Performance Analysis of AMQP Protocol for Patient Health Data in IoT

Case

Dickri Gumelar

, Andrian Rakhmatsyah

and Rahmat Yasirandi

School of Computing, Telkom University, Telekomunikasi Street No.1, Bandung, Indonnesia

Keywords:

Classiﬁcation, Health Data, Publisher, Subscriber, Protocol.

Abstract:

Data classiﬁcation is broadly deﬁned as the process of organizing data based on relevant categories so that it

can be used and protected more efﬁciently. Classiﬁcation of data divided into several classes, one of which

is health data, which needs are essential for data exchange because health data has conﬁdentiality and a large

amount for transfer. Electrocardiogram (ECG) requires timeliness in exchanging data. Currently, the MQTT

Protocol is the most commonly implemented protocol in the IoT area, due to the compatibility of MQTT

with lightweight data to run on it. As for health data, it is far from light data classiﬁcation. There should be

more resilient protocols for use in large scale data or continuous data. In this study, a comparison between

MQTT and other protocols to process health data has conducted. AMQP is an open-source protocol that

provides features suitable for the high requirements for exchanging data on ECG data. This study describes the

performance between the MQTT protocol and AMQP protocol that is principled in the publisher-subscriber,

by comparing the time delay and throughput to measure the data transmission in real-time. This results in

the AMQP protocol delay is less than 1 second per transfer, and the throughput gets an average output of

15209265.86 Bit per second for ten attempts, while the MQTT protocol gets 46 seconds per transfer and

throughput gets an average output of 17592975.29 Bit per second. Proving that in the case of health data with

ECG data as the dataset used, another protocol besides MQTT, which is the AMQP protocol is better in terms

of exchanging data in large and continuous capacity.

1 INTRODUCTION

There have been many studies using the protocol with

the principle of publisher and subscriber. A proto-

col is a set of rules and guidelines for communication

between data in every step and process of communi-

cation to successfully exchanging various messages.

The use of protocols does not have a clear standard.

However, if one chooses, it will have a signiﬁcant in-

ﬂuence on the exchange of data that occurs, and if the

use of the protocol is not following the requirements

will result in fatal events in the ongoing exchange of

data. One of the most popular protocols is the MQTT

protocol. MQTT is commonly used for data exchange

in the world; almost all sectors use it, such as industry,

security, business (Kodali and Gorantla, 2017)(Per-

rone et al., 2017) (van der Westhuizen and Hancke,

2018)(Pereira et al., 2019)(Imane et al., 2018) and

health data is no exception (Sarierao and Prakasarao,

2018)(Terry and Francis, 2007).

Health data is essential because it requires contin-

uous reliability, safety, and availability in the aspect of

data queue for a medical recap because the exchange

of data in health requires proper treatment for suc-

cessful treatment (Das and Ari, 2014).

Not all health data have lightweight data exchange

to be sent at one time; there are also large and busy

data; for example, an electrocardiogram (ECG). An

ECG is a recording of the electrical activity of the

heart, which provides essential information about the

condition of the heart. ECG heartbeat detection is

needed to diagnose heart disease in the early stages

(Naik, 2017). ECG has extensive data and real-time.

Therefore, this study will compare communication

protocols other than MQTT. The MQTT protocol is

one of a machine to machine communications that is

not suitable if it uses in real-time in large amounts of

data. Because in the MQTT Protocol, there are de-

ﬁciencies in terms of data transfer capacity. AMQP

is an open-source protocol that provides features suit-

able for high requirements. The use of appropriate

protocols is one of the crucial things in the case stud-

ies raised (Jaikar and Iyer, 2018). AMQP has suc-

ceeded in becoming a superior communication pro-

tocol in the banking sector. The banking sector re-

quires excellent communication for high-size activi-

Gumelar, D., Rakhmatsyah, A. and Yasirandi, R.

Performance Analysis of AMQP Protocol for Patient Health Data in IoT Case.

DOI: 10.5220/0009907001430150

In Proceedings of the International Conferences on Information System and Technology (CONRIST 2019), pages 143-150

ISBN: 978-989-758-453-4

143

ties with essential and sensitive data (Anusha et al.,

2017). The similarities of the health sector, making

AMQP a feasible protocol to be the proposed proto-

col and compared with MQTT. Therefore, the AMQP

Protocol has a vital role in the reliability, security pro-

vision, and high level of data interoperability.

2 RELATED WORK

2.1 Comparison of AMQP with Similar

Protocols

Protocol able to run according to the needs that exist

in the exchange of data on the IoT (Lee et al., 2018).

but this is only simulated using architecture without

any signiﬁcant results through research. Furthermore,

research on a comparison of existing protocols, such

as MQTT, CoAP, AMQP, and XMPP, shown signif-

icant results and can be a reference for its develop-

ment. Architectural designs built for small scalabil-

ity using MQTT will be reviewed for extensive and

valuable security of data using the AMQP protocol

(Yi et al., 2016) in need for information transfer over

low bandwidth networks that cannot be relied upon

for monitoring. Different requirements make proto-

col performance produce different comparison results

(Pohl et al., 2018). An enormous need appears to inte-

grate the device with an external data source for data

utilization, and assessment collected effectively to get

the best performance out of the requirements (Chaud-

hary et al., 2017).

2.2 ECG (Electrocardiogram)

The development of the medical world cannot be sep-

arated from technological advancements as a tool for

doctors to analyze the function of human organs. In

analyzing the function of human organs, doctors still

read the results of numbers or graphs from assistive

devices. One of the numbers and ﬁgures that doc-

tors analyze is the result of an electrocardiogram. An

electrocardiogram is a test of human heart function.

Electrocardiogram functions to record the electrical

activity of the heartbeat.

To some extent, the Electrocardiogram will also

identify if there is abnormal blood ﬂow. Currently, the

doctor is analyzing the results of electrocardiogram

records manually. After investigating, the doctor can

determine the state of heart function in humans. The

activity of exchanging data on an electrocardiogram

requires real-time. So, this problem is appealing to

be researched. Electrocardiogram recorded data can

turn into training data that can be processed comput-

erized so that a signiﬁcant data exchange can be seen

for large data sizes.

2.3 MQTT (Message Queue Telemetry

Transport)

Message Queue Telemetry Transport (MQTT) is a

lightweight messaging protocol at the application

layer that is suitable for IoT and also machine to ma-

chine. It has a 2-byte header size, that is why it is

has a very small overhead. MQTT (Kim et al., 2015)

follows asymmetrical architecture. MQTT is good for

small data exchanges in the IoT world; its role in mon-

itoring systems in health data is crucial. MQTT server

can be considered as the broker and MQTT clients as

the publisher and subscriber. A broker is a reposi-

tory of topics published by publishers, and customers

subscribe to this topic that has been posted by bro-

kers. There are most brief sessions between the client

server deﬁning the connection of the client and the

server. Yet logically attached to the subject of interest

in the case of a subscription. The user will exchange

messages with this subject when a client subscribes

to an exciting topic. Using this protocol is useful in

situations where small messages requiring less band-

width are exchanged. In small data exchange cases,

MQTT can be the right choice.

2.4 AMQP (Advanced Message

Queueing Protocol)

Advanced Message Queueing Protocol (AMQP) is

[laced at the application layer for protocol-oriented

middleware services in applying the IoT concept

(Hong et al., 2018). AMQP protocol designed to

allow interoperability between different applications

and systems. This AMQP interoperability feature is

vital because it provides various platforms to imple-

ment. AMQP works using the publish-subscribe con-

cept, devices that perform the Publish process are

called Producers, the Subscribe Process are called

Consumer. Producer and Consumer-based AMQP

with a Server that includes Exchange and Queue mes-

sages as a bridge between Producers and Consumers.

The producer gave the message to consumers in the

form of a topic. Producers can distribute which top-

ics will be sent to Consumers on the Server. In this

study, it will be explained that the performance of the

AMQP protocol will be better in the health data case

study because of the large data requirements, com-

pared to other similar Protocols, which will lead to

lower performance.

CONRIST 2019 - International Conferences on Information System and Technology

144

3 DESIGN AND SCENARIO

3.1 System Planning

The design mechanism for the data queue implemen-

tation in sensing nodes consists of hardware, soft-

ware, and brainware. An explanation of the design

of the system mechanism can be seen in Figure 1.

Figure 1: System Mechanism Design.

3.2 Network Architecture

Figure 2: concept exchange Message of AMQP Server.

white

Figure 3: concept exchange Message of MQTT Server.

In the architecture that has been made using two

different protocols, namely the AMQP and MQTT

Protocol, in testing the protocol will use encrypted

ECG data in Excel, the dataset used in this study con-

sists of 11361 ECG data with 202 attributes. These

attributes are used by the publisher to send data to the

server, and the server will receive it as a topic. The

subscriber will request the data that has been sent by

the publisher in the messaging protocol. The perfor-

mance will appear in the server that aims to test the

performance of each protocol.

Figure 4: Basic step Manufacture AMQP Server.

In the basic steps of making AMQP server, the

initial setup of the server used in the AMQP proto-

col uses RabbitMQ, which aims to connect between

client and server, the initial installation of RabbitMQ

has provided the administrator with a username and

password in the form of ”guest” for server access to

be able to see the data exchange performance that oc-

curred. The initial stage of making client access to

the server is the administrator to create a channel so

that the client gets a username and password, then the

channel will set the connection permissions so that

the client username can access the server created, then

the channel will set the virtual host user to connect to

Performance Analysis of AMQP Protocol for Patient Health Data in IoT Case

145

the server, then the user will create an exchange fea-

ture that is contained in rabbitMQ with virtual host

permissions that have been made before, ﬁnally at

this stage create a Queue feature to manage each data

queue that enters the server by connecting Exchange

as a queue regulator if the Queue capacity is Full.

Figure 5: Activity diagram Client Producer AMQP.

In the client producer activity diagram in Figure

5 the ﬁrst step is to prepare an Excel ﬁle electrocar-

diogram (ECG) data that has been encrypted with a

string length in the cell, then the client producer will

read each cell with every row sent with different data

types according to size, then each row of the data is

encoded to release the initial delivery performance.

After all the data that has been sent to the server, the

process is complete and the producer will send a noti-

ﬁcation that the ﬁle has been sent. The client producer

will always run as long as the process is not stopped

(Loo ping Forever).

Figure 6: Activity diagram Server AMQP

In the initial stages of the AMQP Server activity

diagram in Figure 6, the server receives data in the

form of each row in excel sent by the client producer,

the data received will be processed and entered into a

broker that has been given access by the administra-

tor, then the server will start reading data that has been

sent by client producer and output performance data

transfer performed by the client, then the server dis-

tributes the data that has been read to be entered into

the Queue which is managed by the exchange feature

and sent to the client consumer who requests the data

with the same username access.

Figure 7: Activity diagram Client Consumer AMQP.

In the initial stages of the activity diagram in Fig-

ure 7 for the AMQP consumer client, the client will

request data to the server from the same username

given by the administrator, then the server will send

the data requested by the consumer client, the server

will send each excel Row, then the Client consumer

will receive it to be decoded so that can create a new

ﬁle for the results of the receipt that has been done,

if the receipt has been completed, then the client con-

sume will give a notiﬁcation ”Receive ﬁle” on the run-

ning system, but the consumer client that is created al-

ways runs as long as the process is not stopped (Loop-

ing Forever). Based on the sequences diagram above,

the ﬁrst step is to prepare the server for the initial step

to see the exchange of data associated with Wireshark,

then the user will send ECG data that has been previ-

ously encrypted in the form of Excel. Producer will

read the ﬁle with the length of the row that will be

sent to the server and will delivery performance ap-

pears from the producer to the server on Wireshark,

the consumer will send a request to the server to get

the ECG data that has been sent by the Producer, the

request will issue a performance result on the Wire-

shark because the ECG data is sent to the Consumer,

this process will run continuously because it will an-

alyze real-time protocol performance

CONRIST 2019 - International Conferences on Information System and Technology

146

white

Figure 8: Sequences Diagram.

3.3 Testing

In this study, the test was conducted to see the robust-

ness of the protocol of data exchange that took place

in real-time by testing delay and throughput as the

most critical aspects of the exchange that occurred.

3.3.1 Delay Testing

Testing is done to get the time needed from pub-

lisher to subscriber. In this context, the source is pub-

lisher, whose destination is subscriber, from publisher

to server, then server to subscriber. In sending data, it

will take time to travel from the source to a speciﬁc

destination to get the ECG Data delay performance

that arrives.

3.3.2 Throughput Testing

This test is carried out to see the actual data sent per

unit time, the realistic speed of a network, which will

be carried out from publisher to subscriber in send-

ing ECG data that is done continuously. The sending

will be done to show the results to compare the two

performances.

4 RESULT AND DISCUSSION

4.1 Result

Justiﬁcation in research is using extensive data be-

cause the research that has conducted is using small

data. In the form of a string length and the results of

the test do not show a signiﬁcant difference from the

comparison of protocols carried out, therefore in this

study using extensive data. The implementation uses

encrypted ECG data in excel form with the protocol

used is the AMQP protocol and the MQTT protocol,

using the same data exchange concept with the pub-

lisher as the sender and the subscriber as the recipient.

The test is carried out ten times with one time testing

10 ECG data transfers continuously and will be an-

alyzed to see the results of the performance between

protocols.

4.2 Result of Publisher Subscriber

Delay

Packet1 = (timearrived −timesend) (1)

Experiment1 =

Packet1 + Packet2 + .. + Packet10

(2)

AverageDelay =

Experiment1 + .. + Experiment10

(3)

Figure 9: Average Delay MQTT.

AverageDelay/packet =

Delay

Amounto f packet

(4)

In the tests carried out on the MQTT Proto-

col(Figure 10), the output of the delay results from

publisher to subscriber shows the following results.

In tests conducted on the AMQP protocol(Figure 11),

the output of the delay conducted from producer to

consumer shows the following results.

Performance Analysis of AMQP Protocol for Patient Health Data in IoT Case

147

white

Figure 10: Average Delay per Packet MQTT.

Figure 11: Average Delay AMQP.

4.3 Comparison of MQTT Protocol

Delays and AMQP Protocol Delays

averagedelay = averagedelayx10000000 (5)

From the tests conducted and the results of the di-

agram that came out(Figure 12), the average of delay

MQTT shows 42.0727 seconds for each time the pub-

lisher sends a ﬁle to the subscriber. In contrast, the

AMQP Protocol shows the average of each producer

sending the ﬁle to the consumer; it only takes 0.3105

seconds. It shows that the availability of data directly

and continuously AMQP is far superior compared to

MQTT. Because at large enough data size, MQTT is

not able to accommodate the queues that are carried

out continuously by the publisher, whereas AMQP

can accommodate every shipment made by the pro-

ducer.

4.4 Throughput Test Results

Throughput is a measure of how many units of in-

formation a system can process in a given amount of

time and usually represented as an average and mea-

sured in bits per second (bps). The testing is done 10

times ﬁle transfer for 1-time testing, and the results

are taken from the publisher to the subscriber, testing

is done 10 times with a total number of ﬁle transfers

100, here are the results of each test:

Figure 12: Average Delay per Packet AMQP.

Figure 13: Average Time Each Packages.

AverageBps = (

Bytes

TimeSpan

)x8bit (6)

Figure 14: Throughput MQTT.

CONRIST 2019 - International Conferences on Information System and Technology

148

white

Figure 15: Throughput AMQP.

In the table above, the MQTT throughput has

the last average of 10 trials with an average number

of packets 891,331,927.2 Bytes and an average time

span of 208.0996 resulting in an average Bytes / s of

17,592,975.29 or 16.1 Mbit / s. While in AMQP, the

throughput obtained with 208.0996 with the last aver-

age of 14.7 Mbit / s and the average number of packets

of 895,075,801.2 Bytes and Time span of 479.0202.

From the results, there were differences in the aver-

age packet of 3,743,874 Bytes with the AMQP proto-

col sending more packets, but the average shows that

AMQP is smaller issuing throughput.

5 CONCLUSIONS

From this research, it can be concluded that for a big-

ger capacity and continuous availability, AMQP pro-

tocol is better in terms of exchanging data than the

MQTT protocol. The results have been obtained by

showing the time delay and throughput to measure

the data transmission in real-time. Obtain accuracy

when sending health data, with the output that the

AMQP protocol delay that comes out is less than 1

second per shipment while the MQTT protocol gets

46 seconds for each delivery. For the throughput test-

ing, the AMQP protocol gets an average output of

15209265.86 Bit per second for ten attempts, while

the MQTT protocol gets 17592975.29 Bit per sec-

ond. The outputs on testing health data using similar

protocols with AMQP protocol results better than the

MQTT protocol; the availability of the AMQP proto-

col is better for sending extensive data. This research

could speed up the data running process, so there will

be no data fallacy or data lose. It can be concluded

from the initial problem that the AMQP protocol an-

swers to data exchange on health data because it has

various needs, and it is done continuously. The de-

velopment of advanced research that can be done is

more massive data in the security and availability as-

pects, and many implementations can be done in other

sectors to see a better performance other than AMQP

protocol.

REFERENCES

Anusha, M., Babu, E. S., Reddy, L. S. M., Krishna, A., and

Bhagyasree, B. (2017). Performance analysis of data

protocols of internet of things: a qualitative review.

International Journal of Pure and Applied Mathemat-

ics, 115(6):37–47.

Chaudhary, A., Peddoju, S. K., and Kadarla, K. (2017).

Study of internet-of-things messaging protocols used

for exchanging data with external sources. In 2017

IEEE 14th International Conference on Mobile Ad

Hoc and Sensor Systems (MASS), pages 666–671.

IEEE.

Das, M. K. and Ari, S. (2014). Ecg beats classiﬁcation using

mixture of features. International scholarly research

notices, 2014.

Hong, X. J., Yang, H. S., and Kim, Y. H. (2018). Perfor-

mance analysis of restful api and rabbitmq for mi-

croservice web application. In 2018 International

Conference on Information and Communication Tech-

nology Convergence (ICTC), pages 257–259. IEEE.

Imane, S., Tomader, M., and Nabil, H. (2018). Compar-

ison between coap and mqtt in smart healthcare and

some threats. In 2018 International Symposium on

Advanced Electrical and Communication Technolo-

gies (ISAECT), pages 1–4. IEEE.

Jaikar, S. P. and Iyer, D. (2018). A survey of messaging

protocols for iot systems. Int. J. Adv. Manag. Technol.

Eng. Sci, pages 510–514.

Kodali, R. K. and Gorantla, V. S. K. (2017). Weather track-

ing system using mqtt and sqlite. In 2017 3rd Interna-

tional Conference on Applied and Theoretical Com-

puting and Communication Technology (iCATccT),

pages 205–208. IEEE.

Naik, N. (2017). Choice of effective messaging protocols

for iot systems: Mqtt, coap, amqp and http. In 2017

IEEE international systems engineering symposium

(ISSE), pages 1–7. IEEE.

Pereira, E., Pinto, R., Reis, J., and Gonc¸alves, G. (2019).

Mqtt-rd: A mqtt based resource discovery for machine

to machine communication.

Perrone, G., Vecchio, M., Pecori, R., Giaffreda, R., et al.

(2017). The day after mirai: A survey on mqtt se-

curity solutions after the largest cyber-attack carried

out through an army of iot devices. In IoTBDS, pages

246–253.

Pohl, M., Kubela, J., Bosse, S., and Turowski, K. (2018).

Performance evaluation of application layer protocols

for the internet-of-things. In 2018 Sixth International

Conference on Enterprise Systems (ES), pages 180–

187. IEEE.

Sarierao, B. S. and Prakasarao, A. (2018). Smart healthcare

monitoring system using mqtt protocol. In 2018 3rd

International Conference for Convergence in Technol-

ogy (I2CT), pages 1–5. IEEE.

Performance Analysis of AMQP Protocol for Patient Health Data in IoT Case

149

Terry, N. P. and Francis, L. P. (2007). Ensuring the privacy

and conﬁdentiality of electronic health records. U. Ill.

L. Rev., page 681.

van der Westhuizen, H. W. and Hancke, G. P. (2018). Com-

parison between coap and mqtt-server to business sys-

tem level. In 2018 Wireless Advanced (WiAd), pages

1–5. IEEE.

Yi, D., Binwen, F., Xiaoming, K., and Qianqian, M. (2016).

Design and implementation of mobile health monitor-

ing system based on mqtt protocol. In 2016 IEEE

Advanced Information Management, Communicates,

Electronic and Automation Control Conference (IM-

CEC), pages 1679–1682. IEEE.

CONRIST 2019 - International Conferences on Information System and Technology

150