Performance Analysis of Internet of Things Protocols Based Fog/Cloud over High Traffic

Abstract

The Internet of Things (IoT) becomes the future of a global data field in which the embedded devices communicate with each other, exchange data and making decisions through the Internet. IoT could improves the qualityoflife in smart cities, but a massive amount of data from different smart devices could slow down or crash database systems. In addition, IoT data transfer to Cloud for monitoring information and generating feedback thus will lead to highdelay in infrastructure level. Fog Computing can help by offering services closer to edge devices. In this paper, we propose an efficient system architecture to mitigate the problem of delay. We provide performance analysis like responsetime, throughput and packet loss for MQTT (Message Queue Telemetry Transport) and HTTP (Hyper Text Transfer Protocol) protocols based on Cloud or Fog serverswith large volume of data form emulated traffic generator working alongsidewith one real sensor. We implement both protocols in the same architecture, with low cost embedded devices to local and Cloud servers with different platforms. The results show that HTTP response time is 12.1 and 4.76 times higher than MQTT Fog and cloud based located in the same geographical area of the sensors respectively. The worst case in performance is observed when the Cloud is public and outside the country region. The results obtained for throughput shows that MQTT has the capability to carry the data with available bandwidth and lowest percentage of packet loss. We also prove that the proposed Fog architecture is an efficient way to reduce latency and enhance performance in Cloud based IoT.

Full text

doi: http://dx.doi.org/10.4314/jfas.v10i6s.113
J Fundam Appl Sci. 2018, 10(6S), 176-181 176
Performance Analysis of Internet of Things
Protocols Based Fog/Cloud over High Traffic
Istabraq M. Al-Joboury1, Emad H. Al-Hemiary2
Department of Networks Engineering, College of Information Engineering
Baghdad, Iraq
{1estabriq_94, 2emad}@coie-nahrain.edu.iq
Published online: 22 March 2018
Abstract—The Internet of Things (IoT) becomes the future of
a global data field in which the embedded devices communicate
with each other, exchange data and making decisions through the
Internet. IoT could improves the qualityoflife in smart cities, but
a massive amount of data from different smart devices could slow
down or crash database systems. In addition, IoT data transfer to
Cloud for monitoring information and generating feedback thus
will lead to highdelay in infrastructure level. Fog Computing can
help by offering services closer to edge devices. In this paper, we
propose an efficient system architecture to mitigate the problem
of delay. We provide performance analysis like responsetime,
throughput and packet loss for MQTT (Message Queue
Telemetry Transport) and HTTP (Hyper Text Transfer Protocol)
protocols based on Cloud or Fog serverswith large volume of
data form emulated traffic generator working alongsidewith one
real sensor. We implement both protocols in the same
architecture, with low cost embedded devices to local and Cloud
servers with different platforms. The results show that HTTP
response time is 12.1 and 4.76 times higher than MQTT Fog and
cloud based located in the same geographical area of the sensors
respectively. The worst case in performance is observed when the
Cloud is public and outside the country region. The results
obtained for throughput shows that MQTT has the capability to
carry the data with available bandwidth and lowest percentage of
packet loss. We also prove that the proposed Fog architecture is
an efficient way to reduce latency and enhance performance in
Cloud based IoT.
Keywords—Internet of Things; Web of Things; Fog
Computing; Cloud Computing; Edge Computing; MQTT; HTTP;
Tsung.
I. INTRODUCTION
IoT is a new concept and paradigm; in which the real
worlds of things linked to the virtual world, subsequently
enabling anything to anyone [1]. It is becoming a revolution of
devices as well as representing the future of the Internet.
Therefore, it has attracted wide attention from researchers.
The IoT technology consists of two terms "Internet" and
"Things". The firstterm gives the meaning of protocols,
services and networks, whereas the secondterm refers to
sensors, smart devices [2]. The basic idea of IoT is that objects
(such as electronics) connected together to provide an
efficient, low power, and seamless connectivity to humans [3].
Then, new technologies allow objects to be more intelligent,
which can transfer data generated from different things, as
well as, make them recognizable by using IP and RFID. This
leap leads to the integration of IoT and cloud computing.
Furthermore, IoT moves into the unlimited capabilities of IP6
addresses [4]. The elements of IoT can be a physical/digital
entity, which perform various daily tasks for individual users
and then IoT applicationprotocols and technologies used to
achieve IoT vision; for instance, wireless sensor networks
allow objects to measure in real time and data is collected by
using IoT protocols.Smart city provides services to
governments (e.g smart transportation and mobility, smart
building and infrastructure, organizations (e.g e-learning,
manufacturing, smart factories) and humans (e.g smart home,
smart hospital). The common IoT layers are threemodel
categorizing Application, Network, and Perception Layers.
new layers: Business and Middleware Layers are recently
proposed[5].
Fig. 1 Cloud, Fog and embedded devices layers.
Fog Computing is a concept made by Cisco in 2012, that
aiming the realtime applications to handle billions of IoT
devices [6]. It refers as an intermediate layer between cloud
and embedded devices enabling storage, computing and
networking services, the same as Cloud Computing. Itconsists
of servers, routers, switches and access points [7]. Fog
Computing brings all based cloudfeatures and services near to
edge devices "ground" like sensors, smartphones, wearables
Research Article
Special Issue

J Fundam Appl Sci. 2018, 10(6S), 176-181 177
and embedded devices [8] as shown in Fig. 1. Smart cities
including smart hospitals and infrastructures for IoT
environments that they handle big data and stream for realtime
application. Thereby, a real time cityis enableddue to offer
new services for governments and societies, as well as big data
analysis in realtime of infrastructure level and the person's
lifestyle. Here, data generated from IoT devices sent to the
cloud in order to be stored and processed. Cloud computing
enables services (Software as a Service - SaaS) and
(Infrastructure as a Service – IaaS) and Platform as a Service -
PaaS) and provides data processing. It is suitable for
applications that their data is stored and processed in
centralized. Some application such as health care systems
depends on distributed storage and lowlatency, at this point
cloud fails to handle these conditions [9].However, there are
some differences between the two concepts as it discusses
briefly below in Table I.
Table I Comparison between Fog and Cloud[10]
Fog Cloud
Location Local Internet
Data Thousands Hundreds
Latency and
Delay Low High
Storage Distributed Centralized
The rest of this paper is organized as follows:Section II
provides an overview of the web IoT protocol (MQTT and
HTTP) and the difference between them.SectionIII, describes
problem definition of how to manage traffic and protocol
selection. SectionIVcovers related work to the paper.
SectionVproposes IoT architecture and gives the tools and
programs used for testing performance. SectionVI, shows the
results of responsetime, throughput and packet loss of
protocols based on Cloud/Fog.Finally, section VII, concludes
this paper.
II. OVERVIEW OF WEB PROTOCOLS
The MQTT is an application layerprotocol designed for
lightweight M2M (machine to machine) communications,
simple, easy to implement and fast transportation protocol.
MQTT is suitable for resource constrained devices,
lowbandwidth, low latency and reliable networks. Stanford-
Clark and Nipper [11] release the first version of MQTT
protocol in 1999; IBM originally created it. The latest version
of MQTT is 3.1.1 [Nov, 2014] and it is becoming an open
standard protocol.
MQTT is an OASIS (Advancing Open Standards for the
Information Society) runs over TCP/IP protocol. It is
publish/subscribe model based on topics and consists of three
elements: two types of clients (publisher or subscriber) and
one server (called broker). Publishers send messages within a
specific topic, then subscriber clients receive these messages
that refers to the same topic that they subscribed via broker as
shown in Figure 2. Also, publisher do not require the address
of the subscribers[12,13].
Fig.2The operation of MQTT based on Publish/Subscribe model [14].
MQTT has a lower overhead, a synchronous and reliable
with a multiple different levels of quality of services. There
are three types of QoS for a delivery assurance that are used
between client and server[11,15]. There are:
 QoS level 0:the publisher sends the message to the
subscriber through the broker andthe subscriber
receives the message at most once. In addition, the
broker never sends an acknowledgement to the
publisher.
 QoS level 1: the publisher delivers the message to the
clients at least once, and the brokersend back an
acknowledgement if the message is lost.
 QoS level 2:publisher uses level 2 when message lost
or duplicate and this requires fourway handshake to
deliver the message at exactly once, hence this cause
increase in the overhead for this reason level 2 and 3
are not included in this paper
The broker may require authentication(username and
password) from subscribers to allow them to connect, so that
the broker will confirm the privacy by using (Secure Sockets
Layer -SSL)/ (Transport Layer Security-TLS).
HTTP is an application layer protocol based on TCP/IP
suite of protocols. It used to transfer data from client side like
smart phone, personal computer to server side such a Web
server over the World Wide Web. HTTP v2. is last version
[May 2015] [16]. The most commands are GET and POST for
processing data on web. It is request/response model based on
Uniform Resource Locators (URLs) the user request data on
web server, then server not only response to data but all
relevant data to that request. There are some differences
between the two protocols as it summarizes below in Table II.
Table IIDifferences betweenMQTT and HTTP [17]
MQTT
HTTP
Transport
TCP
TCP
Architecture
Client
/
Broker
Client
/
Server
Model
Publish
/
Subscribe
Request
/
Response
QoS
3 Types
None
Messages
Topic
URL
Standard
OASIS
Arch. Style
Encoding
Binary
Different Types
Security
Username and
Password, SSL/TLS SSL/TLS
Pub (topic, data)
Publisher Broker Subscriber
Pub (topic, data)
Sub (topic)

J Fundam Appl Sci. 2018, 10(6S), 176-181 178
III. PROBLEM DESCRIPTION
Smart hospitals generate a large amount of data from
thousands of sensors that it can be useful for monitoring and
analyzing. However, an unprecedented volume of data can
crash storage systems and realtime applications.Cloud
Computing could provide storage “ondemand” and processing
ofsystems, but Cloud could be anywhere and away far from
systems, as well as transferring data from sensors to Cloud and
then giving a feedback to end user and this is a problem for
sensitive healthcare applications because of high delay. Fog
Computing consider to be temporally near to the sensor;
thuswilldecreasedelay [18].There are several of IoT and OSI
application protocols relies on TCP used to communicate and
deliver data. This paper, provides an answer to thesequestions
"Which protocol will be used with low responsetime and
highthroughput?", "Which is the best location for servers that
represents the lowest delay in order to rapidly send
notification to end user" and "Is Fog Computing actually has
better performance than Cloud Computing?".
IV. LITERATURE REVIEW
MQTT and HTTPprotocolsare used in communication
between people and devices especially in the medicalfield.
However, up to our knowledge few papers present the
performance of these protocols under conditions such as over
large volume of traffic and based on Fog and Cloudlayer.The
performance testing of XMPP protocol was tested andthe
evaluation methodology was developed using Tsung traces to
check the requirement need of the protocol [19].Also, the
performance of XMPP server was tested using load distributed
Tsung over high traffic and from honeypot sensors to find the
limit of number of concurrent request [20], but MQTT and
HTTP are not included in the above two papers, as well as
Cloud and Fog layer are not mention. While in [21],the
performance of Web IoT protocols (DDS, XMPP and MQTT)
was compared according to the latency of message delivery
from sensors and throughput, however these protocols arenot
implemented in Fog efficiency concept. Among the above
works, [22] including Fog and the selection of network
management protocols such SNMP, NETCONF and CoAP
were evaluated. But, theyhave mentioned only the
management protocols and implemented using OMNeT++
simulator not the real hardware.And in [23], MQTT,
WebSocket and CoAP application protocols were compared in
IoT scenario based on local via Ethernet and remote server via
internet and cellular network. However, the response time and
throughput of MQTT and HTTP over a huge volume of data
were not included.In [24], they proposed system architecture
to the problem of middleware, scalability and interoperability
between Cloud and sensors. In this system,
publisher/subscriber model was applied using MQTT protocol
and average response time and throughput was measured. In
[25] the overhead and payload size matricesof HTTP and
MQTTwere compared but without the relation to the Cloud
and Fog servers, and then a queuing theory was proposed to
evaluate the performance of MQTT.
V. METHODOLOGY
The main objective of this paper is comparing the
performance of Web IoT protocols with each other, in term of
number of sensors that it can handle with low response time
and packet loss, and then finding the best location for the
servers.
The operation of the proposed IoT architecture is as
follows:
We implement two IoT scenarios as in the figures 3and 4,
and provide the performance analysis of MQTT and HTTP
protocols in six data communication paths: sensors to Fog
(located in Al-Nahrain University, College of Information
__________________________________________________
1http://pulsesensor.com/
2http://tsung.erlang-projects.org/
3http://www.nodemcu.com/index_cn.html
4https://mosquitto.org/
5https://www.mongodb.com/
6https://test.mosquitto.org/
7http://dweet.io/
8http://freeboard.io/
Engineering), sensors to Cloud (located in Ministry of Higher
Education and Scientific Research, Department of Research
and Development) IaaS/PaaS, sensors to Cloud SaaS (located
in different country) and all these steps will be repeated for
http protocol.There is a similarity in some of the settings in the
two scenarios. Such as, we setup one real pulse
sensor1(heartbeat pulse sensor) and emulate the other sensors
using TSUNG2(also called Tsunami).With TSUNG, we solve
the problem of having hundreds or even thousands of sensors
to simulate a real environment— Tsung (also called Tsunami)
is an open source program with GPLv2 (General Public
License version 2) and developed by Erlang, which provides
multi protocols like MQTT and HTTP.For data collection
from sensors, we use NodeMCU3 (also called ESP8266-
12E)programmed using C/C++ programming language. Then,
these sensorsconnect to an IEEE802.11n Access Points. The
last similar settings consist of two type of server Fog server
and Cloud server. So as a whole, APs are connected to Fog
layer by using Ethernet, while connected to remote servers via
Internet, in both cases with constant bandwidth.There are
some different settings in each scenario: in first scenario the
MQTT v3.1.1 protocol with QoS level 0 and 1 is used in the
first scenario, MQTT broker (mosquitto4) is necessary to
mediate the transferring data between subscriber and
publisher, then, data is stored using MangoDB5 temporary
database with Robomongo GUI through Node.js by using TTL
(Time to Live) Fog based. A Path to another MangoDB Cloud
based with same configuration and this is a permanent storage
and also another path to public (mosquitto6) located in
different country as shown in Figure 3. As it shown in Figure
4, HTTP protocol v1.1 and GET command are used to request
data. The Fog layer is LAMP (Linux, Apache, MySQL, PHP)
server used to a temporally store data and Cloud layer is also
lamp server but in contrary it considers a permanent storage.
The final path is to Dweet7 Cloud located in different country
and Freeboard8 for monitoring data or to infrastructure LAMP
Cloud at the same region.

J Fundam Appl Sci. 2018, 10(6S), 176-181 179
Fig. 3IoT architecture based Fog/Cloud and MQTT protocol.
Fig. 4IoT architecture based Fog/Cloud and HTTP protocol.
VI. EXPERMENTAL SETUP AND RESULTS
This section shows experimentalresults from the
performance analysis with comments.
In this paper, one session is programmed in Tsung by
using XML v1.0 language and is executed to handle all
requests of protocols with this session to do authentication and
connection with server side. Also, Tsung is configured to
generate a large number of virtual sensors— or what is called
the average arrival rate— to publish a huge number of
messages only per one physical computer.Finally log level of
Tsung is set to type of debug, so that can handle long logging.
Also, in order to calculate the response time, throughput and
packet loss, every request and message generated by the
MQTT and HTTP protocol are recorded using Tsung. At the
end, the overall running time of test takes 170 minutes.
Thesize of packet contentsof HTTP and MQTT protocols
from sensors using open source network analyzer
Wireshark9as is shown in TableIII.
TableIII Size of Packet Contents (in Bytes)
Message PDU Response
size
MQTT 75 11 2
HTTP 75 79 67
IPerf10is used as a tool to measure network bandwidth
between sensors and Fog, and between sensor and Cloud
(located inMinistry of Higher Education and Scientific
Research). IPerf is a powerful and simple testing tool,
client/server model written in C++, it used to analyze
performance network quality, loss and bandwidth based on
TCP or User Datagram Protocol (UDP). Table IV summarizes
the correlation between location of servers and ISP (Internet
Service Provider) based on available bandwidth.
TableIVPerformance between sensor andFog/Cloud
Metric Type of
Server Bandwidth Protocol
Response
Time
Cloud 20.4 Mbits/sec HTTP
Fog 89.3 Mbits/sec HTTP
Cloud 26.8 Mbits/sec MQTT QoS 0
Cloud 26.8 Mbits/sec MQTT QoS 1
Fog 93.9 Mbits/sec MQTT QoS 0
Fog 94.0 Mbits/sec MQTT QoS 1
Throughput
Cloud 4.11 Mbits/sec HTTP
Fog 6.05 Mbits/sec HTTP
Cloud 6.53 Mbits/sec MQTT QoS 0
Cloud 16.4 Mbits/sec MQTT QoS1
Fog 5.72 Mbits/sec MQTT QoS 0
Fog 7.64 Mbits/sec MQTT QoS 1
The proposed IoT architecture consists of integrated the
simulation and practical work. Each of (mosquitto, MongoDB
and LAMP) of Fog/Cloud basedare installed and configured
on HP ProLiant 380 G7 for Fog server and on HP ProLiant
380 G8 for Cloud server, OS: Ubuntu server 14.04 LTS,
RAM: 32 GB, processor: 32 and 500 GB.Tsung was installed
on different machine with characteristics: OS: Ubuntu 14.04.5
LTS, Memory: 3.7 GB, processor:Intel(R) Core(TM) i3-380
CPU @ 2.53GHz *4, disk: 488.1 GB.
The performance analyses are:
1) Response Time:
The responsetime of protocols is the elapsed time taken by
a web to respond to a request for web services.In this test, the
number of sensors set for requesting and publishing data from
100 to 1500 sensors. In Figure 5, architecture based Fog
shows that sensors requesting a web page as using HTTP is
12.1 times higher than sensors using MQTT protocol, Cloud

J Fundam Appl Sci. 2018, 10(6S), 176-181 180
based HTTP is 4.7 times higher than MQTT where Cloud
located in the same region as the region of sensors, and Cloud
based HTTP achieves 2.5 slower than MQTT and the later
Cloud locatedin different country, and these results compare
withQoS 0 of MQTT, while MQTT QoS level 1 is used with
HTTP the results showed:7.8%, 2.7%, 1.8% as respectively, as
an example if number of sensors is 1000. The reason for that is
the MQTT has long keep a live time for connection to handle
multiple requests and low overhead whereas HTTP opens the
TCP connection for short time. Also, the MQTT has low
overhead size only 2 Bytes in handshake than HTTP. As a
result, it is not efficient that sensors depend on Cloud for
processing data and send feedback to interested persons. Also
Figure 6, shows that MQTT connection based Cloud 3.4 times
slower than MQTT connection based Fog and MQTT based
Cloud (located in different country)151 times higher than
protocol based Fog. And this results are the same with HTTP
protocol.
9https://www.wireshark.org/
10https://iperf.fr/
Fig. 5Requests for MQTT and HTTP Fog/Cloud based.
Fig. 6Connections for MQTT and HTTP Fog/Cloud based.
2) Throughput:
Throughput is the amount of data that server could handle
in period of time. The next Figure 7, shows that the throughput
of two protocols MQTT and HTTP. Throughput performance
shows that HTTP is 7.1 times higher than MQTT protocol
QoS 0 Fog based or 6.38 times higher in Cloud based (located
in the same region). The location of the server does not impact
so much on throughput performance of both protocols and
even if it impacted, factor 1 will be affected.Also, we notice
that the HTTP protocol has reached the saturation level earlier
than MQTT protocol in both cases Fog or Cloud based. The
impact performance of throughput depends on server
capabilities to handle data and load.
Fig. 7Throughput for MQTT and HTTP Fog/Cloud based.
3) Packet Loss:
The packet loss defined as number of packets of data fail
to reach the final destination when they travel through
network.In Figure 8 below, packets loss was compared in
terms of the number of messages that published and requested,
the two protocols MQTT and HTTP, and QoS levels and
location of local and remote servers. The results shows MQTT
QoS 1 packet loss is 6.2 times higher than MQTT QoS 0 Fog
based. Also, HTTP message loss is 49.7 higher than MQTT
QoS 0 Fog based and in case of Cloud based located in the
same regionit would be 41.1 higher than MQTT QoS 0, as an
example if the number of messages per sec is 50,000.And all
this happened because of the following reasons: MQTT has
lowest handshake and lowest PDU, Fog has a lower packet
loss than Cloud because the Fog is local and there is no need
for the network to have routers, these routers are unable to
hold trafficwith limited bandwith, unlikeCloud. In addition,
the path to the Cloud may contain multiple routers connecting
together by links, if one of these links is busy the packets have
to wait in the queue. Also, if the queue is at full capacity the
packets will be dropped. Furthermore, the packet loss impact
on response time of protocols because of the retransmission of
lost packets, thus leads to higher response time.

J Fundam Appl Sci. 2018, 10(6S), 176-181 181
Fig. 8Packet loss for MQTT and HTTP Fog/Cloud based
VII. CONCLUSION
In this paper, the proposed IoT architecture suggest a
middle layer named Fog consists of high speed temporary
storage to enable fast end users reporting. We perform an
experimental setup to analyze two Web IoT protocols: MQTT
and HTTP.We implement these protocols using low cost
embedded devices with private and public servers working as
Fog and Cloud (there are two Cloud: one in the same region
with Fog and other at different country). The work
concentrates on generating large traffic volume from sensor-
like terminal running Tsung tool integrated with real heart
sensor traffic to simulate the required scenarios. The obtained
results of response time and throughput for both scenarios
(Fog based and Cloud based) show that the MQTT protocol
advances the HTTP protocol since the latte one consists of an
extra handshaking and more overhead than MQTT. On the
other hand, using Fog servers as a middleware layer close to
the embedded devices (organization levels) enhances
performance and this is clearly shown in the results obtained.
Fog servers may be designed as close as possible to the end
user devices in distributed layers. While, the throughput in
both scenarios is related directly to the available bandwidth
between the gateway and Fog/Cloud servers.
REFERENCES
[1] H. Sundmaeker, P. Guillemin, P. Friess, and S. Woelfflé, Vision and
challenges for realising the Internet of things. Luxembourg: EUR-OP,
2010.
[2] D. Bandyopadhyay and J. Sen, "Internet of things: Applications and
challenges in technology and standardization," Wireless Personal
Communications, vol. 58, no. 1, pp. 49–69, May, 2011.
[3] G. M. Lee, J. Park, N. Kong, and N. Crespi, "The internet of things:
concept and problem statement: 01," 2011. [Online]. Available:
https://tools.ietf.org/id/draft-lee-iot-problem-statement-01.txt [Accessed:
1- March- 2017].
[4] O. Vermesan and P. Friess, Eds., Internet of things: Converging
technologies for smart environments and integrated ecosystems.
Aalborg: River Publishers., 2013.
[5] A. Al-Fuqaha, M. Guizani, M. Mohammadi, M. Aledhari and M.
Ayyash, "Internet of Things: A Survey on Enabling Technologies,
Protocols, and Applications," IEEE Communications Surveys &
Tutorials, vol. 17, no. 4, pp. 2347-2376, 2015.
[6] I. Stojmenovic and S. Wen, “The Fog Computing Paradigm: Scenarios
and Security Issues,” Proceedings of the 2014 Federated Conference on
Computer Science and Information Systems, 2014.
[7] R. Mahmud and R. Buyya, "Fog Computing: A Taxonomy, Survey and
Future Directions". arXiv preprint arXiv:1611.05539, 2016.
[8] E. M. Tordera, X. Masip-Bruin, J. Garcia-Alminana, A. Jukan, G. Ren,
J. Zhu and J. Farre, "What is a Fog Node A Tutorial on Current
Concepts towards a Common Definition". arXiv preprint
arXiv:1611.09193, 2016.
[9] A. V. Dastjerdi, H. Gupta, R. N. Calheiros, S. K. Ghosh and, R. Buyya,
"Fog Computing: Principles, Architectures, and Applications," Internet
of Things: Principles and Paradigms, Massachusetts, USA, 2016.
[10] Cisco, "Fog Computing and the Internet of Things: Extend the Cloud to
Where the Things Are," cisco public, white paper, 2015.
[11] "MQTT," 2014. [Online]. Available: http://mqtt.org/. [Accessed: 1-
March- 2017].
[12] S. Lee, H. Kim, D.-K. Hong, and H. Ju, “Correlation analysis of MQTT
loss and delay according to QoS level,” The International Conference on
Information Networking 2013 (ICOIN), 2013.
[13] A. W. Gordon, "MQTT-S–A Publish/Subscribe Protocol for Wireless
Sensor Networks," 2010.
[14] P. Masek, J. Hosek, K. Zeman, M. Stusek, D. Kovac, P. Cika, J. Masek,
S. Andreev, and F. Kröpfl, “Implementation of True IoT Vision: Survey
on Enabling Protocols and Hands-On Experience,” International Journal
of Distributed Sensor Networks, vol. 12, no. 4, p. 8160282, 2016.
[15] A. Banks and R. Gupta. "MQTT Version 3.1. 1." OASIS standard, 2014.
[16] M. Belshe, M. Thomson, R. Peon, and rfcmarkup version 1, "Hypertext
transfer protocol version 2 (HTTP/2)," 2015. [Online]. Available:
https://tools.ietf.org/html/rfc7540. [Accessed: 1- March- 2017].
[17] V. Lampkin, W. T. Leong, L. Olivera, S. Rawat, N. Subrahmanyam, R.
Xiang and M. Keen, Building Smarter Planet Solutions with MQTT and
IBM WebSphere MQ Telemetry. IBM, 2012.
[18] A. V. Dastjerdi and R. Buyya, “Fog Computing: Helping the Internet of
Things Realize Its Potential,” Computer, vol. 49, no. 8, pp. 112–116,
2016.
[19] X. Che and S. Maag, "Testing protocols in Internet of things by a formal
passive technique," Science China Information Sciences, vol. 57, no. 3,
pp. 1–13, Feb. 2014.
[20] K. J. Kurniawan, C. Lim and K. I Eng, "XMPP Performance Analysis
using large volume traffic from honeypot sensor," The International
Conference on Innovation, Enterpreneurship, and Technology
(ICONIET 2015), 2015.
[21] Z. B. Babovic, J. Protic, and V. Milutinovic, "Web performance
evaluation for Internet of things applications," IEEE Access, vol. 4, pp.
6974–6992, 2016.
[22] M. Slabicki and K. Grochla, "Performance evaluation of CoAP, SNMP
and NETCONF protocols in fog computing architecture," NOMS 2016 -
2016 IEEE/IFIP Network Operations and Management Symposium, p.p
1315 - 1319, 2016.
[23] S. Mijovic, E. Shehu, and C. Buratti, “Comparing application layer
protocols for the Internet of Things via experimentation,” 2016 IEEE
2nd International Forum on Research and Technologies for Society and
Industry Leveraging a better tomorrow (RTSI), 2016.
[24] M. A. Triawan, H. Hindersah, D. Yolanda, and F. Hadiatna, “Internet of
things using publish and subscribe method cloud-based application to
NFT-based hydroponic system,” 2016 6th International Conference on
System Engineering and Technology (ICSET), 2016.
[25] T. Yokotani and Y. Sasaki, “Comparison with HTTP and MQTT on
required network resources for IoT,” 2016 International Conference on
Control, Electronics, Renewable Energy and Communications
(ICCEREC), 2017.