivity.
3) Security: IoT Edge Computing has made data transfer between the cloud and
devices redundant. We can filter sensitive data locally and move only the data
model to the cloud.
4) Costeffective: A major practical issue with the introduction of the IoT was the
cost of storage, processing power, and network bandwidth. IoT Edge comput-
ing enables local data computing, allowing business organizations to distinguish
between services that need to be performed locally and services that need to be
sent to the cloud. This helps reduce the cost of developing a complete solution
 Agriculture management based on LoRa Edge Computing System 7

for the IoT.

5) Compatibility between modern and obsolete devices: the edge devices function
as a link between current and obsolete machines. This allows outdated machines
to interact with advanced tools for IoT solutions.
6) Increased fault tolerance: A decentralized peripheral computing architecture
allows other network devices to become more resilient. This is a great advantage
as a single server failure in the cloud can result in the defeat of thousands of IoT
devices.
7) Reduced exposure: edge computing minimizes data transfer over the network.
This, in turn, helps to reduce the impact of data during transit. In some cases,
sensitive information such as personal identification information (PII) and the
payment card industry (PCI) may not be sent at all. Such cases can help avoid
some legal complications related to security and confidentiality. Also, data en-
cryption and access control can make it highly secure from familiar threats.

3.5 Simulation results

As a result of the simulation, it was found that the use of edge computing
significantly reduces the load on the channels and can reduce delays in processing
IoT traffic. Figure 3 shows a comparative analysis of the delay time in service
when data passes through the system when the model is a traditional network
without using edge computing and with a system using edge computing. As can
be seen from the figure 2, the delay in the system using edge computing is much
lower compared to systems without the use of edge computing, when data is sent
directly to the server without preliminary processing.

Fig. 2. Delay in service traffic

8 F.Sharofidinov et al.

Without preprocessing, the amount of data uploaded to the server increases

significantly, which increases the queue of data and service time. Since our preci-
sion farming system has very little traffic from sensors, most data can be main-
tained locally. This increases the amount of traffic being serviced and also reduces
the uploading of data to the cloud server. As can be seen in figure 3 and figure 4
at Edge, up to half of the traffic can be serviced, and approximately three times
more data can be serviced compared to systems where Edge is missing.

Fig. 3. The amount of traffic served

Fig. 4. Ratio of edge computing traffic service to the remote cloud

 Agriculture management based on LoRa Edge Computing System 9

4 Proposed monitoring system for precision agriculture

4.1 The basic architecture of the precision farming monitoring
system
The architecture of our monitoring system based on LoRa technology consists
of 4 levels, the components of which are shown in Figure 5. The main tasks of
each level are described in paragraph 3 of this work. In addition to the tasks
described in the third paragraph, in our full-scale experiment of the monitoring
system, artificial intelligence is used, which consists in the application of machine
learning methods to predict the state of the greenhouse based on incoming data
from sensors.

Fig. 5. The basic architecture of precision farming.

Using the Random Forest algorithm to classify a greenhouse

Our prediction model based on machine learning is deployed at the edge level.
This helps to identify actions to improve the plant’s condition correctly. There
are five types of sensors from which data is sent to a remote server. In our
scheme, the data will be preprocessed at the gateway, which is the edge level. The
Random Forest Classifier Model is used to classify the condition of a greenhouse.
A random forest, as its name suggests, consists of a large number of individual
deci-sion trees that act as an ensemble. Each individual tree in a random forest
gives a class forecast, and the class with the most votes becomes the forecast
of our model. The fundamental concept underlying the random forest algorithm
is simple, but strong is the wisdom of the crowd. Data science says that the
reason why the random forest model works so well is that a large number of
relatively uncorrected models (trees) acting as a committee will surpass any of
the individual components of the models.
10 F.Sharofidinov et al.

4.2 Sensor node (Describe sensor node = MCU + Wireless

interface + various sensors)
Currently, there are many development boards to help create a prototype quickly.
With our purpose, a development board equipped with LoRa interface is used
to transmit sensor data and communicate with the gateway. As shown in Figure
6, the development board TTGO LoRa32, which is based on modules such as
ESP32 and LoRa SX1276, was used with some sensor modules to measure tem-
perature, humidty, CO2 index, soil humidity, and light intensity.

Fig. 6. Sensor node

Moreover, the ESP32 module also supports the other communication interfaces
such as WiFi and Bluetooth Low Energy [9][18]. Therefore, this development
board can be used as a single-channel gateway for LoRa communication. A pro-
tocol is defined to transmit messages between the sensor nodes and the gateway.
Before transmitting the sensor data, the node sends a request message to the
gateway. After receiving the sensor data, the gateway also replies to an acknowl-
edgment message to the sensor node. As shown in Figure 7, the payload of a
sensor message is specified with five bytes for measured data from sensors. The
message type is defined in the header. By defining the message transmission
protocol between the gateway and the sensor nodes, we can save the channel
budget. In the payload, the data byte from 1 to 5 represent the measured data
of temperature, humidity, CO2 index, soil humidity, and light intensity, respec-
tively.

Fig. 7. Format of a sensor message

According to the identification of each device, the edge layer and cloud layer
 Agriculture management based on LoRa Edge Computing System 11

can access the device’s data for further processing. In the next subsection, we
will consider some techniques used to preprocess the sensor data and predict the
status in the greenhouse.

4.3 Sensor data processing (Data process on Edge, Prediction

method)

Describe methods of data processing on Edge. (Aggregation, averaging data,

normalizing data) The approach of using the edge layer for preprocessing sensor
data, we can save the bandwidth to the remote server, and reply action to actu-
ator quickly without waiting server responses. However, some noises are added
to measured data from sensor devices. Therefore, to increase accuracy sensor
measurement, a simple filtering tech-nique can be used at the edge layer. The
averaging technique is easy to implement by adding several measures together,
then dividing the total by several measurements. The last data coming from
sensor devices are saved temporarily at the edge layer,
with k measurements, the new data
Xinew newareaveraged : Xinew = xi−k +xk+1i−k+1 +xi

Moreover, the prediction model can be deployed at the edge layer to predict
the con-dition in the greenhouses based on the received sensor data. It helps to
identify actions correctly to improve the plant’s state. There are five types of
sensors sent to the remote server. In our scheme, the data will be preprocessed
at the edge layer. Our prediction model is deployed to classify events into four
classes: 1 – soil without water, 2 – environment correct, 3 – too much hot, 4 –
very cold based on received data. The data frame for each device has a format
in table 1.

Soil Humidity Light Temperature CO2

Air Humidity (%) Class
(Analog) (Analog) (Celsius) (Analog)
550 280.83 21 154 86.7 4
773 695.83 22 150 87,6 1
568 764 34,4 169 61,1 3
130 0 22,1 34 72,9 2

Thus, the input data can be performed as: X=[HS L T CO2 HR] where
HS is soil humidity in analog-digital conversion, L is light intensity in lux, T is
temperature in °C, CO2 is CO2 index in analog-digital conversion and HR is
humidity relative in air.
A random forest classification model is used to classify the greenhouse state.
A random forest, as its name implies, consists of a large number of individual
decision trees that act as an ensemble. Each individual tree in a random forest
gives a class forecast, and the class with the most votes becomes the forecast of
12 F.Sharofidinov et al.

Fig. 8. SHEME

our model. The fundamental concept underlying the random forest algorithm is
the wisdom of the crowd. Data science says that the reason the random forest
model works so well is this: a large number of relatively uncorrelated models
(trees) acting as a committee will outperform any of the individual component
models.
The target output is class,
For each event, the target output is
followed as: classi = f (xi ) = f ([HSI LI TI CO2I HRI ])

5 Experiment and Results

The learning model was evaluated with a different number of samples. By chang-
ing the number of samples used for training, we received the learning curve, as
shown in Figure 9. According to the graph, the training score increases when we
have more samples to learn. With 300 samples, the model can predict the class
of conditions in the greenhouse with high accuracy

Fig. 9. Learning score based on the number of samples.

By monitoring the sensor data after a certain time, we can analyze the factor
which is the most impact on plants. As shown in Figure 10, the most important
factor effected prediction result is soil humidity.
 Agriculture management based on LoRa Edge Computing System 13

Fig. 10. Visualizing important factor

6 Conclusion

Using the paradigm of edge computing and artificial intelligence methods, we

were able to create an effective model of the monitoring system for an agricul-
tural facility. A monitoring system based on machine learning, which is taught
at the edge level, helps farmers to monitor the condition of plants by predict-
ing any case of deviation from normal factors, effectively irrigating plants, and
optimally applying expensive chemicals. Our proposed monitoring system based
on edge computing and machine learning can be used for deployment in both
small home greenhouses and industrial greenhouse networks. The obtained re-
sults prove the effectiveness of using LoRa technology, as well as the advantages
of using artificial intelligence for forecasting in precision farming systems.

References
1. L. Chen et al. A lora-based air quality monitor on unmanned aerial vehicle for
smart city. In 2018 International Conference on System Science and Engineering
(ICSSE), pages 1–5, June 2018
2. General Electric. What is edge computing? [Online]
Available:https://www.ge.com/digital/blog/what-edge-computing. Accessed:
Apr. 18, 2019.
3. He, Y., Richard Yu, F., Zhao, N., Leung, V. C. M., Yin, H. (2017). Software-
Defined Networks with Mobile Edge Computing and Caching for Smart Cities: A
Big Data Deep Reinforcement Learning Approach. IEEE Communications Maga-
zine, 55(12), 31–37. https://doi.org/10.1109/MCOM.2017.1700246.
4. Khakimov, A.; Muthanna, A.; Muthanna, M.S.A. Study of Fog computing struc-
ture. In Proceedings of the 2018 IEEE Conference of Russian Young Researchers in
Elec-trical and Electronic Engineering (EIConRus), Moscow, Russia, 29 January–1
Feb-ruary 2018; pp. 51–54.
5. Perera, C., Qin, Y., Estrella, J. C., Reiff-Marganiec, S., Vasilakos, A. V. (2017).
Fog computing for sustainable smart cities: A survey. ACM Computing Surveys,
50(3). https://doi.org/10.1145/3057266
6. Taleb, T., Dutta, S., Ksentini, A., Iqbal, M., Flinck, H. (2017). Mobile edge compu-
ting potential in making cities smarter. IEEE Communications Magazine, 55(3),
38–43. https://doi.org/10.1109/MCOM.2017.1600249CM
 14 F.Sharofidinov et al.

7. Suresh, V. M., Sidhu, R., Karkare, P., Patil, A., Lei, Z., Basu, A. (2018). Powering
the IoT through embedded machine learning and LoRa. In IEEE World Forum on
Internet of Things, WF-IoT 2018 - Proceedings (Vol. 2018-January, pp. 349–354).
Institute of Electrical and Electronics Engineers Inc. https://doi.org/10.1109/WF-
IoT.2018.8355177
8. Carlin, A., Hammoudeh, M. and Aldabbas, O., 2015. Defence for dis-
tributed denial of service attacks in cloud computing. Procedia computer sci-
ence, 73, pp.490-497. Institute of Electrical and Electronics En-gineers Inc.
https://doi.org/10.1109/NORCHIP.2014.7004716
9. Nguyen Gia, T., Sarker, V. K., Tcarenko, I., Rahmani, A. M., Westerlund, T.,
Liljeberg, P., Tenhunen, H. (2018). Energy efficient wearable sensor node for
IoT-based fall detection systems. Microprocessors and Microsystems, 56, 34–46.
https://doi.org/10.1016/j.micpro.2017.10.014
10. Rahmani, A. M., Gia, T. N., Negash, B., Anzanpour, A., Azimi, I., Jiang, M.,
Liljeberg, P. (2018). Exploiting smart e-Health gateways at the edge of healthcare
Internet-of-Things: A fog computing approach. Future Generation Computer Sys-
tems, 78, 641–658. https://doi.org/10.1016/j.future.2017.02.014
11. M. S. A. Muthanna, P. Wang, M. Wei, A. A. Ateya, and A. Muthanna, “Toward
an Ultra-low Latency and Energy Efficient LoRaWAN,” Lect. Notes Comput. Sci.
(including Subser. Lect. Notes Artif. Intell. Lect. Notes Bioinformatics), vol. 11660
LNCS, pp. 233–242, 2019.
12. A. Dongare, R. Narayanan, A. Gadre, A. Luong, A. Balanuta, S. Kumar, B. Ian-
nucci, and A. Rowe, “Charm: Exploiting geographical diversity through coherent
combin-ing in low-power wide-area networks,” in Proceedings of the International
Confer-ence on Information Processing in Sensor Networks, 2018, pp. 60–71
13. Hammoudeh, Mohammad; Newman, Robert; Dennett, Christopher; Mount, Sarah;
Aldabbas, Omar. 2015. ”Map as a Service: A Framework for Visualising and Max-
imising Information Return from Multi-ModalWireless Sensor Networks.” Sensors
15, no. 9: 22970-23003
14. Carlin, A., Hammoudeh, M. and Aldabbas, O., 2015. Intrusion detection and coun-
termeasure of virtual cloud systems-state of the art and current challenges. Inter-
national Journal of Advanced Computer Science and Applications, 6(6).
15. Gia, T. N., Thanigaivelan, N. K., Rahmani, A. M., Westerlund, T., Liljeberg, P.,
Tenhunen, H. (2015). Customizing 6LoWPAN networks towards Internet-of-Things
based ubiquitous healthcare systems. In NORCHIP 2014 - 32nd NORCHIP Confer-
ence: The Nordic Microelectronics Event. Institute of Electrical and Electronics
En-gineers Inc. https://doi.org/10.1109/NORCHIP.2014.7004716
16. Daraseliya A.V., Sopin E.S., Samuylov A.K., Shorgin S.Y. Comparative Analysis of
the Mechanisms for Energy Efficiency Improving in Cloud Computing Systems //
Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial
Intelligence and Lecture Notes in Bioinformatics). – 2018. – Vol. 11118. – P. 268-
276.
17. Sopin E.S., Daraseliya A.V., Correia L.M. Performance Analysis of the Offloading
Scheme in a Fog Computing System // International Congress on Ultra Modern
Telecommunications and Control Systems and Workshops. – 2019. – Vol. 2018-,
art.no. 8631245.
18. Daraseliya A.V., Sopin E.S., Rykov V.V. On optimization of energy consumption
in cloud computing system // CEUR Workshop Proceedings. – 2018. – Vol. 2332.
– P. 23-31.

View publication stats