![MEC in NFV Architecture [3]](https://www.researchgate.net/profile/Madhusanka-Liyanage/publication/351303283/figure/fig1/AS:1019653715406856@1620115798558/MEC-in-NFV-Architecture-3_Q320.jpg)
Content uploaded by Madhusanka Liyanage
Author content
All content in this area was uploaded by Madhusanka Liyanage on May 04, 2021
Content may be subject to copyright.
1
Driving Forces for Multi-Access Edge Computing
(MEC) IoT Integration in 5G
Madhusanka Liyanage, Pawani Porambage, Aaron Yi Ding, Anshuman Kalla
Abstract—The emergence of Multi-Access Edge Comput-
ing (MEC) technology aims to extend cloud computing capabili-
ties to the edge of the wireless access networks, i.e., closer to the
end-users. Thus, MEC-enabled 5G wireless systems are envisaged
to offer real-time, low-latency, and high-bandwidth access to the
radio network resources. Thus, MEC allows network operators
to open up their networks to a wide range of innovative services,
thereby giving rise to a brand-new ecosystem and a value chain.
Furthermore, MEC as an enabling technology will provide new
insights into coherent integration of Internet of Things (IoT)
in 5G wireless systems. In this context, this paper expounds the
four key technologies, including Network Function Virtualization
(NFV), Software Defined Networking (SDN), Network Slicing
and Information Centric Networking (ICN), that will propel and
intensify the integration of MEC IoT in 5G networks. Moreover,
our goal is to provide the close alliance between MEC and these
four driving technologies in the 5G IoT context and to identify
the open challenges, future directions, and concrete integration
paths.
Index Terms—MEC, IoT, 5G, SDN, NFV, Network Slicing, ICN
I. INTRODUCTION
Internet of Things (IoT) is a thriving ecosystem comprising
of massive interconnections of the exponentially increasing
number of heterogeneous and resource-constrained physical
objects. With its full swing, IoT is geared-up to revolutionize
the way we perceive and interact with the world around us.
Currently, it supports the myriads of application areas such as
healthcare, agriculture, smart cities, automotive, and industries.
In the context of IoT applications following is worth to note.
On the one hand, the increasing number of IoT applications are
designed in such a way that, for data processing and storing,
they need to access the centralized cloud computing facility
[1]. This is because IoT devices are intrinsically resource-
constrained devices (i.e. low battery power, low memory
footprints and less processing power). On the other hand,
IoT is expected to offer real-time scalable applications with
minimal latency and high Quality of Experience (QoE) when
required. Thus there is an evident mismatch between the
implementation and the expectation.
5G, as an underlying technology, has an indispensable role
to play for advancements of numerous other technologies and
Madhusanka Liyanage is with School of Computer Science, Univer-
sity College Dublin, Ireland and the Center for Wireless Communica-
tions, University of Oulu, Finland. e-mail:madhusanka@ucd.ie and madhu-
sanka.liyanage@oulu.fi
Pawani Porambage is with the Center for Wireless Communications,
University of Oulu, Finland. e-mail:pawani.porambage@oulu.fi
Aaron Yi Ding is with the Department of Engineering Systems and Services,
Delft University of Technology, Netherlands. e-mail: aaron.ding@tudelft.nl
Anshuman Kalla is with the Centre for Wireless Communications, Univer-
sity of Oulu, email: anshuman.kalla@ieee.org and anshuman.kalla@oulu.fi
services, IoT being one of them. In general, 5G use cases are
categorized under three broad domains; (i) enhanced mobile
broadband, (ii) massive IoT, and (iii) mission-critical IoT. Each
category requires different types of network features in terms
of mobility, security, policy control, latency, bandwidth and
reliability as highlighted in Figure 1.
The proliferation of new IoT devices to the consumer market
will add a higher burden to the mobile network while they
access cloud servers. Moreover, the need to access the cen-
tralized cloud services via mobile network may limit IoT use
cases that demand low latency and high capacity. This brings
in the importance of edge computing paradigms in the caliber
of MEC, a novel and evolving networking paradigm that is
currently standardized by the European Telecommunications
Standards Institute (ETSI) [2]. The rationale behind MEC is
to extend the capabilities of the cloud to the edge of cellular
networks. In principle, this is realized by placing storage and
computational resources at the Radio Access Network (RAN)
edge, and moving, as and when required, some of the function-
alities offered by the cloud to these additional resources at the
edge. Some of the typical characteristics of MEC technology
are closest proximity,ultra-low latency,location awareness,
and network context information.
Based on the above discussion, it must be evident that
the realization of MEC IoT integration has a huge potential
and should be upheld by many underlying technologies. In
this article, we examine four key enabling technologies, i.e.
Network Function Virtualization (NFV), Software Defined
Networking (SDN), Information Centric Networking (ICN)
and Network Slicing (NS) and illustrate how to utilize them
to accelerate the growth of MEC based IoT systems in 5G
networks. Although several other wireless technologies and
radio access network technologies are relevant to MEC, we
have considered the above key technologies related to the
backhual and core networks of the 5G in this paper.
This paper is structured as follows. Section II highlights
the key issues and benefits for integrating MEC with IoT.
The role of four key enabling technologies NFV, SDN, ICN,
NS are discussed in section III, IV, V, and VI, respectively.
Section VII further illustrates the challenges, integration path,
and future directions. Section VIII concludes the present work.
II. MEC-IOT IN TE GR ATIO N
This section discusses the issues pertaining to IoT and
illustrates how MEC based IoT can address them. In particular,
the key benefits of MEC, in the context of IoT, are highlighted.
Also, the challenges envisioned for MEC-IoT integration are
presented.
2
Undoubtedly, the rapid advancements in IoT are helping
the technology to evolve as a mature technology. Nevertheless,
factors like simultaneous growth in the number of IoT devices,
proliferation in the varied types of IoT applications, and
demand to use versatile connectivity options, have given rise to
the several issues such as scalability, mobility, latency, power
consumption, availability, security and privacy. In general, IoT
leverages cloud computing facility which may face several
challenges such as the single-point-of-failure, reachability,
high wide area network (WAN) latency, and lack of location
awareness. In this situation, MEC is geared-up to play an
alleviating role by providing many mutual advantages [1],
[7]. From IoT’s viewpoint, MEC furnishes computational
resources located very close to IoT devices, thus offering
numerous benefits including computation offloading. From
MEC perspective, IoT (being the widespread use case of MEC)
extends MEC services to all sorts of devices, thereby, it enables
wide adoption and evolution of MEC. Furthermore, the cloud
infrastructure also gets offloaded. Table I summarises the add-
on features that MEC brings-in to overcome many of these
issues of different IoT domains.
MEC offers three key benefits for IoT in 5G era. Figure 1
illustrates the integration of these technologies in 5G networks.
The first key benefit that MEC provides is traffic filtering.
This is because the requests generated by a multitude of IoT
applications are satisfied upfront at the edge of the network,
thereby impeding the traffic that otherwise would be pushed
upstream towards the cloud facility. Of course, this is the case
when the IoT devices do not require services at the global level
(e.g., cloud computing capabilities), so they can be served by
the immediate MEC servers. Thus, using MEC will save cost,
reduce the latency and truncate the traffic volume in the core
network. The second key benefit is that MEC facilitates accel-
erated decision making based on the locally processed data,
which reduces End-to-End (E2E) delay. This is very important
in the critical IoT applications (e.g., remote surgeries, smart
grid, autonomous vehicles, and video conferencing), which
have very high demands of reliability, availability, and low
latency. The third key benefit offered by MEC is enhanced
scalability and lifespan of IoT devices. The rapid increase in
numbers of IoT connections would proportionally increase the
traffic load on the network and decrease the battery life, but
thanks to MEC, the battery drain may not increase since less
transmission time is required between the IoT device and the
MEC server. Besides these three key benefits, MEC offers few
other benefits such as context awareness, local storage and
caching, support for intermittent connectivity, mobility (fast
RAT hand-off), localized privacy and security as presented in
Table I.
Next, we discuss the key challenges that need to be dealt
MEC IoT
Apps
Smart city
eNodeB
Gaming
VR / AR
Retail
Smart home
Smart
healthcare
Smart
farming
Wearable IoT
Smart energy
Massive IoT Applications
High reliability
High Scalability
Large data sets
Security, Privacy, Trust
Critical IoT Applications
High reliability
Low latency
Security, Privacy, Trust
High reliability
Low latency
High bandwidth
Edge of the
mobile network
NFV based MEC Hosts
High Bandwidth IoT Applications
High reliability
High bandwidth
Large data sets
Security, Privacy, Trust
Apps Core Network /
Centralized Cloud
NSSF UDM PCF AUSF NEF
AMF SMF NRF
High Reliability
High Scalability SDN
Controller
ICN Service
Orchastrator
ICN Network
Controller
ICN -AF
ICN -SMF
MEC IoT
Apps
MEC Hosts
MEC
Service
MEC
Orchastrator
Network Slice 1
Network Slice 2
Network Slice 3
NFV based 5G Core Network
Autonomous
vehicles
Remote
consultancy
NFV based MEC
Hosts
High reliability
High bandwidth
NFV based MEC Hosts
ICN Components
ICN-SMF: ICN Session Management Function
ICN-AF: ICN Application Function
ICN-AP: ICN Anchor Point
ICN-DN: ICN Data Network
UPF: User Plane Functions
5G Network Functions
AUSF: Authentication Server Function
UDM: Unified Data Management
NRF: NF Repository Function
PCF: Policy Control function
NEF: Network Exposure Function
AMF: Access and Mobility Management Function
NSSF: Network Slice Selection Function
SMF: Session Management Function
RAN: Radio Access Network
SDN Components
MEC Components
5G Components
ICN Components
MEC Interfaces
SDN Interfaces
ICN Interfaces
5G Interfaces
Data Channel
Cloud Connection
SDN and ICN Enabled
Backhual Switches
eNodeB
MEC IoT
App n
MEC IoT
App 1
Virtulatiation
Infrastuctue
Data Network
UPF
MEC
Services
MEC Platform
MEC Platform
Manager
Local
ICN-DN
ICN-AP
Fig. 1: MEC IoT integration under the umbrella of NFV based 5G core network with SDN, ICN and NS [2] [3]–[6]
3
TABLE I: The alleviating role of MEC in different IoT domains
Added characteristics by MEC
IoT Application
Role of MEC
Low latency
Increased bandwidth
Context awareness
Low power devices (Local Storage)
Ability to operate with Intermittent
connectivity with core Network
Fast Mobility
Caching
Edge analytics
Private or local network
Localized security and Privacy
Smart
home
MEC offers reduced communication latency, easy instantiation and fast reloca-
tion. Moreover, MEC can process sensitive data locally by preserving the privacy.
X X X X X X X X X
Smart
city
MEC caters data processing, storing and retrieval requests at the edge of the
network thereby provides low latency, high availability, location awareness,
mobility management and scalibility.
X X X X X X X X
Healthcare
&
Remote
surgery
EC empowers monitoring and detecting physiological symptoms with unin-
terrupted communication (even for remote areas) and edge-based analytics.
Moreover, the promise of ultra-low latency will furnish technical support to
remote-surgeries.
X X X X X X X X
Autono-
mous
vehicles
MEC can improve the operational functions such as real-time traffic monitoring,
continuous sensing in vehicles, Infotainment applications and security by fulfill-
ing the latency, reliability, fast big data processing, and throughput requirements.
X X X X X X X
AR/VR Migrating computationally expensive tasks to edge servers not only amplifies
the computational capabilities of AR/VR devices and elevates the immersive
experience of users but also extend their battery-life. Moreover, high capacity and
low latency wireles coverage commitments of MEC platforms offer scalability
in terms of users who can experience AR/VR even in highly populated areas
X X X X X X X X
Gaming MEC can improve user experience for delay-sensitive game users by offloading
the resource-intensive applications to the edge servers that are located in the
nearest proximity.
X X X X X X
Retail On-site MEC servers can locally process huge volumes of data generated by
different IoT systems such intelligent payment solutions, facial recognition
systems, smart vending machines.
X X X X X X X
Wearables MEC allows to deploy storage, computing, and caching in close proximity to
satisfy wearable requirements such as scalability, short range and low power
communication.
X X X X X
Environment
Moni-
toring
MEC process the information closer to sensor and removes the burden of sending
raw data over a network with limited bandwidth.
X X X X X
Farming
and
Poultry
On-site MEC servers can analyze collected big data without real-time uploading
to a remote cloud. Thus, MEC can directly reduce the overhead on data access,
synchronization and storage.
X X X X X X X
Smart
energy
By performing computations closer to the source of data generation, MEC
resolves the issues like traffic congestion, delays due to poor or intermittent
connectivity and huge data generation.MEC increases the security and reduces the
attack propagation by enforcing security mechanisms closer to the end devices.
X X X X X X X
Industrial
IoT
(IIoT)
MEC enabling future IIoT applications by addressing the shortcomings of M2M
communication (e.g. latency, peer-to-peer connectivity, resilience, cost, security).
Real-time edge analytics and enhanced edge security properties will help to create
new IIoT services.
X X X X X X X
with for MEC-IoT integration. Security, privacy, and trust
management are three important synergistic research areas
of IoT. Users will be increasingly vulnerable to security
threats as more IoT devices and applications make use of
the edge facility. The fact that users’ data in MEC and IoT
are highly exposed, this may lead to many possible ways
of a security breach in sensitive data. Typically, the IoT
devices are designed with implicit mutual trust, and thus,
data sharing happens without a validation process. In such
a scenario, where all the devices inherently trust each other
and share data, it is difficult to identify a misbehaving device.
The situation is aggravated several folds, specially in the
absence of a perimeter security mechanism (e.g., firewall)
around the network edge. Usually, such security mechanisms
can block threats in MEC but are not used in MEC-NVF
integration. Thus in MEC systems, it is challenging to identify,
authenticate, and authorize devices and the data they generate
from the edge to the cloud and back while maintaining a
latency of milliseconds’ order . Similarly, it is challenging to
achieve a non-negligible impact on caching and computation
4
offloading decisions with the user mobility, which will cause
frequent handovers among edge servers.
The next few sections elaborate on the cardinal role of
four key enabling technologies to achieve the above-mentioned
benefits of MEC-IoT integration.
III. NET WORK FUNCTION VIRTUA LI ZATI ON
NFV technology utilizes the power of virtualization tech-
nologies to decouple physical network equipment from the
functions that run on them [8]. Thus, NFV empowers to
implement different Virtual Network Functions (VNFs) as
software that run on one or more industry standard physical
servers. As a result, VNFs can be relocated and instantiated
at different physical network locations, as and when required,
without the necessity to purchase and install new hardware.
NFV is regarded as one of the key enablers for the deployment
of MEC [9] in 5G-IoT networks.
Next, we discuss, how the NFV and MEC technologies can
be used together to meet the escalating networking demands of
5G-IoT based services. In the core architecture, both MEC and
NFV share similar characteristics. For instance, as depicted
in Figure 2 both MEC and NFV leverage virtualization; the
former utilizes it for running applications at edge servers,
whereas, later applies it to implement virtualized network
functions. Both technologies feature stackable components and
each has a virtualization layer. According to ETSI [2], to
maximise the return on investment and enhance computing
experience, operators may reuse the NVF’s infrastructure and
its management for hosting MEC as well. In other words, MEC
can use the NFVI (NFV Infrastructure) as the virtualization
platform to run mobile edge applications alongside other
VNFs. Therefore, MEC applications also appear as VNFs in
the NFV environment and parts of mobile edge orchestration
can be delegated to the NFVO (NFV Orchestration) [3].
IV. SOF TWAR E DEFI NE D NETWORKING
SDN [10] is an emerging network paradigm that intends to
decouple the control plane functions from that of the data plane
of a physical networking resource. Moreover, it opposes using
vendor specific black-box hardware and instead recommends
the use of commodity switches in the data plane. Transfer-
ring network control functionalities to the centralized entities
has numerous advantages. However, critical IoT applications
demand proximity of the SDN controller to the data plane
to fulfil low latency constraint. In this regard, MEC can
be a pragmatic solution to satisfy the latency requirement.
MEC complements the SDN advancements by transforming
the mobile network into softwarized networks and ensuring
highly efficient network operations and service delivery [4].
Fig. 2: MEC in NFV Architecture [3]
5
Next, we discuss briefly how SDN can support MEC’s
deployment. SDN can orchestrate the network, its services
and devices by hiding the complexities of the heterogeneous
mobile environment for the network service developers. Thus,
SDN has a significant potential for mitigating the limitations
that multi-tier MEC infrastructure tends to face, such as
the high complexity of adopting MEC in existing cellular
infrastructure.
The SDN control mechanism can lower the complexity of
MEC architecture by offering a novel approach that utilizes
the available resources more efficiently. SDN can dynamically
route the traffic between tier-MEC servers and cloud servers
to provide the highest QoS to mobile users. Moreover, SDN
paradigm concentrates the network intelligence at the central
software-based controller. This will relieve the relatively more
straightforward MEC devices from executing complex net-
working functions such as flow management, service discovery
and orchestration.
V. INFORMATION CENT RI C NETWORKING
Information Centric Networking (ICN) is a promising clean
slate future networking architecture that aims to intrinsically
reconcile all the existing issues of TCP/IP networking. ICN
advocates a content-centric model in place of the current
host-centric model. In contrast with end-to-end principle,
ICN takes caching and processing to the core of networks
leading to decoupling of contents from their specific locations.
Further, the named-content proposition of ICN brings content-
consciousness in the network allowing the network to know
the details like what content is flowing through it, what is
cached, and what is requested. Thus, ICN is another network-
ing paradigm that can intrinsically satisfy the ever-increasing
traffic demands along with low latency requirements [11].
Several benefits for 5G-IoT applications can be achieved
by exploiting the synergy between MEC and ICN. ICN can
solve the issues related to the content delivery and application
level reconfiguration in MEC. ICN in the backhaul networks
can provide content-consciousness, traffic aggregation (with
Pending Interest Table), en-route caching, and forwarding
strategy. Thus, it can offer high speed content delivery between
the MEC and central cloud systems.
In MEC 5G, when a service is provided by a non-optimal
service instance, an application level reconfiguration is per-
formed for optimization [5]. However, such application level
reconfiguration can be challenging because it requires session
re-initialization. This leads to an increase in session migra-
tion delay and adversely affects the IoT applications. Using
service-centric networking extension of ICN, the application
level reconfiguration delay can be reduced by minimizing the
network configuration delay and allowing fast resolution of
named service instances [5]. The coexistence of ICN and MEC
can also improve the performance of caching offered by edge
storage. Numerous ICN features like named-content, context
aware and location independent data replication, and data-level
integrated security, can benefit both realtime and non-realtime
5G-IoT applications [5].
ICN can significantly improve the efficiency of session
mobility in MEC based IoT applications with the optimal
operational cost and bandwidth utilization for signaling traffic.
In contrast to the IP anchor-based mobility approach, ICN
could handle session mobility by using application bound iden-
tifier and location split principles which significantly reduces
control and user plane overheads.
VI. NE TWORK SLICING
Network Slicing (NS) is another promising key technology
that provides agile networking platforms based on demand and
service specifications. It allows multiple logical networks to
be created on top of a common shared physical and virtual
infrastructure [12].
Integrated use of MEC and NS, along with other technolo-
gies, in the realm of 5G will ease different IoT domains.
For example consider Massive IoT (MIoT) and delay critical
IoT application domains. Massive IoT (MIoT) demands a
large number of connections for mostly immobile devices
which deal with the exchange of delay-insensitive data. To
enable MIoT applications, the network is expected to sat-
isfy the requirements such as edge analytics, reduction in
communication cost, and network scalability. Network slicing
coupled with the MEC-based edge analytics and faster security
features, can deliver these requirements. MEC provides edge
analytics and faster security assets for network slices due to
its proximity to the end devices. This will lead to massive cost
reduction and the increase in network scalability for the MIoT
footprint. On the other hand, delay critical IoT applications,
such as autonomous driving, Tactile Internet and industrial
Internet, demand ultra low latency, high reliability and traffic
prioritization.
In this regard, the powerful combination of MEC and NS
can fulfil the demands since latency can be reduced by virtue
of MEC and traffic prioritization can be offered by NS.
Figure 3 shows how network slicing can delegate the MEC
resources to different slices based on the tenants’ demands
and achieve efficient network resources utilization. Moreover,
NS can enable dynamic and short life cycles for IoT network
services.
In 3rd Generation Partnership Project (3GPP) towards full
multi-tenancy, MEC has been identified as one of the key
technologies to realize the NS extensions. Thus, the synergy
between MEC and NS is expected to play a critical role in
deploying 5G-IoT applications.
VII. CHA LL EN GE S AN D FUTURE DIRECTIONS
Table II illustrates the pivotal role of the four driving
technologies (i.e. SDN, NFV, ICN and NS) to strengthen
the IoT requirements enabled by MEC, which in turn help
in realizing the various 5G-IoT applications. In this section,
we elaborate on the obstacles, challenges and insights on
future research directions pertaining to each of the four driving
technologies in the context of MEC-IoT integration in 5G
networks.
A. NFV
The main research challenges and obstacles for NFV in-
tegration are the absence of standards, system complexity in
6
MEC
Server
AR
app
Streaming
app
cache
Edge Network Slice
MEC
Server
AR
app
Streaming
app
cache
MEC
Server
AR
app
Streaming
app
cache
MEC
Server
AR
app
Streaming
app
cache
Edge Network Slice
MEC
Server
AR
app
Streaming
app
cache
MEC
Server
AR
app
Streaming
app
cache
5G Core Network Slice
AUSF UDM NRF PCF NEF
AMF SMF
NSSF
MEC
Server
AR
app
Streaming
app
cache
Edge Network Slice
MEC
Server
AR
app
Streaming
app
cache
MEC
Server
AR
app
Streaming
app
cache
MEC
Server
AR
app
Streaming
app
cache
Edge Network Slice
MEC
Server
AR
app
Streaming
app
cache
MEC
Server
AR
app
Streaming
app
cache
Critical IoT Slice
Massive IoT Slice
Personal Entertainment
IoT Slice
Mobile
Communication Slice
Internet and
Other Networks
Other Slices
Other Mobile Networks
Internet
Automotive
Health Infrastructure
Industrial Networks
Command and Control
Smart City
Energy Infrastructure
Retail
Other Critical Infrastructures
Entertainment Services
Utility Providers
Internet
Other Network Services
5G Core Network Slice
AUSF UDM NRF PCF NEF
AMF SMF
NSSF
5G Core Network Slice
AUSF UDM NRF PCF NEF
AMF SMF
NSSF
5G Core Network Slice
AUSF UDM NRF PCF NEF
AMF SMF
NSSF
Fig. 3: Use of Network Slicing and MEC in different 5G-IoT applications
deployment, lack of technical maturity and new security risks.
We ponder on each of them in the subsequent discussion.
Currently, NFV is evolving through the phases of imple-
mentation and hence demands standardization which should
emanate from a collaboration between industry and research
communities. In particular, the interfaces and architectural
components of NFV should be defined at global level. The
absence of this may lead to the rise of compatibility issues
which can impede its widespread adaptation.
Most NFV projects encounter a steep learning curve in
getting their infrastructure operational up to the expected level.
This is due to their heavy dependency on non-standardized
implementations. Moreover, due to the lack of technological
maturity, updates are released frequently. Thus, maintaining a
fully integrated operational deployment model is still hard to
accomplish.
Though latency minimization through optimal utilization of
resources can be achieved with the efficient deployment of
MEC services. However, it is not easy to optimize the MEC
services if they depend on complex system components such
as NFV. The de-facto NFV standard implementations such as
OpenStack are difficult to learn, deploy, and use.
NFV integration results in several new security challenges
because of the following reasons. On the one hand, MEC
introduces software components such as Mobile Edge Platform
Manager (MEPM) and Virtual Network Functions Manager
(VNFM) to NFV’s deployment. These components are not part
of the traditional NFV model and create a ’long chain of trust’.
On the other hand, NFV features such as resource pooling can
lead to sharing security risk between multiple unrelated MEC
domains. For instance, an attack on one VNF might hamper
other VNFs running on the same Virtual Machine (VM) or
physical server.
B. SDN
SDN entails several security threats, including SDN proto-
col weaknesses, information disclosure through interception,
flow poisoning, side-channel attacks, and Denial-of-Service
(DoS) on SDN controller [10]. In contrast to the traditional
black-box type of network devices, SDN uses software pro-
grammable common standard backhaul devices. This will not
only ease the work of network administrators but also allow
malicious attackers to deploy attacks. The integration of SDN
with MEC thus becomes challenging since the possibilities of
attacks are increasing; on the one hand, the impact of SDN
based attacks could result in security degradation of MEC
systems and on the other hand, the impact of MEC threats on
the open-network based SDN becomes much more devastating.
Furthermore, the inter-working between SDN and MEC
will also introduce several connectivity challenges. Similar
7
TABLE II: The role of driving technologies to enhance MEC enabled IoT requirements that support 5G-IoT applications
Related 5G-IoT Applications
IoT Requirements enabled by MEC NFV SDN ICN NS
Smart home
Smart city
Remote surgery
Remote health consultancy
Autonomous vehicles
Augmented Reality (AR)
Virtual Reality (VR)
Gaming
Retail
Wearable IoT
Farming
Smart energy
Industrial Internet
Support for Low Latency X X X X X X X
Resource Optimization X X X X X X X X X X X X X
Dynamic Resource Allocation X X X X X X X X X X X X
Support for Edge Caching X XXXXXXXXXXXXX
Increased Security X X XXXXXXXXXXXXX
Increased Privacy X XXXX XXX
Increased Scalability X X X X X X X X X X
Reduced Operational Cost X X X X X X X X X X X X X X X X
Increase Flexibility X X X X X X X X X X X X X X
Increase Orchestration X X XXX XXXXXX
Dynamic Routing and Traffic Optimization X X X X X X X X X X X
Support for Fast Mobility X X X X X X
Service Diversity X X X X X X X X X X X
to SDN southbound, northbound, and east/west interfaces,
it would be interesting for MEC also to have three such
interfaces, namely, (i) Northbound connections that connect
MEC servers to a cloud service (public or private), (ii)
Southbound connections, that connect MEC servers and the
edge devices and (iii) East/West connections, that connect
MEC servers among themselves, so that MEC servers can
communicate directly without the need of cloud connectivity.
We advocate the necessity to merge similar interfaces to reduce
the signaling overhead. Also, the use of too many interfaces
makes the security of the network enfeeble. Furthermore, it is
indispensable to define clean APIs so that applications and
services can program network functions and SDN network
to optimize the performance. Such APIs are needed, for
instance, to support ultra-low latency applications. Otherwise,
information exchange between MEC-IoT and SDN systems
will introduce additional delays in network operations.
C. ICN
ICN complements MEC since its core functionalities can
efficiently govern the interaction between end-users and MEC,
especially, in the mobile environment [13]. To achieve the best
outcomes of their synergies, proper APIs need to be defined in
order to communicate between the systems. Though there is
on-going research to define NFV and SDN interfaces, however,
the interface for ICN communication with MEC-IoT is yet to
be defined. This requires collaboration between cross-domain
industries as well as standard development organizations.
Furthermore, it is crucial to develop system-level control
orchestrator and coordination architecture to enable coopera-
tion between two systems. Moreover, such architecture should
focus on autonomic system control rather than the traditional
provisioning/configuration or distributed networking systems
control.
The real advantages of MEC-IoT can be achieved by
obtaining context information such as users’ location, other
users in the vicinity, condition and resources in the environ-
ment. Although ICN can provide different context information
(application, network and device level), their simultaneous re-
trievals are still challenging. Most of the current ICN research
is focused on providing the basic functionality rather than
utilizing the available context information to improve network
parameters such as Quality-of-Service (QoS). For integration,
it is required to examine typical scenarios encompassing dif-
ferent IoT and 5G applications (e.g., Tactile Internet, AR/VR,
autonomous driving) with varying context.
Another substantial challenge with the use of ICN for MEC
supported IoT applications is the difficulty of accomplishing
authorization and access control [14]. This is because in ICN,
request for named content can be served from any cache-
enabled-node as long as the security of the cached content
is intact. Thus ICN based communications are unable to make
use of traditional user-to-server authentication mechanisms
based on Access Control List (ACL). Maintaining an individ-
ual access control policy for all the currently available cached
contents at each cache-enabled-node across the network tends
to incur severe overhands both communication and computa-
tion.
8
TABLE III: Technical challenges of MEC-IoT Integration in 5G.
Technical
Aspect
Issue / Challenge Description
Communication Backhaul access Optimized the ommunication between MEC servers and remote cloud servers while offloading
data/process with higher demand of resources.
Inter-node communi-
cation
Orchestrate the communication among IoT devices, MEC servers, and remote cloud servers when
they collaboratively execute multiple jobs.
Flaws with wireless
channels
Multi-path fading, interference, and spectrum shortage should be taken into account for the design of
MEC systems to seamlessly integrate computation offloading and radio resource management.
Limited channel ca-
pacity
Wireless and backhaul access links have a limited channel capacity which should be properly shared
among mobile devices in a similar way of sharing the computing resources at the MEC server.
Computational
Offloading
Decision making MEC servers have to decide whether to execute the relatively simple tasks locally or offload (fully or
partially) to the cloud servers. Low power MEC servers may require to offload data more frequently
by consuming more backhual bandwidth.
Partial offloading A subset of computations is offloaded to the cloud server considering factors like users or MEC
application preferences (e.g., application buffer state), backhaul connections quality (i.e., cloud and
MEC servers), MEC server capabilities, cloud capabilities and availability.
Dependency policies Define dependency of offloadable components of the applications based on their ability to partition data
(e.g., real-time user input has to processed at MEC without offloading) and to predict the execution
time/order of multiple tasks. Eg. sequential, parallel, and general dependencies
Joint computation
and communication
resource allocation
The main goal comprises the minimization of execution delay to ensure the quality of service at the
user end while maintaining high energy efficiency and maximizing the number of served applications.
Eg. allocation of single or multiple MEC servers
Mobility
Management
Connectivity There’s need for smart connectivity with existing networks and context-aware computation using
network resources in IoT environments.
Location
Management
Under normal circumstances, the location of network nodes are confidential, however in the case
of the IoT systems, the location of the nodes needs to be made available without compromising on
security.
Routing group forma-
tion
Multiple MEC server clusters have to create different routing groups for different node clusters in the
IoT environment.
Seamless mobility The seamless execution of applications harnessing capabilities of multiple dynamic and heterogeneous
resources to meet quality of service requirements of diverse applications on IoT nodes.
Mobility context
management
Here the idea is to determine the nature of the operations running on the IoT nodes and the latency
tolerance level of such operations, hence the mobility management entity is able to determine the
optimal mobility handling technique for given use cases.
Migration The movement of IoT nodes from one MEC server clusters to another on a more permanent time
scale.
Scalability Deployment indepen-
dence
The IoT nodes on the MEC server should be capable of conforming to multiple deployment scenarios
with little or no modifications to their predefined architectures.
Resource efficiency Here the goal is to ensure optimal utilization of networking and computing resources in the MEC
system.
Scalable storing A huge amount of data will be constantly generated and circulated around the MEC IoT platform,
hence there’s need for semantic execution environments and architectures that accommodate IoT
requirements and scalable storing and communication infrastructure.
Validity of IoT Sce-
narios
To this purpose, the validity of the different IoT scenarios should be proven as they may have problems
in terms of scalability and adaptability to be applied in such a heterogeneous environment.
Security
and Trust
Management
Denial of
Service (DoS) attacks
Adversaries attack critical networking or computing resources by sending requests at rates beyond
the handling capacity of MEC servers and prevent other nodes from getting access to the resources.
Man-in-the-
Middle (MitM)
attacks
In MEC and IoT integration, in the infrastructure layer, the attacker tries to hijack certain segments
of the network and begins to launch attacks like eavesdropping and phishing on connected devices.
MitM attacks can be launched on multiple VMs.
VM manipulation The attacker can be a malicious insider with enough privileges or a VM that has escalated privileges.
The adversary begins to launch multiple attacks to the VMs running towards the virtual infrastructures.
Trust management Assure the reliability and the trustworthiness among end users, IoT devices and MA-MEC servers.
Privacy Harmonize the local
privacy policies at
global level
When IoTs services are expanding over multiple MEC control regions, it is required to harmonized
the privacy at global level.
Implementation of lo-
cal and dynamic pri-
vacy policies
In current systems, privacy policies and directives are predefined at global level and static over the
duration. It is required to find mechanisms to implement local and dynamic privacy policies.
Foster interoperability Support technology neutrality by avoiding mandated standards or preferences which could prevent
the interoperability.
Update privacy poli-
cies
Existing privacy policies and directives in IoT systems should be modified to support the adaptation
of new technologies such as MEC.
9
D. Network Slicing
To garner the real benefits of NS in MEC IoT integration,
numerous challenges are to be addressed.
The inter-system vertical coordination between NS and
MEC IoT integration need to be structured and modeled for
efficient information sharing. This vertical coordination can
be achieved via two ways. The first method is to define APIs
between management systems of slicing and MEC to share the
available resources in terms of different IoT applications. The
second method is to use physical resource coordination aimed
to handle resources through policy and analytics efficiently.
However, to synchronize the various research and development
activities worldwide, standardization of these interfaces is
required.
If NS powered MEC servers can offer fine-grained network
functions, it would enhance the scalability to support different
vendors. Each coarse grained function at the MEC server
can be further divided into many sub-functions. Nevertheless,
the challenge is defining the granularity of these networking
functions carefully so that they comply with the available
standardized interfaces.
When multiple RATs accommodate the 5G IoT paradigms,
there should be some ways to access them on specialized
or dedicated hardware. Although network slicing may lead
to virtualize RAN instances, it is indispensable to ensure
radio resource isolation and manage efficiency. To assist RAN
virtualization for slicing, Software Defined RAN controllers
can be deployed at the MEC servers.
Even though the high-level description of a concrete slice in
terms of infrastructure and network functions exits, however,
the physical realization of E2E slice orchestration is yet to be
established. MEC servers, as intermediary computing platform
between RAN and core network, can play a vital role to
support E2E slice orchestration by correlating cloud and radio
resources used in different IoT applications.
E. Integration Path
This section explains integration paths and pinpoints tangi-
ble steps to realize the MEC-IoT synergy.
1) Control Level Orchestration: To rectify the potential
benefits of MEC-IoT enabled 5G networks, different tech-
nologies must work simultaneously and in close association as
depicted in Figure 1. However, the fact is, such integration will
face difficulty at the control level. Each technology utilizes its
orchestrator and management entities such as SDN controller,
NFV orchestrator, NS manager and Mobile Edge Platform
Manager (MEPM). In this respect, a synergy between these
control entities is needed to jointly optimize the network
resources and create efficient Service Function Chains (SFCs)
for each user application.
2) Synchronization of Standardization Process: To achieve
orchestration in MEC systems, different technological com-
ponents need to inter-communicate, which require defining
communication interfaces at the architectural-level. However,
as of now, the standardization of different technologies are
coordinated by different organizations, for e.g., MEC and NFV
by ETSI, SDN by ONF, ICN by IETF, IoT by IEEE and Open
Internet Consortium (OIC). Therefore, there is an exigency
of collaborative synthesised efforts by these standardization
bodies. As a good example, ETSI has already started defining
the interfaces for NFV and MEC integration (Figure 2).
3) Hardware Limitations and Platform Dependencies:
The integration of driving technologies demands changes
in the control plane, the data plane and hardware/software
components. For instance, SDN-enabled switches and devices
are needed at the infrastructure layer to implement SDN.
Similarly, ICN enabled switches are needed to enable ICN
functionalities. Production and installation of such multi-
technology hardware will not be easy. To achieve this first,
standardization of different technology should be carried out
so that vendors can start building such multi-technology hard-
ware equipment. Second, extensive hardware resources are
needed to implement multi-technology concepts. Therefore,
these hardware limitations and dependencies must be resolved
to obtain full benefits of integrating technologies.
4) AI as a key integration enabler: Recently, Artificial
Intelligence (AI) and Machine Learning (ML) have been
resorted to create smarter and autonomous wireless systems
[15]. In the 5G context, AI can directly benefit the driving
technologies such as SDN and NFV to be integrated into
MEC and IoT. For instance, AI-based edge orchestrators can
be used for better system and host level management functions
for various NFV based use cases. AI and MEC together (i.e.,
edge automation) will combat low latency for real-time IoT
services, better orchestration, enhanced security, and backhaul
cost savings.
F. Additional Technical Challenges
In addition to the above mentioned challenges, there are
several technical challenges of MEC-IoT integration in 5G.
These challenges can be categorized under communication,
computational offloading, mobility management, scalability,
security and privacy. A summary of these technological chal-
lenges is presented in Table III.
VIII. CONCLUSION
This paper analyzes the feasibility and practical integra-
tion of four technological directions, including NFV, SDN,
ICN and Network Slicing, that can facilitate the MEC-IoT
integration in 5G mobile networks. Besides highlighting the
benefits of using each technology, this paper also identifies
the remaining challenges and presents a pragmatic integration
paths. We believe these solutions will form a solid ground for
network developers and providers to deploy MEC-IoT in 5G
networks optimally.
ACKNOWLEDGMENT
This work has been performed under the framework of
RESPONSE 5G (Grant No: 789658) project which is funded
by European Union and 6Genesis Flagship (grant 318927)
project which is funded by the Academy of Finland.
10
REFERENCES
[1] P. Porambage, J. Okwuibe, M. Liyanage, M. Ylianttila, and T. Taleb,
“Survey on multi-access edge computing for internet of things realiza-
tion,” IEEE Communications Surveys and Tutorials, 2018.
[2] Y.-C. Hu, M. Patel, D. Sabella, N. Sprecher, and V. Young, “Mobile
Edge Computing : A Key Technology Towards 5G,” ETSI White Paper,
vol. 11, no. 11, pp. 1–16, 2015.
[3] “Mobile Edge Computing (MEC); Deployment of Mobile
Edge Computing in an NFV environment,” ETSI Industry
Specification Group (ISG) White Paper, February 2018. [Online].
Available: http://www.etsi.org/deliver/etsi gr/MEC/001 099/017/01.01.
01 60/gr MEC017v010101p.pdf
[4] B. Blanco, J. O. Fajardo, I. Giannoulakis, E. Kafetzakis, S. Peng,
J. P´
erez-Romero, I. Trajkovska, P. S. Khodashenas, L. Goratti,
M. Paolino et al., “Technology Pillars in the Architecture of Future
5G Mobile Networks: NFV, MEC and SDN,” Computer Standards &
Interfaces, vol. 54, pp. 216–228, 2017.
[5] R. Ravindran, A. Chakraborti, S. O. Amin, A. Azgin, and G. Wang,
“Realizing ICN in 3GPP’s 5G NextGen Core Architecture,” arXiv
preprint arXiv:1711.02232, 2017.
[6] N. Alliance, “Description of Network Slicing Concept,” NGMN 5G P,
vol. 1, 2016.
[7] F. Bonomi, R. Milito, P. Natarajan, and J. Zhu, “Fog Computing: A
Platform for Internet of Things and Analytics,” in Big data and internet
of things: A roadmap for smart environments. Springer, 2014, pp.
169–186.
[8] R. Mijumbi, J. Serrat, J.-L. Gorricho, N. Bouten, F. De Turck, and
R. Boutaba, “Network Function Virtualization: State-of-the-art and Re-
search Challenges,” Communications Surveys & Tutorials, vol. 18, no. 1,
pp. 236–262, 2016.
[9] L. Gupta, R. Jain, and H. A. Chan, “Mobile
Edge Computing–An Important Ingredient of 5G Net-
works,” IEEE Software Defined Networks, Newsletter,
2016. [Online]. Available: http://sdn.ieee.org/newsletter/march-2016/
mobile-edge- computing-an- important-ingredient- of-5g- network
[10] D. Kreutz, F. M. Ramos, P. E. Verissimo, C. E. Rothenberg, S. Azodol-
molky, and S. Uhlig, “Software-Defined Networking: A Comprehensive
Survey,” Proceedings of the IEEE, vol. 103, no. 1, pp. 14–76, 2015.
[11] A. V. Vasilakos, Z. Li, G. Simon, and W. You, “Information Centric
Network: Research Challenges and Opportunities,” Journal of Network
and Computer Applications, vol. 52, pp. 1–10, 2015.
[12] K. Samdanis, X. Costa-Perez, and V. Sciancalepore, “From Net-
work Sharing to Multi-tenancy: The 5G Network Slice Broker,”
Communications Magazine, vol. 54, no. 7, pp. 32–39, 2016.
[13] D. Grewe, M. Wagner, M. Arumaithurai, I. Psaras, and D. Kutscher,
“Information-Centric Mobile Edge Computing for Connected Vehicle
Environments: Challenges and Research Directions,” in Proceedings of
the Workshop on Mobile Edge Communications. ACM, 2017, pp. 7–12.
[14] D. Kutscher, S. Eum, K. Pentikousis, I. Psaras, D. Corujo, D. Saucez,
T. Schmidt, and M. Waehlisch, “Information-centric networking (icn)
research challenges,” IRTF RFC 7927, July 2016.
[15] L. Iliadis, I. Maglogiannis, and V. Plagianakos, Artificial Intelligence
Applications and Innovations. Springer, 2018, vol. 520.
