Network Design and Performance Evaluation of MQTT

Based HetNet Using LoRa Network and IEEE 802.11

for Internet of Things
Amin Suharjono1, Roni Apriantoro2, Muhammad Mukhlisin3, Muhammad Anif4, Sidiq Syamsul Hidayat5, Thomas Agung Setiawan6, Sri
Anggraeni Kadiran7
1,2,4,5,6,7Dept. of Electrical Engineering

3Dept. of Civil Engineering

1,2,3,4,5,6,7Politeknik Negeri Semarang
1,2,3,4,5,6,7Semarang, Indonesia
1amin.suharjono@gmail.com, 2roni.apriantoro28@gmail.com, 3mmukhlis2@gmail.com, 4muhammad.anif@polines.ac.id,
5sidiqsh@gmail.com, 6thomas.agung@polines.ac.id, 7rinikadiran67@yahoo.co.id

Abstract— Internet of Things is a future networks Cellular networks are often used in these technologies
development trend that is used in many aspects of our daily because of the flexibility offered. However, the weakness of
activities. Internet of Things applications in the Rural area this network is the considerable level of unreliability, which
quite difficult because the area has a high latency on the mobile reflects the high latency in data transmission, especially in
internet network and it is difficult to find an electricity source. mountainous areas and rural areas with the level of
In addition, the installation of sensors at many points it will availability of connectivity that is volatile and tends to be
take a lot of costs if each node is given their own internet unstable.
connection. Therefore, the wireless heterogeneous networks
(HetNet) is suitable to be applied to the Internet of Things If real-time monitoring equipment is needed which
system in Rural areas with an optimal use approach of the requires the movement of sensors to do mapping or other
advantages possessed by LoRa, IEEE 802.11 and MQTT applications related to mapping, of course, it will be a
protocol at the application layer. LoRa Networks is a network problem if done in areas with cellular internet network
segment with a low data rate and low power consumption that coverage that is unstable with high latency. It can cause
is used as a sensor network, while WLAN is used as the many packets to be expired before reaching the server.
backbone network for link aggregation of LoRa Network. The Therefore we need a protocol that is able to work optimally
experimental results have shown that the network has a delay on an unreliable network with sufficiently high latency, so as
of 66.25 ms and packet loss of 4% for the use of three nodes
to guarantee data transmission
with 15 sensors.
Therefore, we need a monitoring technology that is
Keywords—Heterogeneous Wireless Network, LoRa, MQTT, reliable and has good efficiency in energy costs and
WLAN, Internet of Things. consumption. The design of heterogeneous wireless
networks (HetNet) uses the MQTT protocol, LoRa Networks
I. INTRODUCTION and IEEE 802.11 (WLAN) which are quite well applied to
real-time monitoring systems in Rural areas.
Internet of Things is a vision to connect all the smart
devices in order to interact and exchange data smoothly Heterogeneous Network (HetNet) is a network that
which in its application requires a basic technology platform connects computers and other devices (sensor nodes) with
consisting of a Physical layer and Data-link layer to continue different operating systems and/or protocols. Heterogeneous
transmitting data over an IP-based network or Internet [1]. Network is also used in wireless networks that use different
access technologies, such as wireless networks that provide
Internet of Things offers the Smart Things concept, i.e.
services through WLANs and are able to maintain services
every electronic device such as household appliances,
when switching to cellular networks, this is what is then
medical equipment, natural disaster mitigation equipment,
called the Heterogeneous Wireless Network [3] [4].
etc. is connected to the internet so that monitoring and
management can be done easily and not affected by space Long Range (LoRa) is part of Semtech's proprietary
and time, so that every individual need can be done quickly spread spectrum modulation, which is a LoRa modulation
and efficiently. technique that uses physical layers to communicate
wirelessly under the LoRa Alliance standard. The advantages
Internet of Things in its application can identify, track,
of the LoRa modulation technique compared to other
monitor objects, trigger an event and provide real-time
modulation techniques are the use of the Chirp Spread
automatic notifications [2]. One of the most vital applications
Spectrum (CSS) technique, resistance to interference and a
is in the medical, security and natural disaster mitigation
very low minimum Signal to Noise Ratio (SNR) on the
fields.
receiving side, so that the receiver can demodulate the signal
The role of data transmission technology and network properly. Chirp signal has a linear frequency pulse that is
protocols used in communication networks is very vital to modulated on wideband, which frequency pulses vary based
reduce failures in the data transmission process. The level of on the information encoded. This can improve the sensitivity
availability, reliability, and flexibility in the communication of the receiver by making equal timing and offset frequencies
network is needed to support the convenience offered by the on the transmitter and receiver sides and reduce interference
Internet of Things technology for Machine-to-Machine due to the Doppler effect [5].
communications.

ITC2018 (4-6 Oct 2018, Semarang, Indonesia) 1

International Technical Conference 2018

Message Queuing Telemetry Transport (MQTT) is a can be configured to be able to map the LoRa frame format
communication protocol that runs on TCP/IP stack, designed to the MQTT topic. Messages within the broker are
for Machine-to-Machine (M2M) communication, open distinguished and filtered using topics where each message
sources and lightweight, has a low protocol overhead sent to the broker is tagged or bookmarked by the publisher
(minimum 2 bytes) so that it affects the low power to distinguish each data [9].
consumption and able to work on high latency and small
bandwidth [6].
Research related to network development for Internet of
Things applications is carried out by [1] with the title
Design and Deployment of MQTT Based HetNet Using
IEEE 802.15.4 and IEEE 802.11 for the Internet of Things.
The study aims to develop a heterogeneous wireless network
to optimize the benefits of IEEE 802.15.4 (LR-WPAN) and
IEEE 802.11 (WLAN) networks. Raspberry Pi is used as a
gateway that bridges communication between LR-WPAN
and WLAN networks. The LR-WPAN network is used as
anodes sensor with low energy and low data-rate
specifications, while the WLAN network is used as a
backbone connected to the cloud for link aggregation from
Fig. 1. Energy efficiency techniques in the OSI Layer
the LR-WPAN network. The MQTT protocol is also added
to this network to support reliability and low energy
networks design. But, in this study has a weakness in sensor
networks and gateways that are only able to communicate at
a maximum distance of 100 meters.

In this paper we are developing to create a lower cost

gateway that has a range of distances between sensor nodes
and coordinators over 100 meters and to propose designs
and evaluate the performance of heterogeneous wireless
networks using LoRa Networks and WLAN based on the
value of delay and packet loss at the level QOS 0 on the
MQTT protocol.

II. METHODOLOGY Fig. 2. Heterogeneous wireless network design

A. Network Design
The heterogeneous wireless network is designed using
IEEE 802.11 (WLAN) and LoRa Networks standards which
WLAN is used as a backbone network to connect the LoRa
Network to the internet network. WLAN has a larger payload
size compared to LoRa. The maximum payload size on
WLAN is 2312 bytes per frame larger than LoRa which is
only 250 bytes per frame. This can give a statement that
LoRa can only be used on networks with a low data rate. The
effect of using a low data rate is low energy consumption
with a far range [8].
Meanwhile, the use of WLAN as a backbone is based on
specifications and performance that is capable of transporting
data with large payloads. In addition, WLAN also has a wide
bandwidth, high data rate, and a long distance wireless range.
However, this has an impact on the amount of energy
consumption.
The heterogeneous wireless network in this paper is
designed to combine networks with 2.4 GHz frequency on
WLAN and networks with 915 MHz frequency in LoRa
where both networks are not compatible on radio channels,
bandwidth, modulation schemes and frame formats. To
combine it, a gateway device that is designed as a bridge is
needed. Gateway consists of a microcontroller device that is
serially connected with LoRa Bee as a radio transceiver
device as well as a LoRa coordinator. The gateway module Fig. 3. Flowchart of gateway

ITC 2018 (4-5 Oct 2018, Semarang, Ind) 2

International Technical Conference 2018

MQTT is a type of data-agnostic protocol, which the data size of the data set is greater. Loading one sensor at node 1
that is capable of being transmitted is not limited to certain shows a delay value of 52.23 ms. Meanwhile, loading two
types, but all types of data can be transmitted such as binary, sensors at node 1 results in a delay of 55.39 ms.
text, XML or JSON [9].
The network that is loaded using a node with a total of
MQTT works on the TCP/IP stack at the application five sensors produces a delay of 56.54 ms. When the network
layer level on the OSI system standard [9]. This protocol is loaded using two nodes with the number of sensors as
applies data compression and data reduction techniques to many as ten pieces produce a delay of 56.56 ms. When the
perform energy efficiency [9][7]. It can be seen in Figure 1 network is loaded using three nodes with the number of
[7]. sensors as many as fifteen pieces produce a delay of 66.25
ms. This shows that the addition of the number of nodes can
The MQTT protocol in this paper is implemented on two increase the delay in sending data on the network. The
network technologies namely LoRa Networks and IEEE amount of delay is still categorized as very good (<150 ms)
802.11. Figure 2 is an explanation of the sequence of data according to standards issued by the European
encapsulation in different Protocol Data Units (PDUs). Telecommunications Standards Institute (ETSI).
Nodes sensors that use the LoRa module are physically
connected to the sensor doing data encapsulation on the
LoRa frame. B. Packet Loss Test Results
Figure 5 is the result of testing the packet loss by loading
All data in the LoRa Network is forwarded to the LoRa the number of sensors and nodes mentioned earlier.
Coordinator that is connected to the gateway. The software
module on the gateway is embedded in the NodeMCU. It Based on the test results, it can be seen that the packet
will capture the LoRa frame, then extract the sensor data loss value on the network testing that is loaded by a sensor is
from the LoRa payload. Then the sensor data is encapsulated 3.16%. when the network is loaded using two sensors on a
into MQTT Payload. node, it produces a packet loss of 3.88%
Before the data is sent to the Cloud MQTT Broker, The network that is loaded by a node with five sensors, it
MQTT Payload is encapsulated into TCP Segment. TCP produces a packet loss of 4.73%. When the network is loaded
Segment is then encapsulated on IP Datagram, then by two nodes, it produces a packet loss of 3.61%. When the
encapsulated again into IEEE 802.11 frames. After that, the network is loaded by three nodes, it produces a packet loss of
data is transmitted via the internet and decapsulated by 4%. It shows that the addition of the number of nodes does
MQTT Brokers to store data in the cloud database. Flowchart not always result in the addition of the percentage of packet
how the gateway works can be seen in Figure 3. loss. Besides the delay factor, framing error, Time to Live
(TTL) and throughput can also affect packet loss values. The
B. Testing Method number of packet losses is still in the good category (3% -
14%) based on standards issued by the European
In this paper, we use a test method to calculate packet Telecommunications Standards Institute (ETSI).
data on an actual LTE public cellular network to analyze the
delay and packet loss sent from the gateway to MQTT
Brokers based on data from nodes. In this study, our server
from Mydevices (cayenne.mydevices.com) is used as the
Cloud MQTT Broker. Meanwhile, we use Wireshark to
capture packets sent by MQTT on QOS 0 via the LTE public
cellular network.
We analyze packet loss and delay by capturing packets
between the gateway and MQTT Broker for five minutes.
Delay is calculated based on the average roundtrip delay
value between the gateway and MQTT Broker. While packet
loss is calculated based on retransmission packages during
the testing process within 5 minutes in each part. We
performed the test repetition three times for the purpose of Fig. 4. Delay test results
data validation.
We analyze the delay and packet loss data sent from the
gateway to MQTT Broker based on the loading of the
number of sensors and the number of nodes. In this study, we
used three nodes. Each node consists of five sensors.

III. RESULT AND DISCUSSION

A. Delay Test Results

Figure 4 is the result of testing the delay by loading the
number of sensors and nodes mentioned earlier.
Based on the test results, the addition of the number of
sensors at each node is significant in influencing the delay.
This is because the amount of data sent is increasing, so the
Fig. 5. Packet loss test results

ITC 2018 (4-5 Oct 2018, Semarang, Ind) 3

International Technical Conference 2018

IV. CONCLUSION
We test delay and packet loss on the data sent from the
gateway to MQTT Brokers (Publish) using the public LTE
cellular network. Based on the results of experiments that
have been done, it can be concluded that the delay and
packet loss from the data sent using QoS 0 from the MQTT
protocol on the heterogeneous wireless network which is
loaded by fifteen sensors on three nodes each is 66.25 ms
and 4%.

REFERENCES

[1] A. Mishra, “Design and Deployment of MQTT Based HeTNeT

Using IEEE 802 . 15 . 4 and IEEE 802 . 11 for Internet,” Int. J.
Res. Appl. Sci. Eng. Technol., vol. 4, no. November, pp. 2–12,
2017.
[2] Q. Zhou and J. Zhang, “Internet of Things and Geography
Review and Prospect,” in Proceedings - 2011 International
Conference on Multimedia and Signal Processing, CMSP 2011,
2011, vol. 2, pp. 47–51.
[3] A. Delphinanto, B. A. G. Hillen, I. Passchier, B. H. A. Van
Schoonhoven, and F. T. H. Den Hartog, “Remote Discovery and
Management of End-User Devices in Heterogeneous Private
Networks,” in 6th IEEE Consumer Communications and
Networking Conference, CCNC, 2009, pp. 1–5.
[4] O. Olvera-Irigoyen, A. Kortebi, L. Toutain, and D. Ros, “End-to-
End Available Bandwidth Probing in Hybrid Home Networks,”
in IEEE Workshop on Local and Metropolitan Area Networks,
2011, pp. 431–435.
[5] J. Petäjäjärvi, K. Mikhaylov, M. Pettissalo, J. Janhunen, and J.
Iinatti, “Performance of a Low-Power Wide-Area Network Based
on LoRa Technology: Doppler Robustness, Scalability, and
Coverage,” Int. J. Distrib. Sens. Networks, vol. 13, no. 3, 2017.
[6] B. S. Ullas, S. Anush, J. Roopa, and G. R. M, “Machine to
Machine Communication for Smart Systems using MQTT,” Int.
J. Adv. Res. Electr. Electron. Instrum. Eng., vol. 3, no. 3, pp.
8242–8248, 2014.
[7] A. Ali, G. A. Shah, and J. Arshad, “Energy Efficient Techniques
for M2M Communication: A survey,” J. Netw. Comput. Appl.,
vol. 68, pp. 42–55, 2016.
[8] A. Augustin, J. Yi, T. Clausen, and W. M. Townsley, “A Study
of LoRa : Long Range & Low Power Networks for the Internet of
Things,” Sensors, vol. 16, no. 9, pp. 1–18, 2016.
[9] S. Pal, S. Ghosh, and S. Bhattacharya, “Study and
Implementation of Environment Monitoring System Based on
MQTT,” Environ. Earth Sci. Res. J., vol. 4, no. 1, pp. 23–28,
2017.

ITC 2018 (4-5 Oct 2018, Semarang, Ind) 4