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

net/publication/337439497

Implementation of an SDN Based IoT Network Model for Efficient

Transmission of Sensor Data

Conference Paper · September 2019

DOI: 10.1109/UBMK.2019.8907119

CITATIONS READS

4 811

3 authors, including:

Yasin Inağ Suat Ozdemir

Mus Alparslan University Hacettepe University
5 PUBLICATIONS 5 CITATIONS 150 PUBLICATIONS 3,403 CITATIONS

SEE PROFILE SEE PROFILE

All content following this page was uploaded by Yasin Inağ on 28 September 2022.

The user has requested enhancement of the downloaded file.

 Implementation of an SDN Based IoT Network
Model for Efficient Transmission of Sensor Data
<DVLQøQD÷ 0HKPHW'HPLUFL 6XDWg]HPLU
Computer Engineering Computer Engineering Computer Engineering
Gazi University Gazi University Gazi University
Ankara, Turkey Ankara, Turkey Ankara, Turkey
[email protected] [email protected] [email protected]

Abstract—Internet of Things (IoT) comprises physical Therefore, Software Defined Networking technology (SDN)
devices with an IP address equipped with wireless sensors has been proposed as an enabling technology for IoT [2].
and/or wireless communication devices. Due to the increase in
the number of devices and data production, it is difficult to SDN, often seen as a revolutionary new idea in computer
create manageable, intelligent networks. Models have been networking, separates the control layer from the data layer and
proposed to solve existing problems and improve network provides opportunities for innovation through network
performance. However, traditional network technology is not control, management and network programmability [3]. SDN
capable of solving problems caused by such a large number of can manage heterogeneous networks quickly and
devices and the large data transmission it generates. Software
dynamically. SDN-managed networks have a central
Defined Networking (SDN) provides opportunities for
innovation in network control, management and network
controller with network status information. The controller
programmability by separating the control layer from the data communicates with the switch over a secure channel using the
layer in computer networks. SDN is used in IoT networks to OpenFlow Protocol [4]. Various software applications that
improve network performance, provide more direct control over take control decisions of the switch can run on the controller.
routing, analyze network traffic and manage the network The central controller provides an overview of the network
effectively with the centralized controller. In this study, an topology and the basic physical routing devices perform the
architecture is proposed for efficient transmission of sensor data necessary flow and communication operations by receiving
in SDN based IoT networks. In the model simulated with commands from the controller [5].
Mininet and POX controller, average packet loss and delay are
calculated for varying numbers of switches and path hop counts. The common definition of SDN separates the data layer
It was found that the number of hops had significant impact on from the control layer. The networks created with SDN
delay and loss rate, while the total number of switches in the architecture consist of 3 different layers as shown in Fig. 1.
network had no linear relationship with hop counts. Data layer consists of switches, hubs, sensors etc. which
transmit the data. In this layer, there are IoT devices and basic
Keywords—Software Defined Networking (SDN), Internet of
Things (IoT), Transmission of sensor data network devices that enable the communication of IoT
devices. The devices on this layer do not have to make
I. INTRODUCTION decisions about routing. Decision-making and management
The Internet of Things is the connection of all objects with functions are performed in the control layer, which is the unit
an IP address to the Internet. Generally, physical devices where the network is managed. There are many SDN
equipped with wireless sensors and/or wireless controllers with different features, programming languages
communication devices. These devices are used to create and use cases, such as POX, NOX, OpenDaylight etc.
manageable, intelligent networks. In the networks where these Application layer is the unit where applications such as
devices increase rapidly, different problems arise such as network monitoring, load balancing etc. are developed. It
network management and performance problems. communicates with the control layer and operates
independently from the infrastructure. The control layer
According to a Cisco white paper, 6.4 billion devices are communicates with the application layer via the Northbound
connected to the Internet today, and this number is expected to API and with the data layer via the Southbound API.
be around 50 billion in 2020. These connected devices
generate a large amount of data such as 6.2 Exabyte, which is
expected to rise to 30 Exabyte levels in 2020 [1].
The increase in Internet-connected devices and data
production will make problems such as intelligent network
control and management more important. Several models have
been proposed for current problems in the IoT paradigm.
However, traditional networking technology is not capable of
solving problems caused by such a large number of devices
and the large data transmission it generates. There are different
brands and models of IoT devices and network devices with
various capabilities within the same network. Communication
in such a heterogeneous network raises new challenges.

This work was supported by the Scientific and Technological Fig. 1. SDN Layers
5HVHDUFK&RXQFLORI7XUNH\ 7h%ø7$. *UDQW1R(

ϵϳϴͲϭͲϳϮϴϭͲϯϵϲϰͲϳͬϭϵͬΨϯϭ͘ϬϬΞϮϬϭϵ/

(UBMK'19) 4rd International Conference on Computer Science and Engineering - 682

Authorized licensed use limited to: ULAKBIM UASL - GAZI UNIV. Downloaded on January 04,2022 at 11:39:15 UTC from IEEE Xplore. Restrictions apply.
 In this study, the problem of routing various types of data load balancing in SDN-supported networks. They perform
from IoT sensors effectively in an SDN based network network optimization by redirecting traffic of such
architecture is examined. Data were obtained from 6 different applications. It is stated in the study that the redirection of data
sensors with the objective of estimating the air pollution level with the same characteristics will be successful in network
of the environment. The sensors are connected via a Wi-Fi operations such as prioritization routing in IoT networks.
device and the data is transferred to a computer. This structure, Instead of redirecting all traffic in any problem that may occur
which is formed at 7 different points, is controlled by an SDN in the network, redirecting the data that generates high rate of
controller. The computers are connected to each other with traffic with the same characteristics will reduce the cost.
different OpenFlow supported switches, and each computer
In IoT networks, there are studies on controller position
parses and directs 6 different types of data. In the architecture,
and number of controllers to optimize traffic. N. Handigol et
bandwidth, delay and data loss rate metrics were calculated
al. [12] recommend Plug-n-Server to monitor network status
with varying numbers of switches and path hop counts.
and minimize the response time of the controller. load
The remainder of this paper is organized as follows. The balancing is done according to client requests. These studies
related works are presented in Section 2. The methodology is have scalability problems. In order to increase scalability, a
described in Section 3. The simulation results are discussed in hierarchical controller structure was used. There is a central
Section 4 and lastly, the paper is concluded Section 5. controller that manages the controllers. This structure aims to
make faster decisions and reduce latency in the network [13].
II. RELATED WORKS This is the common solution approach used to solve the
Nowadays, rapidly developing intelligent systems, problem of scalability. However, determining the position and
industry 4.0 technology, etc. are developing with the IoT. IoT number of controllers in the network is a continuing problem.
networks are very difficult and complex to manage due to
Different metrics are used to evaluate the performance of
complex infrastructure and the collection of different types of
SDN managed networks. The metrics used are determined
data. It is necessary to compile the data from different sensors
according to the proposed architecture and user requirements.
and different switches and extract meaningful patterns from
To optimize overall network traffic; metrics such as delay,
the data. Management of the infrastructure, resource
bandwidth, throughput, number of nodes, etc. are examined.
management, transmission optimization are the challenges
In the proposed server optimization architectures, new metrics
faced. Many studies have been done to solve these problems.
such as load balancing, prioritization, etc. have been examined
SDN is one of the most important solutions to these [14].
challenges [6,7]. It separates the data layer from the control
M. Boussard et al. [15] proposed an SDN-based
layer and provides efficient, secure and service QoS in
architecture for managing IoT devices in an intelligent
heterogeneous systems. With SDN, the network is monitored
environment, called Software-defined LANs (SD LANs). SD
and used in maximum efficiency [8].
LAN makes it easy to discover new features in the network
Z. Qin et al. [7] studies are using SDN to facilitate the and group them according to user requirements. When a
management of IoT networks. In this study, special service device is added to the network, it optimizes the network and
requirements are provided by the central controller. Network provides redirection on the local network.
requirements are defined as maximum data rate or minimum
latency or packet loss for each stream. The genetic algorithm III. METHODOLOGY
was used to optimize end - to - end flow performance. The In this section, data from IoT sensors are routed in SDN
report performed routing optimization where all streams used based networks. The data from the sensors are directed based
the same resources and shared the results with SDN. It is stated on content. The model is simulated in a mininet environment
that IoT networks are easier to manage with SDN. with the Pox controller and the results are discussed. In our
experimental setup shown in Fig. 2, the data obtained from 6
Y. Jararweh et al. [9] proposed an architecture managed by
different sensors are separated according to their type and each
SDN according to the different needs of IoT networks. In the
sensor’s data is stored on a separate server. In the proposed
study, which is expected to increase the number of connected
architecture, the hosts to which the sensors are connected are
devices rapidly, security and data storage structures are
both a client to process data and servers on which the data is
controlled by SDN. The study proposes a high-level
stored. Sensors are connected to hosts wireless network. There
architecture that includes all sub-studies of the controller but
are 6 sensors connected to each host and each host is
does not explain the experimental results.
connected to different switches. The hosts are located in
According to J. Liu et al. [10] SDN is not enabled on all different environments so that data from 6 different regions is
network elements in the core network, such as sensor nodes. obtained. From the data, it is aimed to measure the air quality
Therefore, the management of applications by a central of life of the designated location. In this study, the
controller is restricted and applications are embedded in the experimental environment was prepared and the
hardware. This causes some difficulties in heterogeneous communication of sensors and hosts and the management of
network structures. Dynamically improving resource usage, the network with SDN were realized.
data collection, transfer, and processing are among the
problems encountered.
Ghazisaeedi et al. [11] propose the creation of flow
protocols with OpenFlow supported switches according to the
type of network formed in the networks. HTTP applications
account for 47.7% of Internet traffic. Therefore, they argue
that the architecture they propose will be largely successful in

(UBMK'19) 4rd International Conference on Computer Science and Engineering - 683

Authorized licensed use limited to: ULAKBIM UASL - GAZI UNIV. Downloaded on January 04,2022 at 11:39:15 UTC from IEEE Xplore. Restrictions apply.
 Routing is performed if the destination server's
information is present in the switch's flow table. if the
destination server's information is not available in the flow
table, the switch asks the controller. The controller sends the
ARP (Address Resolution Protocol) request to the network to
learn the destination server’s information. The controller
updates the flow table of the switches with the new
information it gets and the packet is forwarded to the
destination server. If the server is listening on the network, it
accepts the packet and the stream is complete. If the server is
not listening to the network, the packet is drop.
A. SDN Controller Implementation
In the proposed architecture, the POX controller is used.
POX is an open source project using python software
language. POX, which manages switches by remote
connection method, is installed on a standalone virtual
machine. POX allows you to manage your network with many
Fig. 2. SDN Based IoT Network Architecture auxiliary functions and set your own rules. In this study,
L2_learning library is used to learn the network and to direct
Fig. 3 shows the flow diagram of the proposed model. The left the flow rules, and OpenFlow.spanning_tree library is used to
side of the flowchart refers to the controller operations and the find the shortest path and provide communication. In this
right side refers to the operations that take place in the hosts. study, communication from the sensors to the hosts is provided
The proposed model is managed by POX controller software. by traditional network methods and the communication
The controller manages the network established in the Mininet between the hosts is managed with the pox controller [16].
environment. The controller identifies the switches, the hosts
connected to it, and updates the communication data in the Forwarding.l2_learning enables OpenFlow switches to act
flow tables of the switches. Xterm allows connection to hosts. as learning switches. This component is used to create the
MAC tables of the switches. The controller updates the MAC
Client and server features are installed on each host with tables of all connected devices by sending an ARP (Address
the Netcat tool. Hosts with server capability start to listen to Resolution Protocol) request to the network [16].
the network with the specified port number. Sensor data is
transmitted to clients. Clients examine incoming data on a OpenFlow (OF) is considered the first of the SDN flow
content-based basis. Different data types direct to different standards. The flow protocol enables the Controller to
servers. After the data is parsed according to its contents, it is communicate with the physical devices. In this way, it is
packaged with the IP address and port number of the provided to configure different devices with flow protocol. OF
destination server and the data packets are forwarded to the enables communication by sending the controller's decisions
servers. to the Northbound API or Southbound API.
OF Spanning tree is used in the discovery component to
create a view of the network topology. It examines all the
roads in the network and forms a tree structure. Disables
unused and overflow nodes in the created tree. As a result, it
prevents infinite loops in the network. This component has
additional options. One of them is the - no-flood option. When
a switch is connected to the network, it prevents overflow on
all ports and enables some ports when needed. Another option
is -hold-down. Stops overflow until it identifies the loop of all
switches in the network. This allows all connections on the
network to be identified.
In this study, the pox controller performs IP and port based
routing. The data from the sensors are transmitted to the hosts
for processing. In the hosts, the presentation of the data is
determined. Destination port and IP are added and sent to the
network. The OpenFlow.spanning_tree tool is used to avoid
any infinite loops in the SDN based IoT network. The
l2_learning class was used to learn the network and create the
MAC table.

TABLE I. SAMPLE OPENFLOW PACKET OUT RULE

Match Action
match.nw_dst=IPAddr("Destination IP") actions.append(of.ofp_action
Fig. 3. The Flowchart of Proposed Model match.tp_dst ="Destination Port" _output(port))

(UBMK'19) 4rd International Conference on Computer Science and Engineering - 684

Authorized licensed use limited to: ULAKBIM UASL - GAZI UNIV. Downloaded on January 04,2022 at 11:39:15 UTC from IEEE Xplore. Restrictions apply.
 The IP address and port number are matched after the The topology established in Mininet environment is
controller receives the data as shown in the table. The resulting managed by remote control with the POX controller. Hosts
matching variable flow mod is sent to the switch. The ARP should work as a client or server in topology. Netcat
request is sent to the system to find out the MAC address of application was used to provide this structure. Netcat is a
the destination IP and determine the communication path. utility that can be installed with a server or client feature to
Data transfer is performed by updating the flow tables. The allow communication between hosts [18]. After setting the
system continues to operate as long as the data flow from the communication method and port numbers, the desired
IoT sensors continues. structure is executed with Netcat tools. In our study, the hosts
are both client and server. For this reason, both the client
B. Experimental Setup
feature with defined IP and the server feature with the port
Data types obtained from sensors; the temperature sensor number to be determined are added to the hosts via Netcat.
is the sensor that measures the ambient temperature with the Servers listen to the network with the specified port number.
Arduino DHT22 sensor and collects data. The humidity sensor Clients start transmitting data to the correct servers.
is a sensor that measures the humidity in the environment.
Data is collected by Arduino's DHT22 sensor. The brightness With the application we developed, clients can process the
sensor is the sensor that measures the light intensity in the data coming from 6 different sensors and mix different data
environment. Data is collected by the TSL2561 sensor. The types and get the necessary content. The contents obtained are
input-output sensor consists of 2 infrared sensors to detect the sent to predetermined servers according to the data type. If the
density of people in the environment. The motion sensor is a data type is outside the specified data types, or if the data is
PIR sensor that measures the mobility in the measured corrupted, it is held on a different server. In future studies, it is
environment. The air quality sensor is measured by the aimed to make sense and rerouting the corrupt data. With
ambient air quality MQ135 sensor. It is defined by the air machine learning and deep learning techniques are intended to
quality sensor as a function of NH3, NOx, benzene, alcohol correct and rerouting data.
and CO3 gases. The communication realized in the proposed topology was
Mininet is an open source project that allows you to create, analyzed with Tcpdump and Wireshark application. Tcpdump
test, and implement software defined networks (SDN). It is a package analyzer program that runs on the command line
enables virtual networks to be tested [17]. In the created of computers running Linux. It allows the user to monitor
architecture, 7 hosts, 1 controller and 7 servers were used. packets transmitted or received over a network to which it is
Table 2 shows the configuration of the hosts and controllers, connected. Tcpdump is free software. Wireshark is a free
and Table 3 shows the configuration of the servers. application that allows network traffic to be monitored via a
graphical interface.
The controller is running local and its IP is 192.168.56.101
by default. Hosts defined IP addresses, netmasks. With IV. EVALUATION RESULTS AND DISCUSSION
OpenFlow 1.3, flow protocols are defined. Which switches are In this section, the results obtained with the proposed
connected to the hosts are shown in Table 2. In this study, the model are examined. In the SDN based IoT network, the data
network performance is examined according to the number of obtained from the sensors can be processed and routed on the
switches. However, the switches and positions to which the clients. In this transmission, the data was transmitted to the
hosts are connected are kept constant. correct servers with the spanning tree algorithm. While 6
Server names are determined by the type of data obtained different types of data are routed to the correct servers,
from the sensors. As previously described, the hosts to which incorrect data is transmitted and stored to the different server.
the sensors are connected also serve as servers. Therefore, The IoT network is managed with a controller. In the literature,
server IP addresses are the same as hosts. The port numbers of the importance of controlling and developing IoT networks
the servers are shown in the Table 3. For convenience, IP with controller is emphasized. In IoT networks, producing
addresses and port numbers are given sequentially. solutions with SDN is more successful than traditional
networking due to the software controller defining new rules
TABLE II. CONFIGURATION OF THE HOSTS AND CONTROLLER and rerouting made easier by remote access.
Name IP Connected In this study, data transmission over the network is
Host1 10.0.0.1/8 Switch 1 examined with Tcpdump and Wireshark. The topology was
Host2 10.0.0.2/8 Switch 7 initially implemented with 20 OpenFlow switches, then tested
Host3 10.0.0.3/8 Switch 9 with 30, 40, 50 and 60 switches respectively. While the
Host4 10.0.0.4/8 Switch 13 number of switches in the topology is increased, the positions
Host5 10.0.0.5/8 Switch 16 of the clients and servers are kept constant and only the
Host6 10.0.0.6/8 Switch 18 number of switches in which data is progressed has been
Controller Local IP All OF Switch increased. The tests were performed with 40,000 data transfers
per host. For each host, the number of hops to all hosts that it
TABLE III. CONFIGURATION OF THE SERVER communicates with was calculated. Data is sent to a host from
6 different sensors. Although the sensors transmit
Server Name Server IP Server Port
simultaneously, they do not send an equal number of data in
Celsius 10.0.0.1/8 1001
total. However, a total of 40,000 data sending restrictions were
Humidity 10.0.0.2/8 1002
GasSensor 10.0.0.3/8 1003 imposed on each host.
Lux 10.0.0.4/8 1004
When the first transfer starts, it takes a certain time for the
Sharp 10.0.0.5/8 1005
Pir 10.0.0.6/8 1006
controller to discover the network and update the flow tables,
Default 10.0.0.7/8 1007 as described earlier. This time was calculated as an average of

(UBMK'19) 4rd International Conference on Computer Science and Engineering - 685

Authorized licensed use limited to: ULAKBIM UASL - GAZI UNIV. Downloaded on January 04,2022 at 11:39:15 UTC from IEEE Xplore. Restrictions apply.
7 seconds. Of course, this time varies according to the number 10
9,6 H1-H2
of switches and the specifications of the computer used. The 9,2
H1-H3
8,8
increase in the number of switches in our test environment has 8,4 H1-H4
8
little effect on this time. When calculating the delay, this time 7,6 H1-H5
7,2
is divided by the total number of transmitted data and added to 6,8 H1-H6

Delay (ms)
6,4
the delay. Because the network monitoring tools calculate the 6
5,6
H2-H3

time it takes for the packet to be sent from the client to the 5,2
4,8
H2-H4

host. 4,4
4
H2-H5

3,6 H2-H6
Table 4 demonstrates how the increase in the number of 3,2
2,8 H3-H4
switches in the network increased the average hop count, 2,4
2 H3-H5
1,6
average delay and average packet loss rate. The results 1,2
H3-H6
0,8
obtained according to the number of different switches 0,4
H4-H5
0 H4-H6
showed that the average delay was 4.4ms and the average 0 10 20 30 40 50 60 70 H5-H6
packet loss rate was 3.77%. It has been observed that 0.4% of Number of switches
data loss occurs due to incorrect data from sensors or incorrect
processing of clients. This result has been obtained by storing Fig. 5. Relationship between number of switches and delay
it on a different server. It is aimed to achieve data estimation
with machine learning methods and reduce the loss rate as a As shown in Fig. 6 and 7, as the number of hops increased,
new study. the average packet loss rate and delay increased. The increase
in loss rate is influenced by the amount of data sent from a
TABLE IV. AVERAGE HOP COUNT, DELAY AND LOSS RATE BY host. Hence different loss rates are observed for the same hop
NUMBER OF SWITCHES count. Another reason for this difference is that the data
Number of Average hop Average delay Average loss occasionally came from the sensors incorrectly.
Switches Count (ms) rate (%)
20 4.30 3.15 3.17
30 4.78 3.50 3.34
40 5.5 4.04 3.60 10,5
10
50 7.06 5.17 4.16 9,5
9
60 8.34 6.11 4.61 8,5
8
7,5
Delay (ms)

7
6,5
As shown in Fig. 4 and 5, as the number of switches 6
5,5
increased, the average packet loss rate and delay increased. 5
According to the literature, it is expected that the average delay 4,5
4
will increase with the increasing number of switches. 3,5
3
However, the increase in the number of switches does not 2,5
2
always indicate that the total number of switches on the 1,5
1
shortest path between the client and host increases. Therefore, 0,5
0
as seen in Fig. 5, there is no linear relationship between the 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14
increase in the number of switches in the topology and the Number of hops
average delay. The increase in the average delay could
partially be attributed to the increase in the time elapsed while Fig. 6. Relationship between number of hops and delay
the controller found a new path and updated the flow tables.

6,8
6,8 6,4
6,4 H1-H2 6
6 H1-H3 5,6
5,6 5,2
5,2 H1-H4
4,8
4,8
Loss rate(%)

H1-H5 4,4
Loss rate(%)

4,4 H2-H3 4
4 3,6
3,6 H2-H4
3,2
3,2 H2-H5 2,8
2,8 H2-H6 2,4
2,4 2
2 H3-H4
1,6
1,6 H3-H5 1,2
1,2 H3-H6 0,8
0,8 0,4
0,4 H4-H5 0
0 H4-H6
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14
0 10 20 30 40 50 60 70 H5-H6
Number of switches Number of hops

Fig. 7. Relationship between number of hops and loss rate

Fig. 4. Relationship between number of switches and loss rate

(UBMK'19) 4rd International Conference on Computer Science and Engineering - 686

Authorized licensed use limited to: ULAKBIM UASL - GAZI UNIV. Downloaded on January 04,2022 at 11:39:15 UTC from IEEE Xplore. Restrictions apply.
 The increase in the number of switches does not show a [6] F.Y. Okay, S. Ozdemir, "Routing in fog-enabled IoT platforms: a
survey and an SDN-based solution," IEEE Internet of Things Journal,
linear relationship with the increase in the number of hops. 5(6), pp. 4871-4889, 2018.
When the number of switches in the topology is configured as [7] Z. Qin, G. Denker, C. Giannelli, P. Bellavista, and N.
20, 30, 40, 50, 60 respectively, the number of hops in the Venkatasubramanian, "A software defined networking architecture for
communication between Host1 and Host 2 is observed as 5, 6, the internet-of-things," in Network Operations and Management
8, 11, 10. The delay, on the other hand, was observed to Symposium (NOMS), 2014 IEEE, pp. 1-9.
progress linearly with the number of hops. [8] P.W. Tsai, C.W. Tsai, C.W. Hsu, C.S. Yang, "Network monitoring in
software-defined networking: a review," IEEE Systems Journal, 12(4),
pp. 3958-3969, 2018.
V. CONCLUSION
[9] Y. Jararweh, A. Mahmoud, A. Darabseh, E. Benkhelifa, M. Vouk, and
In this study, an architecture is proposed for the efficient A. Rindos, ``SDIoT: A software defined based internet of things
transmission of sensor data in SDN based IoT networks. In the framework,'' J. Ambient Intell. Humanized Comput., vol. 6, no. 4, pp.
453-461, Aug. 2015.
proposed model, average packet loss and delay were
[10] J. Liu, Yong. Li, M. Chen, W. Dong, D. Jin, “Software-defined internet
calculated and the changes according to the number of of things for smart urban sensing,” in Communications Magazine,
switches and hop count were examined. For effective IEEE, vol. 53, no. 9, pp. 55-63, September 2015.
communication, it was observed that the number of hops was [11] E. Ghazisaeedi, R. Tafazolli, “Mobile core traffic balancing by
the most effective variable, but the number of hops had no OpenFlow switching system,” in Globecom Workshops (GC Wkshps),
linear relationship with the total number of switches in the IEEE, pp. 824-829, Dec. 2013.
network. The location of the sensors was found to be important [12] N. Handigol, S. Seetharaman. M. Flajslik, N.McKeown, and R. Johari,
“Plug-n.Server: load balancing web traffic using OpenFlow,” Aug
for effective communication. 2009, Demo at ACM SIGCOMM..
It is recommended to use network models, to develop IoT [13] S. H. Yeganeh, A. Tootoonchian, Y. Ganjali, “On scalability of
software-defined networking,” Communications Magazine, IEEE, vol.
services in the best way, to monitor the network to intervene 51, no.2, pp. 136-141, 2013.
in different situations, to develop the network with full-scale [14] K. Sood, S. Yu, Y. Xiang, and S. Peng, “Control layer resource
simulations and to make necessary corrections. The simulated management in SDN-IoT networks using multi-objective constraint”
model is intended to be developed and realized. IEEE 11th Conference on Industrial Electronics and Applications
(ICIEA), 2016.
REFERENCES [15] M. Boussard, D. Bui, L. Ciavaglia, R. Douville, M. Le Pallec, N. Le
Sauze, L. Noirie, S. Papillon, P. Peloso, and F. Santoro. “Software-
[1] “Cisco visual networking index: global mobile data traffic forecast
defined LANs for interconnected smart environment,” pp. 219-227,
update, 2015–2020 white paper,” Cisco.
2015.
[2] Accton making partnership work, “Software defined networking for
[16] "POX Wiki - Open Networking Lab - Confluence",
Wi-Fi” White Paper- Accton, June 2014.
Openflow.stanford.edu, 2019. [Online]. Available:
[3] Faundation, Open Networking, “Software defined networking: the new https://openflow.stanford.edu/display/ONL/POX+Wiki.html.
norm for networks,” ONF White Paper, 2012. [Accessed: 14- Jun- 2019]
[4] B.Lantz, B. Heller, and N. McKeowen,”A network in a laptop: rapid [17] M. Team, "Mininet Walkthrough - Mininet", Mininet.org, 2019.
prototyping for software-defined networks,” in Proc. ACM HotNets, [Online]. Available: http://mininet.org/walkthrough/. [Accessed: 14-
2010. Jun- 2019]/
[5] A. K. Rangistti, H. B. Baldaniya, P. Kumar, and B.R Tamma, “Load- [18] Hernández, M., Gupta, A. and Shenfeld, A. (2019). How to Simulate a
aware hand-offs in software defined wireless LANs,”in Proc. IEEE TCP/UDP Client Using Netcat. [online] Ubidots Blog. Available at:
WiMob, 2014, pp. 685-690. https://ubidots.com/blog/how-to-simulate-a-tcpudp-client-using-
netcat/ [Accessed 14 Jun. 2019].https://ubidots.com/blog/how-to-
simulate-a-tcpudp-client-using-netcat/

(UBMK'19) 4rd International Conference on Computer Science and Engineering - 687

Authorized licensed use limited to: ULAKBIM UASL - GAZI UNIV. Downloaded on January 04,2022 at 11:39:15 UTC from IEEE Xplore. Restrictions apply.
View publication stats