A Class-based Adaptive QoS Control Scheme Adopting Optimization Technique over WLAN SDN Architecture

International Journal of Computer Networks & Communications (IJCNC) Vol.14, No.3, May 2022
DOI: 10.5121/ijcnc.2022.14304 55
A CLASS-BASED ADAPTIVE QOS CONTROL SCHEME
ADOPTING OPTIMIZATION TECHNIQUE OVER
WLAN SDN ARCHITECTURE
Pacharakit Vanitchasatit and Teerapat Sanguankotchakorn
Telecommunications Field of Study, Asian Institute of Technology, Bangkok, Thailand
ABSTRACT
Recently, Software-Defined Networking (SDN), a network architecture approach that enables the network
to be intelligently and centrally controlled by using software applications, has been introduced. Another
important issue in the network management context is Quality of Service (QoS).
This work investigates the QoS provisioning of various traffic classes on an SDN-enabled network. We
propose and implement the class-based adaptive QoS control scheme on SDN-enabled network for various
traffic classes, namely VoIP, Video Streaming and File Transfer. The effectiveness of our proposed scheme
is validated by simulation using Mininet and Ryu controller. The procedure to create the simulation
platform and all details relevant to all software used are described step-by-step in detail. The main
performance evaluation metric is the Maximum Number of Traffic Flows admittable with QoS while
Average Throughput, Latency, Jitter, and Packet Loss Rate are maintained at the comparable level of the
existing work in the literature called JMABC [11]. Our simulation results are illustrated with 95%
confidence interval. According to the simulation results, it is obvious that our proposed class-based
adaptive QoS control scheme adopting the optimization technique significantly outperforms the existing
similar QoS provision scheme in terms of the maximum number of the high priority traffic flows (VoIP)
admittable with QoS while the other evaluation metrics are maintained at the same level.
KEYWORDS
Software-Defined Networking (SDN), Quality-of-Service (QoS).
1. INTRODUCTION
In recent years, the traffic of various applications has been drastically increasing. There is a large
number of applications running on the Internet or communication networks. Some of the most
popular applications are such as web surfing, file transfer, voice, and streaming video. However,
the fundamental structure of the traditional network, namely Internet, computer communication,
and Internet Protocol (IP), has a basic fashion to provide the best effort services which are non-
reliable [1], [7]. Additionally, the existing networks are required to provide different users’
requirements with different fashion of services and resources. Therefore, the mechanism to
allocate the resources and satisfy the users/applications’ requirements is needed. Quality of
Service (QoS) [1,] [3], [6], [7], [12], [16], [17], [19], [20] is one mechanism that can be used in
this case. QoS is a set of technologies that work on a network to guarantee its ability to
dependably run high-priority applications and traffic under limited network capacity [3], [6], [7],
[12]. QoS technologies accomplish this by providing differentiated handling and capacity
allocation to specific flows in network traffic under limited network capacity [16]. The QoS
mechanism sequences packets and allocates bandwidth by queuing and distributing bandwidth

56
depending on the priority of packets. It manages network resources such as bandwidth by
controlling and setting priorities and policies for specific types of data on the network. This
solution helps to manage throughput, packet loss, delay, and jitter on network infrastructure.
There are two models currently used for QoS provisioning [7]: Integrated Services (IntServ) and
Differentiated Service (DiffServ). DiffServ achieves better QoS scalability and is suitable for
large-scale networks, while the IntServ provides a tighter QoS mechanism for real-time traffic.
IntServ model is one of the solutions for real-time traffic but the main problem with IntServ
working properly is all network devices along the traffic path must support it [18].
Recently, Software-Defined Networking (SDN), a network architecture approach that enables the
network to be intelligently and centrally controlled by using software applications, has been
introduced [3], [4], [8], [10], [12-16], [18], [21]. This technology is an approach to network
management that enables dynamic, programmatically efficient network configuration in order to
improve network performance and monitoring, making it more like cloud computing than
traditional network management [10]. It also helps operators manage the
entire network consistently and holistically, regardless of the underlying network technology.
Based on the characteristics of SDN architecture that has a centralized controller [2],[4],[10],
SDN seems to be appropriate for adoption in the current traditional network, where the
complexity as well as the need for QoS provisioning to various applications having different QoS
requirements, are challenging issues [2-4,] [6], [8], [12]. To improve the performance and reduce
the complexity of the traditional network, SDN has been introduced to centrally control the
network by using software applications. This architecture divides the network into a control plane
that contains logicallycentralized controlling components and a data plane that contains only the
network devices, which makes the network more intelligent. The centralized controller can
perceive a global view of the network and users. The network operators can program their
policies or forward rules for their network at any stage in networking [6]. That is, the network
operator can control the network resource management in real-time. The new concept of a
programmable network is the solution for end-to-end network management. Even though the
number of users connecting to the network is increasing, but not all of them need the same
amount of network resources. Implementing an adaptive QoS will improve network resource
management to have higher efficiency and better performance [3].
In this work, we propose the class-based adaptive QoS control scheme on an SDN-enabled
network adopting the optimization technique. The traffic types under consideration consist of
VoIP, video streaming, and file transfer. The optimization technique is utilized in order to
maximize the number of traffic flows that can be admitted into the network with QoS
satisfaction. The simulation is carried out using software called Mininet and Mininet-WiFi [13] to
create the network as well as all network elements, while the Phyton-based software called Ryu is
used to create the SDN controller [2]. The performance of our proposed scheme is evaluated
using the maximum number of traffic flows admittable with QoS as the major metric while the
metrics namely, average throughput, latency, jitter, and packet loss rate are observed. The
simulation results are compared with the existing QoS provisioning scheme called JMABC [11]
and the best effort scheme. According to the simulations results, it is obvious that our proposed
adaptive QoS control scheme on SDN can clearly admit the much higher number of traffic flows
satisfying QoS requirements than JMABC [11] while the other performance metrics, namely,
average throughput, latency, jitter and packet loss rate are maintained at the same level as
JMABC [11], but much higher than Best Effort scheme.
The rest of this paper is structured as follows: In section 2, we describe the related works. Section
3 presents the proposed algorithms where Section 4 describes the simulation software used in this
work in detail. Section 5 illustrates the simulation procedures, network topology, and simulation

57
parameters. Section 6 shows the simulation results as well as a detailed discussion. Finally, we
conclude our work and future work in section 7.
2. RELATED WORK
There are many works in the literature relating to the QoS provisioning scheme in the other
networks [17-20] as well as in SDN architecture that has been carried out [2-
4],[8],[10],[12],[15],[16],[20]. Here, we highlight some works that are of interest and relevant to
our proposed scheme.
In [8], a QoS design called Open QoS was proposed. The OpenQoS is an end-to-end QoS support
that uses centralized control capabilities over the networks. The mechanism of OpenQoS is
similar to the IntServ model which supports QoS from an end-to-end network point of view.
They use dynamic QoS routing where both the QoS requirements and network characteristics are
considered. The experiment is based on a standard OpenFlow controller, Floodlight using video
streaming traffic over a real OpenFlow test network. The quality difference between streaming
with QoS support and without QoS support of a sample test is illustrated. In conclusion,
OpenQoS can minimize packet loss and latency of the flows.
In [3], an adaptive QoS algorithm for data transfers in SDN was proposed. The framework, to
adjust the resources by redistributing the bandwidth to make sure that each flow meets the QoS
requirement, is implemented. To perform the evaluation, virtual switches, virtual hosts, traffic
generators, and one controller was created by using the open-source software called Open
vSwitch. The simulation results of their frameworks are compared with the existing QoS
methods.In conclusion, their framework can outperform those other scheduling methods meaning
that resource redistributing is one of the keys to improving the performance.
In [1], Ahmed, Hamma, and Nasir presented a two-level scheduling algorithm for the WiMax
standard. In the first level, the scheduler allocates bandwidth to different types of traffic based on
traffic demands and QoS requirements. Then, in the second level, they distribute bandwidth
among traffic flows of the same traffic type. The simulation results show that the proposed
solution can provide QoS for all of the traffic types that are supported by the standard. In the end,
it can be concluded the concept of a two-level scheduler can improve the performance of the
network scheduler.
In [11], Leong and Chieng propose a Joint Measurement-based Admission and Bandwidth
Control (JMABC) control scheme which consists of three sub-modules, Feedback Monitoring
Module which is used for monitoring throughput and sending the feedback, Admission Control
Module that is used for traffic policing by comparing throughput of network with their threshold
value, and Packets Scheduling Module is used to put traffic flow into QoS queue. Network
Simulator Tool (NS2) is the software used to evaluate their proposed scheme. In conclusion, their
work illustrated an improvement in terms of throughput and user control.
In [5], Carella, Yamada, Blum, and Luck presented Cross-Layer Orchestrator (CLO) using the
information between the Application and the Network Layers to provide a QoS guarantee on the
end-to-end network. They simulate various types of peer-to-peer traffic and conference video call
traffics with different bandwidth requirements and QoS levels and then measure the bandwidth
utilization of each flow and collect data. It can be concluded, by using information from both
Application and Network layers, that they can create a scheme that is able to provide all data flow
requirements.

58
In [15], Wang, Lin, and Luo proposed a scheme named the QAMO-SDN. Typically, QAMO is
used to achieve QoS in data centers by controlling bandwidth using the reservation method in the
OBS layer and Multipath TCP while maintaining throughput performance. The typical QAMO
was designed for traditional networks and did not support SDN architecture, however, the
proposed method in [15] enhances QAMO-SDN algorithm to make it compatible with SDN
architecture and improve the performance as well. The simulation was developed using C++. As
a result, QAMO-SDN can perform slightly better than QAMO and the traditional TCP network.
In conclusion, their traditional network scheduler scheme can be implemented in SDN
architecture and also can improve performance.
In [21], Kuribayashi S. proposed the scheme to use SDN together with NFV (Network Functions
Virtualization) paradigms to dynamically shape the traffic flow. This scheme can select the
optimal communication flows to be shaped, and the optimal shaping points dynamically. It was
shown that the net cost can be reduced significantly by using this proposed scheme.
The concept of QoS has been proposed not only in SDN but also in various types of networks
[17],[19],[20]. For example, in [17], the QoS in the protocol called OLSR in Mobile Ad Hoc
Network was investigated under two types of traffic. This work proposed the parameters called
weighted connectivity index to look for the next node to forward the packets to. The effectiveness
was shown by simulation using NS2. It was concluded that the proposed scheme can provide
better performance in terms of throughput, average end-to-end delay, packet delivery ratio,
overhead and power consumption than the traditional OLSR. In [19], the generalized multi-
constrained path QoS routing algorithm for mobile ad hoc network (G_MCP) was proposed. The
weighted connectivity index and nonlinear cost function were used. The simulation was carried
out using NS2 adopting OLSR as routing protocol. The simulation results illustrated that the
proposed algorithm provided superior performance in terms of throughput, packet delivery ratio
delay and success ratio than the traditional OLSR. While in [20], the general multi-constrained
QoS routing using weighted metrics (G_MQW) was proposed. The nonlinear cost functions and
relaxed Dijkstra’s algorithm were adopted. The general mathematical closed-form was derived.
The effectiveness of the proposed algorithm was confirmed by simulation using Matlab and
Waxman network topology. According to the simulation results, the G_MQW provided better
performance in terms of success ratio than most of the existing algorithms.
3. PROPOSED ADAPTIVE QOS CONTROL SCHEME
3.1. Overview of Adaptive QoS Control Scheme
In this section, the proposed adaptive QoS control scheme on SDN architecture is described and
implemented. To practically implement our proposed scheme, we divide the control scheme into
two subsections. The first one is the class-based QoS control module used to calculate the
distribution of network resources for each traffic class, and the other is the QoS queue
management module used for traffic policing and scheduling. Figure 1 illustrates the conceptual
model of our proposed class-based adaptive QoS control scheme on SDN architecture.

59
Figure 1. Adaptive QoS Control Scheme on SDN Architecture
From figure1, when the network is on, the SDN controller will collect information about the
network, then run a QoS control algorithm and send a new policy rule to Open-vSwitch. The
scheduling is performed when a new connection flow request had arrived at Open-vSwitch. The
detailed workflow of this algorithm is shown as pseudo-code below.
ALGORITHM ADAPTIVE QOS CONTROL SCHEME
BEGIN
Part 1 - Class-based QoS Control Module
INPUT Network Topology, Number of Traffic Class (C),QoS Requirements
FOR 𝑖 = 1 TO C
𝑏𝑖 = min_bandwidth (𝑖) // Finding Minimum bandwidth of traffic class 𝑖
𝑥𝑖 = max_number (𝑖) // Finding Maximum number of users for traffic class 𝑖
ENDFOR
Part 2 - QoS Queue Management Module
DEFINE 𝑤𝑖 ,𝑞𝑖 = empty queue for traffic class 𝑖
REPEAT
IF New Flow Request Arrives THEN
INPUT F = New Traffic Flow Request
c = getTrafficClass (F)
IF size(𝑞𝑖) < 𝑥𝑖THEN
𝑞𝑖.enqueue(F) // Every flow in 𝑞𝑖are admitted to network
and will be remove automatically, if
connection ended
ELSE
𝑤𝑖.enqueue(F) // Waiting for service
ENDIF
ELSE
FOR 𝑖 = 1 TO C
IF 𝑞𝑖 is NOT full THEN
F = 𝑤𝑖.𝑑𝑒queue()
𝑞𝑖.enqueue(F)
ENDIF
ENDFOR
ENDIF
UNTIL No New Traffic Request AND 𝑤𝑖 ,𝑞𝑖is EMPTY
END

60
3.2. Class-based QoS Control Module with Optimization Technique
The role of the proposed class-based QoS control module is to setup the QoS policy rule to define
the treatment of network flow, which can be described step-by-step as follows:
Step 1: SDN controller collects information about network topology and QoS requirements.
Step 2: Calculate the minimum bandwidth requirement of different types of traffic based on
traffic type and QoS requirements. The calculation of each traffic type depends on its traffic
class. For example, in voice traffic, the minimum bandwidth requirement will be calculated by
using audio codec and sampling rate.
Step 3: Calculate the maximum number of traffic flows for each traffic type by using Integer
Linear Programming (ILP) optimization. To formulate ILP problem, the decision variables,
objective function, and constraints are defined as follows:
maximize
subject to
where N is the number of users in network
B is total system bandwidth of network
C is the number of traffic classes.
𝒃𝒊 is the minimum bandwidth requirement for each traffic class ith
.
𝒙𝒊 is the maximum number of flows of traffic class
ith
satisfying QoS requirement admittable into the network.
Here, we adopt the optimization technique as shown above to maximize the number of flows of
each traffic class that can be admitted into the network by having bandwidth limitation
constraints.
3.3. QoS Queue Management Module
In order to implement our proposed class-based adaptive QoS control scheme in SDN
architecture, the procedure for our QoS management module needs to be defined.
Step 1: SDN controller collects information about the network such as network topology, traffic
classes, QoS requirements, then run a class-based QoS control module to find the minimum
bandwidth based on QoS requirement and maximum number of traffic flows for each traffic class
that can be admitted.
Step 2: SDN controller sends a message to Open-vSwitch to define a new rule for traffic
scheduling.

61
Step 3: Open-vSwitch defines new QoS queue based on traffic class. Each queue must contain
minimum bandwidth requirement and maximum number of traffic flows that can be admitted into
the network. Each QoS queue will be divided into two parts; the active queue for an active flow
and waiting queue for the flow that cannot be admitted under current circumstance.
Figure 2. The conceptual model of QoS Queue
Step 4: Waiting for a new flow request arrives. Then, Open-vSwitch checks the traffic class of
that requesting flow, classifies and sends it to QoS queue based on traffic class. Open-vSwitch
will put the flow into the active queue, if the active queue is available. However, Open-vSwitch
will put it into waiting queue to receive the service later on, if the current number of flows in the
active queue has already reached the maximum value.
Figure 3. Example of QoS management module
Step 5: Checking whether any flows in the active queues end their connections, Open-vSwitch,
then, has to admit the new traffic request from waiting queue to active queue.
4. NETWORK SIMULATION SOFTWARE
In this section, we describe the software used for simulation in our proposed scheme. The
software used mainly in this work consists of two software; Mininet, a well-known network
emulator for SDN architecture and SDN controller, and Ryu, a phyton-based opensource
controller, which is compatible with the Mininet emulator.

62
4.1. Mininet
Mininet is a network emulator that has been developed by Mininet Core Team, [13]. It can create
a network that consists of virtual hosts, switches, controllers, and links. Mininet is used for
development and testing network simulations with OpenFlow protocol on SDN architecture.
4.2. Mininet-WiFi
Mininet-WiFi, [9] is an extension of the Mininet emulator where the functionalities of wireless
station components such as WiFi stations and access points are added. The Mininet-WiFi has
been developed based on standard Mininet code by adding or modifying classes and scripts. This
means that Mininet-WiFi supports all standard Mininet components and can work along with the
SDN controller being supported by standard Mininet.
4.3. Ryu Controller
Ryu is a python-based opensource SDN controller. Ryu provides a well-defined API for
developers to manage network components and applications. Ryu supports various OpenFlow
protocols which are used in the Mininet emulator.
4.4. IPerf
IPerf is a cross-platform network measurement tool that can generate traffic with various
parameters. IPerf also has server and client functionality where we can create data flows to
measure the performance from the end-to-end network.
5. NETWORK TOPOLOGY AND SIMULATION PARAMETERS
In this section, we illustrate all details regarding our simulation, namely system configuration,
network topology, simulation parameters, etc. Here, all our simulations are carried out on a
virtual network where the data plane is simulated on Mininet-WiFi emulator by creating virtual
hosts, Open-vSwitches, and SDN controller which support OpenFlow protocol, while the control
planeis carried out on Ryu SDN controller.
5.1. System Configuration
To set up the network simulation and network environment, it is necessary to install many
software tools. The details of software tools of each element/system used in this simulation are
shown in Table 1 as follows:
Table 1. System Configuration
Systems Details
Operating System Lubuntu 20.04
SDN Controller Ryu 4.3.4
Switches Open-vSwitch 1.3
Network Emulator Mininet 2.2.2
Processor and Memory 3.6 GHz with 16 GB memory

63
5.2. Network Model and Environment
In this paper, we consider mainly on wireless local area network (WLAN). The topology of
network under consideration consists of multiple wireless stations, one access point, one Open-
vSwitch, one SDN controller, and one virtual server, as shown in figure 4.
Figure 4. The Network Model under consideration
The parameters of virtual network environments for our simulation are illustrated in Table 2 as
follows:
Table 2. Network Environment Parameters
Environments Values
Total System Bandwidth 5 Mbps
Simulation Time 200 Seconds
Experimental Area 1000x1000 square meters
Position of User Wireless Stations Uniform Distribution U{−300,300}
User Arrival Time Poisson Distribution (1λ = 10 s)
User Service Time Exponential Distribution (1λ = 80 s)
Noise Distribution Gaussian Distribution – N(0,1)
Propagation Model Log-Distance Path Loss - 𝛾 = 3
Number of Iterations 10 Times
5.3. Network Traffic Classes
In this work, three traffic classes are considered. The details of each traffic classes are shown in
Table 3 as follows:
Table 3. Characteristics of each Network Traffic Classes
Traffic Class Descriptions
Voice over IP G.711 coded VoIP with 20 ms sampling rate payload
Video Streaming H.264 coded Video, 704x480 resolution, 5.5 KB frame size, and 15 FPS
Files Transfer High Speed Download
In our simulation, the concept of traffic priority is also introduced. The priority of traffic classes
is shown in descending order as follows:

64
1) Voice over IP (VoIP)
2) Video Streaming
3) Files Transfer
We assume that each user’s wireless stations randomly request a flow connection to Open-
vSwitch according to network environment parameters shown in Table 2.
5.4. Performance Evaluation Metrics
The performance evaluation metrics used in this work consist of the major ones, which is the
main contribution of this work; the Maximum Number of Admittable Traffic Flows with QoS,
and the minor ones; Average Throughput, Average Latency, Average Jitter, and Average Packet
Loss Rate. The maximum number of admittable traffic flows with QoS is adopted as the main
evaluation metric since our objective is to efficiently allocate the limited resource to the
applications that need QoS guarantee while providing no or less impact to the applications that
QoS deems unnecessary. All performance evaluation metrics and their definitions are given in
Table 4 as follows:
Table 4. Performance Evaluations Metrics
Major Performance
Evaluation Metric
Descriptions
Max. Number of Admittable
Traffic Flows
The maximum number of traffic flows admittable with QoSguaranteed
for each class of traffic.
Minor Performance
Evaluation Metric
Descriptions
Average Throughput The total amount of data which is successfully transferred fromsource
to destination under some certain period of time.
Average Latency The time delay of traffic from source to destination under some certain
period of time.
Average Jitter The variation in the time between data packets arrivals under some
certain period of time.
Average Packet Loss Rate The percentage of packets that are unable to reach destination under
some certain period of time.
The QoS requirements for each class of traffics; Voice over IP, Video Streaming, and File
transfer are adopted from Cisco guidelines, [7] and are shown in Table 5 as follows:
Table 5. QoS requirements of Traffic Flows under Consideration
Traffic type Bandwidth Requirement Average
Jitter
Average
Latency
Packet Loss
Rate
Voice over IP 17~106 Kbps per call depending on
sampling rate
≤ 30 ms < 150 ms < 1%
Video
Streaming
depending on the encoding and the
rate of the video stream.
None < 4 seconds < 2%
File Transfer 20-25% of total bandwidth
(recommendation)
None None None

65
6. SIMULATION PROCEDURES, RESULTS AND DISCUSSION
In this section, the simulation results based on the performance evaluation metrics are shown.
Here, we compare the results achieved from our proposed class-based adaptive QoS control
scheme with the other QoS provision schemes, namely JMABC [11] and the best effort scheme.
All results are plotted using mean values with 95% confidence interval.
6.1. Simulation Procedures
The procedure to run software in our simulation is described step-by-step as follows:
6.1.1. Running Mininet Emulator and Ryu Controller
The first step of the simulation is to run the Mininet emulator along with Ryu SDN controller to
create the network. The network topology of this simulation consists of 1 virtual server, 1 access
point, 1 Open-vSwitch, 1 SDN controller, and 20 devices which are randomly distributed inside
the network area under consideration.
Figure 5. Network Topology showing devices distribution created by Mininet
6.1.2. Running Class-based Adaptive QoS Control Algorithm
Next, we run the software to evaluate the proposed class-based adaptive QoS control algorithm
using three traffic classes shown in Table 3. When the minimum bandwidth requirement of each
traffic class is identified, the maximum number of flows satisfying QoS requirements can be
achieved as shown in Table 6 as follows:
Table 6. Results from Class-based Adaptive QoS Control Algorithm
Traffic Class Minimum Bandwidth Requirement Maximum Number of
Flows
Voice over IP 64 Kbps per flow 10
Video Streaming 640 Kbps per flow 3
Files Transfer 940 Kbps No limit

66
6.2. Simulation Results and Discussion
In this section, all simulation results based on the performance evaluation metrics are illustrated.
The results achieved from our proposed algorithm are compared with the existing QoS provision
in the literature called Joint Measurement-based Admission and Bandwidth Control
(JMABC)[11] and Best Effort scheme.
6.2.1. The Major Performance Evaluation Metric
• The Maximum Number of Admittable Traffic Flows with QoS
The comparison of the maximum number of admittable traffic flows with QoS is shown in Fig. 6
as follows:
Figure 6. Comparison of Maximum Number of Admittable Traffic Flows with QoS
In this work, our main concern is to enhance the performance of the network by increasing the
number of traffic flows admittable with QoS guaranteed using the optimization technique. Our
proposed class-based adaptive QoS control scheme is simulated and compared with the existing
work called JMABC [11] based on various traffic flows; VoIP, Video Streaming, and File
Transfer. In our simulation scenarios, the highest priority is given to VoIP traffic, while the
Video Streaming and File Transfer traffic are given the second and the third priority (Best
Effort), respectively. Based on our simulation results, it is obvious that our proposed algorithm
can admitdrastically the higher number of high priority traffic classes than JMABC [11], while
the second priority as well as the best effort traffic are maintained almost at the same level under
the same resources and environment. This is due to the priority given to VoIP and the
optimization technique adopted to efficiently allocate the resources to the traffic.
6.2.2. Minor Performance Evaluation Metrics
• Average Throughput
Here, the average throughput of our proposed algorithm comparing to JMABC [11] and the best
effort scheme of all traffic classes are illustrated.
20
3
1
5 4
2
0
5
10
15
20
25
VoIP Video Streaming Files Transfer
Number
of
Flows
Traffic Class
Comparison of Maximum Number of
Admittable Traffic Flows with QoS
Proposed Adaptive QoS JMABC

67
Figure 7. Average Throughput of Voice Over IP Traffic under various QoS provision schemes
Figure 7 illustrates the average throughput of VoIP traffic of all algorithms. In the beginning, the
average throughput of all QoS control schemes is directly proportional to the input traffic. Once
the throughput is increased beyond the minimum requirement (64 kbps), or a new traffic flow
requests for the connection with a QoS guarantee, our proposed algorithm and JMABC [11] try to
maintain the existing connections and if possible, admit the new connections satisfying QoS,
while the best effort scheme could not maintain the connection. A new traffic flow has an impact
on network performance as we can see that the throughput of VoIP traffic decreases rapidly in the
case of the best effort scheme, but it has no effect when the QoS control scheme is adopted.
Figure 8. Average Throughput of Video Streaming Traffic under various QoS provision schemes
Similarly, figure 8 illustrates the average throughput of Video Streaming traffic of all algorithms.
From figure 8, which is similar to VoIP traffic, when new traffic arrives, it has no impact on the
performance of Video Streaming traffic. In addition, the Video Streaming traffic that has QoS
control scheme can achieve minimum bandwidth requirement (640 kbps).
0
20
40
60
80
0 20 40 60 80 100 120 140 160 180 200
Throughput
(Kbps)
Time (Sec)
Average Throughput of VoIP Traffic
Best Effort JMABC Proposed Adaptive QoS
0
100
200
300
400
500
600
700
800
0 20 40 60 80 100 120 140 160 180 200
Throughput
(Kbps)
Time (Sec)
Average Throughput of Video Streaming Traffic

68
Figure 9. Average Throughput of File Transfer Traffic of our proposed Adaptive QoS provision scheme
Figure 9 shows the average throughput of File Transfer traffic which is classified as the best
effort traffic. It is obvious that our proposed class-based adaptive QoS control scheme can
achieve the average throughput of the File Transfer traffic nearly 1 Mbps which is approximately
20% of total system bandwidth even though our proposed adaptive QoS provision scheme has
already admitted a large number of high priority traffics. This is because the optimization
technique introduced in our proposed method can enhance efficiently the resource allocation.
• Average Latency
Figure 10. Average Latency of Voice over IP Traffic under various QoS provision schemes
Typically, latency is the most concerned parameter in Voice over IP traffic. The average latency
achieved from our proposed algorithm is measured and compared with JMABC [11] and the best
effort scheme, as shown in figure 10. Our proposed algorithm as well as JMABC [11] can
achieve the recommended latency from Cisco which is lower than 150 ms throughout the time
duration of the experiment. The rationale here is similar to the cases of throughput of VoIP.
0
200
400
600
800
1000
1200
0 20 40 60 80 100 120 140 160 180 200
Throughput
(Kbps)
Time (Sec)
Average Throughput of File Transfer Traffic
Proposed Adaptive QoS
0
50
100
150
200
250
300
350
400
450
0 20 40 60 80 100 120 140 160 180 200
Latency
(ms)
Time (Sec)
Average Latency of VoIP Traffic

69
Figure 11. Average Latency of Video Streaming Traffic under various QoS provision schemes
Similarly, the average latency of Video Streaming traffic in figure 11 also follows the similar
trend as VoIP traffic shown in figure 10.
• Average Jitter
Figure 12. Average Jitter of VoIP Traffic under various QoS provision schemes
Figure 12 displays the average jitter of VoIP traffic of all algorithms. In figure 12, it is apparent
that our proposed adaptive QoS control algorithm provides a lower average jitter than the
minimum requirement (30 ms) which is almost the same as JMABC [11], while the Best Effort
scheme cannot satisfy the QoS requirement.
0
200
400
600
800
1000
1200
1400
0 20 40 60 80 100 120 140 160 180 200
Latency
(ms)
Time (Sec)
Average Latency of Video Streaming Traffic
0
10
20
30
40
50
60
70
80
0 20 40 60 80 100 120 140 160 180 200
Jitter
(ms)
Time (Sec)
Average Jitter of VoIP Traffic

70
Figure 13. Average Jitter of Video Streaming Traffic under various QoS provision schemes
Figure 13 illustrates the average jitter of Video Streaming traffic of all algorithms. Similar to
figure 12, it is apparent that our proposed adaptive QoS control algorithm and JMABC [11]
provide a very lower jitter even there is no rigorous jitter requirement for video streaming traffic.
• Average Packet Loss Rate
Both of our proposed adaptive QoS control algorithm and JMABC [11] provide nearly 0% packet
loss throughout the time duration of the experiment. Only Best Effort traffic has very high packet
loss rate. Therefore, the results of this part are skipped.
7. CONCLUSIONS
This paper had demonstrated the integration of Quality of Service (QoS) provisioning and
Software-Defined Network (SDN) by implementing Quality of Service in terms of Integrated
Service on Software-Defined Network. All details regarding the software as well as
implementation were described.
The class-based adaptive QoS control algorithm proposed in this work was the combination of
class-based QoS control algorithm adopting optimization technique and QoS queue management
in SDN. This adaptive control scheme had offered the method to provide a QoS guarantee to each
traffic class and, additionally, had admitted the maximum number of traffic flows satisfying QoS
requirements into the network. We had shown the effectiveness of our proposed algorithm by
simulation using Mininet and Ryu controller and had compared the results with the existing work
called Joint Measurement-based Admission and Bandwidth Control (JMABC) and the best Effort
scheme.
It was obvious from the simulation results that our proposed algorithm could provide a QoS
guarantee to all traffic under consideration. Additionally, due to the optimization technique
adopted in our proposal, it was clear that our proposed algorithm could provide the much higher
maximum number of admittable traffic flows (approximately 300% higher than the existing work
(JMABC) in the case of VoIP traffic which had the highest priority and almost the same for
Video Streaming traffic which had the second priority) satisfying QoS requirements under the
0
50
100
150
200
250
300
350
0 20 40 60 80 100 120 140 160 180 200
Jitter
(ms)
Time (Sec)
Average Jitter of Video Streaming Traffic

71
same environment as JMABC. Finally, our proposed algorithm could maintain almost equivalent
performance to JMABC in terms of average throughput, latency, jitter, and packet loss rate.
CONFLICT OF INTEREST
The authors declare no conflict of interest.
ACKNOWLEDGEMENTS
The success of this project is only possible because of the support, collaboration and trust of my
advisor and institutions that has helped me to make them come true. Thanks all.
REFERENCES
[1] Ahmed Z., Hamma, S., & Nasir, Z. (2019) “An optimal bandwidth allocation algorithm for improving
QoS in WiMAX”, Multimedia Tools and Applications, Springer Verlag, Vol. 78, June 2019,
pp.25937-25976.
[2] Amiri, E., Alizadeh, E., & Rezvani, M. H. (2020) “Controller selection in software defined networks
using best-worst multi-criteria decision-making”, Bulletin of Electrical Engineering and Informatics,
Vol. 9, No.4, pp.1506–1517.
[3] Boley, J. M., Jung, E.-S., & Kettimuthu, R. (2016) “Adaptive QoS for Data Transfers Using
Software-Defined Networking”, 2016 IEEE International Conference on Advanced Networks and
Telecommunications Systems (ANTS), Bangalore, India, November 6th
-9th
, 2016
[4] Braun, W., & Menth, M. (2014) “Software-Defined Networking Using OpenFlow: Protocols,
Applications and Architectural Design Choices”. Future Internet, Vol. 6, pp.302–336.
[5] Carella, G., Yamada, J., Blum, N., & Luck, C. (2015) “Cross-layer service to network orchestration”,
2015 IEEE International Conference on Communications (ICC), London UK, June 8th
-12th
, 2015
[6] Chen, J., Lv, T., & Zheng, H. (2004) “Cross-layer design for QoS wireless communications”, 2004
IEEE International Symposium on Circuits and Systems, Vancouver, BC, Canada, May 23rd
-26th
,
2004.
[7] Cisco. (2006) “Implementing Quality of Service Over Cisco MPLS VPNs”, Selecting MPLS VPN
Services.
[8] Egilmez, H. E., Dane, S. T., Bagci, K. T., & Tekalpe, A. M. (2012) “OpenQoS: An open-flow
controller design for multimedia delivery with end-to-end quality of service over software-defined
networks”, Proceedings of the 2012 Asia Pacific Signal and Information Processing Association
Annual Summit and Conference, Hollywood, CA, USA, December 3rd
-6th
, 2012.
[9] Fontes, R. R., Afzal, S., Brito, S. H. B., Santos, M., Rothenberg, C. E. (2015) “Mininet-WiFi:
Emulating Software-Defined Wireless Networks”, 2015 11th
International Conference on Network
and Service Management (CNSM), Barcelona, Spain, November 9th
-13th
, 2015.
[10] Kreutz, D., Ramos, F. M. V., Ver ́ıssimo, P. E., Rothenberg, C. E., & Azodol, S. (2015) “Software-
defined networking: A comprehensive survey”, Proceedings of the IEEE, Vol.103, Issue 1, January
2015, pp.14-76.
[11] Leong, P. W., & Chieng, D. (2014) “A Joint Measurement-Based Admission and Bandwidth Control
QoS Provisioning Scheme for WLANS”, International Conference on Frontiers of Communications,
Networks and Applications (ICFCNA 2014), Kuala Lumpur, Malaysia, November 3rd
-5th
, 2014.
[12] Letswamotse, B. B., Modieginyane, K. M., & Malekian, R. (2016) “SDN based QoS provision in
WSN technologies”, Southern Africa Telecommunication Networks and Applications Conference
(SATNAC), George, South Africa, September 4th
-7th
, 2016.
[13] Mininet Core Team. (2017) “Mininet: Rapid Prototyping for Software Defined Networks”, Open
Networking Foundation (2015), Openflow switch specification, Open Networking Foundation(2016),
SDN architecture overview.
[14] Shome, P., Yan, M., Najafabad, S. M., Mastronarde, N., & Sprintson, A. (2015) “Crossflow: A cross-
layer architecture for SDR using SDN principles” IEEE Conference on Network Function
Virtualization and Software Defined Networks (NFV-SDN), San Francisco, CA, USA, November
18th
-21st
, 2015.

72
[15] Tariq, S., & Bassiouni, M. (2015) “Qamo-SDN: QoS-aware multipath TCP for software
definedoptical networks”, 12th Annual IEEE Consumer Communications and Networking Conference
(CCNC), Las Vegas, NV, USA, January 9th
-12th
, 2015.
[16] Wang, P., Lin, S.-C., & Luo, M. (2016) “A Framework for QoS-aware Traffic Classification Using
Semi-Supervised Machine Learning in SDNs’, 2016 IEEE International Conference on Services
Computing, Vol. 6, San Francisco, CA, USA, June 27th
- July 2nd
, 2016, pp. 760–765.
[17] Sanguankotchakorn T., Wijayasekara S.K. andSugino N. (2018) “Performance of OLSR MANET
adopting Cross-layer Approach under CBR and VBR Traffic Environment”, International Journal of
Computer Networks and Communications (IJCNC), Vol. 10, No.6, November 2018.
[18] You, Z., Cheng, G., Wang, Y., Chen, P., & Chen, S. (2019) “Cross-layer and SDN based routing
scheme for P2P communication in vehicular ad-hoc networks”, Applied Sciences, Vol.9, Issue 22,
November 2019.
[19] Kunavut K. and Sanguankotchakorn T. (2012) “Generalized Multi-constrained Path (G_MCP) QoS
Routing Algorithm for Mobile Ad Hoc Networks”, Journal of Communications, Vol.7, No.3, March
2012, pp.246-257.
[20] Poungmanee M. and Sanguankotchakorn T. (2020) “General Multi-constrained QoS Routing using
Nonlinear Functions and Weighted Metrics”, Journal of Communications, Vol.15, No.6, June 2020,
pp.480-495.
[21] Kuribayashi S. (2021) “Dynamic Shaping Method using SDN and NFV Paradigms”, International
Journal of Computer Networks and Communications (IJCNC), Vol.13, No.2 March 2021.
AUTHORS
Pacharakit Vanitchasatit was born in Loei, Thailand on April 14, 1997. He received B.
Eng in computer engineering from Khon Kaen University, Thailand in 2019 and M. Eng in
Telecommunications from Asian Institute of Technology, Bangkok, Thailand in 2021.
Teerapat Sanguankotchakorn was born in Bangkok, Thailand on December 8, 1965. He
received the B. Eng in Electrical Engineering from Chulalongkorn University, Thailand in
1987, M. Eng and D. Eng in Information Processing from Tokyo Institute of Technology,
Japan in 1990 and 1993, respectively. In 1993, he joined Telecommunication and
Information Systems Research Laboratory at Sony Corporation, Japan where he holds two
patents on Signal Compression. Since October 1998, he has been with Asian Institute of
Technology where he is currently an Associate Professor at Telecommunications Field of
Study, School of Engineering and Technology. Dr. Sanguankotchakorn is a Senior member of IEEE and
member of IEICE, Japan.

A Class-based Adaptive QoS Control Scheme Adopting Optimization Technique over WLAN SDN Architecture

More Related Content

Similar to A Class-based Adaptive QoS Control Scheme Adopting Optimization Technique over WLAN SDN Architecture (20)

More from IJCNCJournal (20)

Recently uploaded (20)

A Class-based Adaptive QoS Control Scheme Adopting Optimization Technique over WLAN SDN Architecture