2016 IEEE International Conference on Systems, Man, and Cybernetics' SMC 20161 October 9-12,2016' Budapest, Hungary

A Review on Fog Computing Technology

Firas AI-Doghmant, Zenon Chaczko2, Alina Rakhi Ajayan 3, Ryszard Klempous 4
123 Centre for Innovation in IT Services and Applications (iNEXT), School of Computing & Communications, Faculty of Engineering & IT,
University of Technology, Sydney - 2007, NSW, Australia
4 Faculty of Electronics, Wroclaw University of Technology, Poland
Email: I FirasQaisMohammedSaleh.AI-Doghman-l@student.uts.edu.au, 2 zenon.chaczko@uts.edu.au, 3 Alina.Ajayan@uts.edu.au,
4ryszard.klempous.@pwr.edu.pl

Abstract - Out of the many computing and software oriented in [3] developed an architecture for connecting vehicles to a
models that are being adopted by Computer Networking, Fog "Fog" platform deployed at Road Side Units (RSUs) and
Computing has captured quite a wide audience in Research and M2M gateways. Such a system design, in regard of its
Industry. There is a lot of confusion on its precise definition, characteristics, enables consumer centric services like M2M
position, role and application. The Internet of Things (lOT), Data Analytics with Semantic Web Technologies, loT services
todays' digitized intelligent connectivity domain, demands real discovery and Connected Vehicles Management. Another
time response in many applications and services. This renders method for implementing Fog computing was introduced by
Fog Computing a suitable platform for achieving goals of
[4], wherein an android-like Appstore is to be applied on
autonomy and efficiency. This paper is a justification of the
network devices, downgrading computations from occurring
concepts, interest, approaches, and practices of Fog Computing.
It describes the need for adopting this new model and investigate
at Cloud level to the network, while in [5] an "loT Hub"
its prime features by elucidating the scenarios for implementing which is essentially a generic Fog node positioned along the
it, thereby outlining its significance in the loT world. edges of multiple networks, enhance the networks' capabilities
through the employment of the following entities: border
Index Terms - Fog Computing, Cloud Computing, Internet of router; cross-proxy; cache; and resource directory. However,
Things, Autonomics, Computational Intelligence. the article [6] addresses the utility based matching problem
within the loT domain using Irving's matching algorithm
I. INTRODUCTION
under node-specified preferences for efficient node pairing.

The Internet of Things (loT) epitomizes the cutting edge The Internet of Things and Connected Smart devices have
technology to realize theoretical and practical facets of had an exponential growth in terms of involved technologies,
computer networking, and has initiated a prominent transition market participation and consumer approval, paving the way
in the how communication and interactions of our world. The to the evolution of Fog computing principles, gradually
strong need to remedy the growing concerns regarding dealing amassing productive opportunities in various domains such
with the huge data influx in real-time and functioning the as Vehicular networks, Body Area Networks (BAN), and the
available bandwidth bounds led to the birth of Fog computing, Smart Grid. The advantages of this computing procedure for
which, intensively but not exclusively, operate along the services in several domains are needed to be investigated [1].
network edge. Fog computing is a novel paradigm realizing Fog Computing allows greater support and better response
distributed computing, network services and storage from time to the Internet of things environment, it is suitable for
beyond Cloud Computing Data Centers up until the devices real-time service requests, and to share resource efficiently an
along the network edge. This notion extends the inherent efficient and cooperative utility based pairing strategy between
operations and services of Cloud computing, thus enabling a the high-end loT nodes is needed [6]. The Fog acts as a link
new breed of application. The primary function is to filter and between loT and the Cloud to induce the necessary extra
aggregate data for Cloud data centers and apply logical functionalities for application-specific processing like filtering
intelligence to end devices. Fog Computing is similar to Cloud and aggregation before transferring the data to the Cloud. It
computing in many traits. However, it may be differentiated should be able to decide what is to be sent (the content), how
from the former by its ingrained end devices, the intensive (data format) and when to send (time). During this process, it
spatial distribution achieved and offered mobility support. As also needs to delete some redundant or invalid data, and
Fog-based processing occurs along the network edge, the end- aggregate the complementary data in the space and time
results reflect highly improved location awareness, low dimensions [2].
latency and Quality-of-Service (QoS) in streaming and real The rest of this paper is structured as follows: Section II
time applications [1]. Fog nodes are heterogeneous devices, highlighting the previous work done in some literatures.
ranging from high-end servers, access points, set-top boxes, Section III provides some methodologies used to describe Fog
edge routers to the end devices such as mobile phones, smart Computing practices. Section IV presents a evaluation
watches, and sensors etc. [2]. scenarios for two use cases, followed by deployment case for
In the light of recent research, Fog computing has been Fog. Finally, Section V concludes the paper and provides
effectively demonstrated using several techniques. The study directions for future work.

978-1-5090-1897-0/16/$31.00 ©2016 IEEE SMC2016 001525

Authorized licensed use limited to: INTERNATIONAL ISLAMIC UNIVERSITY. Downloaded on May 08,2021 at 05:56:21 UTC from IEEE Xplore. Restrictions apply.
 2016 IEEE International Conference on Systems, Man, and Cybernetics' SMC 20161 October 9-12, 2016' Budapest, Hungary

II. REVIEW OF RELEVENT LITERATURE thanks to its' flexibility and dynamics, and so may be
In this section, some of the research conducted so far implemented in different environments with varied conditions.
related to Fog Computing is introduced, which helps to build The authors in [10] describe an Adaptive Operations Platform
and support this paper. (AOP) to provide an end-to-end manageability for enabling
Fog Computing infrastructure according to the operational
A significant number of characteristics helps us identify requirements of industrial processes, while [4] outlines a
the Fog as an inconsequential extension of the generic method for transforming the computation environment from
' Cloud'. In terms ofIoT, the work in [7] condenses this vision the Cloud to the Fog by presenting an Appstore kind of system
and defmes the key features of Fog Computing which make it on network devices wherein the user can choose which data
the appropriate platform for a number of critical Internet of should be processed at the edge and which ones on the cloud.
Things (loT) services and applications. It defmes the Fog node This involves tagging packets which are to be processed on
aspects such as: a) Mobility; b) Low latency and location the network device. Untagged packets are then sent directly to
awareness; c) Wide-spread geographical distribution; d) Very the cloud without any intennediate processing. However, the
large number of nodes; e) the predominant role of wireless work in [11] presents the ' reliability defiance ' posed by the
access; f) Strong presence of streaming and real-time current computing paradigms, and expands the debate towards
applications, and g) Heterogeneity. Similarly, [8] discloses the reliable Fog platforms encompassing smart devices' networks
reasons as to why Fog Computing is the most-fitting natural communicating amongst themselves and with the Cloud. [6]
computing platform for loT through the main requirements for delivers an efficient loT-node pairing scheme between the
designing and building a scalable and adaptable loT platfonn same-domain loT nodes in the Fog context. The utility based
which include support of rapid mobility patterns, support of matching or pairing problem within the domain of loT nodes
systems requiring reliable sensing, analysis, control and is addressed utilizing Irving's matching algorithm under the
actuation and management of a vast amount of geographically node specified preferences, to ensure stable loT node pairing.
distributed "things" whilst in [1] the many advantages offered Interoperability, the keystone of loT systems and defmed as
by Fog computing for services in various areas is examined the ability in heterogeneous objects to inter-operate
along with an analysis of the cutting-edge developments and dynamically with minimal human intervention, together with
security issues of this paradigm. It also includes the ideation loT nodes' strong-fort - the diverse range of network and
that some inventions in computation and storage may be physical resources, allow their pairing for sharing resources
inspired in future to address data intensive services based on amongst each other cooperatively to achieve user specific
the interplay between Fog and Cloud. Moving onto [2], the requirements. A refmement is proposed in the classical stable
characteristics and prospects of Fog computing and services in matching or pairing algorithm , which incites Edge computing
regard of healthcare systems is studied. Here the Fog functions with a better utility factor, thereby configuring more proficient
in focus are Switching Networks, Pushing Services and Core device-to-device communication. The shared parking model
Services and the need to support protocols such as ZigBee, created in [12] depicts deployment of the Fog solution and
WiFi, 2G/3G/4G, WiMax, 6LOWP AN, in contrast to the sole Roadside-Cloud concepts into the parking problem. The status
support offered by the Cloud for TCP/IP. The various of a parking slot - vacant or reserved - will be updated in the
communication protocols transfer data in different formats so Fog server installed in local areas, which in turn delivers
the first thing the Fog should do is network switching between information on the managed empty spaces to Road Side Units
loT and the Cloud. Here the Fog will act as an intermediary or RUSs. At each RSUs, the RFPARK communicates with
between end devices and the Cloud, and should provide these fog servers in order to direct drivers to an optimal space.
pushing service to both, while ingesting acquired data and [13] surveys the expansion of the Fog concept onto the
updating processed data onto the Cloud for long-time storage decentralized smart building control, recognizes Cloudlets as a
and deeply digging in parallel. Thus the prime function of the distinctive case of Fog computing, in conjunction with the
Fog is to achieve local data processing, storing and computing Software-Defmed Networking (SDN) scenarios. Cooperative
in devices of weak performance metrics. data scheduling and adaptive Traffic Light problems in the
SDN based Vehicular Networks scene, as well as Demand
Owing to its recent introduction and emergence, there is no
Response in Macrostation and Microgrid based Smart Grids
available standard architecture regarding Fog-based resource
are discussed.
management. [9] presents a simple model for this purpose, by
taking into account resource prediction, resource allocation,
and pricing in a realistic and dynamic way, while also
considering customers' type, traits, and characteristics. This 2.05
Content
model is adaptable to the requirements of different CSPs
Investigated Aspect Traditional Cloud Fog Computing
Prediction Latency 5 seconds 1.5 seconds GETIPOST/PUT/DELETE URL..

~~:s 1~~~~2=. X~X~+=I.="'~~~~_il. ~=~~2='~.~+_"'_"_=_~_J

{
Webpage display
8 seconds 3 seconds
latency
Internet Traffic 75 Kbps IOKbps Cached {I I _~~~~~~~~_
Access

Hardware used Amazon Web Server Raspberry Pi

Fig. 1. Interaction between clients and heterogeneous smart objects with the
Table 1. ComparISon of Results between Cloud and Fog mediation of the loT Hub [51

SMC2016 001526
Authorized licensed use limited to: INTERNATIONAL ISLAMIC UNIVERSITY. Downloaded on May 08,2021 at 05:56:21 UTC from IEEE Xplore. Restrictions apply.
 2016 IEEE International Conference on Systems, Man, and Cybernetics' SMC 20161 October 9-12, 2016' Budapest, Hungary

Th~ evolution of loT, according to its heterogeneous and and overall network performance.
dynamIc nature, had undergone paradigm shifts many times
Ho~ever, [8] e~amines some of the loT's promising and
over in terms of the design of network architectures and many
challengmg scenanos, discusses the inevitable chemistry
apP!oaches .have been presented and proposed. An important
between the Fog and the Cloud Computing in the near future
?eslgn and Impl~mentation of a new Fog node was presented
m [5], whe~e It IS p.laced along the edge of multiple physical a?d . review. some of the technologies that will require
slgmficant Improvement, so as to bolster the applications
networks wIth the aLm of creating an IP-based loT network to
scope for the loT market, whereas [10] presents a deployment
be used as the infrastructure for deploying loT applications.
case for Adaptive Operations Platform (AOP), followed by
Performance evaluation shown that the loT Hub successfully
~valuat.ion sce~ario~ for two use cases. It also develops an
had ma~aged a number of heterogeneous physical networks,
mdustrIaI solutIOn m the context of predictive maintenance
each with several connected devices, with limited resource
applications, using the Fog computing paradigm and the DMo
usage,..in terms of processing capacity and memory, thus
expedltmg the deployment an loT Hub even on low-end technolo?y. Finally, the studied article [13] states the
devices, such as RPis. The article [3] examines an architecture prospectIve role of Fog computing in the important class of
Cyber~Physical Systems (CPS), in the light of stimulating
~or c~nnected vehicles with RSUs and M2M gateways
scenanos, demonstrated by Smart Grids Vehicular Networks
mcludmg the Fog Computing Platform. M2M data processing
with semantics, discovery and management of connected Wireless Sensor and Actuator Netw~rks, Smart Buildin~
Control and that Cloudlets are an important special case of
vehicles are also briefly examined likewise as Consumer-
Fog Computing ..
Centric ~oT ser.vices enabled by the prominent features of Fog
Computmg whIle [14] had the goal of creating a mathematical
III. METHODOLOGY
model for Fog computing and assessing its applicability in the
loT context where it is pivotal to meet the demands of the In this section we will introduce a review of the
latency-sensitive applications rurming along the network-edge methodologies used to describe Fog computing practices.
for analyzing its suitability within the framework of loT. The T~e publishing [12] addresses the problem of Car Parking
work a~so conducts a case study on comparative performance ~specIally along the choice of an ideal slot; amatching method
eva~uatIOn of Cloud Computing with that of the Fog for an
LS used at RFPARK as the problem' s solution. This article has
enVLronment involving a high number of Internet-connected the proposed Parking slots association problem formulated in
devices demanding real-time services. It proves that for the ~he F~g-Cloud environment as a many-to-one matching game
S~N based Vehicular Networks scene, as well as Demand
m which a set of vehicles ' V' has been designated towards a
hLgh number of latency-sensitive loT applications the presence set of parking lots 'L'. The preferences concept formulated the
of Fog Computing enhances network performance in regard of
model in ~er~s of co~on and conflicting interest. A parking
power consumption, decreasing service latencv as well as cost
slot aSSOCIatIOn algonthm detects a stable matching of parking
~r--------------------------------------'
- - - Java Heap Max Alaocellon slots for the vehicles extending requests.
- - Java Heap CUlTent Usage

40000 The methodology used by [4] to propose a method of

moving the computation from the cloud towards the network
was to use. a Raspber~ Pi based Fog device by connecting
three Ardumo boards WIth temperature sensors to the Wi-Fi of
the Rasp?erry Pi, they then measure the temperature of the
surroundmgs every 5 seconds and send it to the router (the
10000 rasp?erry), the raspberry invokes a python script once it
receIves the temperature data that writes the received
OO~--~500~--~IO~OO--~I~500~--~==--~~~---3~OOO-----35~OO temperature values in different files (one file per Arduino
Seconds (5) board). The time series prediction is applied to the data in each
(a) file, the resu It of the time series prediction is then written into
a MySQL database instance running on the A WS cloud the
- - % CPU Usage PHP instance on the cloud reads the values from MySQL and
displays it on a webpage. The results (Table 1) showed that
Fog based architecture has a better response time compared to
the cloud architecture.
The "lo.T Hub" proposed by [5], with a Fog node deployed
along multIple networks' edges is implemented using a Java-
based implementation of CoAP called Californium and other
r~lated.drafts . The loT Hub has been deployed to a Raspberry
PI (RPL) Model B single board computer and used to manage
resources hosted by heterogeneous SOs within a real-world
loT testbed within our department. The added value of the loT
(b ) Hub is its capability to hide completely the diverse nature of
~mart objects which can interact with them using uniform
Fig. 2. Performance evaluation: (a) Heap memory used (dimension: [MB]); (b)
CPU usage [5[. mterfaces and without requiring any prior configuration.

SMC2016 001527

Authorized licensed use limited to: INTERNATIONAL ISLAMIC UNIVERSITY. Downloaded on May 08,2021 at 05:56:21 UTC from IEEE Xplore. Restrictions apply.
 2016 IEEE International Conference on Systems, Man, and Cybernetics' SMC 20161 October 9-12, 2016' Budapest, Hungary

Smart objects' interaction through the loT Hub occurs as

shown in Fig.I. The perfonnance evaluation has shown that
the loT Hub is able to manage a number of heterogeneous
physical networks, each with several devices, with a limited
use of resources such as processing capacities and memory
requirements, thus allowing the deployment of an loT Hub
even on low-end devices such as RPis. The results are shown
in Fig. 2. Meanwhile the model introduced by [9] delivering
an effective proficient resource management loT framework
the Fog entails, was thereafter implemented with NodeplJra
JavalNetBeans 8.0 tools, later evaluated and analyzed with the
help of using CloudSim 3.0.3 toolkit. The choice was due its Fig. 3. Example comparison result showing how proposed matching algorithm can
capability to adapt to the dynamic requirements of dissimilar improve total utility of node pairs 161.
CSPs. It is applicable in various environments of widely
Adaptive Operations Platform
different scenarios with respect to the resource prediction,
allocation and pricing in in a dynamic and flexible way.
The proposed matching algorithm had its performance
criteria addressed in [6] through simulation. The core concept
in Irving' s matching algorithm deals with solving the stable
roommate problem for the purpose of construction of a one-to-
one stable matching scenario. The refmement may then Opem or

initiate a one-to-many cooperative pairing or matching amidst End to end Te~ mu nie8tion Se rvices
the nodes. Refming the Irving' s procedure was to sustain 'I-Provi
-·-si·--'I
onina 1 Bootstrap I 1Mointenance I 'I- I....
- nt-O<'/
--,
quota-based nodal pairing on either one-to-one or one-to-
many pairing. The corresponding performance measures are
portrayed in Fig 3. Comparison results shows the efficiency of
imparted by the Irving' s algorithm for 5 node pairs with quota. loT
Infrastructure
It outperfonns the greedy algorithm where the nodes are
paired by considering the neighboring nodes available for
pairing. Such levels of efficiency is the combination of quota
based approach and the best utility based selection of nodes. It loT
also that the cooperation between loT nodes tremendously End Points

increases the total utility of such pairing actions. This

phenomena also indicates that with a large number of nodes'
Fog Instruments
sets and quota based pairing, overall utility of the entire set of
node domain will increase drastically.
Fig. 4. Architecture of the Adaptive Operations Platform (AOP).
The work indexed in [10] defmes an Adaptive Operations
Platfonn (AOP) to address key limitations that an industrial IV . ASPECTS AND SENARIOS OF FOG COMPUTING
infrastructure poses to data-intensive loT applications. AOP is Since its introduction, many proposals were made to
structured upon the service capabilities of the following implement Fog Computing. The following are some scenarios
layers: the Fog Infrastructure that includes networking which were been introduced in order to implement it.
equipment with particular Fog capabilities and provides for
end-to-end communication services and the Operational A. Scenario 1: Monitoring the status ofa deployed
Support System (OSS) that leverage the Fog Infrastructure to machinery
provide the customary asset management and business support
functions (e.g., inventory, maintenance, provisioning, etc.). To Paper [10] introduced a scenario consist of N sensors
efficiently leverage key features of the Fog Infrastructure, monitoring the status of the deployed machinery to evaluate
AOP includes several functional elements. The Model the performance of Fog components deployed in the network.
Building (MB) functional element combines static infonnation The sensors report the captured values regularly through a
about the failure models of the equipment types found in the router to a central server, as displayed in Fig. 5. The
industrial site along with dynamic data collected during the measurements are stored in a database and are illustrated to an
latter' s operation. The Rule Mapper (RM) functional element operator. The objective of deploying the aggregators and the
tasked with mapping the fused model to a set of traffic Sensor measurements is to investigate potential anomalies in
handling rules understood by the SDN infrastructure. The the monitored machinery. The investigated scenario is tested
Rule Deplorer (RD) functional element which, given a set of using from the one hand a common router (see Fig. 5) and
traffic handling rules and a description of capabilities in the from the other a router/gateway supporting DMo (see Fig. 6),
SDN infrastructure, computes the deployment plan to apply with the aim to compare the centralized scheme to the APO
this set of rules on the suitable elements. This model can be approach. In this scenario, as a first use case the reduction in
modified as in Fig. 4 to describe a more complete loT model. data received by the operator, stored in the Historian, using a

SMC2016 001528

Authorized licensed use limited to: INTERNATIONAL ISLAMIC UNIVERSITY. Downloaded on May 08,2021 at 05:56:21 UTC from IEEE Xplore. Restrictions apply.
 2016 IEEE International Conference on Systems, Man, and Cybernetics' SMC 20161 October 9-12, 2016' Budapest, Hungary

[70, 100]. Obviously, the range can be set differently

according to thresholds. The results obtained using this
approach are illustrated in Fig. 7. As expected, we obtain a
huge decrease in the number of transmitted packets. Fig. 7,
also depicts the CPU utilization of the router. We observe that
the use ofDMo claims only 8% of the CPU for a 113 reduction
in the received traffic.
A second use case aims to demonstrate the capability
of the proposed system to change dynamically the rules when
needed. Specifically, the temperature of N = 6 appliances
inside a rack is wanted to be measured and created an alert to
the operator only when the temperature of some appliances is
above a certain threshold (anomaly). We need also to identify
Fig. 5. Scenario using a normal router 11 01. the appliance that has the anomaly. We use 6 sensors, one for
each appliance, and one sensor which measures the
temperature inside the rack. As in Use Case 1, the normal
behavior of the appliance is when the average value calculated
in 120s is equal to 60. In order to save bandwidth and
resources in the server we have created a rule in the DMo
router to send to the server only measurements from the rack
sensor. Looking at the Fig. 8, in normal conditions we have an
amount of packets received equal to 200. Let us consider that
after a while (at time 720s), the temperature monitored by the
sensor rack increases (from 60 to 80). This provides an
estimate that an abnormal event occurs inside the rack without
designating which appliance has an anomaly.
In this case, the server receives values above the threshold
(b) (set to 60) and using the RM it sends a new rule to the DMo
Fig. 6. Scenario using a router supporting DMo [lOJ. for the router to forward the data coming from all the 6
sensors. The result will lead to a temporary increase in the
router supporting DMo is introduced. One of the N sensors number of received packets, in order to identify the appliance
was considered and supposed that this sensor is monitoring with the anomaly. At this point, the ML identifies the sensor
some operational behavior of a machine (e.g. its operational which is sending data above the threshold and sets a new rule
temperature). Looking at the traffic generated in a time which instructs the DMo router to send data only about this
sensor (i.e., in addition to the rack sensor). The new rule
interval equal to 120s, it is possible to see that the
applies until the situation in the rack goes back to the normal
measurements are distributed as a Gaussian distribution.
conditions. After that, the operator will receive again the
Specifically, most of the values are in the range [50, 70]
measurements of the rack sensor only.
which corresponds to the nonnal behavior of the specific
monitored machine in the scenario. The values out of that B. Senario 2 : a Fog computiong platform
range correspond to potential anomalies of the machine.
Using a common router (Fig.5), all the measurements [15] builds a proof-of-concept fog computing platfonn,
gathered by the sensor are received by the server and stored, consisting of two fog sub-systems where OpenStack was
filling in most of the cases, the database with redundant installed on each of them. Each of the fog sub-systems
information. This also leads to increased bandwidth usage for possesses one router and three servers. The routers are
transferring the data but also it consumes resources and space connected to the Amazon EC2 cloud through WAN, aswell as
in the database. Moreover, to detect the anomalies, a process connected with each other through LAN. The routers are also
has to take into consideration a large amount of collected data integrated with Wireless AP function , so that mobile devices
due to the large number of sensors. In order to reduce the can access the fog as well as the Amazon EC2 cloud through
amount of data stored, it is possible to apply the AOP them. Four OpenStack modules were installed: Keystone,
approach. In this case, the server runs a Ma- chine Learning Glance, Nova, and Cinder. Keystone is for authentication and
(ML) algorithm (e.g., k-Means) that is trained to understand authorization; Glance is for YM image management; Nova is
the normal behavior of the appliance. Then, using the Rule a compute module with simple network functionality; and
Mapper (RM), it sets a rule to receive only data outside of the Cinder is the block-level storage module. A VM offloading
nonnal range in order to identify anomalies. These data will scheme was implemented which can migrate one VM to
be sent as events to the operator. In this case, the ML indicates another fog cluster. The latency and bandwidth provided by
that the average of the received values is close to 60, which fog and cloud were compared and fog computing has stronger
corresponds to the nonnal behavior, and sends to DMo a new advantages for clients. YM migration is essential in fog
rule that specifies to the router to forward to the server only computing; its function was implemented in two ways. First,
values that represent anomalies, e.g. , in the range [0, 50] and Fog 1 takes a snapshot of the YM to be migrated, compresses

SMC2016 001529

Authorized licensed use limited to: INTERNATIONAL ISLAMIC UNIVERSITY. Downloaded on May 08,2021 at 05:56:21 UTC from IEEE Xplore. Restrictions apply.
 2016 IEEE International Conference on Systems, Man, and Cybernetics' SMC 20161 October 9-12, 2016' Budapest, Hungary

environment improves the performance of the network in

1400 10
many aspects. The potentiality of self-management resources
9
paved the way to autonomic Fog Computing which could be
1200
8 an imperative aspect in implementing Fog Computing.
il 1000 7 ;It
c:
> 0 REFE RENCES
.~ 6 -;:;
800
OJ
a:: 5 :x [I] 1. Stojmenovic and S. Wen, "The Fog Computing Paradigm: Scenarios and
'"
Qj 600
4 5 Security Issues," vol. 2, pp. 1- 8, 20 14. [Online]. Available:
~
3
~ https:!/fedcsis.org/proceedings/2014/drp/503.html
'"
"-
400 ~ [2] Y. Shi, H. Wang, H. E. Roman, and S. Lu, "The Fog Computing Service for

•
2
200 Heal thcare," 2015 .
[3] S. Datta and B. Chri stian, "Fog Computing Architecture to Enable Consumer
o o Centric Internet of Things Services," vol. 85, pp. 6-7, 2015 .
Narmal Router Drno [4] Y. N. Krishnan, C. N. Bhagwat, and A. P. Utpat, " Fog Computing- Network
Based Cloud Computing," no. Icecs, pp. 250- 251 , 2015 .
[5] S. Cirani, G. Ferrari, N. lotti, M. Picone, G. Sri , and R. Emilia, "The loT Hub :
Fig. 7. Router CPU utilization and total received packets 1101. a Fog Node for Seamless Management of Heterogeneous Connected Smart
Objects," 2015 .
90. 1....
[6] S. F. Abedin, G. R. Alam, N. H. Tran, and C. S. Hong, "A Fog based System
Model for Cooperative loT Node Pairing using Matching Theory," pp. 309- 314,
8000
1100 2015 .
! 10. [7] F. Bonomi , R. M ilito, 1. Zhu, and S. Addepalli , " Fog Computing and
1000
~bOOO 0 Its Rol e i n the Internet of Things," Proceedings of the first edition of the
0
:E MCC workshop on Mobile cloud computing, pp. 13- 16, 20 12. [Online].
~ 10000 800 I:l

~
~
Availabl e: http://doi .
40.00 &00 S acm .org/ I 0 .1[4 5/2342509.23425 [ 3$\delimiter"026E30F$npapers2:
~ lO.OO ~ Ilpublication /doill O. [[4 5/2342 509.2342513
~ '00
< 10 [8] M. Yannuzzi, R. Mi lito, R. Serral-Gracia, D. Montero, and
10.00
100 M. Nemirovsky, "Key ingredients in an loT recipe : Fog Computing, Cloud
computing, and more Fog Computing," 2014 IEEE 19th International
0.00
Workshop on Computer Aided Modeling and Design o f Communication Links
O~~~~§~~i~~e!~~~~~~~~~~~~~ and Networks (CAMAD), pp. 325- 329, 2014. [Online]. Available:
11....
http:
Ilieeexpl ore.ieee.org/lpdocs/epic03/wrapper.htm?arnumber=7033259
[9] M . Aazam and E.-N. Huh, " Fog Computing Micro Datacenter Based Dynamic
Fig. 8. Packets received and measured values vs. time from different sources [10[. Resource Estimation and Pricing Model for loT," 20 [ 5 IEEE 29th International
Conference on Advanced Information Networking and Applications, pp. 687-
it, and then transfers the compressed data to Fog 2. Fog 2 then 694, 20[5 . [Online]. Availabl e: http:
decompresses the data and re-Iaunches the VM by using the 1lieeexplore. ieee.org/lpdocs/epic03/wrapper.htm?arnumber=7098039 [[ 0] V.
Gazis, A . Leonardi, and K. Mathioudakis, " Components of Fog
snapshot. In the second way, the VM has a "base" snapshot Computing in an Industrial Internet of Things Context," 2015 .
stored on both fogs. The incremental part of the VM' s user [II] H. Madsen, G. Albeanu, B. Burtschy, and F. Popentiu-Vl adicescu, " Re-
snapshot will be transferred instead of the snapshot itself. The liability in the utility computing era: Towards reliable fog computing,"
International Conference on Systems, Signals, na d Image Processing, pp. 43-46,
will only perceive "Transmission time" + "Post-transmission 201 3.
time" and the incremental way is better in this experiment. [12] O. Tran, T. Kim, N. D. Tri, V. Nguyen, N. H. Tran, and C. S. Hong, " A
Also a face recognition application running across a Shared P arking Model in Vehicular Network Using Fog and Cloud Environment,"
smartphone and a fog was implemented, with an app for the pp. 321- 326, 2015.
[13] 1. Stojmenov ic, " Fog computing: A cloud to the ground support for smart
user' s smartphone to capture a face photo and transmit it to a things and machine-to-machine networks," 2014 A ustralasian Telecommunication
remote server which will then try to recognize the face by Networks and Applications Conference (ATNAC), pp. 11 7- 122, 20 14. [Online].
matching it to the local face photo database. The same tasks Availabl e: http://ieeexplore.ieee.org/
were run on fog as well as on the Amazon EC2 cloud. Ipdocs/epic03/wrapper. htm?arnumber=7020884
[14] S. Sarkar, S. Chatterjee, and S. Mi sra, "Assessment of the Suitability of Fog
Computing in the Context of Internet of Things," IEEE Transactions on Cloud
V . CONCLUSIONS Computing, vol. 7161 , no. c, pp. 1- I, 2015 . [Online]. Available:
http://ieeexplore.ieee.org/lpdocs/epic03/ wrapper. htm?arnumber=728678I
A review of the Fog Computing paradigm is introduced in [15] S. Yi, Z. Hao, Z.Qin, and Q. Li, "Fog Computing: Platfonn and
this paper with the aim to resolve some confusion about what Applications," 201 5 Third IEEE Workshop on Hot Topics in Web Systems and
it is and the methodologies used to represent it. The Technologies (HotWeb), pp. 73- 78, 2015 . [Online]. Avail able: http:
introduction of Fog Computing in loT enhances the its 1lieeexplore. ieee.org/lpdocs/epic03/wrapper.htm?arnumber=73 722 86 [1 6] R.
Jhawar and V. Piuri , "Fault Tol erance and Resi lience in
performance and increases the network resources ' efficiency. Cloud Computing Environments," Computer and In fonnaion Securi ty Handbook,
Fog computing is defmed as a computation and services vol. 2, 2013 . [Online]. Avail able: http://spdp.di .unimi .itl papers/JPCISWeb.pd f
extension of the cloud which covers the area from the front- [17] F. Zh u, G. Li, Z. Li, C. Chen, and D. Wen, "A case study of v e aluating t raffic
signal control systems using computational experiments," IE EE Transactions on
end up to the cloud using heterogeneous devices Intelligent Transportation Systems, vol. 12, no . 4, pp. 1220-1226, 2011.
communicating via wireless radio waves or by wired [18] A. Tsitsigkos, F. Entezami , T. Ramrekha, C. Politis, and E. Panaousis, " A
connection as in the case of fiber optics. This paper presents case study of Internet of Things (loT) based on Wireless Sensor Networks
some previous work analyzing Fog characteristics, resource (WSNs) and Smartphone (iPhone)," 2012 .
[1 9] H. Shi , N. Chen, and R. Deters, " Combining Mobile and Fog Computing:
management models and the design of Fog network Using CoAP t o Link Mobile Device Clouds w ith Fog Computing," 2015 IEEE
architectures and approaches. It describes some methodologies International Con ference on Data Science and Data Intensive Systems, pp. 564-
and case studies for implementing Fog Computing. The 571 , 2015 . [Online]. Available: http:
repercussion is that presence of Fog Computing within the loT 1lieeex pi ore. ieee. org/Ipdocs/epic03/wrapper.htm? arnum ber= 7396558

SMC2016 001530

Authorized licensed use limited to: INTERNATIONAL ISLAMIC UNIVERSITY. Downloaded on May 08,2021 at 05:56:21 UTC from IEEE Xplore. Restrictions apply.