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

net/publication/323268025

The Role of Edge Computing in Internet of Things

Article in IEEE Communications Magazine · May 2018

DOI: 10.1109/MCOM.2018.1700906

CITATIONS READS
288 10,595

5 authors, including:

Najmul Hassan Saira Gilani

10 PUBLICATIONS 667 CITATIONS

Bahria University Lahore Campus
37 PUBLICATIONS 1,538 CITATIONS
SEE PROFILE
SEE PROFILE

All content following this page was uploaded by Ibrar Yaqoob on 03 September 2018.

The user has requested enhancement of the downloaded file.

IEEE COMMUNICATIONS MAGAZINE 1

The role of edge computing in Internet of Things

Najmul Hassan, Saira Gillani, Member, IEEE, Ejaz Ahmed, Member, IEEE,
Ibrar Yaqoob, Member, IEEE, Muhammad Imran, Member, IEEE

Abstract—Remarkable advancements in embedded by IoT-devices are essential for organizations that

systems-on-a-chip have significantly increased the number are interested in improving their productivity and
of commercial devices that possess sufficient resources to revenues. However, management and analysis of
run full-fledged operating systems. This increment has
such large amounts of data are cumbersome and
extended the potential of the Internet of Things (IoT).
Many early IoT devices could only collect and send challenging for organizations that rely on conven-
data for analysis. However, the increasing computing tional computing paradigms. Edge computing is
capacity of today’s devices allow them to perform complex gaining popularity in this context because IoT is
computations on-site, resulting in edge computing. Edge becoming common in processing data on the edge
computing extends cloud computing capabilities by of networks [2].
bringing services close to the edge of a network and thus Given that data are rapidly produced at the edge
supports a new variety of services and applications. In
this work, we investigate, highlight, and report recent
of networks, dealing with these data at the edge of
advances in edge computing technologies with respect to the network would be effective. Several approaches,
measuring their impact on IoT. We establish a taxonomy such as cloudlet [3], fog computing [4], and mobile
of edge computing by classifying and categorizing existing edge computing (MEC) [5], provide complementary
literature, and by doing so, we reveal the salient and solutions to cloud computing to reduce data process-
supportive features of different edge computing paradigms ing on the network edge. In short, edge computing is
for IoT. Moreover, we present the key requirements for
the successful deployment of edge computing in IoT and
a general term that represents fog computing, MEC,
discuss a few indispensable scenarios of edge computing cloudlets, and micro clouds. Storage, computing,
in IoT. Several open research challenges are also outlined. and power are regarded as being on the edge of
networks to increase availability, reduce latency, and
Index Terms—Internet of Things, Edge Computing, Fog
Computing, Mobile Edge Computing, Cloudlets eventually overcome cloud computing issues [6].
Edge computing facilitates the processing of delay-
sensitive and bandwidth-hungry applications near
I. I NTRODUCTION the data source [7]. Figure 1 illustrates a layered
Billions of smart devices can now connect to the model for cloud edge-based IoT service delivery.
Internet in the form of the Internet of Things (IoT) Although several studies have been conducted on
due to advancements in networking technologies different edge computing paradigms (i.e., fog, edge,
[1]. According to a Cisco report, these devices will and cloudlets) [8] [9] [3], no study has examined
generate 507.9 ZB of data by 2019. Data generated all of the previously mentioned edge computing
paradigms in terms of IoT. Research in this direction
should be conducted because of the significance of
N. Hassan is with COMSATS University Islamabad, Attock Cam-
pus. (Email: [email protected]) the emerging paradigm of edge computing and its
S. Gillani is with the College of Computing and Informat- role in the success of IoT. This study highlights the
ics, Saudi Electronic University, Jeddah, Saudi Arabia. (Email: role of edge computing in the IoT context.
[email protected])
E. Ahmed and I. Yaqoob are with the Centre for Mobile The contributions of this work are manifold.
Cloud Computing Research, University of Malaya, Malaysia. (Email: • We investigate, highlight, and report recent
[email protected], [email protected]) premier advances in edge computing from the
M. Imran is with the College of Computer and Informa-
tion Sciences, King Saud University, Saudi Arabia. (Email: IoT perspective.
[email protected]) • We categorize and classify edge computing
Corresponding Authors: Ejaz Ahmed and Muhammad Imran literature by devising a taxonomy.
• We outline key requirements for the successful
deployment of edge computing in IoT.
IEEE COMMUNICATIONS MAGAZINE 2

s
red
nd
Cloud Data Centers

Hu
ds
an
us

Edge Gateways
o
Th
s
ion

Smart IoT
ill

Devices
M

Layered Model for Cloud Edge-based IoT Service Delivery

Fig. 1: Layered Model for Cloud Edge-based IoT Services Delivery

• We present a few indispensable scenarios of However, wireless network technologies are not as
edge computing in IoT. reliable as wired technologies.
• We identify and elaborate several open research
challenges. B. Computing Nodes
These contributions are discussed separately in IoT devices have limited processing capabili-
Sections II to V. The concluding remarks are pro- ties, which make them unsuitable for computation-
vided in Section VI. intensive tasks. However, resource-constrained IoT
devices can augment their capabilities by leveraging
II. TAXONOMY OF I OT- BASED E DGE the resources of edge servers. The edge computing
C OMPUTING paradigm relies on different computational devices
Figure 2 depicts a taxonomy of IoT-based edge to provide services to IoT users. These computa-
computing that considers particular features, such tional devices are the core element of IoT-based
as wireless network technologies, computing nodes, edge computing. Computing nodes include servers,
computing paradigms, service level objectives, ma- base stations (BS), routers, and vehicles that can
jor enablers, data types, applications, and attributes. provide resources and various services to IoT de-
vices. The use of these devices is specific to the
A. Network Technologies computing paradigm.
IoT devices send collected data to a locally avail-
able edge server for processing. These devices com- C. Computing Paradigms
municate with edge computing platforms through ei- Various computing paradigms are used in IoT
ther wireless networking technologies, such as WiFi to provide different services depending on diverse
and cellular networking (e.g., 3G, 4G, and 5G), or application requirements. These paradigms can be
wired technologies, such as Ethernet. These network categorized into cloud computing, edge computing
technologies vary in terms of data rate, transmission (i.e., MEC, fog, and cloudlet), mobile ad hoc cloud
range, and number of supported devices. Wireless (MAC), and hybrid platforms. Cloud computing is
networks provide flexibility and mobility to users a centralized computing infrastructure that aims to
who execute their applications on the edge server. provide interruption-free access to powerful cloud
IEEE COMMUNICATIONS MAGAZINE 3

Taxonomy of IoT-based Edge Computing Environment

Network Computing Computing Service Level Major Applications Attributes

Paradigms Objectives Data Types
Technologies Nodes Enablers

WiFi Cloud Latency Networking Non Real- Smart Home Low Latency
Servers Computing time
Minimization Technologies

3G Fog Network Software Soft Real- Health Care Proximity

Base Stations Development time
Computing Management Kits

Cost Resource Rich Hard Real- Video Location

4G Routers Cloudlets Network time surveillance awareness
Optimization
Elements
Dense
Energy Smart Grid Geographical
5G Vehicles MEC Virtualization
Management Distribution

Mobile Network
Bluetooth Resource Smart Cities Context
Ad hoc Awareness
Cloud Management

Hybrid Data Smart Logistics

Ethernet
Computing Management
Environment
Monitoring

Fig. 2: Taxonomy of IoT-based Edge Environment

servers. These servers can rapidly process large when we require the large computing resources of
amounts of data upon receipt from remote IoT cloud computing but cannot tolerate the latency
devices and send back the results. However, real- of the cloud. Variants of edge computing can be
time delay-sensitive applications cannot afford long employed in such applications to overcome the
delays induced by a wide area network. Continuous latency problems of cloud computing.
transmission of voluminous raw data through unreli-
able wireless links may also be ineffective. By con-
trast, edge computing is a decentralized computing D. Service Level Objectives
platform that brings cloud computing capabilities The different service-level objectives for edge
near IoT devices, that is, the network edge. An computing in the context of IoT are as follows]:
important type of edge computing platform is MEC, 1) Latency Minimization: High latency has be-
which brings cloud computing capabilities to the come a crucial problem for IoT-based smart ap-
edge of a cellular network [10]. Computational and plications. An alternative platform, such as edge
storage services in MEC are provided at BS. Unlike computing, that can guarantee timely delivery of
MEC, fog computing employs local fog nodes (i.e., services is required to fulfill the quality of service
local network devices, such as a router or switch) (QoS) requirements of delay-sensitive IoT applica-
available within a limited geographic region to pro- tions (e.g., smart transportation and online gaming).
vide computational services. Fog computing is con- 2) Network Management: A number of phenom-
sidered a premier technology following the success ena, such as inadequate virtualization support, lack
of IoT. Cloudlet is another form of edge computing, of seamless connectivity, and inefficient congestion
in which delay-sensitive and computation-intensive control, degrade the overall network performance.
tasks from IoT devices are performed on a server Therefore, efficient usage of network resources in
deployed in the local area network. Unlike cloud and edge computing is vital for IoT.
edge computing platforms that rely on infrastructure 3) Cost Optimization: The use of an adequate
deployment, MAC capitalizes the shared resources platform for enabling edge computing necessitates
of available mobile devices within local proxim- extensive infrastructure deployment that involves
ity to process computation-intensive tasks. Cloud substantial upfront investment and operational ex-
and edge computing are used together in hybrid penses. Most of these expenses are related to net-
computing. Such infrastructure is usually adopted work node placement, which requires deliberate
IEEE COMMUNICATIONS MAGAZINE 4

planning and optimization to minimize the overall different virtual machines for a number of cloud
cost. Deployment of an optimal number of nodes computing services at the edge of networks.
at appropriate positions can significantly reduce
capital, and optimal arrangement of edge nodes can F. Data Types
minimize operational costs.
One of the primary reasons for edge computing is
4) Energy Management: Energy management is
that its precursor (i.e., cloud computing) was unable
also an important objective of IoT-based edge com-
to fulfill certain requirements of applications deal-
puting. Subscribers need to have strict control over
ing with various data types. These data types can
power management. Energy-efficient IoT devices
be broadly categorized based on delay sensitivity.
and applications are desirable in edge computing.
Hard real-time data cannot tolerate any delay at
According to a study, one trillion IoT nodes need
all, whereas soft real-time data can afford several
sensing platforms that support various applications
bounded delays. Delay-tolerant applications can be
using power harvesting to ensure scalability, reduce
classified as non-real-time.
costs, and avoid frequent battery replacement.
5) Resource Management: Optimal management
of computational resources is crucial in obtaining G. Applications
service-level objectives. Appropriate resource man- A number of applications currently use edge
agement includes coordination of resources, estima- computing.
tion of available resources, and proper allocation of 1) Smart Homes: Smart homes equipped with
workload. a number of IoT devices belong to an emerging
6) Data Management: The large number of IoT application domain of edge computing. The IoT
devices at present are expected to generate large applications envisioned for the monitoring and me-
amounts of data that need to be managed in a timely tering of smart homes will allow subscribers to
manner. Efficient and effective data management obtain automated and precise readings of different
mechanisms are desirable in edge computing. Trans- meters and access invoices accordingly without de-
mission and aggregation of IoT-generated data are lay [11]. These IoT applications are designed for
important concerns in data management. remote monitoring and metering of various utilities,
such as water, electricity, and gas. The data collected
from IoT devices can be transmitted to an edge
E. Major Enablers server for processing instead of sending them to the
The driving forces behind the success of edge cloud, which can lead to real-time data analytics.
computing are different types of technologies. 2) Healthcare: Edge computing has been suc-
Emerging network technologies, such as 4G and cessfully implemented in recent years and is now
cognitive radios, are vital in fulfilling the require- commonly used in different medical appliances.
ments of delay-sensitive applications. These com- Edge computing enables end users to monitor
munication technologies are used in edge comput- and react to health-related data generated by var-
ing for device-to-device and device-to-edge server ious servers. Different architectures that use cloud,
communication. Software development kits with fog, and edge computing have been proposed
appropriate application programming interfaces as- to reap the benefits of collaborative computing
sist in developing and integrating new compatible paradigms. Healthcare applications are usually con-
applications and customizing existing applications sidered delay-sensitive applications in IoT. Initially,
and services. Cloud computing utilizes powerful cloud computing was employed for healthcare ap-
servers for computation-intensive jobs; the same plications but was not highly successful because of
idea has been envisioned to bring cloud capabilities latency issues. The introduction of edge computing
to the edge devices of networks to minimize latency. resolved these issues and made cloud computing
Such servers can help offload computations from realistic for healthcare IoT applications.
small resource-limited mobile devices. Virtualiza- 3) Video Surveillance: Edge computing is cur-
tion is another emerging enabler that allows for rently used for smart video surveillance in different
the creation of logically isolated resources using sectors of life, including domestic security and
similar physical resources. Virtualization is used by anti-terrorism. Different video contents are obtained
IEEE COMMUNICATIONS MAGAZINE 5

and shared from different video cameras and video H. Attributes

sensors. These videos are stored and efficiently
managed for further processing. Different security Edge computing is characterized by certain at-
applications can automatically extract the required tributes, such as low latency, proximity, location
data from the archive of stored video contents. awareness, dense geographical distribution, and net-
Video surveillance is usually performed with the work context information. Mobile operators, content
collaboration of edge and cloud computing. providers, and application developers can utilize
these favorable attributes in their corresponding
4) Smart Grid: Edge computing and IoT are business domain by using them to enhance the qual-
utilized in smart energy management. Such applica- ity of experience for mobile broadband subscribers.
tions automatically observe consumption and distri-
bution patterns. Contributing nodes are used by edge
computing for real-time sensing and processing. III. K EY R EQUIREMENTS OF E DGE C OMPUTING
Cloud computing is also utilized as a collaborative IN I OT
tool to make such applications robust and dynamic
for large amounts of data in the deployment of wide Successful deployment of edge computing in the
area energy networks. Meanwhile, edge computing IoT environment has certain requirements. These
is used for agility and load distribution. requirements are desirable features for the smooth
functionality of any edge computing application or
5) Smart Cities: Edge computing in IoT can service. A system is considered user-friendly and
assist in designing smart cities. Edge computing can feasible when it fulfills the requirements enumerated
be effectively used in lighting control systems of below. Notably, several of the requirements are
streets and roads, air and water quality monitoring, conflicting. Therefore, application designers must
exploring emergency routes in cases of disasters maintain good balance among all of them by con-
or accidents, and automatic watering of different sidering the environment.
gardens in cities.

6) Smart Logistics: Edge computing in the IoT A. Latency

environment helps conventional logistics and offers
new fascinating possibilities that make the flow Delay is one of the basic reasons for preferring
management of things automated and easy. This edge computing over cloud computing in delay-
system enables a smooth flow of transaction be- sensitive applications. Many modern-day applica-
tween the product manufacturer and end consumer tions are delay-sensitive and perform computation
in terms of cost and time. Apart from these appli- on real-time data. Bringing many services and ap-
cations, interested readers may refer to the work of plications from the cloud to the edge of the net-
J. Pan and J. McElhannon presented in [12] that work drastically reduces the latency of IoT appli-
discussed several edge cloud opportunities for IoT cations. Low latency enables real-time communica-
applications. tion, which leads to improved decision making in
IoT-based edge computing.
7) Environment Monitoring: The collaborative
use of edge and cloud computing in IoT can en-
hance the quality of existing monitoring systems. B. Reliability
An automated system will collaborate with sensors
and actuators. Applications have been developed Edge computing is envisioned to be used in
for monitoring critical entities that exert a major almost every field of life. Edge computing plays
effect on the environment. These entities include several roles in critical fields where it is used,
monitoring of gas concentration in air, water level such as healthcare and banking, which require a
in lakes and underground, lighting condition, soil reliable system. Reliability is a critical requirement
humidity, and changes in land position. Environ- of any edge computing system to make the system
ment monitoring is crucial in many fields, such as a practical choice for real-life IoT applications of
agriculture, forestry, and food safety. smart computing.
IEEE COMMUNICATIONS MAGAZINE 6

C. Mobility Support IV. ROLE OF E DGE C OMPUTING IN THE

In IoT, different types of devices are utilized I NTERNET OF T HINGS
for different architectures. Mobility techniques are Edge computing is expected act as a strategic
frequently used in different cloud and edge architec- brain behind IoT. Identifying the role of edge com-
tures and are gaining popularity in applications em- puting in IoT is the main research issue at present.
ployed in hybrid infrastructures. In infrastructures Edge computing is utilized to reduce the amount
with mobile devices, mobility control is always an of data sent to the cloud and decrease service
issue. Meanwhile, mobility support is mandatory for access latency. Figure 3 illustrates the complimen-
certain edge applications. The provision of mobil- tary role of edge and cloud computing in the IoT
ity is important for direct communication between environment. In this section, several major roles of
different modules and for message exchange among edge computing are discussed with IoT scenario
different mobile devices. examples.

D. Real-time interactions A. Data Acquisition

Real-time applications are gaining popularity in Edge devices, including sensors or machines,
all fields of communication. In IoT, certain edge can capture streaming data for rapid analysis and
applications require real-time interaction. Other op- perform immediate actions or processing of the
tions, such as batch processing, are available but data. According to Beckman, we are moving the
may not be suitable for many modern applications algorithm to the data, not the data to the algorithm.
in IoT that use edge computing, such as healthcare Consequently, we can increase productivity and
systems and many other critical and time-sensitive prevent product defects efficiently and rapidly. In a
applications. smart transportation scenario, traffic light cameras
can not only capture data but also analyze the
E. Security collected data and make immediate decisions on
Security is one of the core requirements of all their own to improve the flow of vehicles.
modern systems. The use of edge computing in IoT
is subject to developing secure systems and applica- B. Inferential Controls
tions. The security requirement is almost undefined
in cloud computing; therefore, cloud services are Inferential controls are key components of any
vulnerable to certain types of attacks. Cloud com- edge device. They refer to the capacity of a device to
puting services are easy targets of security threats interpret things in its environment accurately. These
and data breaches because of their specific structure. controls also communicate with an infrastructure
However, in edge computing, security should be that is controlled by other entities. However, bring-
well defined and clearly implemented. Therefore, ing inferential capability to edge devices is difficult
improved data security can be provided because because it depends on contextual information. In a
client data are aggregated at certain access points smart transportation scenario, this inference ability
placed closed to the end user. can provide drivers with highly intelligent naviga-
tion instructions by using GPS and front and back
F. Interoperability cameras.
Interoperability is a key requirement in infor-
mation technology and telecommunication. Appli- C. Data Analysis
cations and services developed for IoT-based edge Edge computing enables real-time data analysis.
computing must be interoperable to ensure compati- Analyzing data in the place of data generation
bility with other applications and hardware compo- can reduce the latency of information generation
nents. Several of the key requirements above are from the collected data. Therefore, edge devices can
mentioned in the perspective of IoT. This set of collect and analyze data from surrounding devices,
requirements is not a full and complete list. Other thus allowing decision makers to deliver actionable
requirements may be applied in different applica- insights faster than before. Edge devices can also
tions. help reduce network bandwidth and cost because
IEEE COMMUNICATIONS MAGAZINE 7

Edge
Server

Thermal
Edge Power
Server

Edge Factories
Server
Human Smart
Gadgets

Cloud

Edge Offices
Edge
Server Server
Smart Cars

Fig. 3: Illustration of Edge Cloud Computing Complimentary Role in IoT Environment

data will be locally analyzed. This can be helpful for E. Enhanced Data Security
many organizations in many industries, including When data are sent abroad for data processing,
manufacturing, healthcare, telecommunication, and data insecurity is increased. Data collection and
finance, such that the need for the IoT concept analysis are performed locally in edge computing.
increases. Therefore, instead of traffic light cameras Extensive routing is not involved, such that identi-
sending data to a central infrastructure for data anal- fying any suspicious activity is easy. Implementing
ysis, they can analyze streaming data themselves, necessary actions before any security breach occurs
communicate with other devices, and make imme- also becomes easy.
diate decisions to accomplish the required tasks.
V. O PEN R ESEARCH C HALLENGES
The following discussion highlights the open re-
D. Decision Making search challenges in the deployment of edge cloud
in the IoT environment.
After locally analyzing the data, the next step for
edge devices is to make critical strategic decisions.
In a smart transportation system, each car generates A. Heterogeneity
a large amount of data every second and requires Heterogeneity in the IoT-based edge computing
real-time processing and correct decisions. For real- environment exists in computing and communica-
time processing, the data cannot be sent to the cloud tion technologies. Computing platforms can have
for processing and decision making because the different operating systems and hardware architec-
response time would be too long in this case. In tures, whereas communication technologies can be
such a case, the data should be analyzed locally heterogeneous with regard to data rate, transmission
on any edge device. In this manner, the car can range, and bandwidth. One of the challenges in edge
make a correct decision on the spot to avoid adverse computing is to develop a solution in software space
situations. that is portable across different environments. This
IEEE COMMUNICATIONS MAGAZINE 8

challenge is crucial because various applications data is computationally difficult. Security risks are
are deployed in edge devices. Several researchers also involved. Therefore, the data should be pre-
have proposed software solutions to resolve this processed at the gateway level, such as noise/low-
issue, but all of these solutions are hardware spe- quality removal, event detection, and privacy protec-
cific. Thus, they cannot resolve the problem in tion. The processed data will be sent to the upper
heterogeneous environments. To solve this issue, layer for future service provision. However, many
programmers should develop a programming model challenges may occur in this process. For privacy
for edge nodes that is supported by task- and data- and security purposes, applications running on edge
level parallelism to facilitate the execution of work-
devices should be blind to these raw data. Therefore,
loads simultaneously at multiple hardware levels. the details of the data should be removed during
The second consideration is to use a language that data preprocessing. However, the usability of the
supports hardware heterogeneity. data can be affected by hiding the details of sensed
data. Defining the extent to which the raw data
B. Standard Protocols and Interfaces should be filtered out is also a challenge because
several applications cannot obtain accurate results
Edge computing is an emerging technology in from such data.
the IoT field. In this heterogeneous environment,
different devices and sensors connect and commu-
nicate with one another and with the edge server via E. Security and Privacy
communication protocols. These devices have their Edge computing acts as a boon to cybersecurity
own interfaces and thus demand specific communi- because data do not travel over a network. However,
cation protocols. Considering that different vendors a highly dynamic environment at the edge of a
manufacture different devices in the IoT environ- network makes the network unprotected. Given that
ment, standard protocols and interfaces should be different devices are connected in IoT, a large array
developed to enable communication among these of potential security threats can be generated. Many
heterogeneous devices. The development of stan- applications are running at the network edge, so the
dard protocols and interfaces in the IoT environment data provided to these applications should be in a
is challenging because of the rapid development of hidden form. Otherwise, any intruder can use the
new devices. open data for illegal purposes. For example, if a
home is connected to IoT, then private data, such as
individual health data, can be stolen. In this case,
C. Availability
how to support the service without harming privacy
Availability in the IoT-based edge computing is a challenge. Applications running on edge devices
environment includes hardware- and software-level should be blind to the raw data. Personal data
provision of resources and services anywhere and can be removed before reaching the edge device.
anytime for subscribed IoT devices. Usually, avail- Several solutions have been provided by researchers
ability comprises three factors, namely, mean time to standardize and store health data [13]. X. Sun
between failure, failure probability, and mean time et al. [14] proposed a hierarchical fog computing
to recovery. Ensuring the availability of resources framework called EdgeIoT. This framework uses a
and services for the growing number of IoT devices proxy virtual machine for securing the privacy of
is a challenging research perspective. However, user content and reducing data traffic in the core
availability can be optimized by maximizing the network. However, security features with enhanced
mean time between failures and minimizing the robustness should be implemented on edge nodes.
failure probability and mean time to recovery.
VI. C ONCLUSION
D. Data Abstraction In this article, we investigated, highlighted, and
With IoT, a number of data-generating devices reported recent premier advances in edge comput-
are connected in the network, and all of these data ing technologies (e.g., fog computing, MEC, and
generators report tremendously large raw data to the cloudlets) with respect to measuring their effect on
edge device. For the edge device, analyzing such big IoT. Then, we categorized edge computing literature
IEEE COMMUNICATIONS MAGAZINE 9

by devising a taxonomy, which was used to uncover [11] E. Ahmed, I. Yaqoob, A. Gani, M. Imran, and M. Guizani,
the premium features of edge computing that can be “Internet-of-things-based smart environments: state of the art,
taxonomy, and open research challenges,” IEEE Wireless Com-
beneficial to the IoT paradigm. We outlined a few munications, vol. 23, no. 5, pp. 10–16, 2016.
key requirements for the deployment of edge com- [12] J. Pan and J. McElhannon, “Future edge cloud and edge
puting in IoT and discussed indispensable scenarios computing for internet of things applications,” IEEE Internet
of Things Journal, vol. 5, no. 1, pp. 439–449, Feb 2018.
of edge computing in IoT. Furthermore, several open [13] F. A. Kraemer, A. E. Braten, N. Tamkittikhun, and D. Palma,
research challenges to the successful deployment of “Fog computing in healthcare–a review and discussion,” IEEE
edge computing in IoT are identified and discussed. Access, vol. 5, pp. 9206–9222, 2017.
[14] X. Sun and N. Ansari, “Edgeiot: Mobile edge computing for the
We conclude that although the deployment of edge internet of things,” IEEE Communications Magazine, vol. 54,
computing in IoT provides numerous benefits, the no. 12, pp. 22–29, 2016.
convergence of these two computing paradigms
brings about new issues that should be resolved in
the future.

ACKNOWLEDGEMENTS
Najmul Hassan was born in 1983 in Ab-
Imran’s work is supported by the Deanship of bottabad, Pakistan. He did his MS in Com-
Scientific Research at King Saud University through puter Sciences from Mohammad Ali Jinnah
Research group No. (RG # 1435-051). PLACE University, Islamabad, Pakistan in 2010. He is
PHOTO currently working in collaboration with Dr. Jin
HERE Li, School of Computer Science Guangzhou
R EFERENCES University China. His research areas are In-
ternet of Things, Edge/Cloud Communication,
[1] A. Al-Fuqaha, M. Guizani, M. Mohammadi, M. Aledhari, and
Wireless Body Area Networks and Wireless
M. Ayyash, “Internet of things: A survey on enabling tech-
Sensor Networks.
nologies, protocols, and applications,” IEEE Communications
Surveys Tutorials, vol. 17, no. 4, pp. 2347–2376, Fourthquarter
2015.
[2] M. Satyanarayanan, P. Simoens, Y. Xiao, P. Pillai, Z. Chen,
K. Ha, W. Hu, and B. Amos, “Edge analytics in the internet of
things,” IEEE Pervasive Computing, vol. 14, no. 2, pp. 24–31,
2015.
[3] M. Jia, J. Cao, and W. Liang, “Optimal cloudlet placement Saira Gillani received her PhD degree in
and user to cloudlet allocation in wireless metropolitan area Information Sciences from Corvinus Univer-
networks,” IEEE Transactions on Cloud Computing, vol. 5, sity of Budapest, Hungary in december 2015.
no. 4, pp. 725–737, Oct 2017. PLACE She is currently serving as an assistant pro-
[4] S. Yi, C. Li, and Q. Li, “A survey of fog computing: concepts, PHOTO fessor in Saudi Electronic University, Jed-
applications and issues,” in Proceedings of the 2015 Workshop HERE dah, Saudi Arabia. Previously, she worked
on Mobile Big Data. ACM, 2015, pp. 37–42. as research scholar in Corvinno, Technology
[5] T. Taleb, S. Dutta, A. Ksentini, M. Iqbal, and H. Flinck, Transfer Center of Information Technology
“Mobile edge computing potential in making cities smarter,” and Services in Budapest, Hungary. Her areas
IEEE Communications Magazine, vol. 55, no. 3, pp. 38–43, of interest include Text Mining, Data Mining, Machine Learning,
March 2017. VANET, Mobile Edge Computing and Internet of Things.
[6] M. Satyanarayanan, “The emergence of edge computing,” Com-
puter, vol. 50, no. 1, pp. 30–39, 2017.
[7] E. K. Markakis, I. Politis, A. Lykourgiotis, Y. Rebahi, G. Mas-
torakis, C. X. Mavromoustakis, and E. Pallis, “Efficient next
generation emergency communications over multi-access edge
computing,” IEEE Communications Magazine, vol. 55, no. 11,
pp. 92–97, NOVEMBER 2017.
Ejaz Ahmed (S’13, M’17) received his
[8] P. Mach and Z. Becvar, “Mobile edge computing: A survey
Ph.D. (Computer Science) from University of
on architecture and computation offloading,” IEEE Commu-
Malaya, Kuala Lumpur, Malaysia in 2016. He
nications Surveys Tutorials, vol. 19, no. 3, pp. 1628–1656,
PLACE is an associate editor of IEEE Communication
thirdquarter 2017.
PHOTO Magazine, IEEE Access, KSII TIIS, Elsevier
[9] E. Ahmed, A. Ahmed, I. Yaqoob, J. Shuja, A. Gani, M. Imran,
HERE JNCA and Elsevier FGCS. His areas of interest
and M. Shoaib, “Bringing computation closer toward the user
include Big Data, Mobile Cloud Computing,
network: Is edge computing the solution?” IEEE Communica-
Mobile Edge Computing, Internet of Things,
tions Magazine, vol. 55, no. 11, pp. 138–144, 2017.
and Cognitive Radio Networks. He has also
[10] X. Huang, R. Yu, J. Kang, Y. He, and Y. Zhang, “Exploring
received a number of awards during his research career.
mobile edge computing for 5g-enabled software defined vehic-
ular networks,” IEEE Wireless Communications, vol. 24, no. 6,
pp. 55–63, Dec 2017.
IEEE COMMUNICATIONS MAGAZINE 10

Ibrar Yaqoob received his Ph.D. (Com- Muhammad Imran is currently working
puter Science) from the University of Malaya, at King Saud University and visiting scien-
Malaysia, in 2017. He has published a num- tist with Iowa State University. His research
PLACE ber of research articles in refereed interna- PLACE interest includes MANET, WSNs, WBANs,
PHOTO tional journals and magazines. His numerous PHOTO M2M/IoT, SDN, Security and privacy. He has
HERE research articles are very famous and among HERE published number of research papers in ref-
the most downloaded in top journals. His re- ereed international conferences and journals.
search interests include big data, mobile cloud, He serves as a Co-Editor in Chief for EAI
the Internet of Things, cloud computing, and Transactions and Associate/Guest editor for
wireless networks. IEEE (Access, Communications Magazine), Computer Networks,
Sensors, IJDSN, JIT, WCMC, AHSWN, IET WSS, IJAACS and
IJITEE.
IEEE COMMUNICATIONS MAGAZINE 11

s
red
nd
Cloud Data Centers

Hu
sd
an
us

Edge Gateways
o
Th
s
ion

Smart IoT
ill

Devices
M

Layered Model for Cloud Edge-based IoT Service Delivery

Taxonomy of IoT-based Edge Computing Environment

Network Computing Computing Service Level Major Applications Attributes

Paradigms Objectives Data Types
Technologies Nodes Enablers

WiFi Cloud Latency Networking Non Real- Smart Home Low Latency
Servers Computing time
Minimization Technologies

3G Fog Network Software Soft Real- Health Care Proximity

Base Stations Development time
Computing Management Kits

Cost Resource Rich Hard Real- Video Location

4G Routers Cloudlets Network time surveillance awareness
Optimization
Elements
Dense
Energy Smart Grid Geographical
5G Vehicles MEC Virtualization
Management Distribution

Mobile Network
Bluetooth Resource Smart Cities Context
Ad hoc Awareness
Cloud Management

Hybrid Data Smart Logistics

Ethernet
Computing Management
Environment
Monitoring
IEEE COMMUNICATIONS MAGAZINE 12

Edge
Server

Thermal
Edge Power
Server

Edge Factories
Server
Human Smart
Gadgets

Cloud

Edge Offices
Edge
Server Server
Smart Cars
View publication stats