This article has been accepted for publication in a future issue of this journal, but has not been

fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/JIOT.2015.2480421, IEEE Internet of Things Journal
IEEE INTERNET OF THINGS JOURNAL 1

Software Defined Wireless Networking Opportunities

and Challenges for Internet of Things: A Review
Keshav Sood, Student Member, IEEE, Shui Yu, Senior Member, IEEE and Yong Xiang, Senior Member, IEEE

Abstract—With the emergence of Internet of Things (IoT), there

is now growing interest to simplify wireless network controls. This
is a very challenging task, comprising information acquisition,
Software Defined Wireless
information analysis, decision making and action implementation Networking is a solution !
on large scale IoT networks. Resulting in research to explore Unlimited data transmission

the integration of Software Defined Networking (SDN) and IoT

for a simpler, easier, and strain less network control. SDN is a Network tools to Collect, Store,
promising novel paradigm shift which has the capability to enable Internet of Things Process, and Forward massive data
are missing.
a simplified and robust programmable wireless network serving
an array of physical objects and applications. This review article
starts with the emergence of SDN and then highlights recent Fig. 1. IoT networks and SDN.
significant developments in the wireless and optical domains with
the aim of integrating SDN and IoT. Challenges in SDN and IoT
integration are also discussed from both security and scalability To address this need of the network, to support the onslaught
perspectives. of connected zillions of devices and to remain competitive,
Keywords—Internet of Things, SDN, Software Defined Wireless service providers are required to look into other alternatives
Networks (SDWN), SDN Use Case. such as Software Defined Networking (SDN), to increase
their bandwidth and reinforce their networks [4]. SDN’s wide
acceptance from industry ensures SDNs ability to develop a
I. I NTRODUCTION tighter connection within the ecosystem of IoT that provides
cyberspace to every object. We also emphasise that SDN and
M OBILE carrier networks are approaching a tipping
point. Emerging mega trends in the ICT domain has
reached a significant level of integrating the internet into every
IoT are evolving parallel, intersecting and perhaps dependent
on each other. Figure 1 reflects the need to simplify the
object in a network. With the evolution of the Internet of network control mechanisms in IoT.
Things (IoT), mobile networks will handle an influx in big Significant benefits of integrating SDN and IoT include:
data, massive network traffic, and new types of connected a) SDN has a potential to intelligently route traffic and
devices including industrial machines, thermostats, sensors, ac- use underutilized network resources. This will significantly
tuators, smart cars, wearables, and smart appliances. They will enhance network’s ability and therefore it will be much easier
share the same network with PCs, tablets, and smartphones, for networks to prepare for the data onslaught of IoT. This
which are already bandwidth sensitive. Presently there are 9 will eliminate bottlenecks to efficiently process the data gen-
billion connected devices and the number is expected to rise erated by IoT without placing a large strain on the network,
to 24 billion by 2020 [1]. With such significant involvement of especially on Wi-Fi network. b) SDN integration with IoT
connected devices, carriers are already experiencing complex will simplify the information acquisition, information analysis,
control on elements and overloaded networks [2] [3]. If the decision making, and action implementation process. c) The
networks are not prepared, this flood of Internet-of-things, deployment of SDN in IoT will provide visibility of the
where the things are producers of traffic, not just consumers network resources and management of access based on user,
in the network, could leave the network paralysed [1]. Fur- group, device and application that eventually enables the ability
thermore, IoT devices are getting wirelessly connected to to exchange data capacity between users and even devices.
the Internet serving diversity of applications where no single d) Researchers are designing intelligent algorithms in SDN
wireless standard can adequately prevail. In this case choosing to build effective traffic pattern analyser, which simplifies the
the right wireless connectivity and to form a potential control tools of data collection from IoT devices. This facilitates the
on IoT wireless device is another challenging task, while the design of novel debugging tools. IoT networks will benefit
traditional network is insufficient to meet this challenge. with the integration of Software Defined Wireless Networking
(SDWN) technology to strengthen networks controlling ability.
Manuscript received March 30, 2015; revised August 27, 2015. e) With SDWN, IoT networks can become more agile and
Keshav Sood, Shui Yu and Yong Xiang are with School of Information scalable based on demand. In Figure 2, we provide a typical
Technology, Deakin University, VIC 3125, Melbourne, Australia. Email: high level view of integrated SDN and IoT architecture. SDN
{ksood, syu, yxiang} @deakin.edu.au.
Copyright (c) 2012 IEEE. Personal use of this material is permitted.
deployment in wireless segment is known as SDWN, therefore
However, permission to use this material for any other purposes must be the meaning of SDN and SDWN in this paper is same.
obtained from the IEEE by sending a request to [email protected]. Efforts have been made to investigate SDWN in the context

2327-4662 (c) 2015 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See
http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/JIOT.2015.2480421, IEEE Internet of Things Journal
IEEE INTERNET OF THINGS JOURNAL 2

IoT Device IoT Controller IoT Device Carrier Grades SDN, Data
Operator Centre, Migration, Enterprise

Information Modelling,L4-7
Services,
Open Networking Services
SDN Controller Networking Northbound Interfaces,
started in 1960s Foundation
Architecture and
established in 2011
Framework

OpenFlow (2008) Extensibility, Forwarding

MPLS (1990)
Abstraction, OF-Conf.,
Specification Optical Transport, Protocol
SDN Enabled DCAN & Open
Ethane (2006) Independent Forwarding,
Testing and Interoperability,
SDN Enabled Heterogeneous Signalling (1997) Wireless and Mobile
Networks 4D (2004)
Heterogeneous SDN Enabled Liaisons, Proof of Concept,
Publications,
Networks Heterogeneous ForCES Market SDN Solution Showcase, Skill
Networks (2003) Certification,
Workshops

Fig. 2. Typical architecture of integration of SDN and IoT, a high level view. Fig. 3. Early efforts of SDN and major technical communities in ONF.

of infrastructure based SDN enabled Wi-Fi networks such

as OpenRoad [5], Odin [6], OpenRadio [7], OpenRAN [8],
SoftRAN [9], CellSDN [10], SoftMoW [11] and so on. The functions moved from software to hardware such as ASICs,
field of SDWN is in its infancy stage and there are still many Field Programmable Gate Array (FPGA) and Ternary Content
important challenges to be addressed in order to control IoT Addressable Memory (TCAM). As time passed networking
network with a unified protocol. Therefore in this review paper, of devices has become increasingly complex. This is due
we aim to provide the recent developments of SDWN that can in part to the existing independent and autonomous design
bring lot of research opportunities in IoT. of devices that makes it necessary for so much intelligence
The remainder of the paper is organized as follow. The being placed inside each device. This made the functional-
early efforts of SDN is presented in Section II and SDWN ity in some ways very simple but made the device more
opportunities in IoT are detailed described in Section III. In complicated because of the difficult handshake and trade-offs
Section IV the SDWN open research challenges are elaborated between handling packets in hardware versus software [10].
mainly from the prospectives of security and scalability. In With time, researchers attempted to move control off the device
Section V, we conclude the paper. placed into centralized controller that is having a full network
view and the ability to make optimal forwarding and routing
decisions [14]. Control software means the intelligence that
II. S OFTWARE DEFINED NETWORKING : E ARLY EFFORTS
determines optimal paths and responds to outages and new
Immigrant Paul Baran, a researcher working at Rand Cor- networking demands. Forwarding responsibilities implemented
poration US in 1960s proposed to transmit the voice signals of in hardware tables, filtering based on ACLs and traffic prioriti-
phone in form of packet data that could travel autonomously zation are enforced locally on device remain on the device [13].
through the network [12]. To further increase the packet for- The forwarding table on hardware device is available to
warding intelligence for different reasons such as developing be programmed by external software controller. Above the
fine grained traffic forwarding decision to save bandwidth and controller the network application runs, implementing higher
increase network performance, Policy Based Routing (PBR) level functions, involving to make decisions to best manage
methods were proposed [13]. At that stage a new term Flow the traffic and network.
was generated to describe particular set of traffic between
two end points that receive the same forwarding treatment. There is a steady progression of solutions and ideas
PBR defines a set of criteria (commonly known as match- around advancing networking technology prior to OpenFlow.
action criteria in SDN) that determines whether an incoming The early efforts include, MPLS (1990) to separate control
packet corresponds to a particular flow or not. This was software, establishing semi-static forwarding paths for flows
a centralized approach of programming forwards rules that in traditional routers, Devolved Control of ATM Network
has provided further under piping for SDN technology. In (DCAN) to separate control and forwarding plane in ATM
this regard we can incorporate PBR at the ground level of Switches (1997) and Open Signaling (1997) began with
SDN. Figure 3 illustrates the early efforts of SDN and major ATM Switches [13]. Forward and Control Element separation
Open Networking Foundation’s (ONF) technical communities (ForCES) (2003), 4D named after four plane decisions (2004)
(Operator, Services, Specifications, and Market) responsible and Ethane (2006) are all known as precursors of SDN [13].
for various tasks. Although all these solutions adequately and automatically re-
From its birth, SDN has been based on the notion of configure the edge network, the static and manually configured
constructing forwarding tables defining actions to take on core of the network remains the same [15]. The long awaited
flow rather than having forwarding tables merely map des- solution of this problem is now available in the form of
tination address to output port [14]. Over time networking OpenFlow (2008) [16].

2327-4662 (c) 2015 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See
http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/JIOT.2015.2480421, IEEE Internet of Things Journal
IEEE INTERNET OF THINGS JOURNAL 3

TABLE I. SDN S IMULATION T OOLS

1. First packet arrives, usually every first
packet misses the entry.
Tool/Reference Methodolgy
2. Flow forwarded to controller to get
action to be taken on particular flow.
Mininet [18] Using lightweight OS containers to emulating hosts and
3. Controller installs new flow entry.
switches in a network.
4. Flow path.
SDN Centralized
W3 [19] A gdb-like debugger for OpenFlow based networks.
Controller

3
FatTire [20] Provides complier target fault tolerance requirement.
3
fs-sdn [21] Enabling direct use of OpenFlow controller components.
3
3

1 4 4 4
Laptop Lantza et al. designed Mininet to support collaborative
Laptop
OpenFlow Switch OpenFlow Switch OpenFlow Switch network research [18], which is a flexible, deployable, inter-
active, scalable, realistic and at the most share-able software
Fig. 4. SDN OpenFlow switching. simulation tool. Virtual Machines (VMs) can be created to test
desired network behaviour with OpenFlow. This is a python
based open source tool. At the same time, it is very critical in
A. OpenFlow and Open Networking Foundation SDN to detect and debug the failures in large scale network.
OpenFlow was developed by researchers to begin experi- Scott et al. argued that concise and specific policies and more
ments and innovate with new protocols every day. The spec- sophisticated tools are required for the testing of SDN large
ifications of OpenFlow encourage vendors to implement and scale network [19]. Because SDN software stacks itself in a
enable OpenFlow in switches and other products. OpenFlow complex distributed system and the working of SDN is, in a
protocol delineates to use it between the controller and switch, challenging, synchronous and failure prone environment. They
hence a unified control protocol comes into existence. The developed a tool called W3 to troubleshoot bugs in SDN con-
basic switching operation of OpenFlow, as shown in Figure trol software. W3 stands for “What network problem exist?”
4, is a switch evaluates every incoming flow independently, “Where the problem arises first in the software?” and “When
finds a matching flow against it, and performs the associated the triggering event happened?” Corresponding checking and
action. If no match is found, the switch forwards packet to Simulation based causal interference are the two approaches
controller for getting instructions on how to deal with packet. designed in W3. This tool successfully tested on various SDN
The SDN controller populates the switch with flow table platforms but still needs many rounds of improvement to fully
entries. Typically controller updates switches with new flow reap its benefits in large scale enterprise domain. Reitblatt
entries as new flow patterns are received. Wild card rules are et al. [20] developed a declarative language “FatTire” for
also accepted. This technology is known as Software Defined developers to express fault tolerance requirements and provide
Networking. complier that targets SDN fast-failover mechanism. Further,
Before 2011 the OpenFlow standards and protocol versions fs-sdn has been recently designed by Mukta et al. [21] that
had been designed by Stanford University. Open Networking offers simulation at large scale in comparison to Mininet.
Foundation [17], a new autonomous body was formed in 2011
by companies including Google, Facebook, Cisco system and III. S OFTWARE DEFINED NETWORKING OPPORTUNITIES
Microsoft to design the standards for OpenFlow. Various re- IN I OT
search groups are working towards designing SDN OpenFlow In this section, only key research efforts of SDWN over the
specification, configuration, management protocols and so on. period between 2008 and 2014 are discussed.
The Extensibility working group focuses on SDN deployment
in wireless (IEEE 802. XX) and telecom sector. It is important
A. SDN in Wireless Networks
to note that ONF has introduced, 64 OpenFlow products to
market, and more than 20 members have demonstrated inter- Researchers in [16] developed OpenFlow protocol to run
operability of OpenFlow standards till date. In the following experiments in a uniform way in real time running network,
sub-section we will shed some light on various open source aiming to develop a new switch feature that can potentially
tools simulating the ideas of SDN. extend programmability into campus network. Further [22] has
proved the conceptual idea of OpenFlow by implementing a
test bed with OpenFlow at Stanford University Campus that
B. TestBed and Troubleshooting Tools allows heterogeneous network experiments to be concurrently
Table I highlights the most common SDN simulation tools conducted in production environment so that multiple networks
used by researchers nowadays. But till 2010, to choose appro- will act as a single network. In [5] OpenRoad is embedded
priate prototyping tool to simulate ideas had been a challenge into mobile/cellular networks with a vision to attract virtual
because real test bed was expensive and out of reach from service providers to independently allow seamless handover
most scientists. between different wireless technologies. Also, the authors

2327-4662 (c) 2015 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See
http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/JIOT.2015.2480421, IEEE Internet of Things Journal
IEEE INTERNET OF THINGS JOURNAL 4

of [5] extended SDN to Wi-Max services and described the analyse the performance of distributed SDN enabled IoT net-
brief idea of SDN and virtualization in carrier domain. A major works. Such an effort, to deploy distributed architectures, was
breakthrough in SDN came into existence when Jian et al. [23] evaluated in [29]. Authors significantly analysed Quality of
from Google presented a design, implementation and evalua- Service (QoS) of streaming applications. Extensive simulations
tion of the B4 project, a private WAN connecting Google’s were done to prove that distributed architectures have much
data centers across globe. The authors of [23] described the more advantages than centralized SDN. We emphasize that
three years of B4 experience in production domain, lessons the path inflation factor must be investigated before deploying
learned and significant future research areas in their tutorial SDN IoT distributed topology to avoid latency.
paper. Ruckert et al. [30] pointed out that current broadband net-
Suresh et al. developed Odin to introduce programmability work architecture uses tunnelling concept to subscribe traffic
in WLAN networks [6]. With Odin the association of client through a single aggregation point irrespective to different
with Access Point (AP) can be made by SDN controller. Odin types of services. This implies huge bandwidth requirements
introduces Light Weight Access Points (LVAP) that runs over and high end-to-end latency. A proof-of-concept approach
controller to make decision of association and disassociation was discussed to show the feasibility of the proposed novel
of client with AP while ensuring seamless handover. The lim- SDN based flexible traffic management architecture. Feng et
itations of this work are: a) The association and disassociation al. [31] proposed a price-based joint allocation model to fairly
processes do not count the load on APs, b) Odin runs over allocate the bandwidth and flow table space. In its analysis,
single controller hence it will be difficult to prioritise the the maximum mean forwarding rate and minimum mean delay
multiple applications running over single controller. In [24] time is calculated. The flow volume at each port is calculated
Odin’s work was extended with the proposition of a novel and according to that volume of flow, bandwidth is allocated
load aware hand off scheduling algorithm running over SDN to flow. Authors proved that the mean delay of SDN network
centralized controller for SDN based WLAN networks. In [25] is minimum if the mean delay of each OpenFlow switch is
the impact of OpenFlow SDN over wireless networks was minimum [31].
evaluated, particularly in terms of end-to-end delay, throughput These research efforts are also highlighted in Table II. All
and jitter. these efforts provide opportunities to design novel architec-
Costanzo et al. [26] analysed SDN in IEEE 802.15.4 tures in IoT control by SDN. The controller sense the user
networks (Low Rate-Wireless Personal Area Networks, LR- application type and allocate the node or switch according to
WPANs). They elaborated the opportunities of SDN and bandwidth requirement. The SDN controller has full network
requirements to implement SDN with different scenarios in wide view and usage of each element/node and thus it can
this network domain, i.e., Internet of Things domain. Authors easily and efficiently distribute the bulk of data to different
argued that requirements alike SDWN must support duty cycle, nodes. SDN also provides an opportunity to virtualize the IoT
provides flexible definitions of rules, and effectively track the networks, as demonstrated by Lee et al. [32]. They designed
node mobility in network data aggregation where sensor nodes meSDN architecture to demonstrate WLAN virtualization with
transmit and receive tremendous amount of data in IoT sector. SDN.
[27] showed an SDN based system architecture in WLAN
tested for fast handover to improve streaming video. The
controller and streaming server are connected to OpenFlow B. SDN Enabled Hybrid Architecture
Virtual Switch (OVS) which is further connected to APs.
Controller updates forwarding table of OVS to route traffic SDN can also bring new opportunities in IoT by designing
from streaming server to the station to connect several APs hybrid network architecture. It has the potential to control
simultaneously. At this point, SDN appears as a viable alter- circuit and packet switching by a unified control protocol.
native integrated architecture that facilitates the possibilities Gudla et al. presented a unified control architecture with
of creating novel IoT SDN integrated services, architectures, OpenFlow to dynamically control packet and circuit switch-
data-driven protocols and more efficient applications to cover ing networks [33]. Latency and link-up time is reduced
the actual requirements of programmable network. in comparison to traditional networking methodologies. The
In [28] a similar approach to [27], i.e., Media Independent OpenFlow switch functionality was implemented in NetFPGA.
Handover mechanism using OpenFlow was investigated. In An independent testing of OpenFlow in circuit and packet
this work, the mobile device initiates the signaling process switching was also conducted. The authors in [34], designed
to handover to another service point. The mobile device sends the OpenFlow enabled hybrid network architecture. They ex-
signals to SDN central controller which then sends information tended the OpenFlow table matching rule entry by introducing
about the best available controller and the best available service two additional tuples, i.e., Channel and Transport class for
point, back to mobile device for association and handover. wavelength/timeslot and guaranteed circuit/packet switching
From this work, it seems that novel load aware algorithms respectively. [35] has demonstrated the improved path set
can be proposed that may sense the load on wireless sensors. up times and control stability with SDN in optical transport
This may facilitates re-routing of the traffic to less loaded node technologies. From these research efforts, one can infer that
in coverage zone. A complete IoT network can be divided into novel hybrid architecture in IoT will have lesser effect on
zones and controlled by logically centralized controller. At this convergence time that is affected by the limited information
point, in order to maintain service quality, it is essential to of the node to recalculate the route.

2327-4662 (c) 2015 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See
http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/JIOT.2015.2480421, IEEE Internet of Things Journal
IEEE INTERNET OF THINGS JOURNAL 5

TABLE II. OVERVIEW OF K EY SDWN R ESEARCH BETWEEN 2008 AND 2014

Technological Domain
Wireless/Telecom:W,
Year Project Theme/Reference Overview
Optical: O,
Hybrid:H
2008 OpenFlow: Enabling Innovation in Campus Networks [16] Proposed a new feature in the switch with OpenFlow protocol. H

First test bed that allows heterogeneous network experiment to be

2009 The Stanford OpenRoad Deployment [22] concurrently conducted in production environment so that multiple W
networks will act as a single network.
2010 OpenRoad: Empowering Research in Mobile Networks [5] Extended SDN in mobile networks. W

2010 Experimental Demonstration of OpenFlow Control of Packet H

Proposed unified control architecture in circuit and packet switching.
and Circuit Switching [33]

2010 A Network in a Laptop [18] Designed simulation software Mininet for SDN application testing. H

2011 Open Networking Foundation [17] A new body to design and standardise SDN protocols. –

2012 W3, a troubleshooting tool in SDN [19] A troubleshooting tool to detect bugs in SDN networks. W

2012 Base stations can be remotely programmable to enable operators and W

OpenRadio: A Programmable Wireless Data plane [7]
vendors to upgrade and optimise the network more easily.
Introduce Light Weight Access Points (LVAP) that runs over con-
2012 Towards Programmable Enterprise WLANs with ODIN [6] troller to make decision of association and disassociation of client W
with AP while ensuring seamless handover.
Elaborate the opportunities of SDN and requirements to implement
2012 Software Defined Wireless Networks: Unbridling SDNs [26] SDN with different scenarios in IEEE 802.15.4 networks (low rate- W
wireless personal area networks, LR-WPANs).
2012 Proposed extensions to controller platform, switches and base sta- W
Towards Software Defined Cellular Networks [36]
tions.

2012 A design of a Software Defined Middle Boxes networking framework W

Towards Software Defined Middle-Box Networking [37]
capable of supporting future scenario.

2012 Deconstructed the scalability concerns in SDN networking and H

On Scalability of Software Defined Networks [44]
argued that they are not unique to SDN.

2012 A SDN Approach for Handover Management with Real-time W

Demonstrating OpenFlow for improving streaming video.
Videos in WLANs [27]

2012 Empowering SDWN through Media Independent Handover Media Independent handover mechanism was demonstrated using W
Management [28] OpenFlow.
Extension of the OpenFlow table matching rule entry by introduc-
2013 Design and Test of a Software Defined Hybrid Network Archi- W
ing two more tuples, i.e., Channel and Transport class for wave-
tecture [34]
length/timeslot and guaranteed circuit/packet switching respectively.
2013 Demonstrated improved path set up times and control stability when O
SDN Optical Network Technology and Infrastructure [35]
SDN is implemented directly in optical transport technologies.

2013 SDN Networks for Telecom Operators: Architecture and Appli- The aggregate and core layer of wireless backhaul can also be W
cations [38] virtualized depending on the SDN controller’s ability.

2013 Designed Open and Flexible network management system where user W
OpenRAN [8]
is free to join the strongest network nearby.

2013 Abstracts all base stations in a local geographical area as a virtual W

SoftRAN [9]
big base station comprised controller and radio appliances.

2013 Enables data migration from native layer to overlay layers and thus W
Applying SDN to Telecom Domain [39]
requires encapsulation and de-encapsulation.
A declarative language for developers to express fault tolerance
2013 FatTire [20] requirements and provide complier that targets SDN fast-failover H
mechanism.
2014 Modelling and Evaluation of SDN Scalability [51] Mathematical tools for evaluating the scalability of SDN network. W, O and wired

2014 A proof-of-concept approach that can show the feasibility of pro- W

Traffic Management in Broadband Access Networks [30]
posed novel SDN based flexible traffic management architecture.

2014 Fair Network Resources Allocation and Scheduling [31] The flow volume at each port is calculated for bandwidth calculation. W

2014 Distributed QoS architecture for Multimedia Streaming with Proved that distributed architectures have much more advantage than W
SDN [29] centralized SDN.

2014 Load aware hand off scheduling algorithm for SDN based WLAN W
Extension of ODIN [24]
network is proposed.

2014 Evaluated the impact of OpenFlow SDN over wireless networks W

Performance Analysis of SDN over Wireless Networks [25]
particularly in terms of end-to-end delay, throughput and jitter.
An meSDN architecture that demonstrates that SDN control on
2014 SDN Mobile extension [32] mobile client enables WLAN virtualization and application aware W
QoS improves power efficiency.
2014 Application Aware Data Plane Processing [50] Implemented apps in the OVS switches for Stateful switch actions. W, O and wired

2327-4662 (c) 2015 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See
http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/JIOT.2015.2480421, IEEE Internet of Things Journal
IEEE INTERNET OF THINGS JOURNAL 6

C. Other Relevant Research Work of SDWN have recently provided effective general media aware security
architecture that jointly considers the security services and
Bansal et al. [17] presented a key conceptual novel design multimedia traffic characteristics in the IoT context.
for programmable wireless data plane, OpenRadio, to provide Figure 5 shows the web search popularity, as measured by
modular and declarative programming interface within wireless the Google search trends during the last 10 years for the terms
stack. Network operators are always in need to dynamically Internet of Things, Wireless Sensor Networks and Ubiquitous
adjust spectrum and power at base stations according to traffic Computing. This evidence clearly supports our vision inspires
requirements. According to the proposed design, base stations the need of SDN and IoT integration. This trend motivate
can be remotely programmable to enable operators and vendors researchers to deploy novel and simplified network control
to upgrade and optimize the network more easily. An operators mechanism in IoT. Recent research presented above inspires
are able to define and set a protocol based on matching scientists to innovate novel SDN enabled architectures in IoT
subsets of traffic streams and then can specify actions on domain and eventually beat the challenges caused the rapid
them. Bansal et al. [17] defended the feasibility of this idea by growth of things in wireless cyberspace.
arguing that PHY and MAC layers are shared across different
protocols (different versions of LTE, 4G, 3G and Wi-Fi).
In [36] it is argued that today’s cellular networks do not have
fine-grained control over routing and altering traffic to direct
destination. Middle boxes, (i.e., network appliances deployed
by different vendors) are also problematic causing serious
performance issues in carrier domain. The authors of [36]
proposed extensions to controller platform, switches and base
stations. They have made some changes in carrier domain
to implement SDN architecture in SDWN. The argument
made in [37] is that current Middle Boxes (MB) (such as
Firewall, Load balancer, Intrusion prevention etc.) are clumsy Fig. 5. Google search trends since 2004 for terms Internet of Things, Wireless
and unsuitable to handle future networks. A Software Define Sensor Networks, Ubiquitous Computing. [1]
Middle Boxes networking framework capable of supporting
future scenarios was proposed in [37].
Currently we are forced to associate with network even if D. Testing Novel OpenFlow Enabled IoT Products and Design
the network performs quite poorly and even if we have other The rapid growth of devices and online services for infor-
more appropriate networks around. This is a closed and non- mation transaction over the network consolidated the concept
flexible approach. In [8] a conceptual overview was presented of IoT. The rigidity of traditional architecture is inefficient
for the design of Open and Flexible network management in this concept, suggesting rethinking new ways to use the
system known as OpenRAN. The term Open indicates that infrastructure and technology. Software Defined Networking
users are free to join the strongest network nearby. This term provides an alternative to the current problems of traditional
here does not indicate the Open from OpenFlow, rather the idea networks. It gives potential opportunities to allow administrator
is built on OpenFlow SDN technology. Further [9] proposed to have a global view of the network, as well as to control
SoftRAN which was the extension of OpenRoad. Authors in the network according to the need of individual organiza-
the paper has provided a wireless programming interface in tion and its users. Organizations are looking for deployment-
RAN network called SoftRAN that abstracts all base stations in ready solutions or novel IoT products running with IPv6 for
a local geographical area as a virtual big base station comprised their current networking environment. They are also looking
controller and radio appliances. for future compatibility built on the OpenFlow specification.
SDN research in telecom sector is also gaining momentum. Open Networking Foundation is a user-driven organization
Wang et al. [38] pointed out that the aggregation and core dedicated to the promotion and adoption of Software Defined
layer of wireless backhaul can be virtualized depending on Networking through open standard development. The Wireless
the SDN controller’s ability. Whereas [39] argues to enable and Mobile project in ONF collects use cases and determines
data migration from native layer to overlay layers thus requires architectural and protocol requirements. The aim of the project
encapsulation and de-encapsulation. is to extend ONF based technologies to carrier networks such
Researchers also argue that conventional network manage- as backhaul network, cellular Evolved Packet Core (EPC) and
ment algorithms (especially multimedia traffic management so on. Further they aim to provide a unified access and manage-
algorithms) are inappropriate for security-aware multimedia ment across enterprise wireless and fixed networks including
applications [41]. For a security-critical multimedia service IoT domain. The ONF OpenFlow Conformance certification
architecture in the IoT context, there are various other chal- is the highest level of assurance available in the market by
lenges including traffic classification and analysis for various ONF today to validate product conformance. For this ONF has
multimedia applications streaming over IoT, developing novel approved six labs till 2014 to test the novel design applications
architectures for media-aware traffic security, designing and and products etc. [17]. This highly encourages researchers to
evaluating the proposed security-critical traffic management design and test the novel SDN enabled IoT products and ideas
scheme, etc. To alleviate these challenges, the authors of [42] on live and real platform.

2327-4662 (c) 2015 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See
http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/JIOT.2015.2480421, IEEE Internet of Things Journal
IEEE INTERNET OF THINGS JOURNAL 7

TABLE III. D ECONSTRUCTING SDN S ECURITY I SSUES

Selfish Attacks Flow Table
Generate More Flows Overloading
and Consumes Larger
Security Challenges
Mounts DDoS
Bandwidth SDN Switch can be
Attacks
Compromised and To prevent an authenticated application from being hacked or authenticate an
Behave like Bot application’s access to the control plane.
How to mitigate the DDoS threat to prevent server, controller or switch to overload?
To test novel designs in order to evaluate clients not nodes,and evaluate payload not
packet.
SDN Network is SDN Security Advantages
Unable to Mitigate
and Detect
High Volume of Aggregated Flows OpenFlow Access Point Advanced
SDN allows for the decoupling of the control plane from the data plane. If hackers
towards OpenFlow Access Point
Persistent Threats were to reach the data plane in an SDN ecosystem, they would be unable to use
the data because the controls would no longer be embedded.
Match-Action Criteria The distributed protocols, which are more resilient and harder to attack because they
Failed to Revel are not concentrated.
Matching Tuple to
SDN is capable of automatically quarantining an endpoint or network that has been
Invoke Deep Packet
Inspection
infected with malware.
Red arrows indicate selfish insider
attacks or malicious flows Security Application in SDN [43]
Green arrows indicate legitimate flows
DDoS mitigation applications (Defence flow, Defence Pro, APSolute vision), aiming
to mitigate L2, L3 and L7 attacks.
SE-Floodlight (providing integration of a security mediation kernel into the
Fig. 6. Security concerns in SDN network. BigSwitch Floodlight OpenFlow controller) to avoid controller hacking. Additionally
it provides role-based authorization and strong security constraints enforcement.
SDN Security Actuator, a middle-ware abstraction service that provides flexibility
to integrate legacy INFOSEC security products and technology into an OpenFlow
IV. O PEN RESEARCH CHALLENGES network stack. It also enables security services to communicate high level threat
response directives, which are further translated into Stateful OpenFlow flow rule
A. Secuirty of SDN and IoT Network insertions to be sent to SE-Floodlight.
OF-BotHunter, a sample OpenFlow Security application that provides interfaces
The reason that security is always a major issue in cy- with OpenFlow network stack via the SDN Security Actuator. Worm propagation,
berspace is that measures are often considered only af- an application to detect malware for mobile devices in SDN Stack.
ter launching a new technology. The IoT network, i.e, a BlueCat DNS director, an application by HP to deliver network-driven enforcement
of DNS policies which allow security infrastructures to gain complete visibility and
network of physical objects is more sensitive to security, control with IP applications.
contains embedded system to communication machine to Ecode evolve, an SDN orchestrator to facilitate dynamic service provisioning with
machine. In IoT there are potential risks to network such built-in QoS and DDoS mitigation.
F5 BIG DDoS Umbrella, a network application that allows network customers to
as; AES public/private key exchange methods, protecting DNS and SSL DDoS protection at the network edge that is closer to the attacker.
attached TPC/IP networks from intrusion through your de- GuardiCore Defense Suite for software-defined data centers, detecting and mitigat-
vice and protecting pre-shared keys from reverse engineer- ing advanced persistent threats, malware propagation and insider attacks.
KEMP Adaptive Load Balancer application, providing end-to-end visibility of
ing through an MCU debugger. (http://www.link-labs.com/ network path for optimal routing of applications.
internet-of-things-securitychallenges). Figure 6 represents ba- Real Status Hyperglance, providing simplified context aware hybrid cloud and SDN
management.
sic security concerns in SDN domain.
SDN and IoT integration will doubtlessly simplify the
network control using common protocol in every technological
domain, but SDN also poses some risks. For example, in SDN switch can be compromised and can serve as a bot (a compro-
logically centralized controller controls the switch which eval- mised host used to perform malicious task). In real world,
uates every incoming packet based on match-action criteria, attackers can use compromised hosts, such as a botnet, to
as shown in Figure 4. In order to temporarily store match- start DDoS attack. If the switch is compromised by botmaster,
action rules to take decision on incoming packet or flow, each then effective mechanisms viable in network to detect the
switch possesses a flow table. Unfortunately these flow tables Advanced Persistent Threats (APT) are missing in the current
are implemented using expensive and power hungry TCAM. SDN technology. In this case a SDN switch may become a
As a result, the flow table size is limited. Usually the flow bot and serves as a bridge to disrupt the network. Similar
table size cannot scale beyond few hundreds entries or rules. things can happen in IoT network which is large in scale
Therefore SDN switch can only handle limited number of and composed of millions of devices. Furthermore, enhancing
flows per time [40]. This limited flow table size is a potential the controller’s intelligence software may increase controller
weakness of SDN and is vulnerable to attackers. High volume vulnerability to hackers and attack surfaces. If attackers have
of traffic can very easily consume the table capacity and thus access to the controller, they can damage every aspect of the
the switch is overloaded. Continuing high volume of traffic network and eventually knock down the whole network.
(flow) may make switch disabled or knock it down. As a SDN security risks come out because of the absence of inte-
result, the later arriving packets may all be dropped, failing gration with existing security technologies and SDN’s inability
to be forwarded. Therefore attackers can easily knock down to poke around every packet. Therefore it is essential that a
the switch and thus disable the network services by mounting packet has to undergo a deep inspection for risk assessment
Distributed Denial of Service (DDoS) Attack. before routing to certain levels.
To hamper network security to extreme level, the OpenFlow SDN security requires to support the authentication and

2327-4662 (c) 2015 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See
http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/JIOT.2015.2480421, IEEE Internet of Things Journal
IEEE INTERNET OF THINGS JOURNAL 8

authorization classes of the network administrators at every

plane but it may prevent the access to flow management
policies. Thus SDN must construct novel security mechanism,
different from the traditional ones. Although it is still at Factors Constructed Scalability
Concerns: SDN Controller
an early stage in the context of the Internet of Things and
1. Link bandwidth between end node and
Software Defined Networks, it is clear that change is afoot. In edge device i.e. user and router/AP
SDN, research is gaining momentum to secure both the control respectively.
2. Link bandwidth between edge device
and data plane [15]. Table III highlights the research progress and controller such as router, AP etc.
3. Type and length of flow/s.
that deconstructs various SDN security issues. 4. Time taken by flow to remain in
We argue that SDN integrity with IoT will simplify the wireless shared medium.
5. Switch/s flow table capacity.
information process, information acquisition, information anal- 6. Network topology. Router Wireless
7. Path inflation factor if the network Access
ysis, decision making, and action implementation. This will controllers are physically distributed Point
ease the IoT network management. Secure flow automation and logically connected.
8. Controllers computational capacity.
needs central management of data forwarding which is very
complex without SDN. As the IoT network diameter grows,
much more simplified and secure mechanisms are required.
We also argue that as organizations depend more on machine-
generated data for real-time business processes, it is essential to
ensure the trust of data. Strong detection mechanisms to detect
rogue devices trying to interact with the IoT infrastructure both Fig. 7. Factors constructed scalability concerns in SDN network
from device and network vectors must be deployed. As the
IoT connects many devices together, it provides many decen-
tralized entry points for malwares. Cheap devices or things consumed by flow length propagated in air that further depends
in physically compromised locales are fragile to tampering. on the type of traffic. Mobile traffic (mostly UDP) is inelastic
Whereas SDN’s centralized approach is significantly better and insensitive to bandwidth such as browsing youtube videos,
than traditional networks. Moreover SDN has full potential online gaming, if the flow size is large then it will affect
to globally view the traffic patterns, mobility of nodes, and the other legitimate users to join the network. Larger the
change in traffic volumes. Therefore security policies can be flow length, more is the time taken by the particular flow in
easily implemented in SDN enabled IoT networks. air which eventually increases the collision rate, delay and
To conclude, based on SDN’s potential to classify/slice jitter in WLAN networks. Therefore it is necessary to taken
traffic and to attain a full global view on it, we emphasize into account the type and length of flow while constructing
that SDN integration with IoT can bring novel and simplified scalability concerns.
ways to deal with such critical issues.
TABLE IV. R ECENT R ESEARCH D IRECTIONS TO E NHANCE SDN
S CALABILITY
B. Scalability of SDN and IoT Networks
Simplifying Abstraction
Soheil et al. [44] deconstructed the scalability concerns
in SDN networking and argued that they are not unique to Beehive [44]: aiming to build a programming abstraction to enable the SDN platform
SDN, and can be overcome by deploying multiple controllers to automatically infer how applications maintain their state and how they depend on
one another. This simplifies the implementation process of distributed applications.
and switches. But based on our literature review we highlight
various factors that construct scalability concerns, shown in Controller Design [15]
Figure 7. The reasons for these concerns are described below.
Distributed controllers such as Flat structure multiple controllers, e.g., ONIX
1) As shown in Figure 4, flow handling may generate Recursive controller design, e.g., Xbar
additional network overhead because any new flow entry is Hierarchical controller design, e.g., Kandoo
treated like an alien (missed entry in flow table), and packets Distributed Architectures [15]
are then forwarded to controller for designing new flow entry.
This forces the controller to install new rules into the flow DIFANE: providing scalable solution keeping all traffic in the data plane.
DevoFlow: decreases number of interactions between switch and controller.
table to process the flows. The controller takes additional DISCO: providing the intercommunication between E2E network services.
time, because of its limited processing capability, to generate
new flow entry and then populates that to switch. This whole Research on Future Internet Oriented SDN Architecture [45]
process may add extra latency in end to end flow transaction. ALICANTE: a media ecosystem deployment through ubiquitous application aware
2) We argue that the bandwidth between the switch and network environment with the aim of providing flexible access to multimedia
controller is an additional significant resource which cannot services.

be ignored. It is observed that scientists consider the link

bandwidth, the computational power or flow handling capacity IoT networks are more sensitive to bandwidth and thus
of controller, and the switch’s capability to handle the flow as the scalability concern must be taken care more strongly in
the three main resources in SDN networking. We emphasize this territory with SDN. While analyzing the scalability of
that the scalability concerns are also dependent on the time IoT network it is important to consider the network topology,

2327-4662 (c) 2015 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See
http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/JIOT.2015.2480421, IEEE Internet of Things Journal
IEEE INTERNET OF THINGS JOURNAL 9

average service rate of controller, average arrival rate of for protocol non-compliance or defined criteria to decide
initiation requests, the path inflation factor that depends on whether the packet may pass or if it needs to be routed to a
the distance of the distributed controllers, channel capacity and different destination, is critical and essential in security. Also,
flow size. to distinguish the suspicious flows for the purpose of collecting
Another research challenge for SDN and IoT under scala- statistical information needs DPI. Therefore, more advanced
bility theme is the optimum controller placement concern that firewalls may need to sometimes examine and act on fields
influences every aspect of the decoupled plane. For example, that are not available to OpenFlow match and action. In order
high propagation delay in WLAN networks limits availability to set up the flow rule to invoke DPI for security reasons
and convergence time. This has practical implications on one question to be answered is, “which matching tuple will
software design, affecting whether controllers can respond to detect the packet that might needs special treatment?” Besides,
events in real-time or must push forwarding actions to for- whether the SDN edge switch has enough processing power on
warding elements in advance. This can create another issue of board (for DPI) rather than sending flows to controller for risk
controller placement in the designed network topology and the assessment is another challenge. These concerns are not unique
number of controllers for processing flows. Random placement in IoT networks. At this moment we are not aware about any
for a small k-value in the k-median problem, a clustering research outcome addressing this issue besides [50].
analysis algorithm, will result in an average latency between
1.4x and 1.7x greater than that of the optimal placement [46]. D. Packet Drop at AP
A reliability-aware controller placement problem has been
proposed in [47]. Authors of [46] proposed a latency-aware Dynamic change in user application and SDN application
controller placement problem with the objective to provide to action the flow can take time to set up rules to route traffic.
an initial analysis for further study of the formulation of For example, in security check application, the SDN manager
fundamental design problems. forwards such packets to the controller. The controller contacts
We have observed that, in SDN the research is continuously the malicious sites and only after verification sends flow entries
growing to overcome scalability issues. Table IV highlights the back to edge device. If this process takes too much time, the
recent research to enhance scalability of SDN oriented internet further packets will be dropped [40] by edge node because of
architectures. We argue that without SDN it will be difficult for limited flow handling table capacity. The path inflation may
IoT network to effectively process the real time data because add additional latency in wireless segment [51]. The edge
the problem lies in the nature of the IoT itself. It connect network device runs the risk of being overloaded with flow
remote nodes and provide a data stream between nodes and entries [40]. Minor latency because of network overhead in
decentralized management systems. The amount and type of wired network will introduce high latency in wireless Time
big data differ than other sets of data comes from social media. Division Multiple Access (TDMA) scheduling. To address this
Some features of the IoT data can be summarized as follows. problem the following mechanisms can be studied thoroughly.
a) The IoT data tends to arrive as a steady stream and at a a) Immediately offload the traffic to different AP or LTE might
constant pace, although it could arrive in batches like test logs be a better option in this situation [52]. b) Because of already
that can be processed and passed on straight away. b) The data limited capacity, AP will not support multiple flow tables.
comes in very large quantities and accumulates very fast. c) Thus dividing the flow table stack in different groups and
The real value of data can sometimes only be uncovered using state them as prior hierarchy gives less space to the multiple
effective analytics. d) The data is rarely used for production flow table stack. This ensures that only applications having
purposes. e) It can be deleted very quickly, unless there is a higher priority, e.g., VoIP and multimedia applications, can
need for compliance reasons. stay longer in multiple flow table groups of AP. So far, we
From the above concerns, we observe that the traditional have not been aware of any significant outcome on this topic
storage architecture, processing and management software will from the SDN research community.
treat IoT big data in the same manner as they treat other From our literature review we emphasize that SDN can
unstructured data. Therefore we conclude that although SDN minimize the data centre investment. Further the programmable
has some research issues about scalability, SDN architecture, network enables IoT devices to talk each other without much
which is different and significantly better than traditional hardware investment. Adequate uptime of the services, with
architecture, can effectively handle IoT big data streams. the integration of SDN and IoT, can be ensured. We argue
that the reliability can be guaranteed by Service Level Agree-
ments (SLA) that providers have to meet irrespective of the
C. Deep Packet Inspection technology they use in their networks. Table V illustrates SDN
Another limitation of OpenFlow is that the Deep Packet and IoT apps1 developed by Intel and Meru networks2 in this
Inspection (DPI) is unfortunately not supported in standard direction.
OpenFlow [49] because currently defined match fields in order To uplift the SDN integration to a higher and at broader
to evaluate packet are limited to the packet header only. level, Web of Things (WoT) seems a novel opportunity for
The common usage of DPIs includes lawful intercept, 1 http://searchsdn.techtarget.com/feature/SDN-to-support-Internet-of-
targeted advertising, and copyright enforcement. In security Things-devices
it is sometimes essential to thoroughly investigate the data 2 http://www.crn.com/news/networking/300074651/meru-networks-
part of the packet as it passes an inspection point. Searching launches-sdn-app-store-management-platform.htm

2327-4662 (c) 2015 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See
http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/JIOT.2015.2480421, IEEE Internet of Things Journal
IEEE INTERNET OF THINGS JOURNAL 10

TABLE V. I NDUSTRY I NITIATIVES TO I NTEGRATE SDN AND I OT

[4] http://www.techcentral.ie/software-defined-networking/
[5] K. K. Yap, M. Kobayashi, R. Sherwood, T. Y. Huang, M. Chan,
SDN, IoT apps
N. Handigol, N. Mckeown, “OpenRoad: Empowering Research in Mo-
Intel is creating an SDN environment to run IoT applications, developing open bile Networks,” Newsletter ACM SIGCOMM Computer Communication
source code for packet processing and for aspects of the orchestrator to allow it to Review, vol. 40, no.1, pp. 125-126, 2010.
understand server capabilities. [6] L. Suresh, J. S. Zander, R. Merz, A. Feldmann, T. Vazao, “Towards
Meru Network Manager for managing, configuring and monitoring a Meru wireless Programmable Enterprise WLANs with Odin,” Proceedings of the 1st
LAN. This will manage connection of mobile device on-boarding and guest access.
ACM SIGCOMM Workshop on Hot topics of Software Defined Networks
Meru Spectrum Manager for identifying sources of wireless interference.
Meru Services Assurance Manager, for performing predictive ”health checks” on (HotSDN12), pp. 115-120, 2012.
network applications. [7] M. Bansal, J. Mehlman, S. Katti, P. Levis, “OpenRadio: A Pro-
In May 2015, Huawei unveiled the world’s first SDN based IoT Solution named grammable Wireless Dataplane,” Proceedings of the 1st ACM SIG-
AgileIoT. This consists of Agile IoT getaways, operating system, and an Agile COMM Workshop on Hot Topics on Software Defined Networks
Controller. (HotSDN12), Finland, pp. 109-114, 2012.
IoT Challenges That SDN Can Solve [8] M. Yang, Y. Li, D. Jin, S. Ma, L. Zeng, “OpenRAN: A Software-defined
RAN Architecture via Virtualization,” Proceedings of ACM SIGCOMM
SDN automation meets Internet of Things. conference on SIGCOMM, China, pp. 549-550, 2013.
SDN can develop a scalable distributed system to manage the flow of events (data
[9] A. Gudipati, D. Perry, L. E. Li, S. Katti, “SoftRAN: Software Defined
flows from zillion of IoT devices).
SDN can effectively provide frictionless integration of new IoT components (de-
Radio Access Network,” Proceedings of the ACM SIGCOMM Work-
signed and deployed by multiple vendors) into the distributed systems and the shop on Hot Topics on Software Defined Networks 2013 (HotSDN13),
various data flows in a scalable manner. What primitives a network need in order China. pp. 25-30, 2013.
to support the variety of protocols in IoT?, and the solution is OpenFlow. [10] ftp://ftp.cs.princeton.edu/techreports/2012/922.pdf
[11] M. Moradi, L. E. Li, Z. M. Mao,“SoftMoW: A Dynamic and Scalable
Software Defined Architecture for Cellular WANs,” Proceedings of the
SDWN researchers. In WoT, using normal web application 3rd Workshop on Hot Topics on Software Defined Networks, HotSDN,
pp. 201-202, 2014.
designing tools, the real-world objects or things can be repre-
[12] P. Goransson, C. Black, “Introduction,” in Software Defined Network-
sented as resources that can be accessible via web technolo- ing; A Comprehensive Approach, 1st ed., Elsevier, pp. 1-20, 2014.
gies. This will minimize the need of waiting for other new [13] P. Goransson, C. Black, “The Genesis of SDN,” in Software Defined
components in networks, installing new infrastructure, or to Networking; A Comprehensive Approach, 1st ed., Elsevier, pp. 37-57,
redesigning the way we build our applications [53]. 2014.
[14] P. Goransson, C. Black, “How SDN Works,” in Software Defined
V. C ONCLUSION Networking; A Comprehensive Approach, 1st ed., Elsevier, pp. 59-79,
2014.
In this review article, we have presented the current key [15] B. A. A. Nunes, M. Mendonca, X. N. Nguyen, K. Obraczka, T. Turletti,
research efforts on Software Defined Wireless Network. We “A Survey of Software Defined Networking: Past, Present, and Future
emphasize that integration of SDN in IoT network can poten- of Programmable Networks,” Communications Surveys and Tutorials,
tially bring exciting opportunities. We also highlighted that the IEEE, vol. 16, no.3, pp. 1617-1634, 2014.
traditional network tools to collect, store, process, and forward [16] N. McKeown, T. Anderson, H. Balakrishnan, G. Parulkar, L. Peterson,
J. Rexford, S. Shenker, J. Turner, “OpenFlow: Enabling Innovation in
massive data, are inefficient to meet critical future IoT network Campus Networks,” Newsletter ACM SIGCOMM Computer Communi-
needs whereas SDN can significantly simplifies the network cation Review, vol. 38, no.2, pp. 69-74, 2008.
control and management needs. Further we described critical [17] https://www.opennetworking.org/openflow-
security and scalability issues of SDN network that are also conformancecertificationlabs
common in IoT network. We conclude that SDN technology [18] B. Lantz, B. Heller, N. Mckeown,“A Network in a Laptop: Rapid
is gaining much attention from researchers from both industry Prototyping for Software-Defined Network,” Proceedings of the 9th
and academia. Significant growth over coming years has been ACM SIGCOMM Workshop on Hot Topics on Networks (HotNets),
USA, pp. 1-6, 2010.
observed from industry such as Google and Juniper. The next
generation of technology is almost ready to reap the benefits [19] C. Scott, A. Wundsam, K. Zarifis, S. Shenker, “What, Where, and When,
Software Fault Localization for SDN,” Technical Report UCB/EECS-
of controlling networks with a unified control protocol almost 2012-178, EECS Department, University of California, Berkeley, 2012.
in every technological domain. Recent industry initiatives to [20] M. Rietblatt, M. Canini, A. Guha, N. Foster, “FatTire: Declarative Fault
integrate SDN and IoT technology are also presented. Tolerance for Software-Defined Networks,” Proceedings of the ACM
SIGCOMM Workshop on Hot Topics on Software Defined Networks
(HotSDN13), China, pp. 109-114, 2013.
R EFERENCES
[21] M. Gupta, J. Sommers, P. Barfors, “Fast, Accurate Simulation for SDN
[1] J. Gubbi, R. Buyya, S. Marusic, M. Palaniswami, “Internet of Things Prototyping,” Proceedings of the ACM SIGCOMM Workshop on Hot
(IoT): A Vision, Architectural Elements and Future Directions,” Future Topics on Software Defined Networks 2013 (HotSDN13, China, pp. 21-
Generation Computer Systems, vol. 29, no. 7, pp. 1645-1660, 2013. 36, 2013.
[2] W. H. Chin, Z. Fan, R. J. Haines, “Emerging Technologies and Research [22] K. K. Yap, M. Kobayashi, D. Underhill, S. Seetharaman, P. Kazimian,
Challenges for 5G Wireless Networks,” Wireless Communications, N. Mckeown, “The Stanford OpenRoad Deployment,” IEEE Proceed-
IEEE, vol. 21, no.2, pp.106-112, April 2014. ings of 4th International Workshop on Experimental Evaluation and
[3] P. M. Julia, A. F. Skarmeta, “Extending the Internet of Things to IPv6 Characterization, WiNTECH09, China, pp. 59-66, 2009.
with Software Defined Networking,” white paper http://www.euchina- [23] S. Jain, A. Kumar, S. Mandal, J. Ong, L. Poutievski,
fire.eu/wp-content/uploads/2014/06/SKARMETA-A.-Extending-the- A. Singh,S. Venkata, J. Wanderer, J. Zhou, M. Zhu, J. Zolla,
Internet-of-Things-to-IPv6-with-Software-Defined-Networking.pdf U. Holzle, S. Stuart, A. Vahdat, “B4, Experience with a Globally-

2327-4662 (c) 2015 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See
http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/JIOT.2015.2480421, IEEE Internet of Things Journal
IEEE INTERNET OF THINGS JOURNAL 11

Deployed Software Defined WAN,” Newsletter ACM SIGCOMM [40] I. F. Akyildiz, A. Lee, P. Wang, M. Luo, W. Chou, “A Roadmap for
Computer Communication Review, vol. 43, no.4, pp. 3-14, 2013. Traffic Engineering in SDN-OpenFlow Networks,” Computer Networks,
[24] A. K. Rangisetti, H. B. Baldaniya, P. B. Kumar, B. R. Tamma, Elsevier, vol.71, pp. 1-30, 2014.
“Load-Aware Hand-offs in Software Defined Wireless LANs,” IEEE [41] L. Zhou, H. C. Chao, “Multimedia traffic security architecture for the
10th International Conference on Wireless and Mobile Computing, internet of things,” Network, IEEE, vol.25, no.3, pp.35-40, May-June
Networking and Communications (WiMob), pp. 685-690, 2014. 2011.
[25] G. Araniti, J. Cosmas, A. Lera, A. Molinaro, R. Morabito, A. Orsino, [42] L. Zhou, R. Q. Hu, Y. Qian, H. H. Chen, “Energy-Spectrum Efficiency
“OpenFlow over Wireless Networks: Performance Analysis,” IEEE Tradeoff for Video Streaming over Mobile Ad Hoc Networks,” Selected
International Symposium on Broadband Multimedia Systems and Broad- Areas in Communications, IEEE Journal on, vol.31, no.5, pp.981-991,
casting (BMSB), pp. 1-5, 2014. May 2013.
[26] S. Costanzo, L. Galluccio, G. Morabito, S. Palazzo, “Software Defined [43] http://www8.hp.com/au/en/hp-news/press-
Wireless Networks: Unbridling SDNs,” Software Defined Networking release.html?id=1798074.VQoZD6O4aUk
(EWSDN), 2012 European Workshop on, pp. 1-6, Oct. 2012. [44] S. H. Yeganeh, A. Tootoonchian, Y. Ganjali, “On Scalability of
[27] P. Dely, A. Kassler, L. Chow, N. Bambos, N. Bayer, H. Einsiedler, Software-Defined Networking,” Communications Magazine, IEEE, vol.
C. Peylo, D. Mellado, M. Sanchez, “A Software Defined Networking 51, no.2, pp. 136-141, 2013.
approach for Handover Management with Real-time Video in WLANs,” [45] FP7 ICT project, MediA Ecosystem Deployment Through Ubiqui-
Journal of Modern Transportation, vol. 21, no.1, pp. 58-65, March tous Content-Aware Network Environments, ALICANTE, No248652,
2013. http://www.ict-alicante.eu.
[28] C. Guimaraes, D. Corujo, R. L. Aguiar, F. Silva, P. Frosi, “Empower- [46] B. Heller, R. Sherwood, N. McKeown, “The Controller Placement
ing Software Defined Wireless Networks through Media Independent Problem,” First Workshop on Hot Topics in Software-Defined Networks,
Handover Management,” Global Communications Conference (GLOBE- pp. 7-12, 2012.
COM), IEEE, pp. 2204-2209, Dec. 2013. [47] Y. N. Hu, W. D. Wang, X. Y. Gong, X. R. Que, S. D. Cheng,“On the
[29] H. E. Egilmez, A. M. Tekalp, “Distributed QoS Architectures for Placement of Controllers in Software-Defined Networks,” The journal
Multimedia Streaming over Software Defined Networks,” Multimedia, of China Universities of Posts and Telecommunication,Elsevier vol. 19,
IEEE Transactions on, vol. 16, no.6, pp. 1597-1609, Oct. 2014. pp. 92-97, October 2012.
[30] J. Rukert, R. Bifulco, M. Rizwan-Ul-Haq, H. J. Kolbe, D. Hausheer, [48] S. H. Yeganeh, Y. Ganjali, “Beehive: Towards a Simple Abstraction for
“Flexible Traffic Management in Broadband Access Networks using Scalable Software Defined Networking,” Proceedings of the 13th ACM
Software Defined Networking,” Network Operations and Management Workshop on Hot Topics in Networks, HotNets-X111, pp. 1-13, 2014.
Symposium (NOMS), IEEE, pp. 1-8, May 2014. [49] G. Finnie, “The Role of DPI in SDN Work”, QOSMOS Technology,
[31] T. Feng, J. Bi, K. Wang, “Joint Allocation and Scheduling of Network White Paper. pp. 1-14, Dec 2012.
Resource for Multiple Control Applications in SDN,” Network Opera- [50] H. Mekky, F. Hao, S. Mukherjee, Z. L. Zhnag, T. V. Lakshman,
tions and Management Symposium (NOMS), IEEE, pp. 1-7, May 2014. “Application-aware Data Plane Processing in SDN,” Proceedings of
[32] J. Lee, M. Uddin, J. Tourrilhes, S. Sen, S. Banerjee, M. Arndt, the 3rd ACM SIGCOMM Workshop on Hot topics of Software Defined
K. H. Kim, T. Nadeem, “meSDN: Mobile Extension of SDN,” Proceed- Networks (HotSDN14), pp. 13-18, 2014.
ings of the fifth International Workshop on Mobile Cloud Computing [51] J. Hu, C. Lin, X. Li, J. Huang, “Scalability of Control Planes for
and Services (MCS14), pp. 7-14, 2014. Software Defined Networks: Modeling and Evaluation,” Quality of
Service (IWQoS), IEEE 22nd International Symposium of, pp. 147-152,
[33] V. Gudla, S. Das, A. Shastri, G. Parulkar, N. McKeown, L. Kazovsky,
May 2014.
S. Yamashita, “Experimental Demonstration of OpenFlow Control of
Packet and Circuit Switches,” Optical Fiber Communication (OFC), [52] M. Manic, D. Wijayasekara, K. Amarasinghe, J. Hewlett, K. Handy,
Collocated National Fiber Optic Engineers Conference, Conference on C. Becker, B. Patterson, R. Peterson, “Next Generation Emergency
(OFC/NFOEC), pp. 1-3, March 2010. Communication Systems via Software Defined Networks,” Research
and Educational Experiment Workshop (GREE), 2014 Third GENI, pp.
[34] W. Cerroni, G. Leli, C. Raffaelli, “Design and Test of a Software 1-8, March 2014.
Defined Hybrid Network Architecture,” Proceedings of 1st ACM edition
Workshop on High Performance and Networking (HPPN13), pp. 1-8, [53] D. Raggett, “The Web of Things: Challenges and Opportunities,” IEEE
USA, 2013. Computer Magazine , vol. 48, no. 5, pp. 26-32, May 2015.
[35] M. Channegowda, R. Nejabati, D. Simeonidou, “Software Defined
Optical Networks Technology and Infrastructure: Enabling Software-
Defined Optical Network Operations [invited],” Optical Communica-
tions and Networking, IEEE/OSA Journal of, vol.5, no.10, pp. A274-
A282, June 2013.
[36] L. E. Li, Z. M. Mao, J. Rexford, “Toward Software-Defined Cellular
Networks,” Software Defined Networking (EWSDN), 2012 European Keshav Sood worked with Terminal Ballistic Re-
Workshop on, pp. 7-12, Oct. 2012. search Lab as a trainee. He did his B.Tech in
[37] A. Gember, P. Prabhu, Z. Ghadiyali, A. Akella, “Toward Software- Electronics Engineering with distinction and M.tech
Defined Middle box Networking,” Proceedings of the 11th ACM SIG- degree in Optical Fibre Engineering. Currently he is
COMM Workshop on Hot Topics on Networks (HotNets) USA, pp. 7-12, towards his PhD degree in School of IT at Deakin
2012. University, Melbourne, Australia. He is a recognised
Professional Engineer by Engineers Australia. He
[38] J. Q. Wang, H. Fu, C. Cao, “Software Defined Networking for Telecom is also a Student Member of IEEE. He served as
Operators: Architecture and Applications,” Proceedings of the 8th TPC member of various IEEE conferences includes
International Conference on Communication and Networking in China, IEEE INFOCOM, IEEE Bigdata service, and IEEE
CHINACOM, pp. 828-822, 2014. ITNAC. His current research domain are network
[39] G. Hampel, M. Steiner, T. Bu, “Applying Software-Defined Network- security and flow management in SDN.
ing to the Telecom Domain,” Computer Communications Workshops
(INFOCOM WKSHPS), IEEE Conference on, pp. 133,138, 14-19 April
2013

2327-4662 (c) 2015 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See
http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/JIOT.2015.2480421, IEEE Internet of Things Journal
IEEE INTERNET OF THINGS JOURNAL 12

Shui Yu (M05-SM12) is currently a Senior Lec-

turer of School of Information Technology, Deakin
University. He is a member of Deakin University
Academic Board (2015-2016), a Senior Member of
IEEE, and a Member of AAAS, the Vice Chair of
Technical Subcommittee on Big Data Processing,
Analytics, and Networking of IEEE Communication
Society, a Member of IEEE Standard Committee of
Big Data. Dr Yu’s research interest includes big data,
networking theory, cyber security, and mathematical
modelling. He has published two monographs and
edited one book, more than 100 technical papers, including top journals and
top conferences, such as IEEE TPDS, IEEE TC, IEEE TIFS, IEEE TMC, IEEE
TKDE, IEEE TETC, and IEEE INFOCOM. Dr Yu initiated the research field
of networking for big data in 2014. His h-index is 21. He is currently serving
the editorial boards of IEEE Transactions on Parallel and Distributed Systems,
IEEE Communications Surveys and Tutorials, IEEE Access, and a number
of other international journals. He has served more than 50 international
conferences as a member of organizing committee, such as publication chair
for IEEE Globecom 2015 and IEEE INFOCOM 2016, TPC co-chair for IEEE
ATNAC 2014, IEEE BigDataService 2015, IEEE ITNAC 2015.

Yong Xiang (SM12) received the Ph.D. degree in

Electrical and Electronic Engineering from The Uni-
versity of Melbourne, Australia. He is a Professor
and the Director of the Artificial Intelligence and
Image Processing Research Cluster, School of In-
formation Technology, Deakin University, Australia.
His research interests include signal and system
estimation, information and network security, multi-
media (speech/image/video) processing, and wireless
sensor networks. He has published more than 110
refereed journal and conference papers in these areas.
He is an Associate Editor of IEEE Signal Processing Letters and IEEE Access.
He has served as Program Chair, TPC Chair, Symposium Chair, and Session
Chair for a number of international conferences.

2327-4662 (c) 2015 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See
http://www.ieee.org/publications_standards/publications/rights/index.html for more information.