Journal of Physics: Conference

Series

PAPER • OPEN ACCESS You may also like

- Automatically reconfigurable optical data
Application Scenarios based on SDN: An center network with dynamic bandwidth
allocation
Overview Xuwei Xue, Kristif Prifti, Bitao Pan et al.

- Network Management in Software-Defined

Network: A Survey
To cite this article: Tong Li et al 2019 J. Phys.: Conf. Ser. 1187 052067 Zaid Ibrahim Rasool, Ridhab Sami Abd Ali
and Musaddak Maher Abdulzahra

- Design and Research of SDN Unified

Controller in Large Data Center
Jiye Wang, Hui Liu and Cong Yu
View the article online for updates and enhancements.

This content was downloaded from IP address 41.238.36.248 on 26/10/2024 at 14:26

ISPECE IOP Publishing
IOP Conf. Series: Journal of Physics: Conf. Series 1187 (2019) 052067 doi:10.1088/1742-6596/1187/5/052067

Application Scenarios based on SDN: An Overview

Tong Li 1,2, Jinqiang Chen2, Hongyong Fu 2

1
University of Chinese Academy of Sciences, Beijing 100049, China.
2
Technology and Engineering Center for Space Utilization, Chinese Academy of
Sciences, Beijing 100094, China.
[email protected], chjinq@ csu.ac.cn, fuhongyong@ csu.ac.cn

Abstract—As a new network concept, SDN subverts the vertical structure of traditional
network integration, decoupling the control plane and data plane of the network, replacing
distributed control with centralized control, and transforming the closed network into an open
one. Programming platform to promote the innovation of network applications, in order to
deeply study the application of SDN, we put forward the concept of "SDN+", which combines
SDN with different network scenarios to provide a new paradigm for network applications. The
article analyzes the architecture of SDN and investigates the application of SDN in different
scenarios.

1. INTRODUCTION
The digital society established by the Internet connects almost everything and can be accessed anytime,
anywhere. The hierarchical structure of the traditional network promotes the rapid development of the
Internet. The traditional network adopts distributed control, and the control plane and the data plane are
closely coupled. However, for the complexity and scalability of the current network requirements, the
traditional network gradually exposes architectural defects, including three aspects: 1）closed network
deployment management is difficult; 2）flow control requires high network operation and maintenance
costs; 3）distributed architecture limits network configuration and network innovation.
In 2006, SDN (Software-Defined Networking) was born in the Stanford University Clean Slate
project funded by the GENI project in the United States. The purpose of this project is to reinvent the
Internet and change the existing network infrastructure. The concept of SDN came into being.
SDN is essentially a new type of network architecture or a network design concept, which uses IT
technology to soften the network. Moreover, SDN is application-centric and software-based. The
comparison between SDN and traditional network architecture is shown in Figure 1. The SDN network
architecture has three layers, which are the application layer, the control layer and the data layer, from
top to bottom, as shown in Figure 2. The core of SDN technology is to separate the control plane and
data plane of the network, and make the network open and programmable through centralized control.
By separating the control plane in the traditional network equipment from the hardware, using the
gradual generalization and standardization of the underlying hardware, the entire virtualized network
layer is loaded on the physical network, laying the foundation for the introduction of virtualization
technologies such as NFV. The application layer is composed of some applications, which are
connected to the control layer through the northbound interface, thereby programming the underlying
device and developing various business applications according to the needs of the user. The control
layer is responsible for formulating the flow table to issue commands to the data layer, maintaining

Content from this work may be used under the terms of the Creative Commons Attribution 3.0 licence. Any further distribution
of this work must maintain attribution to the author(s) and the title of the work, journal citation and DOI.
Published under licence by IOP Publishing Ltd 1
ISPECE IOP Publishing
IOP Conf. Series: Journal of Physics: Conf. Series 1187 (2019) 052067 doi:10.1088/1742-6596/1187/5/052067

topology state information of the entire network, and dynamically allocating network resources; the
data layer is mainly responsible for forwarding data packets, and is connected to the control layer
through the southbound interface.
App1 …… APPn APP1 …… APPn

Control
Layer
Traditional network architecture: SDN network architecture:
distributed control centralized control Infrastructure
Layer

Fig. 1. Comparison of traditional network architecture and SDN network architecture.

Access Attack
Application Layer Control Detection

Routing Load
Algorithms Balancers
Northbound API
Security Topology
Mechanisms Manager
control layer
Stats Flow table
Manager Manager

Southbound API

Data
data layer Forwarding

Fig. 2. SDN network architecture.

2. THE OVERVIEW OF SDN APPLICATION SCENARIO

SDN originally originated from the experimental innovation of the campus network. The research team
led by Professor Nick McKeown of Stanford University proposed the concept of Openflow [15], which
promoted the birth of SDN; SDN succeeded in Google's B4 [6] network, in the early stage of
development of SDN, the application scenarios are mainly focused on: data center network,
interconnection between data centers, government and enterprise networks, telecom carrier networks,
and Internet company business deployment. The article will outline the initial application of SDN and
some innovative application scenarios in the later stage.

2.1. SDN+WAN
Paper [6] proposes a software-defined WAN architecture B4, as shown in Figure 3, which is one of the
first and largest SDN/OpenFlow deployments. This is the first publicly available commercial case for
SDN-based data center interconnection and the first publicly available SDN-based distributed controller
application case. This network uses a three-tier architecture consisting of switch hardware layer, site
controllers layer, and global layer. The hardware layer of the switch uses switches made by Google.
The switches run the OpenFlow protocol and support new services such as centralized TE, which is
mainly responsible for forwarding traffic. The site controller layer is composed of OpenFlow Controller

2
ISPECE IOP Publishing
IOP Conf. Series: Journal of Physics: Conf. Series 1187 (2019) 052067 doi:10.1088/1742-6596/1187/5/052067

(OFC) and Network Control Server (NCS). The OFC cluster runs on the NCS, and the OFC maintains
the network status according to the instructions of the network control application and switch events.
The global layer consists of an SDN gateway and a central traffic engineering server, responsible for
centralized and unified control of the entire network. The SDN gateway collects link information from
the OFC and abstracts it to the TE Server to learn the global path information. The forwarding entry
information generated by the TE server is translated by the SDN Gateway and sent to the underlying
switch through the OFC. The centralized traffic engineering of the architecture allocates bandwidth
according to the level of demand and priority. The bandwidth utilization of the link is improved by
more than 3 times and the long-term is close to 100%，therefore the link cost is greatly reduced. This
proves that SDN can be successfully applied to large-scale networks interconnected by data centers.
And it also describes some lessons in the research and future research directions.

Fig. 3. B4 architecture overview[6] .

2.2. SDN+ Cellular Networks

For the limitation of the current cellular network, as shown in Figure 4, the authors in [7] designed an
SDN-based architecture of the programmable control plane in cellular network, and analyzed how the
introduction of the SDN architecture can better solve the problems of the cellular network, such as
reducing the extra equipment in the network. Cellular providers can effectively monitor traffic at
different levels of granularity; SDN will provide common control protocols across different cellular
technologies, making mobility management easier and reducing latency; SDN can make distributed
implementation of QoS and firewall policies based on network-wide views, and manage traffic
scheduling through multi-hop QoS classes in the network; SDN can support network virtualization. The
mobile virtual operator can subscribe to the corresponding service according to its own needs; the SDN
has a global view of the base station, and supports centralized control of the base station. And the SDN
controller can allocate radio resources more efficiently to handle new users. The scalability of
controllers, switches and base stations is proposed from four aspects: 1) The architecture will
automatically convert policies based on user packet processing rules of IP addresses and network
locations. 2) Under the direction of the controller, a software agent capable of performing simple local
operations should be run on the switch. 3) The cellular network will support deep packet inspection,
header compression and message-based control protocols. 4) The architecture virtualizes base station
resources, allowing each "slice" of controllers to control radio resource management, admission control,
and mobility control.

3
ISPECE IOP Publishing
IOP Conf. Series: Journal of Physics: Conf. Series 1187 (2019) 052067 doi:10.1088/1742-6596/1187/5/052067

Fig. 4. Cellular SDN architecture[7] .

2.3. SDN+Mobile Cloud

Paper [8] proposes several mobile cloud architecture designs based on SDN and Ad hoc networks. As
shown in Figure 5, different wireless technologies are used in the control plane and data plane
respectively: the remote wireless connection is used for LTE in the control plane, and the high
bandwidth wireless connection is used for Wi-Fi in the data plane. The global SDN controller is the
core intelligent component of the architecture and is responsible for populating the flow table of the
SDN wireless node to control the way the traffic moves in the network. The SDN wireless node is
equivalent to the switch in the SDN network. Each SDN wireless node has an alternate local SDN
controller to ensure normal communication of the network when there is a problem with the global
controller. The global SDN controller and the SDN wireless node have two connection examples,
which are a continuously connected global SDN controller and an intermittently connected global SDN
controller. In the MANET system, each SDN wireless node constructs a node connection graph by
exchanging beacon messages, and the SDN controller can make intelligent routing decisions based on
general algorithms or flow-based routing information according to the path information provided by the
node connection graph. The article also describes the mobile cloud instance based on SDN frequency
selection. According to the interface frequency configuration, the wireless node can be divided into:
wireless nodes with wireless interfaces configured to specific frequencies and wireless nodes with
reconfigurable wireless interfaces, In addition, there are three potential application cases: wireless
network virtualization, reserved traffic, and frequency hopping. The article proves the feasibility of
SDN-based mobile cloud architecture by emulating on NS-3. Based on SDN routing, it is better than
traditional Adhoc routing protocol, which can achieve good packet transmission rate. Finally, the author
proposes three future exploration directions: 1) recovery mechanism when SDN controller is lost; 2)
standby SDN wireless architecture; 3) frequency selection use case.

Fig. 5. SDN-based mobile cloud overview[8] .

4
ISPECE IOP Publishing
IOP Conf. Series: Journal of Physics: Conf. Series 1187 (2019) 052067 doi:10.1088/1742-6596/1187/5/052067

2.4. SDN+VANET
Authors in paper [9] proposed the SDN-based VANET architecture and its operation mode as shown in
Figure 6. According to the degree of control of the SDN controller, the architecture is divided into three
modes: centralized control mode, distributed control mode and hybrid control mode. The three
advantages of the architecture are analyzed: 1) Path selection: When the SDN controller senses that data
traffic may become unbalanced, it can initiate a rerouting process to improve network utility and reduce
congestion. 2) Frequency/Channel Selection: SDN-based VANET can dynamically coordinate the
channel/frequency used by SDN wireless nodes with multiple available wireless interfaces. 3) Power
selection: Due to the awareness, SDN-based VANET can dynamically change the wireless interface
power and its transmission range according to the perception. Finally, the SDN-based routing and
traditional MANET/VANET routing protocols are compared by simulation, which proves the feasibility
of software-defined VANET.

Fig. 6. Software-Defined VANET Communications[9] .

2.5. SDN+Air-Space-Ground Integration Network

An SDN-based air-ground integration network described in paper [10], consists of a three-layer
structure of an air-based network layer, a space-based network layer and a ground-based network layer.
Each layer works independently and is interconnected. The space-based network layer is mainly
composed of satellites. The air-based network layer is composed of UAV Ad-hoc network. The ground-
based network layer is mainly composed of some cellular networks and digital cluster systems. In
addition to the space-based system, the other two layers are equipped with SDN controllers to
implement unified centralized control for each layer of the network. In the UAV Ad-hoc network, each
cluster head is equipped with a micro SDN controller, and has real-time global information of the
cluster network, and other UAVs do not need to carry any processing and decision-making equipment,
greatly improving the endurance capability of the UAV Ad-hoc network. The ground station controller
can be used as the master controller to command the entire UAV Ad-hoc network. If the link between
the UAVs and the ground station fails, the ground-based satellite relay can be used to contact the
ground station. The cooperation of the three layers of the network ensures the normal operation of the
entire network. In this paper, the delay is used as an indicator to simulate the performance of AODV
routing protocol and OLSR and DSR routing protocols. At the same time, the traditional air-space-
ground integrated network and the SDN-based air-space-ground integrated network are compared by
simulation. The results show that the transmission of SDN-based air-space-ground integrated network is
better, the end-to-end delay is shorter, and the overall performance is superior.
Integrated space and terrestrial network architecture envisioned in paper [11] is based on the SDN
architecture. The application layer consists of a number of spatial tasks. The control plane consists of
the controller nodes of the space base and the ground, and is represented by the network operating
system in the logical architecture. The control plane is the core of the network and has an overall global
view of the network to implement routing decisions and management and control of the entire spatial
information network. The data plane consists of satellite nodes and is mainly responsible for data

5
ISPECE IOP Publishing
IOP Conf. Series: Journal of Physics: Conf. Series 1187 (2019) 052067 doi:10.1088/1742-6596/1187/5/052067

processing and forwarding. The physical architecture consists of a space-based backbone network, a
data forwarding layer, and a ground information port. The space-based backbone network consists of
three GEO satellites, the data forwarding layer consists of MEO and LEO satellites in the middle and
low orbit, and some ground station controllers are set up in the ground information port. The controller
node of the ground information port and the GEO satellite node in the space-based backbone network
together form a control plane in which multiple controllers cooperate. In a multi-controller deployment
scenario, each controller is responsible for its own control area, with network status information of the
local area, manages and maintains the normal operation of the network in the area, and communicates
between the controllers through the east-west interface to generate the global topology view of entire
network. Since there are still large technical difficulties in placing the controller on the satellite, the
SDN controller is placed on the ground in the early stage. The global view of the SDN architecture
provides a more reasonable routing strategy for the satellite network. The centralized resource
allocation and management guarantees higher service transmission quality. The open and
programmable features make the configuration of the satellite network more flexible and convenient.
The unified standards and interfaces well solve the problem of heterogeneity of the spatial network. The
separation of the data plane and the control plane simplifies the architecture of the satellite function and
the design cost, effectively reducing the complexity of satellite management.

2.6. SDN+ Satellite network

OpenSAN [12], software-defined satellite network, is composed of three planes: management plane,
control plane and data plane, as shown in figure 7. The management plane consists of network
operations and control centers; the control plane consists of three GEO satellites that cover the globe;
the data plane consists primarily of satellite infrastructure and terminal routers. The management plane
and the control plane are connected through the SDN Northbound Interface (NBI), and the data plane
and the control plane are connected through the SDN Southbound Interface (CDPI). The control plane
has a global topology view, which is responsible for monitoring the network status and sending
dynamic information of the network to the management plane. The management plane (NPCC) can
provide functional modules such as routing policy, user management, virtualization, security, resource
utilization and network management. The management plane calculates a new flow table through the
global network information and sends it to the data plane through the control plane. The data plane only
needs to complete the forwarding of the data packet according to the flow table. Since the motion of the
satellite is periodic, we can predict the changes in the entire network using the corresponding
prediction-based algorithms (such as neural networks). By decoupling the data plane and control plane,
OpenSAN saves the resources of satellite nodes, overcomes the difficulty of satellite network closure
and expands, provides efficient and fine control, and supports the flexibility of future advanced network
technologies and services.

Fig. 7. The Architecture of OpenSAN[12] .

6
ISPECE IOP Publishing
IOP Conf. Series: Journal of Physics: Conf. Series 1187 (2019) 052067 doi:10.1088/1742-6596/1187/5/052067

Paper [13] proposes a Software Defined Satellite Network Architecture (SDSN), which has three layers:
the application layer, the control layer, and the infrastructure layer, as shown in figure 8. The
application layer is the Network Operations Control Center (NOCC), which has functions such as
routing policy calculation and application generation. The control layer is a ground station that
translates instructions from the upper NOCC into simple data structures and sends them to the
infrastructure layer. The infrastructure layer is made up of satellites, and the satellites only need to
implement the simplest forwarding and hardware configuration functions, which reduces the
development and design cost of the satellite. The SDSN distributes routing tables and configuration
commands through a single-layer inter-satellite link (ISL) and GEO satellites to implement a satellite
control model with a global network view. When the network update is small, single-layer ISL
forwarding is adopted; when the network update is large, GEO satellite broadcasting is adopted. This
new hybrid control structure enables real-time and fast network configuration, reducing the number of
ground stations and greatly improving the update speed of traditional satellite systems. The software-
defined network technology decouples the data plane and the control plane, and obtains a global
network view through centralized control, and deploys a more granular management strategy to
optimize the calculation of the global routing path.

Fig. 8. The SDSN architecture[13] .

2.7. SDN+ UAV network

In paper [14], an SDN-based UAV backbone network framework was designed. Figure 9 shows the
illustration of the SDN-based UAV network in this paper and figure 10 shows the functional
architecture of a UAV network. The control core includes a UAV controller and an SDN controller.
The UAV controller is responsible for managing information such as flight control, location, battery
storage, etc. The SDN controller is responsible for network information management. The UAV
controller and the SDN controller can interact to work together and share information to make optimal
decisions. Based on the two core issues of information management and resource management of
UAVs, an SDN controller monitoring platform is proposed, which includes four modules, namely
monitoring display, traffic management, link management and strategy. According to the analysis
results of the monitoring platform, a load balancing algorithm is further proposed in the policy module
to fully utilize the UAV network resources and maintain the desired network traffic balance.

7
ISPECE IOP Publishing
IOP Conf. Series: Journal of Physics: Conf. Series 1187 (2019) 052067 doi:10.1088/1742-6596/1187/5/052067

Fig. 9. Illustration of the SDN-based UAV network[14] .

Fig. 10. Functional architecture of the UAV network[14] .

3. CONCLUSIONS
Traditional networks have given the Internet a vibrant life but have also hindered the longer-term
development of the Internet. The new dynamic network architecture of SDN has successfully
transformed traditional networks into diverse application-oriented platforms. By decoupling the control
plane and data plane in the network architecture, the hardware cost of the network equipment can be
effectively reduced, so that the network can be flexibly large-scale programming and simplified
management like the computer infrastructure; the centralized control logic can make the network have a
global view information can be globally optimized and resource-adapted. The deployment and
maintenance of network nodes will be more agile; open and programmable network can promote more
business network service innovation. However, the current SDN technology is not fully mature,
including the standardization and unification of various interfaces, the compatibility of heterogeneous
networks, multi-controller coordination and some security issues. The expansion of SDN applications is
also facing more challenges. In this paper, we surveyed 7 different areas to analyze SDN-based
application scenarios. The "SDN+" architecture provides a new way of thinking for networks with
different applications, expanding the intelligence of the network and achieving maximum optimization.
The application of SDN is no longer limited to the campus network, the WAN network between data
centers, but has a broader space, not just the ground network, and even some good application ideas
have been made in space-based networks. In the future, we will focus on exploring more application
scenarios of "SDN+" and optimized routing algorithms that are more suitable for SDN architecture.

8
ISPECE IOP Publishing
IOP Conf. Series: Journal of Physics: Conf. Series 1187 (2019) 052067 doi:10.1088/1742-6596/1187/5/052067

REFERENCES
[1] Diego Kreutz ，Fernando M. V. Ramos, Paulo Esteves Verı´ssimo, et al, “Software-Defined
Networking:A Comprehensive Survey,” IEEE, vol. 103, issue 1 ,2015.
[2] H. Kim, N. Feamster, “Improving network management with software-defined networking,” IEEE,
vol. 51, issue 2, pp. 114–119,2013.
[3] B. Nunes, M. Mendonca, X.-N. Nguyen, et al, “A survey of software-defined networking: Past,
present, future of programmable networks,” IEEE, vol. 16, issue 3, pp.1617–1634,2014.
[4] M. Casado, N. Foster, and A. Guha, “Abstractions for software-defined networks,” ACM
Commun. vol. 57, issue 10, pp. 86–95, 2014.
[5] McKeown, Nick，Anderson, Tom，Balakrishnan, Hari Balakrishnan, et al, “OpenFlow:Enabling
Innovation in Campus Networks,” ACM SIGCOMM Computer Communication Review, vol.
8, issue 2, pp. 69-74 ,2008.
[6] Sushant Jain, Alok Kumar, Subhasree Mandal, et al, “B4: Experience with a globally-deployed
software defined wan,” ACM SIGCOMM,vol. 43,issue 4,pp. 3–14,2013.
[7] Li Erran Li ， Z. Morley Mao ， and Jennifer Rexford, “Toward Software-Defined Cellular
Networks，” IEEE Computer Society, pp. 7-12, 2012.
[8] Muhammad Usman, Anteneh A. Gebremariam, Fabrizio Granelli, and Dzmitry Kliazovich,
“Software-Defined Mobile Cloud: Architecture, Services and Use Cases,” IEEE CAMAD,
2015.
[9] Ian Ku, You Lu, and Mario Gerla, “Towards Software-Defined VANET: Architecture and
Services,” in Proc. 13th Annual Mediterranean Ad Hoc Networking Workshop, pp. 103-110,
2014.
[10] C Chen ， SS Xie ， XX Zhang, ZY Ren “A new space and terrestrial integrated network
architecture aggregated SDN,” Journal of China Academy of Electronics and Information
Technology, vol. 10, issue 5, 2015.
[11] FM Xu, ZJ Tong, CL Zhao, ZC Qin, “Architecture, Technology and Challenges of Software
Defined Integrated Space and Terrestrial Network” ZTE TECHNOLOGY JOURNAL, Vol.
24, issue 2, 2018.
[12] Jinzhen Bao, Baokang Zhao, Wanrong Yu,et al, “OpenSAN: A Software-defined Satellite
Network Architecture,” Acm Sigcomm Computer Communication Review, 2014.
[13] Zhu Tang ; Baokang Zhao ; Wanrong Yu, et al, “Software Defined Satellite Networks: Benefits
and Challenges,” IEEE Computers, Communications and IT Applications Conference, 2014.
[14] Xiao Zhang, Haijun Wang and Haitao Zhao, “An SDN Framework for UAV Backbone Network
towards Knowledge Centric Networking,” IEEE Conference on Computer Communications
Workshops, 2018.
[15] P. Dely, J. Vestin, A. Kassler, N. Bayer, H. Einsiedler, and C. Peylo, “CloudMAC-An OpenFlow
based architecture for 802.11 MAC layer processing in the cloud,” in Proc. IEEE Globecom
Workshops, pp. 186-191, 2012.
[16] Y. Zaki, L. Zhao, C. Goerg, and A. Timm-Giel, “A novel LTE wireless virtualization framework,”
in Proc. Int. Conf. Mobile Netw. Manage, pp. 245-257,2010.
[17] K. Pentikousis, Y. Wang, and W. Hu, “MobileFlow: Toward software defined mobile networks,”
IEEE Commun. Mag., vol. 51, issue 7, pp. 44-53, 2013.
[18] N. Feamster, J. Rexford, and E. Zegura, “The road to SDN,'' Queue, vol. 11, issue 12, 2013.
[19] Y. Li and M. Chen, “Software-defined network function virtualization: A survey,” IEEE Access,
vol. 3, pp. 2542-2553, 2015.
[20] B. Raghavan et al., ‘‘Software-defined internet architecture: Decoupling architecture from
infrastructure,’’ in Proc. 11th ACM Workshop Hot Topics Netw., pp. 43–48, 2012.
[21] Miao Y, Cheng Z, Li W, et al., “Software Defined Integrated Satellite-Terrestrial Network: A
Survey.” International Conference on Space Information Network. Springer, Singapore,
2016.