International Journal of Scientific & Engineering Research, Volume 8, Issue 4, April-2017 154

ISSN 2229-5518

Edge Computing: Needs, Concerns and

Challenges
Vaishali Sharma1 Vipul Pratap Rai2 Dr. K K Sharma3
Sagitech Solution San Jose University Lingaya’s University
California, USA California, USA Faridabad, Haryana
[email protected] [email protected] [email protected]

Abstract— In numerous parts of computing, there has been a continuous issue between the centralization and decentralization
aspect which prompted to move from mainframes to PCs and local networks in the past, and union of services and applications
in clouds and data centers. The expansion of technological advances such as high capacity mobile end-user devices, powerful
dedicated connection boxes deployed in most homes, powerful wireless networks, and IoT (Internet of Things) devices along
with developing client worries about protection, trust and independence calls for handling the information at the edge of the
network. This requires taking the control of computing applications, information and services away from the core to the other the
edge of the Internet. Relevance of cloud computing to mobile networks is on an upward spiral. Edge computing can possibly
address the concerns of response time requirement, bandwidth cost saving, elastic scalability, battery life constraint, QoS, etc.
MEC additionally offers, high bandwidth environment, ultra-low latency that gives real-time access to radio networks at the edge
of the mobile network. Currently, it is being used for enabling on-demand elastic access to, or an interaction with a shared pool
of reconfigurable computing resources such as servers, peer devices, storage, applications, and at the edge of the wireless

IJSER
network in close proximity to mobile users. It overcomes obstacles of traditional central clouds by offering wireless network
information and local context awareness as well as low latency and bandwidth conservation. In this paper, we introduce edge
computing and edge cloud, followed by why do we need edge computing, its classifications, various frameworks, applications
and several case studies. Finally, we will present several challenges, concerns and future scope in the field of edge computing.

Index Terms— Mobile Edge Computing (MEC), Internet of Things (IoT)

——————————  ——————————

1 INTRODUCTION
A. Understanding of Edge Computing? when it comes to computation of data generated at an edge
device. In addition to requesting service and content from the
Emergence: Back in the early days, when Mainframe comput- cloud, edge devices also perform the computing tasks from
ers were used, the end devices or clients were completely the cloud. Edge can take the responsibility of processing, stor-
dumb. They were meant for only taking the input and giving ing and caching data and offload both requests and responses
output to the screen. The entire load of computation was on from the cloud to the end user. Given those responsibilities to
the server. Later on, in the eighteenth century, client server be handled by the edge device, it is to be equipped well to
architecture was introduced. During those day’s clients com- perform efficiently as well as for maintaining security, reliabil-
paratively had better intelligence still most of the processing ity, privacy etc. Recent studies show that the results of data
was done on server side. Recently, in 2015, the concept of In- analysis that were retrieved from edge computing were in
ternet of things came up which transformed the end devices near real time. It reduced the maintenance as well as the oper-
into smart devices. Now when millions of end users send te- ational costs as the number of data center’s that were needed
rabytes of data to the server, the computation power of a stan- to be maintained were reduced. In addition to that, it reduced
dalone centralized server is not enough to process huge data. the network traffic as well and also improved latency.
Also there is not enough bandwidth to allow all these devices
to communicate constantly with the server. This led to the Use Cases: Usually, Digital surveillance cameras used to send
concept of edge computing i.e. pushing intelligence to the continuous data, earlier to the back end server 24 hours a day
edge of the network. 7 days a week and keeping a note of the notion but now the
entire statistics is stored on client side and goes to the server
B. Why do we need Edge Computing? only when there is a notion. Envision has a large network of
20,000 wind turbines. There are 3 million sensors which are set
Edge computing is required as the computation power of a up on these turbines and they produce up to 20 terabytes of
single server is not enough to handle terabytes of data coming data at a time. By implementing, edge computing, envision
from the end users. Also there is not enough bandwidth to has reduced its analysis of data from minutes to few seconds,
allow all these devices to communicate constantly with the which increased the efficiency of wind turbines’ production by
server. Secondly, cloud computing is not always efficient 15 percent. Besides that, there are many other real world im-
IJSER © 2017
http://www.ijser.org
International Journal of Scientific & Engineering Research, Volume 8, Issue 4, April-2017 155
ISSN 2229-5518
plementations of edge computing as well. Palo Alto city in edge computing and combine it with cloud so as to reap the
California, is working towards making the parking spaces benefits of both as distributing the computation across mul-
smart. tiple edge devices and at the same time the user pays how
much resources he uses on the cloud, has the flexibility to
C. Needs, Challenges and concerns access the servers for which he has subscribed for from any-
where from any device at any point of time.
Needs: In many scenarios, it makes sense to compute and store
data on the edge devices rather than pushing everything to What are the differences between an edge cloud and conventional
cloud. For example, considering the case cloud?
of important matters related for human survival like health
and safety concerns, due to restricted access in cloud, these In conventional cloud technology, when a device attempts to
crucial information is not shared which could have benefitted communicate with any other device, the communication hap-
many individuals. By allowing computation and storage of pens through the cloud for example if someone sends a text
data locally, edge computing can implement collaboration message to another person, it first goes to the cloud, gets
across stakeholders. Not only that, with the help of sensors stored there and then reaches the other person whereas in
and controllers, homes can be made smart. But terabytes of edge cloud, the text message directly reaches the receiver
data regarding the whereabouts of home if uploaded to the without having to pass through the cloud. Another difference
server outweighs the computation power and bandwidth re- between edge cloud and conventional cloud is that cloud is
quired for processing it. But this problem can be outwitted if used when the location where the application is deployed is
this huge amount of information can be executed and stored known whereas edge cloud is used when the application is
locally. In a similar way the idea of building smart cities for deployed all over the world. So, in general the cloud where
the favorable and safe existence requires involvement of cut- any service which is deployed in cloud and the end users do
ting edge technology, human coordination and connection not participate in processing of any data is conventional cloud
between smart cities. whereas a cloud hosted service from any of the vendors such

IJSER
as AWS, Azure or Open shift with the computation distri-
Challenges: There needs to be efficient discovery mechanisms buted across edge devices is called as edge cloud.
to search appropriate edge nodes to decentralize computation
tasks. In this case, manual search won’t make much difference E. Understanding of Edge Cloud Computing.
as there will are thousands of end devices. Along with the dis-
tribution of computation and storage of information, there What is an Edge Cloud? (Definitions): When, a service imple-
should be a mechanism to keep the system up and running menting edge computing is deployed in cloud, it is known as
even if one of the edge node fails. This needs technologies Edge Cloud. Edge cloud is mostly beneficial over traditional
such as high availability where for every edge node, there will approaches where the data center is geographically far away
be a replica (either in active or passive mode so that if the from the user.
main edge node fails, it can immediately take over without
causing any disruption in the system. Also, each device needs What are the differences between an edge cloud and conventional
to be confined once, it fails to avoid failure of the entire sys- cloud? (Definitions): In conventional cloud technology, when a
tem. device attempts to communicate with any other device, the
communication happens through the cloud for example if
Concerns: Offloading of tasks among the end users is a chal- someone sends a text message to another person, it first goes
lenge as, when the end devices are selected for task offloading, to the cloud, gets stored there and then reaches the other per-
it is done in a random manner without considering the capa- son whereas in edge cloud, the text message directly reaches
bilities or location of the edge nodes. Also, it’s needed to be the receiver without having to pass through the cloud. Anoth-
ensured that the edge devices achieve intended throughput er difference between edge cloud and conventional cloud is
and perform efficiently even if they are handling extra load. that cloud is used when the location where the application is
Other concerns include how to use edge nodes securely and deployed is known whereas edge cloud is used when the ap-
reliably etc. plication is deployed all over the world. So, in general the
cloud where any service which is deployed in cloud and the
D. Understanding of Edge Cloud Computing. end users do not participate in processing of any data is con-
ventional cloud whereas a cloud hosted service from any of
What is an Edge Cloud? (Definitions): When, a service imple- the vendors such as AWS, Azure or Open shift with the com-
menting edge computing is deployed in cloud, it is known as putation distributed across edge devices is called as edge
Edge Cloud. Edge cloud is mostly beneficial over traditional cloud.
approaches where the data center is geographically far away
from the user. F. Needs, Challenges and Concerns

Why do we need Edge Cloud? (Motivations and needs): When it comes to edge cloud, the major challenge for pro-
grammers is to write codes in different programming lan-
We need Edge cloud so that we can implement guages to suit to the platform and environment of various end
IJSER © 2017
http://www.ijser.org
International Journal of Scientific & Engineering Research, Volume 8, Issue 4, April-2017 156
ISSN 2229-5518
devices for example the language to choose, the frameworks to
select etc. In addition to that an efficient naming convention
needs to be maintained as there are multiple applications run-
ning in the edge nodes and for a vendor to provide a service to
these nodes, there has to be standardization in programming,
frameworks, data transfer etc. Overall there needs to be diffe-
rentiation of services, flexible design for extension to other
devices when needed, isolation of a node when it fails and
maintaining reliability throughout the lifecycle of the system.

2 EDGE CLOUD AND SERVICES CLASSIFICATION

2.1 Edge Cloud classifications

Edge computing is pushing the computing, storage, analytics,

data, applications and services towards the edge rather than
having a centralized structure where everything is done by the
central unit. There are many ways to implement the edge
computing paradigm. Edge computing cloud networks can
take many shapes and form i.e. infrastructures and frame-
works. The way an edge computing infrastructure is laid out
depends on the application it is going to serve and the edge Figure 1 Mobile edge computing [11]
clouds purpose. Therefore, when classifying the edge compu-

IJSER
ting cloud, it is best to classify different edge architectures Introducing a mobile edge platform at the cellular network
with relation to the use cases. This way of classifying the edge allows the application to execute their need for computation
cloud and its services helps to bring a clear view about differ- directly at the cellular tower. This reduces the amount the data
ent architectures used while implementing edge cloud compu- to be carried to the remote data center. These Mobile Edge
ting. clouds can be hosted with commodity hardware i.e. with regu-
lar CPU’s, memory hardware and network interfaces. Mobile
1. Edge clouds at mobile towers: operators define the type of the traffic that should be for-
warded to the cloud and the type of the traffic that should be
Many mobile applications rely upon the data provided by processed at the Mobile Edge clouds. This decision can be tak-
the remote servers, data centers. To provide data to the mobile en by considering of privacy and the application provider’s
applications and services mobile networks work to transfer demands. The application developer demands can be specified
the data from the mobile phone to datacenter and vice versa. by sending the additional information at the header of a HTTP
This leads to huge amount of data transferred back and forth Request by the mobile application which are requesting data
from the datacenters. It is estimated that bandwidth require- and processing. Also, the traffic originating from the devices
ments will continue to double each year [mec1] and the advent can be marked as transparent and blocked to make sure only
of 5G networks only partially solve the problem of bandwidth the transparent data would be processed at the edge location.
constraints. To mitigate the bandwidth problem mobile edge If the packets are marked for processing at the Mobile Edge
clouds are being proposed. The mobile Edge clouds are lo- cloud and the application VM is not present at the Mobile
cated at the cellular network base stations. These mobile edge Edge server those packets are routed to mobile networks core
clouds change the cellular network base stations from for- handling the requests. As the packets are rerouted transpa-
warding points to computational edge cloud locations.” Mo- rently developers can use the same URL to access the Mobile
bile edge computing eco-system has four main stakeholders: Edge Computing infrastructure or backend server if needed.
1) Mobile end user Equipment (UE) 2) Network operators
maintaining the base station 3) Internet Infrastructure provid- 2. Edge computing as IOT devices aggregator:
ers 4) Application services provider” [11].
Many of the IOT devices are sensors and data collection
units which send lot of data to the cloud for process. “CISCO
estimated that there will be 50 billion devices which will be
connected to the internet and produces data touching zetta-
bytes” [14]. So, the regular centralized cloud based solution is
not a feasible idea. To help reduce the bandwidth load pro-
duced by IOT devices edge computing at the IOT level is pro-
posed.

IJSER © 2017
http://www.ijser.org
International Journal of Scientific & Engineering Research, Volume 8, Issue 4, April-2017 157
ISSN 2229-5518

Figure 2 Edge computing for IOT [15]

Figure 3 In premise Edge [15]

IJSER
IOT data filtering and aggregation can be used to in indus-
trial mining and oil rigs, where the individual sensors produce Thus, this kind of on premise edge computing server acts as
high quality, bandwidth consuming data constantly. But not a backup in case of outages and failures. These on premise
all the data produced by those sensors are necessary to main- edge computing servers can also use as the extension of the
tain and manage the industrial process running. There is a cloud where you can process and store the sensitive material
good need of filtering and aggregating the data before sending for a public cloud.
it to remote datacenter for processing, Edge computing here
help to weed out the unwanted data from the streams of data .
provided by the sensors and aggregate them to a format best
suited for processing and utilization 2.2 Edge Cloud Services classification

3. On-premise Edge computing:

The need to keep up or increment accessibility of IT and its

systems is quite often a top of mind concern. Cloud computing
has dependably been a centralized design. Edge computing
changes cloud computing into a more circulated computing
cloud engineering. The primary preferred standpoint is that
any sort of disturbance is restricted to just a single point in the
system rather than the whole system. A Distributed Denial of
Service DDoS assault or a dependable power blackout for in-
stance would be constrained to the edge computing device
and the neighborhood applications on that device instead of
all applications running on a centralized cloud server farm.
Organizations that have moved to off-premise cloud compu-
ting can exploit edge computing for expanded redundancy
and accessibility. Business basic applications or applications
expected to work the center elements of the business can be
duplicated nearby. “An approach to think about this is a resi- Figure 4 Edge service classification [11]
dential area utilizing an extensive shared water supply as
principle source as represented in Figure. Ought to this water
supply be hindered because of an interruption created in the 1) Augmentation:
primary supply or dispersion organize, there is a crisis water
tank situated in the town” [16]. Edge computing clouds are very ideally placed to identify
the location of the user since the requests they receive are
mostly local. When a request is received by an edge server the
IJSER © 2017
http://www.ijser.org
International Journal of Scientific & Engineering Research, Volume 8, Issue 4, April-2017 158
ISSN 2229-5518
very fact that this edge server received the request narrows ting servers, servers are fit for isolating traffic streams and
down the geographical location from which such a request diverting traffic to different goals. A use of this class is inter-
may originate. Using this information as a base the edge serv- facing enterprise users straightforwardly by means of edge
er can further identify the exactly location of the device by stations sent on enterprises housetops to the enterprise net-
reply to the device from where the request originated or can work. For instance, this applies on games/music occasions
define a protocol to identify the exact location of the device. where cameras getting extra perspectives communicate their
This can be done with very less bandwidth usage compared substance among users in the cell. Moreover, Local Breaking
with the using a centralized server. This ability of the edge considers nearby redistribution of data bolstered into the cell,
cloud can be used to verify the proximity and location of the for instance, notices and data identified with the land area of
service requesting device. The location data which is identified the base edge station. In this manner, Edge computing servers
by the local edge cloud can be passed along to the main core communicate privately produced and locally important sub-
server. stance inside the phone. Traffic is routed by going around In-
2) Offloading: ternet switches, prompting to lower correspondence delay for
User Equipment and Internet service providers. Besides, net-
Even today many mobile and desktop applications are li- work transfer speed usage and power utilization is decreased,
mited by the hardware which is available to them locally at since traffic is not routed through the core network. Decreased
their machines. This had presented itself as a bottleneck for network usage has a positive contact concerning core network
designing applications which are processor and memory in- correspondence delay.
tensive. This situation has resulted in making tradeoffs be-
tween the performance of the applications and having low 5) Edge content delivery
latency when delegating the workloads to remote servers.
This kind of the situation where you should choose to have Edge cloud servers offer assets for the organization of extra
a low performing application or high latency can be eliminat- content delivery services at the network edge. Content gener-
ed by using the edge cloud computing. ally facilitated by Internet services/Content delivery networks

IJSER
is currently moved more to the network edge. Edge cloud
Offloading essentially means transferring the processing servers work as nearby content delivery hubs and serve re-
and the memory intensive task to remote servers. But by using served substance. Storing methods, not just with regards to
edge cloud servers to offload we can use the required compu- Mobile Edge Computing, can be delegated being either res-
tational capacity with very low latency compared to offload- ponsive/transparent or proactive. Transparent Caching: Cach-
ing to remote server. This will increase the computational, ing is transparent if neither the user equipment nor the Inter-
storage, bandwidth and battery capacities of the mobile and net service provider know about the storing Edge cloud serv-
desktop computational devices. Rules can be set at the user er. Proactive Caching: Content is non-transparently reserved
devices when to offload the workloads. These rules may in- before it was asked for, since it is relied upon to produce high
clude situations like offloading when there are not enough network use later. Both methodologies can be utilized either
computational resources available at the local machine to per- isolated or shared. In the isolated situation, each cache works
form the task on time. To reduce or improve the battery con- autonomously of other caches: Content as of now stored by
sumption. New form of pay to use offloading edge cloud other Edge cloud servers is not shared. In the common situa-
computing servers can take form. tion, Edge cloud servers collaborate and get content from oth-
er Edge cloud servers.
3) Content scaling
Edge computing empowers downscaling of user created traffic In fact, edge computing decreases content delivery network
before it is directed through the portable core network. Con- usage furthermore. Like distributed database management
tent scaling can likewise be connected to traffic sent by Inter- frameworks (DDBMS), edge content delivery points at putting
net servers. Scaling user equipment created content before it is away data near where they are often asked. This sort of data
conveyed to far away data centers diminishes transfer speed limitation prompts to a lessening of computational complexi-
requests of core network. For instance, picture sharing locales ty, contrasted with incorporated database frameworks. Also, it
like flickr and facebook downscale user produced content likewise diminishes access delays with deference to inactivity,
keeping in mind the end goal to diminish stockpiling requests. since correspondence ways are kept short
Downscaling user equipment content straightforwardly at the
edge likewise decreases Internet service providers core net-
work usage. Moreover, Edge mists additionally empowers 6) Aggregation
ongoing scaling of Internet infrastructure – if traffic clog hap-
pens at data centers of website, Edge cloud computing servers Rather than routing all User equipment data to core routers
can downscale traffic with a specific end goal to both decrease independently, Edge cloud computing servers are fit for total-
worry of Internet service providers core network and incre- ing comparable or related traffic and, in this manner, lessen
ment network speed. network traffic. For instance, numerous Big Data applications
like Car2Car arrangements create a considerable measure of a
4) Local connectivity comparative and locale related occasion notices which can be
With traffic being routed through Edge distributed compu- aggregated. This likewise applies in the setting of checking
IJSER © 2017
http://www.ijser.org
International Journal of Scientific & Engineering Research, Volume 8, Issue 4, April-2017 159
ISSN 2229-5518
applications where numerous gadgets measure comparable
data that can be aggregated at the edge. Because of the way
that the amount of data got by Internet service providers
would diminish, aggregation has a beneficial outcome as far as
Internet service provider’s transmission capacity usage, con- Figure 5: Mobile edge Computing Architecture [17]
trol utilization, and versatility. Nonetheless, postpone incre-
ments since data should be prepared by Edge cloud compu- There are three essential segments in the architecture:
ting servers. Since core network traffic diminishes, the same 1) Edge gadgets incorporate all sort of gadgets (both cell
applies for data transfer capacity usage, control utilization. phones and IoT gadgets) associated with the system.
Working Edge cloud computing servers comes with extra 2) Edge cloud is the less ingenious cloud sent in each of the
power utilization cost, in any case, add up to control utiliza- versatile base station. Edge Cloud have the duty of customary
tion is relied upon to diminish therefore of lower core use. system movement control (both sending and separating) and
facilitating different portable edge applications (edge medi-
cinal services, savvy following and so on.)
3) open cloud is the cloud foundation facilitated in the Inter-
3 EDGE CLOUD INFRASTRUCTURES AND FRAMEWORKS net.

A. Edge Cloud Infrastructure B. Edge Cloud (Swarm Architecture)

Edge Cloud Generic Architecture
Mobile edge computing provides access to IT services and
cloud computing services to the mobile subscribers in the
close range of Radio Access Network (RAN). Its motive is to
reduce the latency by bringing the storage capacity and com-
putation to the edge network from the core WAN. It can also

IJSER
be defined as “Mobile Edge Computing is a model for enabl-
ing business oriented, cloud computing platform within the
radio access network at the close proximity of mobile sub-
scribers to serve delay sensitive, context aware applications.”
[17]
Versatile Edge Computing offers constant RAN data (like sys-
tem load, client's area) to the application designers and sub-
stance engineers. These constant system data are utilized to
give setting mindful administrations to the portable suppor-
ters, in this way advancing client's fulfillment and enhancing
Quality-of-Experience (QoE). Versatile Edge Computing stage Figure 6: Swarm Architecture of Edge Cloud [18]
builds the edge obligation and permits calculation and admin-
istrations to be facilitated at the edge, which lessens the sys- The original sensor nets were application-specific and limited
tem idleness and transfer speed utilization of the endorsers. in scope and reach, the swarm represents a much broader vi-
Arrange administrators can permit the radio system edge to be sion, potentially connecting trillions of sensory and actuating
taken care of by outsider accomplices, this will permit to devices worldwide into a single platform abstraction. This
quickly convey new applications and edge administrations to enables the true emergence of concepts such as cyber-physical
the versatile endorsers, undertakings. and cyber- biological systems, immersive computing, and
augmented reality, and will have a tremendous impact in do-
mains such as advanced healthcare, improved energy efficien-
cy, environment-friendly living, mobility management, en-
hanced security, and many others.
Unleashing the full force of the swarm however requires leaps
forward at all levels of the reflection chain, from the gadget
level through universal remote availability up to the applica-
tion levels.
The "vanishing" part of the swarm – requiring that hubs are
incorporated with and are a necessary piece of their condition
– be it dividers, protests, garments or even the human body
requests execution systems that go a long way past the custo-
mary silicon hardware. It might incorporate "More than
Moore" (silicon consolidated with different advancements) or
"Past Moore" (natural, organic, mechanical) systems.
One property that is normal between most swarm hubs is that
they give "detecting modalities" improving the hyper-world
IJSER © 2017
http://www.ijser.org
International Journal of Scientific & Engineering Research, Volume 8, Issue 4, April-2017 160
ISSN 2229-5518
with a rich arrangement of channels to interface with the a "vitality relative" framework [19] with current day ad-
physical world. Giving these capacities in a smaller than nor- vancements and designs is hard, as most systems administra-
mal and ultra-low vitality design requires inventive detecting tion or figuring abilities are enhanced for a constrained opera-
approaches, utilizing nanoscale NEMS, natural and bio- tional range. For instance, most remote connections perform
compound innovations. just well as far as vitality per bit for a settled and very much
The independent and self-managing nature of the hubs oblig- characterized information rate. The production of vitality rela-
ing them to work for their lifetime of vitality put away or tive advancements, segments and frameworks is of quintes-
reaped consolidated with their size imperatives requires sence to the acknowledgment of viable swarm stages.
hardware for calculation and correspondence whose vitality
utilization is request of greatness beneath the cutting edge of C. Heterogeneous Network for Mobile Edge Cloud Compu-
today. While inventive gadgets can in part help to address the ting:
hole, creative correspondence and calculation models are es-
sential. What's more, vitality conveyance and capacity textures In MECC (Mobile Edge Cloud Computing), MDs associate
must be considered. with the Internet and afterward to the cloud through the re-
One of the essential parts of the "swarm" is that future scaling mote associations. So, the remote wireless system is a critical
is not in the "quantity of transistors on a chip", however in the segment of the MECC engineering. With SMDs turning out to
"quantity of segments associated in the frameworks", that is be progressively universal and portable application economy
"Moore's law meets Metcalfe's law". The origination, investiga- keeping on demonstrating amazing development, future re-
tion, plan, sending and administration of swarm frameworks mote wireless systems need to bolster detonating versatile
require procedures that contrast significantly from our present information activity and ought to be advanced for versatile
understandings. All the more definitely, swarm frameworks broadband movement. The presentation of MECC puts addi-
are non-direct, non-deterministic, versatile and complex. tionally strain on remote systems since extra information re-
Additionally, the troublesome applications that are sent on the lated with cloud benefits between the MDs and the cloud must
dispersed swarm/versatile/cloud stage should be considered be transported by means of the remote systems with potential

IJSER
in light of the attributes of the stage prompting to various util- QoS as well as power imperatives.
ity measurements and algorithmic methodologies. The mean- HetNets (Heterogeneous Network) offer promising answers
ing of the reflection layers that consider a compelling mapping for these difficulties. We discuss about the structure of HetNet
of capacity on stage while guaranteeing that utility measure- for MECC, distinguishing the principle blocks. We particularly
ments are met is a to a great extent open issue. concentrate on the conventional unified cloud, of which the
figuring asset pool is put in the remote cloud and MDs can get
to the assets by associating with the current remote system. As
represented by Fig. 2.4, the HetNet for the most part compris-
.. es of two segments, i.e., macrocells and little cells, where the
previous give versatility while the last lift scope and limit.

Figure 7: Swarm OS creates virtualized platform abstraction

[18]
Figure 8: Access Nodes in Heterogeneous Network for Mo-
It is fundamental for the worldwide framework to be "giganti- bile Edge Cloud Computing [20]
cally adaptable", which means being fit for performing profi-
ciently over an expansive execution run. Acknowledging such 1. Small Cells
IJSER © 2017
http://www.ijser.org
International Journal of Scientific & Engineering Research, Volume 8, Issue 4, April-2017 161
ISSN 2229-5518
Since SMDs and cloud applications are eager for a rapid and
stable availability to the cloud, the little cells improve as a de-
cision.

2. Pico Cells
The low-fueled radio gets to hubs, which have a scope of
around 200 meters or less, are sent in Pico cells. Their get to is
interested in every cell Md. Generally, the pico cells are uti-
lized to give hotspot scope in shopping centers, air terminals
or stadiums.
Figure 9: Framework of heterogeneous network for MECC
3. Femto Cells [20]
A femto cell has a little, low power base station, normally in-
tended for MDs in a home or private company, whose get to The previous is the aggregate power devoured in the MD and
hub is alluded to as Home eNB in LTE. Like Wi-Fi, the scope the last is the aggregate postponement by means of the HetNet
extend for a femto cell is under 100 meters. The entrance to a for transmitting the calculation assignments to the cloud and
femto cell can be either controlled to a constrained arrange- getting the calculation comes about because of the cloud. De-
ment of MDs in the femto cell's get to control list (shut get to spite the fact that there is some current research on the QoS
mode), or open to every single cell Md (open get to mode). system for MCC we concentrate on the communications be-
tween the offloading choice capacity of MCC and the radio
4. Distributed Antenna System asset administration capacity of HetNet, which haven't been
A distributed radio wire framework is a system of spatially tended to in earlier work to the best of our insight
isolated receiving wire hubs associated with a typical source
by means of a vehicle medium that gives remote administra-

IJSER
tion inside a geographic region or structure. It can make little
virtual cells by circulating radio wires of large scale eNBs 4. Edge Cloud Applications and Services
crosswise over whole cell. The receiving wires are associated
with a typical handling unit by means of fiber. A. Geo Targeting

5. Transfer Nodes Users are distributed to the different region of the world and
The transfer hubs are low power base stations that can give each of those regions has their own preferences, business,
scope and limit improvement to large scale cells at the cell products and user requirement. So, to provide the region/user
edge. A transfer hub is associated with its Donor eNB (DeNB) specific data Edge cloud computing is proving to be an impor-
by means of a radio interface. Because of the insecurity of tant. Edge computing customizes the experience for each user
DeNB scope, MDs in a few areas of the large scale cell may like a different version of the homepage, serve different con-
have a disappointment in access to the DeNB. The arrange- tent and change the language preference, etc., based on the
ment of transfer hubs can take care of this issue successfully. geographical location of each user. This can be done by main-
The HetNet offers to the offloading choice capacity a scope of taining the IP-to-Geo databases for all the locations to provide
remote transmission administrations with various service low latency in lookup and response time.
classes and financial costs, and the offloading choice is made
considering both the offloading pick up and the cost of utiliz-
ing the Het-Net when a Service Level Agreement (SLA) is
built up with it. In particular, the SLAs depend on the as-
sessed normal values or most pessimistic scenario values with
a given infringement likelihood of two qualities, i.e., control
utilization and remote transmission delay.

Figure 10: Geo Tagging Based in Real-time Ge-IP database

lookups [26]

IJSER © 2017
http://www.ijser.org
International Journal of Scientific & Engineering Research, Volume 8, Issue 4, April-2017 162
ISSN 2229-5518
B. Geo-Blocking
This is another application of Edge computing, which works
similar to the Geo Targeting. It blocks or prevent the certain
content to accessed at a certain location. Such traffic filtering
requires to be taken care at the edge itself to provide better
filtering of traffic to comply with the rules, licensing restric-
tions etc.

service.

Figure 12: Augmented Reality Service [25]

Also, it’s better to analyze and process the location, camera

information on a mobile edge cloud rather than centralized
cloud as because there might be need to refresh and update
the information at a faster rate, depending upon a user moves
and in context of AR service. “In other words, augmented real-
ity data requires low latency and a high rate of data
processing in-order to provide the correct information to the
user's device depending on the location and orientation of the

IJSER
user. Performing such data processing on the MEC platform
Figure 11: Geo Blocking Based in Real-time Ge-IP database also has the advantage of collecting metrics, anonymized me-
lookups and Custom rules [26] ta-data, etc., in-order to analyze the service usage and help to
C. Augmented Reality improve the service in order to provide a better user expe-
Augmented Reality is becoming possible because of the mo- rience.” [25]
bile network support for low latency computing and high data
rates. Augmented Reality is nothing but the combination of D. Intelligent Video Acceleration
real-world and supplementary computationally produced End client Quality of Experience (QoE) and use of radio sys-
tactile information, for example, video, Graphics, GPS or tem assets can be enhanced through intelligent video accelera-
sound information. Augmented reality enhances the user ex- tion. In general Hypertext Transmission Protocol over the TCP
periences to an individual point of interest. For example, con- protocol is used for media, file streaming or downloading.
sider a person visited to an exhibition hall, museum or city Accessible limit can differ depends the size of the requested
landmark holding their cell phone towards a specific purpose data. TCP might not able to adjust sufficiently quickly to ra-
of enthusiasm with the application identified with their visit pidly changing conditions in the radio access network (RAN).
enacted. The camera captures the image of the user interest This may prompt to under-use of valuable radio resources and
and based on the analysis algorithm the application shows to a user experience.
extra data identified with what the person is viewing. The figure below shows an example of the intelligent video
Augmented Reality Service requires an application to examine acceleration service.
the output from the camera, and the location in-order to en- In this, a radio examination application, which lives in a MEC
hance the user’s experience by giving extra de- server, furnishes the video server with a sign on the through-
tails/information to the user more than what a user will re- put assessed to be accessible at the radio downlink interface.
ceive without it. This data can be utilized to help in deciding the TCP conges-
To provide accurate information the application need to know tion control, “for example, in selecting the initial window size,
the precise location of users, the direction in which they face. setting the value of the congestion window during the conges-
This is accomplished by either by location positioning sensor tion and avoidance phase, and adjusting the size of the con-
or/and by camera view. And after the analysis of data, the gestion window when the conditions on the ‘radio link’ dete-
application can give extra real time data to the user. Also, in riorate” [25].
the event of that the User moves, the data will be refreshed.
So, hosting augmented reality services on the mobile edge will
be advantageous as supplementary data relating to a point of
intrigue are profoundly local and is frequently irrelevant
beyond the point of interest. The figure below shows how a
MEC platform can be used to provide an Augmented Reality

IJSER © 2017
http://www.ijser.org
International Journal of Scientific & Engineering Research, Volume 8, Issue 4, April-2017 163
ISSN 2229-5518

Figure 13: Intelligent Video Acceleration Service [25]

The video server may utilize this data to help TCP Congestion
control choices. The content’s time-to-start as well as video- Figure 14: Connected Vehicles Service [25]
stall occurrences can be reduced and additionally video-slow
down events can be diminished, enhancing enhanced video F. Internet of Things Gateway
quality and throughput. In the IOT world, device interact with each other and with the
internet to provide the efficient service. In-order to do so, it
E. Connected Cars sends messages over the network and aggregate the message
Recently there have been huge developments in the vehicular to ensure low latency. Also for IOT applications real time ca-
network, which involves communication between vehicles pability is required.
and RSU (road side units) which intends to increase safety, IOT device generally has very limited storage and
efficiency, and convenience of the transportation system. Such processing capabilities. Thus, there is always a requirement of
communication can also be utilized to offer some incentive another system which can store and process the messages and
included services, for example car finder, parking location etc. give desirable instruction to other IOT devices. Such system
So, with the increasing number of vehicles the amount of data should have high processing capabilities and high storage.
flow will also increase, which cause latency in data delivery,
analysis and further delay in services.
For such scenarios, Mobile edge cloud computing can be
proven advantageous.

IJSER
“Mobile Edge Computing can be used to extend the connected
car cloud into the highly distributed mobile base station envi-
ronment, and enable data and applications to be housed close
to the vehicles” [25]. Such effort minimizes the round-trip time
of data. It additionally gives a layer of abstraction from both
the core network and internet applications. MEC Applications
and servers can be deployed on base stations to provide RSU
(road side unit) functionality so that it can receive the com-
munication messages for vehicles, sensor and can analyze and
can further propagate received hazard warnings and other Figure 15: IoT Gateway Service [25]
latency-sensitive messages to other vehicles in the area. Such
information can help other vehicles to react immediately re- Different gadgets are associated over various types of net-
ducing the large-scale hazards. Figure below shows scenario work, for example, 3G, LTE, Wi-Fi or other radio advance-
of connected car. Here the MEC application on the base station ments. There is a need for a low latency aggregation point to
receives the information from the car in the hazard region and manage the various protocols, dispersion of messages and for
thus propagate the warning to the cars that are nearby in the the processing and analytics. For such scenarios, mobile edge
affected area. Also, RSU will further send the information to cloud can prove to beneficial.
the centralized cloud server for further processing and report- Mobile Edge computing can be utilized to associate and
ing. control gadgets remotely, analysis of data and provide real
time provisioning and analytics. MEC provides collection and
dissemination of IoT service into the exceptionally conveyed
Edge base station, which allows applications to react imme-
diately. This reduces the RTT of the data, thus providing high
speed service and improving efficiency.

G. Big Data at the Mobile Edge

Geographically Distributed Database as a Service (DBaaS)
provides replication, storage and access of data sets globally to
enable faster performance.
Server side GUID database make use of user cookie infor-
mation to identify visitor and stores their information. Gener-
ally, cookie size is limited. “While a server side GUID data-
base can support vastly larger data sets than cookies, it often
results in performance problems when the local data center
processing a request must reach 'off-box' to a distant data cen-
IJSER © 2017
http://www.ijser.org
International Journal of Scientific & Engineering Research, Volume 8, Issue 4, April-2017 164
ISSN 2229-5518
ter housing the visitor data needed to process the request” security, and in addition streamlining measurements. edge
[26]. In such cases, Mobile edge cloud can be used as Geo- processing is here, and we trust this paper will convey this to the
graphically Distributed Database as a Service (DBaaS) which consideration of the group.
can provide the highest end-to-end performance irrespective
of each User’s geographical area and ensuring a reliable end-
user experience across all data centers at all times.

Figure 16: Big Data Service at Mobile Edge [26]

IJSER
H. Dynamic Creative Optimization for Ad Tech Figure 17: Dynamic AD Service [26]
Generating dynamically customize ad creative templates in
real time is one of the Mobile edge cloud services. This pro-
vides highly personalized messages without the need to create
and store a large set of data.

5 CONCLUSION
These days, an ever-increasing number of administrations are
pushed from the cloud to the edge of the system since preparing
information at the edge can guarantee shorter reaction time and 6 CHALLANGES
better unwavering quality. Besides, transmission capacity could
likewise be spared if a bigger segment of information could be 6.1 Privacy and Security
taken care of at the edge as opposed to transferred to the cloud. Edge-driven Computing goes past endeavors on utilizing
The thriving of IoT and the universalized cell phones changed the E2E (End-to-End) encryption and client driven privacy
part of edge in the processing worldview from information cus- frameworks that attempt to ensure clients data in the cloud.
tomer to information producer/consumer. It would be more pro- Edge-driven models will challenge analysts in new ways. Past
ductive to process or back rub information at the edge of the sys- encryption to ensure private information, more secure inter-
tem. In this paper, we thought of our comprehension of edge fi- mediaries will be required for meet, correspondence, and get
guring, with the reason that registering ought to occur at the to control utilizing different procedures like re-encryption or
nearness of information sources. At that point we list a few cases quality based encryption among others. Besides, novel secure
whereby edge registering could prosper from cloud offloading to middleware for privacy mindful data sharing must be made to
a keen domain, for example, home and city. We additionally lift edge-driven frameworks. Many existing takes a shot at
cloud security, for example, scrambled information stores,
present collective edge, since edge can interface end client and
inquiries over encoded information, homomorphic systems
cloud both physically and consistently so not exclusively is the
could add to the making of novel edge-driven administrations.
traditional distributed computing worldview still upheld, addi-
An essential difference with customary cloud security research
tionally it can associate long separation arranges together for in- is that Edge-driven Computing may accept the presence of
formation sharing and cooperation because the closeness of in- trusted or most confided in, stable assets playing out some
formation. Finally, we set forward the difficulties and openings correspondence, diligence, inquiries, and indeed, an even cal-
that merit chipping away at, including programmability, naming, culation for applications sent and controlled in the edges.
information reflection, benefit administration, protection and Edge-driven computing may likewise consider the conjunction
IJSER © 2017
http://www.ijser.org
International Journal of Scientific & Engineering Research, Volume 8, Issue 4, April-2017 165
ISSN 2229-5518
of put stock in hubs with pernicious ones in distributed edge- [3] P. G. Lopez, A. Montresor, D. Epema, A. Datta, T. Higashino, A. Lamnitchi,
based overlays. This will again require secure directing, repeti- M. Baecellos, P. Felber, E. Riviere, “Edge-centric Computing:Vision and Chal-
tive steering put stock in topologies and past P2P examine lenges”,vol. 45, no. 5, pp. 37-42, Oct. 2016.
connected to this novel setting. Another key difference is that [4] “Feeling Smart City Through Edge Computing,” Huawei Enterprise Blog, 09-
Edge-driven Computing keeps the convergence of data when Feb-2017. [Online]. Available: http://eblog.huawei.com/feeling-smart-city-
contrasted with centralized computing. Past cloud security through-edge-computing. [Accessed: 07-Mar-2017]
inquiries about on fracture of data consolidated with encryp- [5] R. Kempf, “Subscription,” Bizety, 20-Dec-2016. [Online]. Available:
tion may unite with decentralized overlay advances to guaran- https://www.bizety.com/2016/12/20/iots-rise-of-edge-computing. [Ac-
tee fitting data assurance for sensitive data. Besides, secure cessed: 07-Mar-2017].
cloud queries and calculation over divided data and lists in [6] “Feeling Smart City Through Edge Computing,” Huawei Enterprise Blog, 09-
overlay systems may make totally new models regarding the Feb-2017. [Online]. Available: http://hweblog.com/feeling-smart-city-
privacy of sensitive data. through-edge-computing. [Accessed: 07-Mar-2017].
[7] “Edge computing,” Wikipedia, 04-Feb-2017. [Online]. Available:
https://en.wikipedia.org/wiki/Edge_computing. [Accessed: 07-Mar-2017].
6.3 Scalability [8] elithecomputerguy, “Edge Computing Introduction,” YouTube, 14-Mar-2016.
Scalability is a repeating research challenge both in distributed [On-
and distributed computing settings. The design of architectures line].Available:https://www.youtube.com/watch?v=RXVcQHMNuQU&t=
that scale to many clients must consider issues like adaptation to 263s.[Accessed: 07-Mar-2017].
internal failure, churn, flexibility and numerous others. In P2P, [9] intelswnetwork, “The IoT Opportunity: Near-Real-Time Analytics from Edge
beat and dynamism entangle the possibility of these models and to Cloud,” YouTube, 28-Oct-2015. [Online]. Available:
their general administration accessibility. In distributed compu- https://www.youtube.com/watch?v=RhetYBuk7mI. [Accessed: 08-Mar-
ting, scaling and flexibility are repetitive points, what's more, 2017].
even real cloud suppliers might be overcome by the huge dissent [10] R. Roman, J. Lopez and M. Mambo. Mobile edge computing, fog et al.: A
of administration assaults. Edge-driven Computing, in any case, survey and analysis of security threats and challenges. Future Generation

IJSER
changes totally the versatility challenges introduced some time Comput. Syst. 2016.
recently. Stir is not such a constraining variable any longer, be- [11] M. T. Beck et al. Mobile edge computing: A taxonomy. Presented at Proc. of
cause of the utilization of stable cloud assets. A noteworthy test is the Sixth International Conference on Advances in Future Internet. 2014, .
a right trade between computing and correspondence duties be- [12] P. Garcia Lopez et al. Edge-centric computing: Vision and challenges. ACM
tween edge devices, trusted servers, and untrusted administra- SIGCOMM Computer Communication Review 45(5), pp. 37-42. 2015.
tions. Given that the control is at the edges, scaling issues are still [13] Y. Lin et al. Enhancing edge computing with database replication. Presented
exceptionally significant. Building gigantic overlays joining cell at Reliable Distributed Systems, 2007. SRDS 2007. 26th IEEE International
phones with constrained batteries with stable cloud resources Symposium on. 2007, .
require unique consideration for correspondence conventions [14] W. Shi and S. Dustdar. The promise of edge computing. Computer 49(5), pp.
among hubs. Besides, cloud edge administrations should likewise 78-81. 2016.
be efficient and consider the heterogeneous hubs they should [15] F. Bonomi et al. Fog computing and its role in the internet of things. Presented
serve. Another research test is the mix of scalability with security at Proceedings of the First Edition of the MCC Workshop on Mobile Cloud
in huge overlays. Edge models requiring security will force non- Computing. 2012,.
insignificant overheads due to encryption, that must be managed [16] S. Carlini. The drivers and benefits of edge computing. Schneider Electric–
to give adaptability. Data Center Science Center pp. 8. 2016.
[17] E. Ahmed, A. Ahmed, "A survey on mobile edge computing," 2016 10th
7 FUTURE WORK International Conference on Intelligent Systems and Control (ISCO), Issue
Date: 7-8 Jan. 2016
In future edge computing would unravel large portions of the [18] J. M. Rabaey, "The swarm at the edge of the cloud - A new perspective on
most troublesome issues confronting robotics and computing wireless," 2011 Symposium on VLSI Circuits - Digest of Technical Papers,
infrastructure. A swarm of air-and land-based drones inspecting Honolulu, HI, 2011, pp. 6-8.,URL:
a remote forest fire, a collapsed building or a limitless tract of http://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=5985988&isnu
farmland is today tested by a powerlessness to associate and mber=5985982
transmit extensive quantities of information over wireless sys- [19] L. Barroso, U. Hozle, "The Case for Energy-Proportional Computing", IEEE
tems or to get directions from a central controller in a timely Computer, vol. 40, no. 12, December 2007.
manner. These issues are exacerbated by the befuddling territory [20] L. Lei, Z. Zhong, K. Zheng, J. Chen and H. Meng, "Challenges on wireless
of debacle environments however; edge computing dodges these heterogeneous networks for mobile cloud computing," in IEEE Wireless
impediments. Communications, vol. 20, no. 3, pp. 34-44, June 2013.doi:
10.1109/MWC.2013.,URL:
REFERENCES http://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=6549281&isnu
[1] W. Shi, J. Cao, Q. Zhang, Y. Li and L. Xu, "Edge Computing: Vision and Chal- mber=6549270
lenges," in IEEEInternet of Things Journal, vol. 3, no. 5, pp. 637-646, Oct. 2016. [21] X. Chen, L. Jiao, W. Li and X. Fu, "Efficient Multi-User Computation Offload-
doi: 10.1109/JIOT.2016.2579198 ing for Mobile-Edge Cloud Computing," in IEEE/ACM Transactions on
[2] H. Liu, F. Eldarrat, H. Alqahtani, X. D. Foy, A. Reznik, "Mobile Edge Cloud Networking, vol. 24, no. 5, pp. 2795-2808, Oct. 2016. doi:
System: Architectures, Challenges, and Approaches" in IEEE Systems Journal. 10.1109/TNET.2015.2487344 URL:

IJSER © 2017
http://www.ijser.org
International Journal of Scientific & Engineering Research, Volume 8, Issue 4, April-2017 166
ISSN 2229-5518
http://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=7307234&isnu
mber=7590206
[22] M. Gerla, "Vehicular Cloud Computing," 2012 The 11th Annual Mediterra-
nean Ad Hoc Networking Workshop (Med-Hoc-Net), Ayia Napa, 2012, pp.
152-155. doi: 10.1109/MedHocNet.2012.6257116 URL:
http://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=6257116&isnu
mber=6257105
[23] A. Davis, J. Parikh, and W. E. Weihl. 2004. Edgecomputing: extending enter-
prise applications to the edge of the internet. In Proceedings of the 13th inter-
national World Wide Web conference on Alternate track papers & posters
(WWW Alt. '04). ACM, New York, NY, USA, 180-187.
DOI=http://dx.doi.org/10.1145/1013367.1013397
[24] Gabril Brown, “Mobile Edge Computing Use Cases and Deployment Op-
tions”, at Juniper Networks, July 2016.
[25] Y. C Hu, M. Patel, D. Sabella, N. Sprecher and V. Young, “Mobile Edge Com-
puting A key technology towards 5G” at ETS. Sep, 2015.
[26] "Edge Computing Use Cases", Mirror-image.com, 2017. [Online]. Availa-
ble:http://www.mirror-image.com/services/edge-computing/edge-
computing-use-cases/. [Accessed: 06- Mar- 2017].

IJSER

IJSER © 2017
http://www.ijser.org