1536-1284/19/$25.00 © 2019 IEEE IEEE Wireless Communications • December 2019 64
ABSTRACT
IoT has revolutionized the modern way of life
with the advent of intelligent systems. The tradi-
tional network architecture employed in the IoT
domain is unable to define sufficient solutions of
these challenges for a cost-effective and seam-
less workflow. The intelligent IoT system ensures
scalability in challenging or hostile environments.
With the emergence of the SDN domain offering
programming ability of the control plane, many
of these challenges seemed surmountable. This
article presents a synergized overview of the chal-
lenges faced by the traditional domain and how
they can be overcome by the upcoming domain
of SDN-IoT. A thorough analysis of the practical
adoptions showcase the feasibility of the solution
in a real-time environment. The article examines
the state of the art and highlights some of the key
open points in the domain, based on shortcomings
of the current state of SDN-IoT, that can be taken
up for future research.
INTRODUCTION
The recent advancements made in the Internet
of Things (IoT) domain paradigm are heralding
remarkable transformations for improving the way
of human life. The plethora of smart devices inte-
grated with their actuation and sensing abilities,
environmental awareness, and real-time analy-
sis are making every aspect of modern society
smarter and more efficient. For the purpose of
achieving these goals, IoT devices must be able to
interconnect and concurrently provision services
that are backed by the back-end data storage sys-
tems, while processing and assimilating the huge
data generated by the various sensors and actu-
ators [1].
The fact that the IoT domain employs a huge
number of devices that can collect information
in a real-time environment and at such a high
frequency is perceived as its greatest strength.
However, this also acts as a major bottleneck
for the network. The complexity, heterogeneity,
and associated limitations of the various devic-
es require complex and specific tools for man-
agement and for improving the performance
of the network. The critical aspects associated
with the performance and scaling of the net-
work are often attributed to the characteristics
of the devices and the proprietary architectures
adopted by them. These issues cannot be solved
by the mere introduction of a “gateway” like
structure; challenges also emerge in the form
of data aggregation, reliability, privacy, security,
and trustworthiness. The capability and proto-
col mismatch between the IoT devices accen-
tuates when there is a huge number of devices
deployed in a complex environment with varying
protocols and varying designs. These problems
become even more significant in real-time sce-
narios like robotics and self-driven cars, where
the exchange of real-time information also needs
the applications to become scalable, efficient,
seamless, and cost-effective. Unfortunately, how-
ever, the current state of IoT technology alone
cannot provision such requirements and over-
come such challenges in an efficient manner [2].
Current state-of-the-art architecture for IoT
devices is not capable of supporting features like
mobility, higher scalability, and heavy traffic all at
the same time along with the above mentioned
functionalities. Moreover, with the number of
connected IoT devices slated to grow exponen-
tially over the next few years, it will become even
more tedious to manage the mammoth amount
of data generated by these devices without the
inherent features of elasticity and flexibility of
the network. In absence of these features, such a
flood of data and traffic could paralyze the entire
network.
For overcoming these challenges faced by the
IoT paradigm, upcoming technologies like software
defined networking (SDN), OpenFlow architec-
ture (based on SDN), and network function vir-
tualization (NFV) are gaining major traction. The
technology domain of SDN, specifically, has been
receiving a lot of attention from the research
community and has also proven its mettle in large
deployments of data center networks optimizing
the needs of IT and network resources [3]. SDN
architecture aims at making networks flexible and
agile. The primary objective of SDN is to bring
about an improvement in network control by
empowering service providers and enterprises to
provide faster response to ever-evolving business
requirements. In a software-defined network, the
network administrator can control and monitor traf-
fic without making any changes to the individual
switches of the network. These switches are direct-
ed by the centralized SDN controller to provision
network services based on their requirement, irre-
spective of the connections between the devices
and the server.
Pritish Mishra, Deepak Puthal, Mayank Tiwary, and Saraju P. Mohanty
Sofware Defned IoT Systems:
Properties, State of the Art, and Future Research
FUTURE COMMUNICATION TRENDS TOWARDS INTERNET OF THINGS SERVICES AND APPLICATIONS
Pritish Mishra and Mayank Tiwary are wtih SAP Lab; Deepak Puthal is with Newcastle University;
Saraju P. Mohanty is with the University of North Texas.
Digital Object Identifier:
10.1109/MWC.001.1900083