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

net/publication/362641021

SDN‐Based Cloud Combining Edge Computing for IoT Infrastructure

Chapter · July 2022

DOI: 10.1002/9781119857921.ch14

CITATIONS READS

7 756

6 authors, including:

Jyoti Sn Manish Snehi

MECP Infosys
58 PUBLICATIONS 279 CITATIONS 27 PUBLICATIONS 197 CITATIONS

SEE PROFILE SEE PROFILE

Devendra Prasad Sarita Simaiya

Chitkara University Chitkara University
41 PUBLICATIONS 345 CITATIONS 87 PUBLICATIONS 1,547 CITATIONS

SEE PROFILE SEE PROFILE

All content following this page was uploaded by Jyoti Sn on 21 December 2022.

The user has requested enhancement of the downloaded file.

 14. SDN-based Cloud combining edge computing for IoT
Infrastructure

Jyoti Snehi1, Manish Snehi2, Devendra Prasad1, Sarita Simaiya1, Isha Kansal1, Vidhu
Baggan1
1
Chitkara University Institute of Engineering and Technology, Chitkara University, Punjab,
India
2
Department of Computer Science and Engineering, Punjabi University, Patiala, Punjab,
India

[email protected]
[email protected]
[email protected]
[email protected]
[email protected]
[email protected]
Abstract:
Traditional network protocols have proven to be a significant impediment to network
administration. Computer networks are evolving at a breakneck pace. SDN enables central
control of networks through software applications that leverage open APIs. This chapter’s
authors have explored the significance of virtualization in resolving fundamental challenges
associated with SDN and NFV in the cloud technologies for IoT infrastructures and services to
propose an SDN-based IoT architecture that makes better exploitation of data from IoT
applications and network operations that a standard SDN based network lacks. Additionally,
they have extended the control plane with IoT-specific capabilities so that the control logic can
be tailored to specific IoT applications, allowing for heterogeneous IoT systems to be
supported. By building a robust knowledge-driven SDN and IoT integrated architecture that
employs edge cloud technologies to bring cloud solutions perilously close to the Internet of
Things, the authors investigated the possibility of SDN being used to enhance IoT networks.
The convergence of SDN-enabled programmability and standard cloud computing enables the
IoT system to be both agile and cognitive, evolving it into an autonomous system. The
contributors of this chapter conducted research and highlighted cutting-edge cloud services in
the domain of service transition from centralized to distributed cloud solutions, as well as the
challenges provided by these highly distributed systems. In this study, authors have discussed
network challenges that include integrating a cloud technology in SDN network with the
Internet of Things and suggested that by incorporating a knowledge plane into the devices,
along with a knowledge-bound API; IoT proxies may be realized, which results in increased
data efficiency through IoT devices and data.

Keywords: Cloud computing, Edge computing, Fog computing, IoT, Internet of Things, SDN,
Software-defined networks, NFV, Network Function Virtualization
14.1. INTRODUCTION
Cloud computing ushers in a new age of computer technology. Software, platform, and
infrastructure services are the most prevalent types [1]. SaaS software service provides cloud-
based email, social networking, and scheduling services. Developers build applications using
cloud PaaS platform services. IaaS provides cloud customers with virtual machine servers [2].
Using IaaS enables for SaaS and PaaS [3]. Nevertheless, Cloud data centres are geographically
positioned remote and centralized at end-users systems. [4]. Software-Defined Networks
(SDN) a new standard for network administration, separates the packet processing data plane
of the routing processing control plane [5]. As SDN use rises, so do new demands and
opportunities. SDN is now generally available as a cloud-based setting [6]. However, the
concept applies to both conventional and computer networks [7]. Storage, network processing,
confidentiality, edge, and cloud systems are all included. The first commercial telephone
exchange was introduced in 1878, which marked the beginning of SDN. Then, throughout
1891, Almon Strowger invented the phone line dial, which used the same unsecured route to
send data and dial signals to the telephone exchange. To separate data and control, Central
Network design predictors were introduced in early 1980. In the 1990s, the adaptive network
was developed, in which access points are programmed to perform specific transactions on
messages getting passed through them. By decoupling the control plane, SDN enabled the rapid
modernization applications by allowing centralised control, digitization, and the formation of
a reconfigurable network. SDN centralises network control, whereas Network Function
Virtualization (NFV) virtualizes network components. As a result of the growth of pervasive
computing, new technologies such as wireless and communication networks, mobile platforms,
embedded systems, wearable technology, language processing, radio frequency tags,
intelligent machines, middleware, and application services have emerged.[7]

The SDN paradigm creates a dynamic, coordinated system [8]. In a Software-Defined

Network, each data and control plane is distributed [9]. SDNs allow applications to monitor
network services and traffic. Consistent centralized control guarantees better resource
utilization, information exchange, and programming operational processes across SDN nodes
[10]. A controller in SDN responds to changing requirements by working with applications,
devices, and routers [11]. The number of Internet-connected objects associated with the IoT
has been rapidly increasing. To maintain the data explosion, adequate internet infrastructure
must be in place. One issue is that Devices are spread all over the world. All of these
geographically dispersed devices must be connected to the network infrastructure. This is a
major task that will necessitate a significant investment on the part of any single service
provider. Because customers are enrolled in different vendors and are geographically
distributed, each provider could have its own unique IP address to serve subscribers. As new
technology advances, hardware quickly gets obsolete, resulting in massive recurring costs for
each service supplier. A unified communication architecture that can adapt to changes is
required. This is especially true when the percentage of subscribers for each service supplier
varies. The demand for network services will fluctuate dynamically. Services providers can
request facilities from the mutual network design, which they can then return about subscriber
consumption. As a result, the sharable design ensures that adequate services are available to
providers based on their current needs, while also looking at potential anticipated demands.
Another critical issue is the network reconfiguration required to meet evolving mobility
patterns, such as bandwidth and latency requirements. As commercial applications are
presented to the system to serve users, security and support deployment rules will keep
evolving. Changes to packet handling rules are required, and new data transmission may
necessitate high-layer processing. The placement of routers, networking devices, and other
specialized portals must be altered as a result of the new rules. In a shared network
environment, where many network operators share network infrastructure, such restructuring
becomes a problem. All of these issues necessitate the detachment of system resources as well
as strict confidentiality between service providers. Technological innovations such as SDN and
NFV virtualization will make this possible. Future SDN and NFV designs are discussed in this
article. A cost analysis is used to evaluate the benefits of such able to share SDN as well as
NFV-based network models.
The use of technology, as well as the anticipated request for it, is increasing at an
alarming rate. To keep up with demand, innovation is actively evolving more efficient and
accurate transmission network solutions. These technological concepts are critical because old
ones are quickly becoming obsolete. This is due not only to advances in technology but also to
growing concerns and new requirements. Every day, networks add new users and network
services. Increased user numbers, as well as new types of services, place a substantial burden
on available bandwidth, which is rapidly increasing. The number of connected devices has
increased dramatically, far exceeding the capacity of communication systems to handle such a
wide range of devices now or in the coming years. A new approach to data communications
network technology is required to support an enormous growth in the number of
interconnections that can be provided by an active network solution. Furthermore, such a
system has to assist with prospective internet applications that are being developed in
expectation of possible requirements that may necessitate larger chunks of networking devices.
The network design must supervise the constant movement of users and devices and end up
making their data available to them and in a timely way that is consistent with the level of
service they have chosen. The network design must be robust, with fast recovery methods for
resolving network inadequacies and other causes of issues. It is essential to consider the
numerous variables related to the expense of deploying connections to provide trillions of
connected devices in the IoT technology.
End devices are available in a wide range of types and sizes. Specialized network
characteristics and facilities are required to operate these various IoT devices. When it comes
to data flow, each type of application necessitates its access control that takes priority
confidentiality, integrity, and basic safety. Previously, it was common to practice to incorporate
specialized network services onto equipment to speed up network operations. These specialized
devices must be distributed properly across the system concerning the various services offered
by network operators. For service providers, the cost of acquiring locations and
communications networks across a vast geographical area is extremely costly. In some cases,
leasing system resources from other container-based services to provide services to customers
is more helpful and cost-effective. This criterion introduces a new criterion for network
resource sharing between multiple service providers. These infrastructure provider
organizations will provide server resources to access network connections on a need-to-know
basis through appropriate partitioning and isolation methods. Virtualization software is a vital
part of resolving each of these issues. By implementing the virtualized network concept, all
network components will be configurable via a highly standardized graphical interface and will
be remotely accessible from any of the central locations selected. Because all nodes in the
network will be entirely software-based, the new architecture will streamline all internal and
external networks. The role of a particular network component can be easily adjusted by
changing the node's network function [10]. Edge computing has different connotations in
different industries. By far the most subsequent euphemism for cloud applications is edge
computing. A distributed computing definition relocates data and manages storage close to
where they are needed, improving response time and reducing bandwidth consumption. Edge's
technology has been in development for decades. Edge computing can be dated directly to 1990
Akamai launched the first network. This network distributed servers closer to users to provide
cached images and movies. Ubiquitous Computing, the technology underpinning the Internet
of Things, first appeared in 1997, offloading resource-intensive programs to local servers. To
avoid long-distance server downloads, overlay Peer-to-Peer networking used proximity routing
in the early 2000s. Amazon's Elastic Compute Cloud, the first open-source platform, was
launched around 2006, allows users to quickly rent computer resources. Cloudlet, a mobility-
enhanced narrow cloud-based data center intended to serve resource-intensive software
products, was launched in 2009. In 2010, Cisco introduced Fog Computing, a distributed cloud
services architecture that employs intelligent edge nodes to handle massive amounts of
handling, storage, and communication.[7].
Current network protocols are an important stumbling block in network administration.
Computer networks are rapidly evolving, and this trend is likely to continue. SDN enables
network regulation to be centralized through the use of application software that leverages open
APIs. When it comes to trying to overcome fundamental difficulties related to SDN and NFV
in the public cloud for IoT critical infrastructure, such as integrating a cloud services SDN
network with the IoT, virtualization research is critical. The term "Edge Computing" refers to
the latest developments of bringing computing capacity closer to edge routers. It is being driven
by the rapid development of AI, as well as a diverse range of small footprint appliances and
mobile devices. These end systems require supercomputing supplementation, and because
legacy virtualization is not the best fit, additional options must be considered. The author
conducted a literature review of the software platforms of the Edge Computing framework,
particularly regarding Cloudlets, Fog Nodes, and MEC, in this paper. With this effort, authors
keep hoping to present a clear overall picture for the Edge Of the network space while also
emphasizing the technological differences between the various options. While Edge
Computing is pledging and actively trying to fill a technological void, it still encounters a set
of technical challenges before becoming completely omnipresent. It is critical to emphasize
that enhancing the computational power of network edge through Edge Devices does not
replace the cloud; instead, they are complementary belief systems that should be used in
tandem. However, deploying additional computation power at the channel's edge and
establishing complementarity while making them accessible elsewhere is not an easy task. To
realize the envisioned instances of ubiquitous computing, we must develop a system that
conceals all internal problems from users, particularly application developers and service
providers. To that end, SDN emerges as a credible contender for organizing networks, services,
and structures while trying to shield end users from the intricacies of this heterogeneous
network, owing to its network control capabilities.
Edge computing enabled by 5G unlocks new opportunities for every business. It helps
bring computation and data computation closer to the analysis point, enabling improved data
control, cost savings, faster profound knowledge and actions, and uninterrupted operations. By
2025, 75percent of the total corporate data will be evaluated at the edge, up from 10% today.
Even though these edge devices are more durable than ever, the overwhelming majority of
them are incapable of supporting real-time utilization scenarios while adhering to predefined
Quality of Service (QoS) standards [12]. Cloud computing is widely regarded as a support
system advancement for low-capacity devices and systems because it provides a pool of
computing and storage resources. On the other hand, wearable devices and the Internet of
Things are optimized for time-sensitive applications. Due to the delay introduced by attempting
to reach them over the WAN, legacy cloud hosting may be the only option, as the majority of
these systems require a high level of QoS due to their agility, the framework's purpose, and
real-time expectations. The progress of IoT and its applications necessitates the establishment
of strict QoS standards. Extending real-time connections necessitates the utilization of external
computational resources at the channel's edge. SDN-based Cloud with edge devices for IoT
Infrastructure allows for more data exploitation from IoT systems and network processes than
traditional SDN-based channels. The network control has also been reinforced with IoT-
specific features, enabling control logic to be tailored to the specific IoT applications and
promoting the operation of huge and varied IoT systems. The integration of SDN-enabled
automation and standard cloud services enables the IoT technology to evolve into a self-
contained, adaptable, and intelligent system. The authors of this chapter carried out an
investigation and provided their study results in this research study using cutting-edge
information technology in the context of service transition from centrally controlled to
distributed cloud services, as well as the challenges presented by these widely spread systems.
According to the study, IoT proxies could be realized by trying to insert a knowledge plane
into the device and pairing this with a knowledge-bound Application, resulting in increased
data efficiency via the use of IoT devices and data. The authors of this study recognized
potential network issues [13].

14. 1.1 ARCHITECTURE OF SDN VS. TRADITIONAL NETWORKS

The goal of SDN is to decouple hardware and software [14]. OpenFlow benefits from
data plane and control plane separation. Routines and content determine a packet's destiny [15].
The physical infrastructure gets a control layer. So, the SDN controller here only follows basic
rules. The OpenFlow switches exchange statistics with the Control plane [16]. A Control to
Data-Plane Interface (CDPI) may provide unrestricted visibility and control over its data route
and processing[17]. The SDN Northbound Interface enables controllers to interact with SDN
requests. The first fundamental feature of SDN is the division of planes into data planes
incorporated in the device and a centralized controller implements the control plane. An
incoming packet can be forwarded, dropped, consumed, transformed, or replicated by the data
plane. It determines the optimum output port by undertaking an address table lookup in the
very fast hardware, which operates at terabits per second. The control plane contains the
methodologies used to program the data plane. The network control is removed from the
switching device and placed on a central system [18].

Figure 14.1: Traditional Network Architecture

In a traditional communication network as shown in Figure 14.1., network devices

feature control plane capabilities that are used to decide where to send frames or packets
entering the computer network, such as a switch, based on programmable rules. When adding
new apps or changing the rules governing how traffic flows all over the network in this context,
modifications must be performed directly to the physical device. With a typical network
architecture, the data center manager could stuff a rack with components that were almost
certain to connect with database servers and application servers. These components can be
located anywhere inside the network virtualization infrastructure. Additionally, virtualization
stretches the boundaries of IP addressing.
 Figure 14.2: SDN Architecture

SDNs enhance network agility and most significantly, network control centralization. [19]. The
SDN design in Figure 14.2. gives programs a broader perspective of the information system.
SDN splits the network into three layers and independent of underlying resources provides
programmatic access to network services [20]. OpenStack is a unique programmable network
infrastructure control and management tool that provides Open standards-based access to basic
IT resources [21].

14. 1.2 SDN/NFV Tiers

The Internet's fast expansion has created issues of heterogeneity, scalability, and
interoperability. SDN and NFV are two prominent virtualization technologies for
communication channels. As a result, the network layer's physical base serves as a relaying
layer, while the control system discovers the flow of information and administrative
information [22]. Traditional network architecture relies on specialized network equipment. A
software-based network function may be arranged on top of a shared physical structure [23].
For example, by virtualizing the SDN controller, NFV may assist SDN. So, combining SDN
and NFV enhances both. SDN comprises the NFV platform, control module, and forwarding
device at the network edge [24]. Cloud platforms provide virtual hardware, virtual network
operations, virtual platform, and finally virtual applications and software as a service [25].
Cloud networks are intended to be global, have a limitless capacity, be very reliable and secure.
Infrastructure organizations would leverage the cloud to broaden the deployment of virtual
hardware. Network service providers would practice global acceptability to build virtual
systems and Virtual network functions (VNFs). Network service providers would build
networks for operators of networks [26]. The network administrators may perform applications
and settings related to software for their customers [27]. It helps in utilizing the internet to
socialize, use the software at home or workplace, receive news and other information, or have
video conferences [28].
Network Function Virtualization (NFV) is a technique for transitioning functions from
specialized hardware to a virtual environment. NFV eliminates the need for specialized
hardware in favor of commodity servers [29]. On-demand virtual network services allow
security, load balancing, data collection and analysis, and other SDN applications virtual
network services (VNFs). This increases scalability and flexibility while maximizing network
device computation, memory, and networking capacity [30]. SDN and NFV are mutually
exclusive technologies. As a result, the primary benefits of both technologies overlap [31].
NFV may aid in the virtualization of the SDN controller and more SDN-based network
applications where SDN may perform VNF-driven traffic engineering decisions [32].

Figure 14.3: SDN/NFV Tiers

Figure 14.3. presents SDN/NFV tiers and depicts the current network situation and
future anticipated virtual network possibilities. Consider the current state of the conventional
network. All of these disparate devices are distributed across the network in this scenario. Due
to the fixed network functionality of each device, the device's placement inside the network
must be carefully selected. If the network structure has to be altered after a period to suit
additional requirements, these devices must be reorganized, reconnected, and then configured
separately. Because each device's operation is integrated within it, each device has its interface
to allow that must be obtained separately. New machines must be introduced on occasion, and
older ones should be removed, wasting valuable network equipment resources. All of these
elements contribute to the cost of infrastructure, processes, and administration. A large number
of the infrastructure, as seen, is made up of a bank of available commercially servers and
switches constructed as a hardware pool. These hardware pools can host a wide range of virtual
connected devices with different characteristics. Virtual network capabilities defined in
software share information with the hardware resources via a common interface. This enables
a provider to establish, remove, and change network functions on an as-needed basis. This
increased flexibility enables system resources to be allocated to real service provider
requirements. Virtual networks constructed on top of these software-implemented network
operations are very adaptable, allowing for easy reconfiguration and adaptation to new network
topologies. A single administration interface is available, and the whole network may be
managed from one or many central locations. As a consequence, network operations and
administration are much easier and more cost-efficient than managing specialist hardware-
based networks.SDN devices contain forwarding functionality, forwarding information is
stored in a flow table. The flow table is stored on the access point and includes a set of flow
tables and actions to take when a systematically collected one of the entries. If the SDN system
receives the data, it performs the configured action. When it does not establish a connection, it
has the option of dropping the packet or passing it to the control system. It is responsible for
implementing control plane functionality. It exposes a network abstraction to the SDN software
running above. The controller enables the SDN software to describe flows on modules and to
react to packets transmitted to the control system by routers. It keeps an overview of the entire
network.
The model's tiers serve different functions. It includes basic traffic forwarding hardware
and software. Unlike traditional network devices, IoT devices create traffic internally[33]. It
manages data routing and virtualizes resources for NFV Applications. It has a control plane
and virtualized components [34]. The Device Manager keeps track of nodes that can receive
control plane signals. The other modules will have access to the nodes' IP and MAC addresses.
It also uses device manager data to map device location and connectivity [35]. The Statistics
manager collects basic statistics from data-control plane communications. The Flow manager
keeps track of the network's data flows as well as creates bespoke control plane modules. The
NFV Applications (storage, networking, and compute) may be downloaded, installed, or
configured. These are pro applications that include NFV applications [36]. As a result of
virtualization techniques, this layer allows many user programs to use the same hardware
resources. Users may share virtualized resources to execute programs separately. Security,
QoE, and data analysis are NFV applications [37]. Separate data and control planes, network
function software, and resource virtualization all require infrastructure layer cooperation.
Because of this, the orchestration and management layer may act on other infrastructure layers.
It is in charge of VNF and control applications resources allocation. Three layers are VNF
Manager, Orchestrator, and Virtual Infrastructure Manager (VIM). VIM can handle storage,
networking, and compute resources. The Orchestrator gets these resources [24]. The VNF
architecture has three fundamental components. A physical hardware with memory, CPU, and
storehouse. A virtual hypervisor layer which is in charge of supervising resources such as the
computer system, retention, and storage capabilities and a Virtual machine that is a software
program that mimics the design and functionality of a physical platform. As a result, a single
virtual machine can host multiple VMs. The number of virtual machines that can be featured
on a virtual machine is determined by the hardware's resources [38].
Cloud platforms enable the operation of virtual networks, virtual applications, and
immersive cloud computing. Rather than using a network hypervisor, the virtual layer is loaded
into memory. In the significant transformation, the hypervisor creates VMs on legitimate
hardware. These open APIs can be used to create VNFs from higher-level languages. The
VNFs serve as the primary management software for a distinct server farm. A software
supervisor can dynamically allocate and manage resources. On the other hand, VNF-enabled
devices can be controlled centrally via an SDN network, enabling both optimization techniques
to coexist. Thus, the primary goal of next-generation interconnection is to enable remote
creation, regulatory oversight, and management of virtualized networks composed of
interconnected VNFs embedded in source code [39]. The network topology can be changed
without the need for human intervention. In response to demand, the capacity of VNFs can be
increased or decreased. This eliminates the risk of over or under-provisioning resource
management for a specific customer, allowing users to maintain the service level set out in the
contract at all times. Complex network rules can be achieved through incredibly fine network
bandwidth administration and configuration. Implement and test connection rules in real-time
without disrupting service. New network technology can be easily incorporated into the
software without incurring significant costs. The communication system must be completely
revamped to realize the concept of future knowledge networks depicted in Figure 14.3.
Integrating virtual machine instances into physical hardware will require a completely new
design. Numerous theories can be applied to required specifications and network situations by
lending from existing information. Two very important processes are regularly used in software
to construct what is widely regarded as a VNF network. These two strategies, as illustrated in
Figure 14.3, are distinct but mutually reinforcing. The foremost difference between the two
methods is that one incorporates the control plane into the physical device while the other does
not. The other method decouples the control method from the hardware device, allowing the
device to focus entirely on data transmission or another purpose for which it was designed. The
design incorporates network regulation into the physical device, adhering to hardware solutions
manufacturing practices. This is a temporary fix rather than a comprehensive rewrite of the
architectural style. The actual hardware is concealed behind a software platform that manages
the resources in this design. This application virtualization prevents applications from
accessing the physical device directly. These are known as virtual machines, that are managed
by a hypervisor in this design. An additional minor component of the software level is
concerned with cross-platform compatibility. When these two layers are combined, they enable
virtual provisioning. VNFs are created by modeling the precise functionality of the targeted
network element using virtual resources. VNFs are simply specialized network functionality
elements created virtually in software to perform the same function as highly specialized
network hardware. This method enables VNFs to make use of the hypervisor's coherently
accessible resources. The capacity of a VNF can be precisely adjusted by provisioning logically
accessible segments of hardware appropriately. The second critical layout under evaluation
recommends an overhaul of the existing manufacturing process for network devices. It does
not endorse the device's network control to be removed. As a result, it enables the generation
of overall connected devices capable of performing nothing more than message transmission.
A customized software controller will manage all network devices. Through a southbound API,
the software control systems communicate with the machines. The program controller will
expose a northbound API that will be accessible to network applications programmed to deliver
specific network functions. These computer systems will generate the necessary VNFs, which
will be managed by the software controller. The deployment is driven by controller-triggered
sequences and external inputs. External sources of information could include internet backbone
monitoring systems, Netflow, intrusion prevention systems, or BGP peers.

14. 1.3 Objective of Chapter

The researchers of this study have surveyed Edge Computing with SDN solutions with use
case scenarios. The integration of Edge Computing with SDN is being researched and extracted
to determine the future direction. The importance of virtualization in settling fundamental
issues associated with SDN and NFV in cloud technologies for IoT infrastructure services was
investigated, and an SDN-based IoT architecture was proposed that allows for better data
exploitation from IoT systems and network activities than standard SDN-based networks. In
summary, the following are the paper's original contributions:

1. Investigating the feasibility of using SDN to enhance IoT networks by creating a

powerful knowledge-driven SDN and IoT integrated architecture that employs edge
cloud technologies to bring cloud solutions proximate to the Internet of Things.
2. Augmenting the control plane using IoT-specific features, allowing control logic to
be adapted to specific Edge computing and supporting heterogeneous IoT systems.
The combination of SDN-enabled programmability and conventional virtualization
enables the IoT system to be both nimble and cognitive, eventually evolving into an
autonomous system.
3. Introducing cutting-edge technology in the domain of service migration from
centralized to distributed cloud services, as well as the difficulties posed by these
widely-scattered systems.
4. Introducing future network challenges, such as assimilating a cloud technology SDN
network well with the Internet of Things, and suggesting that by implementing a
knowledge plane into the devices, as well as a knowledge-bound API; IoT proxies
may be realized, resulting in increased data reliability through IoT devices and data.

14.1.4 Organization of Chapter

The rest of this article is organized as follows. Section II presents the issues related to
SDN cloud and NFV techniques for IoT, using use case scenarios as examples. Section III
examines and presents in detail the SDN cloud trying to combine edge computing for
Infrastructure and its proposals. Section IV focuses on the propagated awareness of SDN-based
IoT networks that ability to leverage edge cloud technology. Section V summarises the
discussion and makes future recommendations. Section VI summarises the contributions,
provides future directions, and wraps up the research.
14.2 CHALLENGES WITH SDN-BASED CLOUD AND NFV TECHNOLOGIES FOR
IoT

Next-generation connections enable remote establishment, control, and long-term

planning of virtualized networks composed of connected software-defined VNFs. The network
topology can be changed without requiring human intervention. The capacity of VNFs can be
decreased or increased to meet demand. This eradicates the concern of over or under-
provisioning resources for a specific customer, letting users always meet the contract-defined
service level. By trying to manage and arrange the network with extreme precision, large
complex norms can be accomplished. Execution and testing of communication rules in real-
time without interrupting with service. New network technology has been successfully
integrated into the software without incurring significant costs. To implement the concept of
prospective functionality networks displayed in Figure 14.3., the network infrastructure must
be completely redesigned. An entirely new design has been developed to incorporate virtual
machines into physical hardware. Borrowing from existing knowledge, multiple concepts have
been applied to current regulations and network situations. Typically, two remarkable
approaches are used in software to build a VNF element. As depicted in Figure 14.3., these two
strategies are distinct but interrelated. The fundamental difference between these two is that
one encompasses the network control into the hardware machine, whereas the other is not. The
other method relinquishes control of the hardware device, allowing it to concentrate entirely
on transmitting data or another objective for which it was designed. The design utilizes network
restriction into the physical server while adhering to industry standards for hardware
manufacturing. This is a band-aid solution, not a complete overhaul of the architecture. In this
design, the actual device is hidden behind the same piece of software that appears to be able
the physical resources. This virtualization will act as a buffer between the software and the
hardware. This is known as virtual equipment, which in this design are referred to as
hypervisors. Cross-platform compatibility is a small aspect of the application level. When these
layers are combined, it becomes possible to procure device resources virtually. VNFs are
formed by simulating an element's exact functionality using virtual resources. VNFs are merely
specialized network configuration elements that are generated virtually in the application to
replace highly specialized dedicated servers designed for the same purpose. This method
enables VNFs to access logically accessible resources on the hypervisor. A VNF's capacity can
be accurately adjusted by appropriately provisioning rationally available sections of hardware
resources. The second critical design under consideration proposes an overhaul of the current
network device manufacturing process. It does not endorse the removal of the device's control
plane. As a result, network control is decoupled from either device, allowing for the
development of overall connected devices competent of nothing other than message
transmission. All network devices will be managed by a dedicated software controller.
Additionally, the term "network operating system" is used to refer to this. The software control
mechanisms will communicate with device objects via a southbound API [33].
These network components will generate the required VNFs, which the software
controller will manage. [40]. SDN-IoT management defines how applications engage and
support each other. [41]. It enables network analysts to specify how users and SDN controllers
regulate control methods. SDN stability for the Internet of Things specifies the number of
security setups for network access, node, and a variety of other control scheduled. Recently,
the emphasis has shifted from broad SDN concepts to particular aspects of virtualization. [42].
In the new SDN virtualization paradigm, all IoT network components are merely forwarding
devices with no intelligence to manage or route data traffic. The same physical substrate may
host several virtual networks. Virtual networks form real network architecture. Infrastructure
virtualization comprises a computer, hypervisor, and network resources. [43]. SDN software
managers may dynamically assign and release resources. A central SDN controller can control
VNF-enabled devices in both topologies [44]. The number of Internet-connected devices is
growing at an alarming rate. The network infrastructure should be capable of managing the
onslaught of data. One problem is that IoT systems are distributed globally. All of these distant
devices should be connected to the network infrastructure. Any single service provider has a
tremendous task and infrastructure investment. Each service provider faces significant ongoing
expenses as new technology develops. To evolve in response to changing traffic factors like
capacity and latency. As new performance applications are attached to the network, the rules
governing security measures and service provisioning will evolve. Packet handling rules may
need to be changed, and new traffic may need high-layer processing. As a consequence, new
rules dictate moving firewalls, load balancers, and additional special-purpose gateways. In a
shared network situation, when several service providers share a network infrastructure,
reconfiguration is a problem [45].
Thus, the goal of next-generation communications networks is to enable the remote
establishment, control, and management of networks of virtual servers composed of connected
SDN VNFs. VNFs' capacity may be raised or lowered to ensure their sustainability. This
eliminates the risk of over- or under-provisioning a client's resource base, allowing users
always to maintain the service agreement. Large complex rules can be implemented through
extremely precise network management and configuration. Instantly deploy and test internet
backbone controls without affecting service. New network technologies can be seamlessly
implemented into the software without incurring significant additional costs. To implement the
notion of futuristic capability, the infrastructure must be completely redesigned. Integrating
hypervisors with physical hardware necessitates the development of an entirely new
architecture. Many of the ideas can be tailored to current requirements and network conditions
by attempting to draw on existing knowledge. Two remarkable technologies are frequently
used in applications to construct what is commonly referred to as a VNF network. The
fundamental distinction between these two strategies is that one incorporates the control plane
into the device, whereas the other does not. While the method isolates the control system from
the dedicated device, allowing the device to focus exclusively on the transmission process or
some other purpose. The architecture enables the integration of network management into the
hardware machine, consistent with established device manufacturing practices. The actual
hardware is concealed beneath a virtual machine that manages the physical resources used by
the design. This application virtualization prevents applications from gaining direct access to
the underlying hardware. This is referred to as virtual machines, and the virtual machine is
referred to as a hypervisor in this architecture. Another insignificant component at the
application level is in charge of cross-platform compatibility. When these two layers are
combined, it is possible to virtualize device resources. VNFs are established by simulating the
desired network component's exact operation using virtual resources. VNFs are virtualized
versions of specialized network functional components that perform the same functions as
highly specialized network hardware. This way, VNFs can access the hypervisor's logically
available sources. The capacity of VNFs can be precisely controlled by providing logically
available portions of physical network resources appropriately. The second crucial design
under consideration proposes a complete overhaul of the network equipment manufacturing
process. It makes no suggestions regarding the device's network control being removed. As a
result, it decouples the network control from the devices, enabling the development of generic
network equipment capable of performing nothing except data forwarding. All network devices
will be managed by a dedicated software controller. Through a southbound API, the program
controllers will communicate with physical devices. The program control system will expose
a north-bound API that network programs can use to perform specific network tasks. These
computer systems will generate the required VNFs, which will be handled by the software
controller [46].
The implementation method should make appropriate use of resource allocation policy
initiatives to address the usage issue. As a result, techniques for energy-efficient data centers
must be proposed. Service over- and under-subscription Customers are constantly compelled
to subscribe to resources with a higher priority to meet their immediate needs. The real-time
resource is more expensive. As demand for specialized real-time services increases,
particularly for specific application requests, certain data centers are likely to be over- and
underutilized. Two extremely common initiatives to resolve these issues are to use SDN-based
dynamic authorization mapping to distribute requests across data centers and to use load
balancing. Mobility of virtual machines in an unbroken chain Virtual machine migration is
always complicated in data centers, but it becomes even more complicated when multiple DCN
providers are involved. The critical feature of the Internet of Things is its seamless
connectivity. This must be accomplished in DCN via the creation of requests by VMs or
containers. SDN and NFV are great examples of technologies that enable rapid transfer while
adhering to the required quality of service and service-level agreements [47].
Cloud computing is an excellent network solution for delivering a variety of services,
including virtualization as a service, virtual network processes, platform as a service, and
implementations and software as a service. Cloud networks are intended to enable global
service, to have the theoretically infinite capacity that is adaptable to changing demand, to be
extremely resilient and secure, and to provide metered service. Infrastructure providers will
leverage the cloud environment to launch their virtual hardware resources as a service, making
them globally accessible to their clients. Network service providers will use this global
availability service to create virtual networks and VNFs for their clients, who will act as the
network operators. The network service providers will construct networks that will be used to
conduct business by their network operator clients. Network operators will be able to build
applications and software services for their end-users, who will eventually become network
service consumers. These network services may include Internet access for social networking,
the use of specialized software in their home or office environment, the acquisition of news
and information, and video conferencing. Cloud networks by definition meet the criteria for
next-generation networks, which include infrastructure resource pooling and sharing, on-
demand service, a simple interface. When dispersed data centers are connected via high-speed
connections, cloud networks are formed. These data centers are equipped with a large number
of virtualized server pools and network equipment.
Due to the scale of the infrastructure deployment, it will be cost-effective to conduct
research and development to create more powerful software virtualization methodologies and
to continually improve them over time. Due to the enormous cost savings expected from cloud
networks, all actors are constantly looking for ways to invest in this framework to improve and
expand the services offered through this model. The primary benefit of cloud networks is that
they open up enormous opportunities for improving the services delivered on a global scale
with little investment. These cloud networks make it possible to rapidly deploy pretty advanced
scientific services such as grid computing and self-driving cars. On the other hand, they provide
less-advanced users with utility-based computing. Cloud networks are focusing their efforts on
making network services more accessible across the board, from infrastructure to end-user
applications. Network topologies, both physical and logical, are optimized for redundancy and
availability. The purpose of network protocols is to ensure that data is transmitted securely and
authentically to authorized users without jeopardizing its confidentiality or authenticity. These
network protocols must transmit data within the specified period and without exceeding the
maximum delay allowed. Each network user has their own unique set of requirements. Users'
requirements are constantly changing and are not always consistent in quantity due to their
fluctuating demands.
Cloud networks enable multitenancy, or the ability to run various applications on the
same hardware and software infrastructure, by leveraging the previously mentioned software
virtualization framework. The challenging task here is to properly allocate all of the underlying
resources. Simultaneously, the cloud framework should ensure the complete isolation of
multitenant applications. The addition of new applications should have no adverse effect on
the performance of currently running applications. To that end, the cloud network must be
elastic to adapt smoothly to changing network loads. Scalability is a critical issue in today's
networks. Everyone is spending more time online and is looking for faster data transfer speeds,
increased storage capacity, and increased computational capability. Without the installation of
new network equipment, real-time processing of this massive amount of data will be
impossible. Resources are expensive and consume a great deal of energy, stressing the
environment. Rapid technological advancement is necessary to keep up with the exploding
demand that has become unstoppable in any circumstance. Solutions that are affordable,
rapidly deployable, and energy efficient are required. These networks will be constructed using
a combination of highly unstructured, semi-structured, and structured data. Each of these
instances necessitates data manipulation to extract the necessary knowledge and information
for daily life. Cloud networks built on a robust software virtualization framework are the way
to go when it comes to addressing scalability concerns and dealing with massive data
generation. As a result, our primary objective is to gain an understanding of how to manage the
massive amounts of data generated by IoT and other types of devices [48]. The aim is to design
a cloud-based software virtualization platform that is very scalable. Challenges associated with
cloud SDN and NFV technologies for IoT are tabulated in Table 14.1.
 Table 14.1: Challenges associated with SDN-based cloud and NFV technologies for IoT infrastructure

Challenges for Description References

cloud-based SDN
and NFV
technologies
SDN Scalability and The control plane's logical centralization and frequent dependence [49]
dependability on a on the SDN-specific controller occupy a single point of failure.
global scale
SDN Increasing As IoT traffic increases, user demand will never cease to exist. [50]
Demand
SDN Implementation SDN deployment is still in its infancy. [51]
SDN Management It is necessary to manage a variety of techniques, tools, and activities [29]
to guarantee excellent quality for the end-user.
SDN Features The ability to decouple the forwarding plane by the data plane results [52]
in several abstractions.
SDN Economic Energy consumption and high-security networking lead to increasing [11]
Factors costs.
SDN Protocol and SDN standardization over OpenFlow is indeed not effectively [27]
standardization scalable, consistent, or effective to handle all plausible suspects and
regulatory issues in hybrid cloud-fog.
SDN synthesis inside In the ETSI NFV variants, the intent of the SDN controller's [53]
NFV ecosystem communication with the rest of the NFV ecosystem remains unclear,
particularly for distributed contexts such as cloud-fog methods.
SDN Government Governments are expected to impose more limitations on network [54]
policy adaption: providers in the future, both in terms of co2 emissions and privacy
protection.
SDN slicing element Slicing may acknowledge a significant enabler technology during [20]
for 6G 6G.
Complete migration Complete migration to SDN infrastructure may provide several [55]
to SDN-based support difficulties, not the least of which are financial limitations.

Consistent control Having a single administrator for businesses and data centers [56]
channels for SDN necessitates the use of a centralized controller.
SDN control plane's The network managed by the SDN controller and the control plane [57]
resilience and fault is the most vulnerable aspect of any SDN system.
tolerance
SDN fault detection In an SDN application environment, the controller decides how [1]
in changing network packets dispatching mechanisms sustain a dynamic network
topologies topology.
SDN Trust and SDN raises particular concerns regarding the protection and security [46]
security alliance between SDN applications, including controllers.
SDN Efficiency High computational cost incurs due to rule enforcement, overlapping [35]
network rules, and the memory constraints abilities of OpenFlow-
enabled tools.
SDN and Machine To employ machine learning techniques to promote the efficiency [41]
Learning and dependability of SDN by utilizing real-time data obtained from
the data and control planes.
Integration of SDN Other technologies, including multi-agent systems and blockchain, [51]
alongside other may be included in the SDN-based cloud fog deployment paradigm.
technologies toward
cloud-fog systems
 In a classical SDN, each controller is in charge of a distinct domain. Typically,
controllers in a hierarchy communicate via proprietary protocols. Additionally, communication
with other network elements may be investigated to guarantee that communication is effective
and optimized. In classical SDN, a backup controller is used to avoid a central point of failure
at the control plane. Maintaining a running back controller for each controller, on the other
hand, may become cost-prohibitive, necessitating research into cost-effective redundancy
options. Another crucial field of investigation is the effective use of virtualized network
functions. The placement of a control system or other control plane elements is largely
unknown, owing to the widespread use of a network node controller. The number of control
systems and taking a good look at IoT systems in Software-defined IoT networks may
necessarily require a close inspection of the controller placement in the topology. IoT networks
will consist of thousands of devices connected via a single Virtualized domain. As a result,
controller scalability is critical. This includes not only scalable frameworks, but also
development tools, capabilities, data storage, and controller-level processing. Given the
diversity of controller types, scalability and pairing at a large scale will be an exceedingly
interesting area of research. Another fascinating challenge will be synchronizing controllers
and their regulations. Additionally, it will be intriguing to examine the domain's prerequisites
and then use only the necessary types of Software characterized controllers. Vertical versus
horizontal implementation of controllers and related virtual network functions may also
provide interesting design options. Controller virtualization is a vital part of software-defined
IoT systems. Virtualizing and coordinating the operations of multiple controllers is a complex
process. Similarly, placing virtualized components in core or edge networks will be an exciting
area of research. Numerous edge networks may exist within a single Program defined -IoT
domain, each supporting lots of unique mobile IoT nodes with a high degree of variability and
limited resources. While several solutions have been proposed to address mobility in SDNs,
addressing it in a hybrid infrastructure context using a variety of physical projects would
introduce novel research challenges. Further research should be conducted on the rapid and
reliable discovering of configurations in the portable domain, path settings, hand-over, and
other scalability issues. Configuration of the device: The edge and connect directly networks
of a software-defined Internet of Things network will be composed of heterogeneous portable
devices. Configuring them according to the application layer's guidelines is a significant issue.
Additionally, extensive research is required before developing a cohesive framework. Virtual
functions: The Software-Defined-Internet-of-Things ecosystem will heavily rely on the
virtualization of different networking services. As a result, controlling them in the control and
data planes, distributing them, virtualizing them, and integrating them with another layer upon
layer and APIs constitutes a significant area of research.

14.3. LITERATURE SURVEY

The majority of architecture designs are typical hybrid architectures that use both
technologies but emphasize NFV approaches. Several approaches discuss the architectural
problems of SDN-IoT, including its scalability, heterogeneity, fog connection; migration to
SDN, and interoperability issues. SDN may be an ideal option to overcome the limitations of
present IoT network infrastructures due to the diverse nature of the IoT. The fundamental
concept is to distinguish the control plane from the data plane, which increases the network
flexibility, openness, and programmability, as well as its usability. OpenFlow is a critical
Southbound protocol for implementing the SDN concept. SDN introduces several novel
capabilities to IoT network topologies, including network programmability, centralized
control, and flexible management. To begin, SDN has the potential to ease the management of
IoT architectures. The SDN controller offers central management of IoT devices via remote
configuration services. Second, SDN is naturally expandable, allowing for the rapid addition
of new IoT devices. SDN's abstract capabilities enable IoT applications to access data, analyze
and manage devices, and simultaneously add additional sensors and network control appliances
without exposing the underlying infrastructure details. Additionally, SDN enables dynamic
reconfiguration of network equipment and data traffic, enabling the IoT to adapt components
flexibly in response to changes in data flows. Numerous IoT applications have stringent
timeliness, reliability, and security requirements. Due to the limits of limited bandwidth and
compute resources, the Internet of Things requires traffic preloading. As a result, it must build
a platform that unifies connection, compute, storage, and applications at the edge networks. To
address these criteria, the edge computing paradigm arises. Control latency must be less than a
few milliseconds in IoT applications that demand real-time processing, such as autonomous
driving. If control servers are moved to the cloud, they will be unable to meet the stringent
control delay requirement. As a result, certain analysis and control operations must be moved
to the network's edge to achieve the low latency requirement. Integrating IoT with SDN and
Edge Computing has the potential to produce synergistic benefits. IoT applications can fully
leverage their advantages in omnipresent sensing and data collection, which could drive their
expansion. Along with reduced latency, edge computing enables dispersed management of IoT
devices. Thus, the collaboration of distributed intelligent systems on the edge can ensure the
stability of local system services in the IoT. For instance, extending an SDN to the network
edge and building a distributed control plane leads to governing IoT device mobile access.
Additionally, because significant computing resources are deployed on edge servers in the form
of virtual machines and are used by IoT. SDN controllers may manage these edge resources.
IoT data not only assists in determining the operational state of IoT devices but also opens up
several options for other IoT applications. The server stores and processes the data via a variety
of data processing frameworks. The processed data can then be used for a variety of purposes,
including data mining, offline/online analysis, anomaly detection, and smart IoT administration
and services. Recently, researchers proposed several SDN-based schemes for efficient data
collection and network flow monitoring in the context of edge networking, which can be
applied for IoT applications.

Rahman et al., 2021 [58] proposed architecture for intelligent and efficient management of
the smart industry with IoT networks during COVID-19 and demonstrated how to use SDN-
enabled layers such as data, control, and application to effectively and automatically monitor
IoT data from a remote location. Through specific performance evaluations on appropriate
simulation setup and environment, the proposed convergence between SDN and NFV provided
an efficient control mechanism for managing IoT sensor data and offered robust data
integration on the surface and the devices required for Industry 4.0 during the COVID-19
pandemic.
Isyaku et al., 2020 [59] addressed a flow monitoring technique to reduce network
communication costs A heuristic-based optimization strategy is utilized in their work to
consolidate polling requests from nodes and responses to optimize communication costs while
allowing tracking of the network's global view.
Beigi-Mohammadi et al., 2020 [60] offered a unique autonomic solution for cloud web
application administration to optimize profit At the heart of the solution is an optimization
component that evaluates both the revenue and cost models and employs a variety of run-time
optimization models to determine the optimal plan of action if adaptation is required. Our
proposed solution, which makes use of software-defined capabilities such as compute and
networking programmability, assists applications in optimizing the allocation of resources to
best suit their business objectives.
Gama et al., 2019 [61] presented a multi-tiered fog/cloud architecture to take advantage of the
augmented brain-machine interface Playing a multi-player and online brain-computer interface
activity demonstrates one of the possibilities that Fog Computing promises: real-time
engagement.
Francesca Meneghello et al., 2019 [62] published and studied several attacks against real-
world IoT devices in the literature to highlight the current security flaws in commercialized
IoT solutions and to emphasize the need for security as an important aspect of IoT system
design. They have concluded their work with a reasoned assessment of the IoT technologies
under consideration in terms of a set of relevant security qualities, including reliability,
anonymity, security, privacy, permissions, authentication, validation, resilience, and self-
organization.
Yurchenko et al., 2018[63]. created an OpenNetVM component that manages network
functions (NFs) on-demand in lightweight Microservices. Through its support for a high-
performance NFV architecture, OpenNetVM allows interoperability between NFs. Their
proposed NF management module makes optimal use of service chaining to manage packets
Bera et al., 2017 [64] published and studied several attacks against real-world IoT devices in
the literature to highlight the current security flaws in commercialized IoT solutions and to
emphasize the need for security as an important aspect of IoT system design. They have
concluded their work with a reasoned assessment of the IoT technologies under consideration
in terms of a set of relevant security qualities, including reliability, anonymity, security,
privacy, permissions, authentication, validation, resilience, and self-organization.
Maksymyuk et al., 2017 [65]presented an SDN NFV-based industrial 5G tactile Internet
solution. The system enables IIoT heterogeneity by computerizing physical network functions
into VNFs that are dynamically generated, monitored, updated, and removed through a layered
architecture with embedded intelligence with cloud support.
Antonakakis et al., 2017 [66] presented an SDN IoT architecture built on SDN-enabled
virtualized components. They used VNFs hosted on SDN/NFV edge nodes. Their framework
provides rich user context, minimal latency, guaranteed bandwidth, and rapid deployment of
IoT devices through edge devices. The MANO plane manages the network's infrastructure and
operations. Additionally, NFV enables the creation of virtual gateways, enhancing scalability,
mobility, and deployment speed. While the models presented in this study are theoretically
valid, they have not been tested or implemented.
Du et al., 2016 [67] discussed an IoT traffic management prototype that utilized context-aware
forwarding/processing. In an IoT network, scalability, security, reliability, and computational
constraints are mitigated by contextual information disseminated from sensor and application
layers. Thus, programmable switches allow MVNO software-defined data plane services.
Context-aware packet forwarding/processing and IoT device identification and connection are
provided by the IoT Gateway software on the FLARE platform.
Vilalta et al., 2016 [68]proposed an NFV edge node structure based on SDN. The edge node
is provided with an OpenFlow switch that controls the edge SDN controller. Cloud/Fog
Network Orchestrator, for practice with edge cloud and transit controllers, and Inter SDN
Orchestrator, during edge and data processing SDN controllers.
Li et al., 2015 [69]suggested a top-down strategy. It is split into the application, control, and
infrastructure layers. By centralizing control and virtualizing various IoT services, SDN and
NFV may help address issues like interoperability, heterogeneity, scalability, and security. The
authors want to continue exploring the structure and elements of each perspective of an SDN
and NFV-based IoT ecosystem in the prospect.
Boussard et al., 2015 [70] suggested a software-defined LAN-based connecting system for
heterogeneous devices in the network while managing the devices' control logic As a result,
the network's gadgets are interconnected, creating a smart environment. The authors introduced
two kinds of controller architecture: network controllers and virtual object controllers. The
network controller is analogous to the typical SDN controller.
Huang et al., 2014 [71] provided a traffic aggregation-based admission control technique in
SDN Network calculus that has been used to optimize admission control while also checking
the network's allocated buffer space and bandwidth. As a result, the proposed approach ensures
that the efficiency of previously admitted flows is not impacted by the addition of a new flow
to the network. Furthermore, the proposed technique uses less buffer space through data
aggregation.
Malboubi et al., 2014 [72] For optimal data aggregation, an SDN-based network aggregation
and measurement methodology were developed. Two optimization issues are defined: traffic
flow aggregation and de-aggregation of essential flows from the aggregated ones. Both the
aggregation and de-aggregation operations make use of ternary content addressable memory.
The critical flows are examined to effectively assess network activity using a multi-armed
bandit-based optimization approach. It is discovered that the suggested approach can measure
network behaviour using aggregation and de-aggregation approaches.
Batallé et al., 2013 [73] presented a new NFV-based virtualized routing protocol. When an
OpenFlow network system performs inter-domain forwarding, it guarantees that overhead
signalling is managed and minimized. It reduces the number of linked and deployed devices,
thus saving cost.
Patouni et al., 2013 [74] demonstrated that by employing SDN-based technologies, network
service virtualization can be accomplished for enhanced network services above traditional
invasive networking paradigms. SDN-based hypervisor techniques for sensor management
have been presented. The suggested approach discusses wireless service integration, and sensor
management while leveraging the benefits of SDN.
Zhen Liu et al., 2005 [75] presented a network flow-update technique while taking into
account available throughput and flow-table capacity restrictions. Heuristic optimization is
employed in this approach to adjust the flow-table rules efficiently.
Table 14.2: SDN-based Cloud edge technologies for IoT Infrastructure and their use cases.

Edge Use cases References

Computing
technology
Cognitive A Cloudlet-based approach enables end users to analyse sensor data. Along Ha et al., 2014
Assistance with the control and user-guiding processes, the server-side is comprised [76]
of a slew of virtual machines, each of which is responsible for a unique
sub-service that is critical to the framework's Optical Character
Recognition.
Body Area Fog Computing has a focus on healthcare applications. Fog servers collect Shi et al., 2015
Networks data to perform needed data filtering as well as data aggregation to give [77]
real-time interactivity. If ECG sensors acquire urgent data and perform
immediate processing, this allows for immediate response.
Hostile Sensors deployed in hazardous environments, such as a military base, are Yaseen et al.,
Environments subject to attack. Aside from improving response time using Fog 2016 [78]
and Military Computing, Cloud services are also useful in a military environment since
they reduce DoS susceptibility in combat operations by focusing it on a
single link hop.
Speech Multilingual processing applications that rely on speech as an input for Bahtovski et al.,
Processing translation services require a continuous Internet connection and a large 2015 [79]
pool of resources. Cloudlets can provide the necessary resources while
alleviating the strain on low-capacity mobile systems.
Smart Grid Fog computing is a foundational technology for the smart grid concept, in Peng et al.,
which energy load trying to balance services are run on edge nodes that 2018 [80]
investigate alternative energy sources based on demand and accessibility.
For business analytics research, the cloud provides worldwide coverage
and stores data for months, if not years. Fog collectors may analyse the
data they collect or may send a subset of it to higher-tier systems for
visualisation or real-time reporting.
IoT and CitySee is a monitoring system in the largest WSN environment. After Liu et al., 2013
Wireless preprocessing, the data acquired by the sensors is transmitted to the sink [81]
Sensor node, which subsequently transmits it to the Cloudlet. CitySee's goal is to
Networks provide services to users, hence Cloudlets are used inside this architecture
to create a set of APIs for consumers to acquire then use the sensed data.
Video Gabriel is a Cloudlet-based cognitive assistant application. After detecting Satyanarayanant
Streaming a wearable glass in the neighborhood, the video is transferred to the et al., 2014
and Analysis Cloudlet. The footage is then processed to provide instruction to the user. [82]
Optionally, the video can be sent to cloud servers for centralized activities
such as error reporting. It is envisaged that transferring the video analysis
process to the Cloudlet will improve the experience for users by lowering
latency below a specific threshold.
Augmented To take advantage of the augmented human-machine interface, Zao et al. Gadal &
Reality suggested a multi-tier fog/cloud architecture. Playing a multi-player and Mokhtar, 2017
online human-computer interface game demonstrates one of the [83]
advantages that Fog Computing promises i.e real-time engagement.
Gabriel implementations have the aesthetic appeal of augmented reality
while also meeting the logic and computing needs of artificial intelligence.
 Intelligent There are significant limits in vehicular networks that have a direct impact Meneguette et
Traffic Lights on the overall performance of the design, such as high speeds, unstable al., 2014 [84]
and wireless connections, and highly dynamic topology. It is difficult to
Connected achieve reduced delay while satisfying the user experience within these
Vehicles constraints. By deploying processing resources and keeping
communication between vehicles at the edge through fog technology,
performance can be improved while the inherent limits are relaxed.

Table 14.2. discusses SDN-based Cloud edge technologies for IoT Infrastructure and their use
cases

14.4. KNOWLEDGE-DRIVEN SDN-BASED IOT ARCHITECTURE THAT

LEVERAGES EDGE CLOUD

Edge Computing is a free and open platform for managing, analyzing, controlling, and
processing data at the Internet of Things' network edge. This satisfies the needs of entities that
detect and control their surroundings in terms of connectivity, computation, storage, and
application installation. Edge computing systems enable the rapid development and support of
real-time applications that rely on high-performance computing. SDN effectively virtualizes
IoT networks, enabling driverless traffic rerouting, device configurations, and bandwidth
allocation, thereby improving efficiency and simplifying management. Increased network
visibility as a consequence of automated security risk identification, enforcement of
regulations, and access management are just a few of the advantages. Through automation,
defense authentication, and status monitoring, the SDN platform allows the security control of
sensors, communication interfaces, IoT gateways, and many other devices. Figure 14.4.
describes knowledge-driven SDN-based IoT architecture that leverages edge cloud technology.

The exponential growth of internet-connected devices in recent years, as well as the

growing need for real-time computing, are driving the development of edge-computing
systems. In today's technological landscape, there has been a significant increase in
heterogeneous device connections, necessitating the need for remote device control via
dependable IoT. The Internet of Things is still in its infancy, and it is impossible to predict all
possible future scenarios. As a result, gateways are critical for implementing the required
access and communication systems and ensuring the broadest possible integration.
Additionally, gateways ensure interoperability between various types of objects and physical
systems, providing users with a true IoT experience. Additionally, as described in, gateways
may perform security functions, such as providing confidentiality and integrity services,
thereby increasing efficiency. As a result, multiple gateways will be deployed to provide
intelligence for approved entities and applications across multiple application domains. With
the proliferation of objects and routers, there will undoubtedly be a strong requirement for these
gateways to be managed and controlled from the network and information technology
infrastructure. To improve overall efficiency and decrease latency, the controller cloud located
in edge nodes may become mandatory.
 Network virtualization and SDN may eventually fall under the purview of network
operators. As demonstrated in the literature and the preceding section, combining NFV and
SDN provides infrastructure operators with an opportunity. Third parties also have direct
access to infrastructure services, which they can leverage to develop their products.
Infrastructure services can also be made available to a broad range of third-party actors,
including application developers, service providers, and even virtual operators for the Internet
of Things. As a result, gateways are critical for implementing the required access and
communication systems and ensuring the broadest possible integration. Additionally, gateways
ensure interoperability between various types of objects and physical systems, providing users
with a true IoT experience. Additionally, as described in, gateways may perform security
functions, such as providing confidentiality and integrity services, thereby increasing
efficiency. As a result, multiple gateways will be deployed to provide intelligence for approved
entities and applications across multiple application domains. With the proliferation of objects
and routers, there will undoubtedly be a strong requirement for these gateways to be managed
and controlled from the network and information technology infrastructure. To improve overall
efficiency and decrease latency, the controller cloud located in edge nodes may become
mandatory. Network virtualization and SDN may eventually fall under the purview of network
operators. As demonstrated in the literature and the preceding section, combining NFV and
SDN provides infrastructure operators with an opportunity. Third parties have direct access to
infrastructure services, which they can leverage to develop their offerings. Infrastructure
services can also be made available to a broad range of third-party actors, including application
developers, service providers, and even virtual operators for the Internet of Things. When
analyzing heterogeneous IoT data, an SDN is used to establish the existence of networked
virtualization, recollection, and connection resources throughout the operation.

Machine learning has been applied to IoT data monitoring and management activities
via cloud and edge/fog computing. One notable distinction is that knowledge-guided SDN
intelligence is shared across IoT network entities such as SDN controllers, IoT control systems,
and IoT proxies. SDNs powered by knowledge are capable of modeling across IoT devices, in
contrast to fog and cutting-edge technology powered by AI, which is limited to a single cloud
service or edge node. In comparison to traditional SDN-based communication networks, a
knowledge-driven SDN-based infrastructure makes optimal use of IoT platforms and
operational data. Additionally, it includes network control capabilities that are tailored to IoT
technology. The control system is well-equipped to deal with the heterogeneity inherent in IoT
networks. The gradual development of SDN and the interpretation of IoT endows the IoT
system with versatility and intelligence, enabling it to be self-sufficient. The management and
application-specific service planes, the network management and connectivity planes, and the
knowledge plane all contribute to the architecture. The control plane is made up of SDN and
IoT control systems. The SDN controller is in charge of optimizing network resources and
deploying new network-level services rapidly. IoT controllers are experts at deploying and
provisioning IoT services. Unlike SDN controllers, IoT controllers may be application-wide.
The network device may provide an overview of the entire network to IoT controllers. A central
or several distributed IoT control systems may also interact with an SDN controller.

Figure 14.4: Knowledge-driven SDN-based IoT Architecture leveraging edge cloud

Routers, gateway, sink, and IoT devices constitute the infrastructure. Routers and
gateway are used to connect networks. Additionally, gateways may store and manage local
data as part of controllers. Additionally, control logic and functionality add the gateways. The
knowledge plane constitutes applications and data about the global network. Controllers use
the global communications state to generate flow tables that manage data forwarding in
infrastructure extension gateways and routers. IoT devices and applications generate
geographic time-series data. The knowledge plane may use a variety of data processing,
assessment, and learning methods to extract meaningful data for use by other planes. The
management and application-specific service, control, and infrastructure planes connect the
knowledge-bound API. According to their respective objectives and duties, the three planes
may access a range of various kinds and degrees of data representation, abstractions, and
content. The knowledge plane is capable of adapting the data it collects and the methods it
uses to the requirements of the three planes. For instance, an SDN controller may receive raw
network data in real-time to set up routers, while IoT controllers ought to have access to
concise abstract representation to manage devices and plan activities. Additionally, massive
amounts of data may be sent to an IoT network operator, but just data that is relevant to that
IoT service provider. Configuration data may be delivered directly to the IoT device or
through a gateway, depending on the device's capabilities. The knowledge plane may aid the
next three planes in becoming intelligent and capable of self-adaptation to their environments.
Controllers for the control plane of the IoT that are application-specific allow the programming of
IoT applications and services. As a result, the SDN controller may significantly reduce the overall
resource usage of the SDN IoT network. Additionally, it enables the network to manage application-
specific control while retaining network-specific information, resolving the resource utilization
issues caused by IoT application isolation, and lack of improved network resource utilization
flexibility. Networks and applications may exchange system and application knowledge by using
the knowledge-based API. This increases the knowledge and autonomy of decision-making planes
and adapts the network to its environment deciding the behavior of IoT applications. Application-
specific functionality may also be implemented at the network level, improving network flexibility
and reducing service delivery time. These stages may be carried out via the use of several data
analytics techniques and algorithms. Machine learning methods may be appropriate for IoT data that
is dynamic. Models are updated continuously to reduce storage essentials and respond to new
applications. Additionally, decentralized deep learning and distributed refinement have the potential
to improve model accuracy.
SDN-based Cloud combining edge computing for IoT Infrastructure consists of the IoT
device layer, the edge layer, and the cloud layer. Various smart devices and systems are linked
to heterogeneous entry points, which are linked to Switches and routers to request various
services. Local SDN controllers are linked to the OpenFlow switches. The southbound protocol
aided by OpenFlow switches allows IoT devices to communicate with SDN controllers. For
heterogeneous network devices and access methods, the southbound interface protocols need
to be extended to support smarter and more friendly multi-network. A three-tier SDN-based
Cloud combining edge computing for IoT Infrastructure access can collect data which are then
transmitted to edge servers to be processed. The SDN controllers are deployed close to the
network edge, which divides the network into several partitions. The controllers can centrally
manage the IoT devices within their partitions. The resources of storage, computing, and
communication are also available in the edge server rather than all in the cloud. The computing
resource typically lands on the edge server in the form of a virtual container. The controller can
also support the unified management functions of various VMs, including image file management,
resource management, and life cycle management. Besides, the edge server can support third-party
applications or platforms to perform specific data analysis and processing tasks on the IoT data. For
example, we can deploy the TensorFlow platform, which significantly facilitates the realization of
machine learning on IoT data analysis. The processing results are consumed by applications and
may also be transmitted to the cloud for offline batch processing. Although the management of IoT
in the cloud server is weakened, it still has powerful computing resources and centralized control on
the distributed edge servers including SDN controllers. The cloud server can provide offline analysis
and support non-real-time applications. Besides, the cloud server sends decisions and configurations
to the edge servers and handles unsolved tasks uploaded by them. Data collection and sensing The
data generated by IoT devices need to be transmitted to edge servers.

JSON is used to encapsulate the logic of IoT services. To implement the SDN southbound interface,
we propose using OpenFlow V1.5. The OpenFlow protocol is based on the TCP protocol. TCP 6633
is the standard protocol for OF V1.0, and TCP 6653 is the standard protocol for OF V1.3+. To
establish an OF connection, IP connectivity between the controller and the switches is required.
Only after a successful TCP 3-way handshake is an OF channel formed. To begin the OF channel
communication, the switch sends a "HELLO" packet to the controller. The switch also transmits
information such as the highest version of OF that it supports. The controller responds to the hello
message with the most recent OF version it supports. The switch then negotiates the highest level
of OpenFlow version support that they both support.

The controller sends a "FEATURE REQUEST" message once the version has been agreed
upon. This message asks for the switch's supported OF capabilities, such as the number of supported
flow tables, actions, and so on. The switch responds with a "FEATURE REPLY" message that lists
all of its capabilities as well as its Datapath ID. After this, it is said that the OpenFlow channel is
successfully established between the switch and the controller. The connection between the
controller and switch is essential as it is the only way for a switch to communicate with a controller.
Rather than a TCP connection, a protocol such as TLS can be used to secure this connection. For a
viable TLS connection, the control system and switch must have the appropriate certificates and
keys. This prevents the OF channel from being snooped on.

Flow tables are used to hold flow entries that instruct the SDN switch on how to handle a
packet that arrives at an incoming port. The switch will compare specific parameters such as IP
address, destination port, MAC address, and VLAN ID and will decide the correct matching
flow from the table and perform the associated action. The packet may be dropped, forwarded to a
different port, flooded, or sent to the controller for further inspection. If a switch doesn't get access
to a particular packet, it may have a default or "TABLE MISS" entry. This arrival has the least
concern, and the packet may be dropped or sent to the controller. When the control system receives
this type of payload from a transition, it forwards it to the program running at the protocol stack,
which practices it and informs the controller whether a new flow entry should be added to the
switch's flow table. When this occurs, the controller creates a flow on the switch. The switch will
deal with another packet of the same type at the data layer, as it has access, and adequate actions
will be taken.To support the proposed Architecture, we recommend the architecture's major modules
and perform performance analysis on them. Modules may be installed on both physical and virtual
devices. A test network could be established and several test facilities could be run in it, with
performance evaluations performed on virtual machines involves in the implementation. The control
system could be incorporated using POX and an OpenFlow framework. On POX, a forwarding
feature can be implemented that configures the flow tables in routers in response to instructions
from IoT control plane applications, as well as the data storage and caching in gateways and data
processing and storage centers. The access points could be built on top of Open. The gateways can
store the collected data and process it according to the methodologies compiled by the IoT
developers. The processed data can then be distributed to the routes on demand. We recommend
using Open vSwitch 2.16.1 in implementation and tests. Gateways and routers both can be equipped
with almost the same functions. JSON is used to implement the SDN northbound data exchange.
JSON is used to encapsulate the logic of the IoT services.

14.5. DISCUSSION AND FUTURE RECOMMENDATION

Knowledge-based SDN-IoT provides a resilient and intelligent deployment approach for IoT
services. A knowledge-based API will be required for the knowledge plane. For the control,
and infrastructure planes, varying degrees of knowledge abstraction may be needed. Thus,
APIs that are knowledge-based should be categorized. Additionally, the northern and southern
bound APIs should be extended to include IoT controllers. A novel southbound API is needed
to address the limits of IoT devices as well as the rise of programmability. For efficient data
exchange, intra-plane communications between IoT controllers, SDN controllers, and IoT
proxies should be investigated. Critical IoT performance parameters like latency,
performance, scalability, jitter, etc. must be considered by SDN. The availability and
reliability of services are essential and the combination of SDN with IoT introduces additional
security issues that need to be addressed. This increases operational overhead. The optimal
trade-off is application-dependent. Knowledge plane methods and algorithms should be
investigated, taking into consideration IoT characteristics. For instance, application-specific
analysis of data and learning algorithms. When assessing traffic and performance, large
amounts of heterogeneous data streams should be carefully addressed. While IoT proxies may
greatly simplify the deployment of IoT collective knowledge, its size and breadth should be
carefully managed. Additionally, new problems for network management may arise as a result
of the introduction of IoT controllers and the knowledge plane. The efficiency with which
worldwide network maintenance and policies control may be accomplished should be
investigated further. The following developments and industries seem to be complementary
to SD-Cloud. Reorganizations of companies exacerbate the complexity of new security,
regulation, and auditing requirements. Employees increasingly use business applications
through personal devices, while information technology departments must safeguard firm
information and assets. The data is sent to the end user's system through numerous servers
located in different corporate data centers. This increases machine-to-machine
communication between connected devices. Numerous computers exchanging big datasets
necessitate increased bandwidth.

14.6. CONCLUSION

SDN may significantly improve the scalability and knowledge of IoT management and
management by using the data produced by IoT applications. This chapter provides in-depth
research on the evolution of SDN and IoT and highlights the privileges and drawbacks of
integrating SDN and IoT within IoT data. By considering a knowledge-based SDN-managed
IoT architecture, authors have assessed the opportunities for knowledge-based SDN to
intensify IoT in terms of versatility and application-specific performance of SDN IoT
networks. This chapter provides a detailed overview of existing SDN-based technologies in
the context of IoT applications, intending to provide users with seamless, cost-effective, and
dependable service delivery. Edge networking and access networking are two of the
networking aspects of IoT that are discussed. This survey indicates that the use of SDN-based
solution strategies in IoT applications is potentially useful in meeting the requirements for
establishing an IoT environment, even though there are several constraints to supporting the
network's massive connections. Various forms of implementation and IoT provider control
can be improved by combining a knowledge plane into frameworks, and often a knowledge-
bound API and Cloud computing proxy, and systems can be efficiently used in the same way
as other kinds of network’s resources. Other issues with IoT systems, such as shared
knowledge and distinguishable deployment islands, are simple to resolve. Furthermore, the
characteristics of knowledge-based SDN-based IoT, as well as its connection to the Smart
cloud and computing capabilities, were discussed. Finally, we discussed the challenges and
possible areas for future research associated with improving IoT via knowledge-based SDN.

14.7. REFERENCES

1. Al-mansoori, A.: A Survey on Big Data Stream Processing in SDN Supported Cloud
Environment. (2018). https://doi.org/10.1145/3167918.3167924.
2. Badotra, S., Panda, S.N.: A review on software-defined networking enabled iot cloud
computing. IIUM Eng. J. 20, 105–126 (2019).
https://doi.org/10.31436/iiumej.v20i2.1130.
3. Souri, A., Norouzi, M., Asghari, P., Rahmani, A.M., Emadi, G.: A systematic literature
review on formal verification of software-defined networks. Trans. Emerg.
Telecommun. Technol. 31, 1–23 (2020). https://doi.org/10.1002/ett.3788.
4. Cui, L.: When big data meets software-defined networking: SDN for big data and big
data for SDN. (2020). https://doi.org/10.1109/MNET.2016.7389832.
5. Tamanna, T., Fatema, T., Saha, R.: SDN , A Research on SDN Assets and Tools to
Defense DDoS Attack in Cloud Computing Environment. 1670–1674 (2017).
6. Zeng, D., Gu, L., Pan, S., Guo, S.: Software Defined Systems - Sensing, Communication
and Computation. (2020). https://doi.org/10.1007/978-3-030-32942-6.
7. Evers, B.: The Evolution And Benefits Of IoT, SDN And Edge Computing, SD-IoT,
data center network%0A2.3.1 Efficient flow handling%0AIn the data center network
there are two types of flows,%0Along-lived and short-lived, which are also known as
the%0Aelephant- and mice-flows respectively. It is necessary to%0Ahandle these tw,
last accessed 2021/10/14.
8. Belgaum, M.R., Musa, S., Alam, M., Mazliham, M.S.: Integration challenges of
Artificial Intelligence in Cloud Computing, Internet of Things and Software-defined
networking. MACS 2019 - 13th Int. Conf. Math. Actuar. Sci. Comput. Sci. Stat. Proc.
(2019). https://doi.org/10.1109/MACS48846.2019.9024828.
9. Cui, L., Yu, F.R., Yan, Q.: When Big Data Meets Software-Defined Networking: SDN
for Big Data and Big Data for SDN. 58–65 (2016).
10. Alenezi, M., Almustafa, K., Meerja, K.A.: Cloud based SDN and NFV architectures for
IoT infrastructure. Egypt. Informatics J. 20, 1–10 (2019).
https://doi.org/10.1016/j.eij.2018.03.004.
11. Polat, H., Polat, O.: Detecting DDoS Attacks in Software-Defined Networks Through
Feature Selection Methods and Machine Learning Models. (2020).
12. Edge computing platform solutions for 5G, https://www.ibm.com/in-en/cloud/edge-
computing?utm_content=SRCWW&p1=Search&p4=43700055270654453&p5=b&gcl
id=CjwKCAjw2bmLBhBREiwAZ6ugo-kh49b3m3-C-
djGx6Ublmd3D4J1nKHpfuJ5URzUEah6P3H-
s9cDnBoC5goQAvD_BwE&gclsrc=aw.ds, last accessed 2021/10/19.
13. Baktir, A.C., Ozgovde, A., Ersoy, C.: How Can Edge Computing Benefit from Software-
Defined Networking: A Survey, Use Cases, and Future Directions. IEEE Commun.
Surv. Tutorials. 19, 2359–2391 (2017). https://doi.org/10.1109/COMST.2017.2717482.
14. Govindarajan, K., Meng, K.C., Ong, H.: A literature review on Software-Defined
Networking (SDN) research topics, challenges and solutions. 2013 5th Int. Conf. Adv.
Comput. ICoAC 2013. 293–299 (2014). https://doi.org/10.1109/ICoAC.2013.6921966.
15. Sivanathan, A., Gharakheili, H.H., Loi, F., Radford, A., Wijenayake, C.: Classifying IoT
Devices in Smart Environments Using Network Traffic Characteristics.
16. Horvath, R., Nedbal, D., Stieninger, M.: A Literature Review on Challenges and Effects
of Software Defined Networking. Procedia Comput. Sci. 64, 552–561 (2015).
https://doi.org/10.1016/j.procs.2015.08.563.
17. Shamim, S., Miah, M., Sarker, A., Bahar, A., Sarker, A.: Simulation of Minimum Path
Estimation in Software Defined Networking Using Mininet Emulator. Br. J. Math.
Comput. Sci. 21, 1–8 (2017). https://doi.org/10.9734/bjmcs/2017/30609.
18. Lv, Z., Xiu, W.: Interaction of Edge-Cloud Computing Based on SDN and NFV for Next
Generation IoT. IEEE Internet Things J. 7, 5706–5712 (2020).
https://doi.org/10.1109/JIOT.2019.2942719.
19. Muñoz, R., Vilalta, R., Yoshikane, N., Casellas, R., Martínez, R., Tsuritani, T., Morita,
I.: Integration of IoT, Transport SDN, and Edge/Cloud Computing for Dynamic
Distribution of IoT Analytics and Efficient Use of Network Resources. J. Light.
Technol. 36, 1420–1428 (2018). https://doi.org/10.1109/JLT.2018.2800660.
20. Ahvar, E., Ahvar, S., Raza, S.M., Manuel Sanchez Vilchez, J., Lee, G.M.: Next
Generation of SDN in Cloud-Fog for 5G and Beyond-Enabled Applications:
Opportunities and Challenges. Network. 1, 28–49 (2021).
https://doi.org/10.3390/network1010004.
21. Bhatia, J., Dave, R., Bhayani, H., Tanwar, S., Nayyar, A.: SDN-based real-time urban
traffic analysis in VANET environment. Comput. Commun. 149, 162–175 (2020).
https://doi.org/10.1016/j.comcom.2019.10.011.
22. Verma, J., Bhandari, A., Singh, G.: Review of existing data sets for network intrusion
detection system. 9, 3849–3854 (2020).
23. Khan, S., Ali, M., Sher, N., Asim, Y., Naeem, W., Kamran, M.: Software-Defined
Networks (SDNs) and Internet of Things (IoTs): A Qualitative Prediction for 2020. Int.
J. Adv. Comput. Sci. Appl. 7, 385–404 (2016).
https://doi.org/10.14569/ijacsa.2016.071151.
24. Snehi, M., Bhandari, A.: Vulnerability retrospection of security solutions for software-
defined Cyber–Physical System against DDoS and IoT-DDoS attacks. Comput. Sci.
Rev. 40, 100371 (2021). https://doi.org/10.1016/j.cosrev.2021.100371.
25. Oladunjoye, O.: Software Defined Networking– The Emerging Paradigm To Computer
Networking. 38 (2017).
26. Bera, S., Member, S., Misra, S., Member, S., Vasilakos, A. V, Member, S.: Fo r p er s
on e on Fo r p. 1–15.
27. Alam, I., Sharif, K., Li, F., Latif, Z., Karim, M.M., Biswas, S., Nour, B., Wang, Y.: A
Survey of Network Virtualization Techniques for Internet of Things Using SDN and
NFV. ACM Comput. Surv. 53, (2020). https://doi.org/10.1145/3379444.
28. Snehi, J., Bhandari, A., Snehi, M., Tandon, U., Baggan, V.: Global Intrusion Detection
Environments and Platform for Anomaly-Based Intrusion Detection Systems. 817–831
(2021). https://doi.org/10.1007/978-981-16-0733-2_58.
29. Mijumbi, R., Serrat, J., Gorricho, J.L., Bouten, N., De Turck, F., Boutaba, R.: Network
function virtualization: State-of-the-art and research challenges. IEEE Commun. Surv.
Tutorials. 18, 236–262 (2016). https://doi.org/10.1109/COMST.2015.2477041.
30. Szabó, R., Kind, M., Westphal, F.J., Woesner, H., Jocha, D., Császar, A.: Elastic
network functions: Opportunities and challenges. IEEE Netw. 29, 15–21 (2015).
https://doi.org/10.1109/MNET.2015.7113220.
31. Verma, J., Bhandari, A., Singh, G.: Review of existing data sets for network intrusion
detection system, (2020). https://doi.org/10.37418/amsj.9.6.64.
32. Nadeau, T.D.: What are the hard (and interes-ng) open research problems in the SDN
Space?
33. Datta, P., Sharma, B.: A survey on IoT architectures, protocols, security and smart city
 based applications. 8th Int. Conf. Comput. Commun. Netw. Technol. ICCCNT 2017.
(2017). https://doi.org/10.1109/ICCCNT.2017.8203943.
34. Caraguay, Á.L.V., Ludeña-González, P.J., Tandazo, R.V.T., López, L.I.B.: SDN/NFV
architecture for IoT networks. WEBIST 2018 - Proc. 14th Int. Conf. Web Inf. Syst.
Technol. 425–429 (2018). https://doi.org/10.5220/0007234804250429.
35. Omnes, N., Bouillon, M., Fromentoux, G., Le Grand, O.: A programmable and
virtualized network & IT infrastructure for the internet of things: How can NFV & SDN
help for facing the upcoming challenges. 2015 18th Int. Conf. Intell. Next Gener.
Networks, ICIN 2015. 64–69 (2015). https://doi.org/10.1109/ICIN.2015.7073808.
36. Jyoti, S., Manish, S., Rupali, G.: Virtualization as an Engine to Drive Cloud Computing
Security Virtualization as an Engine to Drive Cloud. (2020).
37. Snehi, M.: Security Management in SDN Using Fog Computing: A Survey. In:
Strategies for e-Service, e-Governance, and Cyber Security. CRC Press (2020).
38. Badotra, S., Panda, S.N.: SNORT based early DDoS detection system using
Opendaylight and open networking operating system in software defined networking.
Cluster Comput. 0123456789, (2020). https://doi.org/10.1007/s10586-020-03133-y.
39. Molina, E., Jacob, E.: Software-defined networking in cyber-physical systems: A
survey. Comput. Electr. Eng. 66, 407–419 (2018).
https://doi.org/10.1016/j.compeleceng.2017.05.013.
40. Dataset, U., Jing, D., Chen, H.: SVM Based Network Intrusion Detection for the. 1–4
(2019).
41. Wang, A., Zha, Z., Guo, Y., Chen, S.: Software-Defined Networking Enhanced Edge
Computing: A Network-Centric Survey. Proc. IEEE. 107, 1500–1519 (2019).
https://doi.org/10.1109/jproc.2019.2924377.
42. Zhao-hui, M., Li, Z.G., Ze, W.M., Chen, W.X.: Research on DDoS Attack Detection in
Software Defined Network. 2018 Int. Conf. Cloud Comput. Big Data Blockchain. 1–6.
43. Mantur, B., Desai, A., Nagegowda, K.S.: Emerging Research in Computing,
Information, Communication and Applications. Emerg. Res. Comput. Information,
Commun. Appl. (2015). https://doi.org/10.1007/978-81-322-2550-8.
44. Baggan, V., Sarangi, P.K., Prasad, D., Snehi, J.: Augmenting border gateway protocol
with multi-protocol label switching for enhancing network path restoration. Proc. 2020
9th Int. Conf. Syst. Model. Adv. Res. Trends, SMART 2020. 306–309 (2020).
https://doi.org/10.1109/SMART50582.2020.9337076.
45. Abubakar, A., Pranggono, B.: Machine learning based intrusion detection system for
software defined networks. Proc. - 2017 7th Int. Conf. Emerg. Secur. Technol. EST
2017. 138–143 (2017). https://doi.org/10.1109/EST.2017.8090413.
46. Dang, V.T., Huong, T.T., Thanh, N.H., Nam, P.N., Thanh, N.N., Marshall, A.: SDN-
based SYN proxy - A solution to enhance performance of attack mitigation under TCP
SYN Flood. Comput. J. 62, 518–534 (2019). https://doi.org/10.1093/comjnl/bxy117.
47. Kumar, G.D., Science, A., Venkata, C., Rao, G.: Leveraging Big Data Analytics for
Real-time DDoS Attacks Detection in SDN. (2018). https://doi.org/10.18231/2454-
9150.2018.0287.
48. McKeown, N., Anderson, T., Balakrishnan, H., Parulkar, G., Peterson, L., Rexford, J.,
Shenker, S., Turner, J.: OpenFlow: Enabling Innovation in Campus Networks.
SIGCOMM Comput. Commun. Rev. 38, 69–74 (2008).
https://doi.org/10.1145/1355734.1355746.
49. Kreutz, D., Ramos, F.M.V., Verissimo, P.E., Rothenberg, C.E., Azodolmolky, S., Uhlig,
S.: Software-defined networking: A comprehensive survey. Proc. IEEE. 103, 14–76
(2015). https://doi.org/10.1109/JPROC.2014.2371999.
50. Sooraj, V.H., Krishnan, P.: SDN based Intrusion Detection System for OpenStack
 Cloud. Int. J. Innov. Technol. Explor. Eng. 8, 2443–2449 (2019).
https://doi.org/10.35940/ijitee.i8929.078919.
51. Scott-Hayward, S., O’Callaghan, G., Sezer, S.: SDN security: A survey. SDN4FNS
2013 - 2013 Work. Softw. Defin. Networks Futur. Networks Serv. (2013).
https://doi.org/10.1109/SDN4FNS.2013.6702553.
52. Tang, T.A., Mhamdi, L., McLernon, D., Zaidi, S.A.R., Ghogho, M.: Deep learning
approach for Network Intrusion Detection in Software Defined Networking. Proc. - 2016
Int. Conf. Wirel. Networks Mob. Commun. WINCOM 2016 Green Commun. Netw.
258–263 (2016). https://doi.org/10.1109/WINCOM.2016.7777224.
53. ETSI - Standards for NFV - Network Functions Virtualisation | NFV Solutions.
54. Miller, S.T., Busby-Earle, C.: Multi-perspective machine learning a classifier ensemble
method for intrusion detection. In: ACM International Conference Proceeding Series.
pp. 7–12. Association for Computing Machinery (2017).
https://doi.org/10.1145/3036290.3036303.
55. Huawei Software Defined Network (SDN) Solution.
56. He, D., Chan, S., Ni, X., Guizani, M.: Software-Defined-Networking-Enabled Traffic
Anomaly Detection and Mitigation. IEEE Internet Things J. 4, 1890–1898 (2017).
https://doi.org/10.1109/JIOT.2017.2694702.
57. Software Defined Networking (SDN) Market Research Report- Global Forecast 2023.
58. Rahman, A., Chakraborty, C., Anwar, A., Karim, M.R., Islam, M.J., Kundu, D.,
Rahman, Z., Band, S.S.: SDN–IoT empowered intelligent framework for industry 4.0
applications during COVID-19 pandemic. Cluster Comput. 4, (2021).
https://doi.org/10.1007/s10586-021-03367-4.
59. Isyaku, B., Mohd Zahid, M.S., Bte Kamat, M., Abu Bakar, K., Ghaleb, F.A.: Software
Defined Networking Flow Table Management of OpenFlow Switches Performance and
Security Challenges: A Survey. Futur. Internet. 12, 147 (2020).
https://doi.org/10.3390/fi12090147.
60. Beigi-Mohammadi, N., Shtern, M., Litoiu, M.: Adaptive Load Management of Web
Applications on Software Defined Infrastructure. IEEE Trans. Netw. Serv. Manag. 17,
488–502 (2020). https://doi.org/10.1109/TNSM.2019.2948969.
61. Gama, E.S., Immich, R., Bittencourt, L.F.: Towards a multi-Tier fog/cloud architecture
for video streaming. Proc. - 11th IEEE/ACM Int. Conf. Util. Cloud Comput.
Companion, UCC Companion 2018. 13–14 (2019). https://doi.org/10.1109/UCC-
Companion.2018.00022.
62. Meneghello, F., Calore, M., Zucchetto, D., Polese, M., Zanella, A.: IoT: Internet of
Threats? A Survey of Practical Security Vulnerabilities in Real IoT Devices. IEEE
Internet Things J. 6, 8182–8201 (2019). https://doi.org/10.1109/JIOT.2019.2935189.
63. Yurchenko, M., Cody, P., Coplan, A., Kennedy, R., Wood, T., Ramakrishnan, K.K.:
OpenNetVM. 1–2 (2018). https://doi.org/10.1145/3185467.3190786.
64. Bera, S., Misra, S., Vasilakos, A. V.: Software-Defined Networking for Internet of
Things: A Survey. IEEE Internet Things J. 4, 1994–2008 (2017).
https://doi.org/10.1109/JIOT.2017.2746186.
65. Maksymyuk, T., Brych, M., Dumych, S., Satria, D., Jo, M.: An IoT based monitoring
framework for software defined 5G mobile networks. Proc. 11th Int. Conf. Ubiquitous
Inf. Manag. Commun. IMCOM 2017. 7–10 (2017).
https://doi.org/10.1145/3022227.3022331.
66. Antonakakis, M., April, T., Bailey, M., Bernhard, M., Arbor, A., Bursztein, E., Cochran,
J., Durumeric, Z., Halderman, J.A., Arbor, A., Invernizzi, L., Kallitsis, M., Network, M.,
Ma, Z., Mason, J., Menscher, D., Seaman, C., Sullivan, N., Thomas, K., Zhou, Y.,
Antonakakis, M., April, T., Bailey, M., Bernhard, M., Bursztein, E., Cochran, J.,
 Durumeric, Z., Halderman, J.A., Invernizzi, L., Kallitsis, M., Kumar, D., Lever, C., Ma,
Z., Mason, J., Menscher, D., Seaman, C., Sullivan, N., Thomas, K., Zhou, Y.:
Understanding the Mirai Botnet. USENIX Secur. 1093–1110 (2017).
https://doi.org/10.1016/j.religion.2008.12.001.
67. Du, P., Putra, P., Yamamoto, S., Nakao, A.: A context-aware IoT architecture through
software-defined data plane. Proc. - 2016 IEEE Reg. 10 Symp. TENSYMP 2016. 315–
320 (2016). https://doi.org/10.1109/TENCONSpring.2016.7519425.
68. Vilalta, R., Ciungu, R., Mayoral, A., Casellas, R., Martinez, R., Pubill, D., Serra, J.,
Munoz, R., Verikoukis, C.: Improving security in internet of things with software
defined networking. 2016 IEEE Glob. Commun. Conf. GLOBECOM 2016 - Proc.
(2016). https://doi.org/10.1109/GLOCOM.2016.7841889.
69. Li, J., Altman, E., Touati, C., Li, J., Altman, E., Touati, C.: A General SDN-based IoT
Framework with NVF Implementation To cite this version : A General SDN-based IoT
Framework with NVF Implementation. (2015).
70. Boussard, M., Bui, D.T., Ciavaglia, L., Douville, R., Pallec, M. Le, Sauze, N. Le, Noirie,
L., Papillon, S., Peloso, P., Santoro, F.: Software-Defined LANs for Interconnected
Smart Environment. Proc. - 2015 27th Int. Teletraffic Congr. ITC 2015. 219–227 (2015).
https://doi.org/10.1109/ITC.2015.33.
71. Huang, J., He, Y., Duan, Q., Yang, Q., Wang, W.: 07036969. 1182–1186 (2014).
72. Malboubi, M., Wang, L., Chuah, C.N., Sharma, P.: Intelligent SDN based traffic
(de)Aggregation and Measurement Paradigm (iSTAMP). Proc. - IEEE INFOCOM.
934–942 (2014). https://doi.org/10.1109/INFOCOM.2014.6848022.
73. Batallé, J., Riera, J.F., Escalona, E., García-Espín, J.A.: On the implementation of NFV
over an OpenFlow infrastructure: Routing function virtualization. SDN4FNS 2013 -
2013 Work. Softw. Defin. Networks Futur. Networks Serv. (2013).
https://doi.org/10.1109/SDN4FNS.2013.6702546.
74. Patouni, E., Merentitis, A., Panagiotopoulos, P., Glentis, A., Alonistioti, N.: Network
virtualisation trends: Virtually anything is possible by connecting the unconnected.
SDN4FNS 2013 - 2013 Work. Softw. Defin. Networks Futur. Networks Serv. (2013).
https://doi.org/10.1109/SDN4FNS.2013.6702545.
75. Zhen Liu, Bridges, S.M., Vaughn, R.B.: Combining static analysis and dynamic learning
to build accurate intrusion detection models. 164–177 (2005).
https://doi.org/10.1109/iwia.2005.6.
76. Ha, K., Chen, Z., Hu, W., Richter, W., Pillai, P., Satyanarayanan, M.: Towards wearable
cognitive assistance. MobiSys 2014 - Proc. 12th Annu. Int. Conf. Mob. Syst. Appl. Serv.
68–81 (2014). https://doi.org/10.1145/2594368.2594383.
77. Shi, Y., Ding, G., Wang, H., Eduardo Roman, H., Lu, S.: The fog computing service for
healthcare. 2015 2nd Int. Symp. Futur. Inf. Commun. Technol. Ubiquitous Heal. Ubi-
HealthTech 2015. 70–74 (2015). https://doi.org/10.1109/Ubi-
HealthTech.2015.7203325.
78. Yaseen, Q., Albalas, F., Jararweh, Y., Al-Ayyoub, M.: A fog computing based system
for selective forwarding detection in mobile wireless sensor networks. Proc. - IEEE 1st
Int. Work. Found. Appl. Self-Systems, FAS-W 2016. 256–262 (2016).
https://doi.org/10.1109/FAS-W.2016.60.
79. Bahtovski, A., Zdravkova, K., Gusev, M.: Performance of cloudlet-based multilingual
dictionary. ACM Int. Conf. Proceeding Ser. 02-04-Sept, (2015).
https://doi.org/10.1145/2801081.2801092.
80. Peng, K., Leung, V.C.M., Xu, X., Zheng, L., Wang, J., Huang, Q.: A survey on mobile
edge computing: Focusing on service adoption and provision. Wirel. Commun. Mob.
Comput. 2018, (2018). https://doi.org/10.1155/2018/8267838.
 81. Liu, Y., Mao, X., He, Y., Liu, K., Gong, W., Wang, J.: CitySee: Not only a wireless
sensor network. IEEE Netw. 27, 42–47 (2013).
https://doi.org/10.1109/MNET.2013.6616114.
82. Satyanarayanant, M., Chent, Z., Hat, K., Hut, W., Richtert, W., Pillai, P.: Cloudlets : at
the Leading Edge of Mobile-Cloud Convergence ( Invited Paper ). 6th Int. Conf. Mob.
Comput. Appl. Serv. 1–9 (2014).
83. Gadal, S.M.A.M., Mokhtar, R.A.: Anomaly detection approach using hybrid algorithm
of data mining technique. Proc. - 2017 Int. Conf. Commun. Control. Comput. Electron.
Eng. ICCCCEE 2017. (2017). https://doi.org/10.1109/ICCCCEE.2017.7867661.
84. Meneguette, R.I., Maia, G., Madeira, E.R.M., Loureiro, A.A.F., Villas, L.A.: Autonomic
data dissemination in highway Vehicular Ad Hoc Networks with diverse traffic
conditions. Proc. - IEEE Symp. Comput. Commun. (2014).
https://doi.org/10.1109/ISCC.2014.6912549.

View publication stats