2722 IEEE SENSORS JOURNAL, VOL. 20, NO.

5, MARCH 1, 2020

WaterGrid-Sense: A LoRa-Based Sensor Node

for Industrial IoT Applications
Oratile Khutsoane , Student Member, IEEE , Bassey Isong, Member, IEEE ,
Naison Gasela, and Adnan M. Abu-Mahfouz, Senior Member, IEEE

Abstract —A Wireless sensor-network is used in the

industry to monitor critical parameters. However, its industrial
usage in harsh environments presents some challenges to the
wireless communication signals such as noise, interference,
etc. which contribute to poor quality of service. This paper
presents analysis of WaterGrid-Sense, as a full-stack node
based on LoRa deployed on a smart water management sys-
tem. It has a smart interface platform and the ability to monitor
and control in real-time. To determine its effectiveness for
industrial usage, we conducted experiments to evaluate its
power consumption rate and the reliability of the communica-
tion link while deployed in a harsh environment. The results
obtained show that the sensor node can operate effectively
with a battery for a long period of time until a cut-off voltage of 3.2 V. Moreover, despite the harsh environment, the received
signals were sufficiently reliable. Thus, a WaterGrid-Sense could be deployed for industrial usage with great reliability
and less maintenance costs.
Index Terms — LoRa, LoRaWAN, IWSN, Industrial applications.

I. I NTRODUCTION Things (IoT). Industrial IoT (IIoT) consists of the autonomous

large number of sensors which are deployed to monitor crit-
W IRELESS sensor networks (WSN) is a network tech-
nology equipped with sensor nodes to sense, process
and transmit data wirelessly to help comprehend the behavior
ical parameters such as physical conditions, motor efficiency,
vibration, etc. [2], [5]. The data sensed, locally processed, and
of the monitored environment and to respond to resulting transmitted is critical to important decision making. Moreover,
events [1]–[3]. WSN has important attributes such as self- in order to be in line with the envisioned IIoT applications
organizing and local processing abilities, making it a driving and take the advantages provided by WSN, the research com-
force for industrial monitoring, controlling, and tracking activ- munity has positioned itself in advancing the nature of WSN
ities. In particular, WSNs have been widely used in several from sensor nodes designs, to communication technologies and
applications such as medical health care, energy, emergency protocols. WSNs are gradually replacing legacy wired systems
recovery, agriculture, smart buildings and cities, military res- due to their low-cost, rapid deployment, flexibility, ease of
cue, industrial automation and so on [1], [2], [4]. maintenance, scalability etc. [1], [2].
Currently, WSNs have invaded the industrial communication Albeit, WSN has received much interest and adoption
systems and have been considered the future of Internet of from research and development (R&D), several challenges
in node designs, communications technologies, and protocols
Manuscript received September 18, 2019; revised October 24, 2019; still lingers. Wireless sensors as the engines of WSN are
accepted October 25, 2019. Date of publication November 4, 2019;
date of current version February 5, 2020. This work was supported characterized by their small size, have resource constraints,
in part by the FRC and in part by the Department of Computer Sci- low processing power at the perception layer, and inefficient
ence at the NWU-Mafikeng and CSIR, South Africa. The associate energy consumption [1], [2]. Due to these challenges, R&D
editor coordinating the review of this article and approving it for pub-
lication was Prof. Subhas C. Mukhopadhyay. (Corresponding author: activities have been channeled to: 1.) network communications
Oratile C. Khutsoane.) technologies that are adaptive and do not consume extensive
O. Khutsoane, B. Isong, and N. Gasela are with the Depart- power from network nodes while transmitting data. 2.) Design
ment of Computer Science, North-West University, Mafikeng 2745,
South Africa (e-mail: [email protected]; [email protected]; of intelligent low-cost sensor nodes that are resource efficient
[email protected]). and resource aware. 3.) network topologies that are efficient,
A. M. Abu-Mahfouz is with the Council for Scientific and intelligent and fault tolerant, and 4.) energy harvesting tech-
Industrial Research (CSIR), Pretoria 0184, South Africa (e-mail:
[email protected]). niques to prolong sensor nodes’ lifespan resulting in low
Digital Object Identifier 10.1109/JSEN.2019.2951345 maintenance costs. These are important and have to be taken

1558-1748 © 2019 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.
See https://www.ieee.org/publications/rights/index.html for more information.

Authorized licensed use limited to: VIT University. Downloaded on May 18,2022 at 11:47:02 UTC from IEEE Xplore. Restrictions apply.
KHUTSOANE et al.: WATERGRID-SENSE: LoRa-BASED SENSOR NODE FOR INDUSTRIAL IoT APPLICATIONS 2723

into consideration when designing sensor node as they are nodes. Some of these works are discussed in the rest of this
critical to IIoT solutions [1], [2]. Moreover, a low-cost and section:
low-power sensor is ideal as most IIoT consist of monitoring Gungor and Hancke [1] explored the challenges of node
and tracking. In addition, an energy-harvesting source is a design in IIoT and some countermeasures. Some design princi-
necessity to node development. In terms of node software, ples were defined and described technical approaches to assist
authors in [6] advised on the use of various operating systems R&D when designing sensor nodes for IIoT deployments.
to ensure flexibility. In the same vein, Stoopman et al. [11] presented a system and
The recent developments in wireless communication links circuit design, of an external radio frequency (RF) powered
have given birth to low-powered networks classified as 2.4 GHz-based complementary metal oxide semiconductor
low power wide area networks (LPWANs), enabling wide (CMOS) transmitter integrated into modern autonomous wire-
area communication for low powered devices. LPWANs less sensor nodes. It uses the dedicated RF signal for energy
are characterized by low-cost devices for low-cost net- harvesting and frequency synthesis to eliminate inductors in
work deployment like IoT, low power consumption, easy to circuit design, enabling low-complexity, low power, and area
deploy network infrastructure nationwide, secure and extended efficient solution.
coverage [6], [7]. An example is the long-range low power Liu et al. [12] designed a low-power sensor node to
wireless technology platform called LoRa [6]. LoRa uses address the challenge of monitoring building temperature
LoRaWAN protocol with the aim to eliminate repeaters, reduce and light intensity for energy-efficient building design. The
device cost, increase battery life on devices, improve network sensor integrates detecting light intensity, temperature, motion
capacity, and support a large number of device connectiv- tracking, and compressive image acquisition in a single board.
ity. Currently, there are lots of developments in LPWAN Moreover, Lu et al. [13] developed the world’s smallest sensor
networks [7] and one important application is in water dis- node with ultralow power consumption and buried pump
tribution networks. interconnection technology. They explored applicable practical
This paper presents a LoRa/LoRaWAN based device known approaches for green sensor node integration with existing
as WaterGrid-Sense, which is a smart interface platform with systems and assembling them.
the ability to monitor and control in real-time, the components Paul et al. [14] designed an always-on always-off energy-
in smart networks. We evaluated the power consumption of harvesting wireless sensor node that features a near-threshold
WaterGrid-Sense while in normal operation over time and voltage IA-32 microcontroller. It is based on edge comput-
observed the performance of LoRa when the end-device is ing, having a tightly integrated sensor interface, an onboard
deployed indoors while communicating with the gateway processing unit, and onboard communication in a single
(GW) located outdoors. We carried out experiments with the board. Accordingly, Cheong et al. [15] also designed a WSN
device in terms of the power consumption and the communica- node based on ZigBee technology for ultraviolet detection
tion link and the data collected were analyzed, and presented. of flames in WSN safety applications using a spectroscopic
The performance evaluation confirms the effectiveness and the technique. In the same vein, Kan and Chen [16] designed and
impact of LoRa on energy efficiency and link reliability. implemented a small-sized wearable inertial sensor node for
The remaining parts of this paper is organized as follows: body motion analysis. To achieve the smallest node size, they
Section II presents the related works on Sensor nodes designs, printed an Inverted-F antenna and integrated it on a four-layer
Section III presents WaterGrid-Sense Node, Section IV Printed Circuit Board (PCB).
presents empirical evaluation and Section V presents data Somov et al. [17] designed a real-world wireless sensor
collection method. Lastly, we present results and analysis in node for gas sensing. To evaluate their sensor, they com-
Section VI and Section VII concludes the paper. pared it with an identical platform that uses the Wheat-
stone bridge. To reduce power consumption, a voltage divider
II. R ELATED W ORKS was used instead of a traditional Wheatstone bridge. Also,
Chen et al. [9] proposed a novel sensor node called Multi-
IIoT presents some challenges such as interference, noise, Module Separated Linear underwater sensor node to operate in
multipath fading, shadowing and so on which make the underwater WSNs. They focused on robust coverage and good
establishment of quality of service (QoS) difficult [3], [8]. communication performance. Furthermore, Imran et al. [18]
Industrial environments are hasher as a result of unpredictable investigated the use of SRAM-based Field-programmable gate
variations of temperature, pressure, humidity, and others [9]. array (FPGA) design for duty-cycled wireless vision sensor
Its nature of deployments requires sensor nodes that can handle networks. They present a low-complexity, energy efficient,
such environmental conditions. Moreover, the sensors have and reconfigurable vision sensor node using a design matrix,
to be of low-cost with certain limitations such as low rate which includes task partitioning between the server and the
and low processing power while preserving QoS [3], [10]. node addressing both processing and communication energy
Authors in [4] showed that at present, there are no full-stack consumption.
LoRa network nodes available for IoT deployment. R&D have The above discussions are some of the research works and
been using either single board computers and attaching a LoRa innovations on node design mainly for WSN. Reference [1]
module, or plug-and-sense devices that are expensive and not emphasized on node power consumption, [14] used energy
flexible for deployment. Several works have been performed harvesting techniques from external sources to power their
and some are ongoing in regards to the development of WSN nodes using solar cell while [11] harvested energy from

Authorized licensed use limited to: VIT University. Downloaded on May 18,2022 at 11:47:02 UTC from IEEE Xplore. Restrictions apply.
2724 IEEE SENSORS JOURNAL, VOL. 20, NO. 5, MARCH 1, 2020

Fig. 1. Top and bottom view WaterGrid-Sense V2.1.

surrounding magnetic fields. Harvesting is advantageous to

sensor nodes and prolong node lifetime in terms of power.
Moreover, in terms of node size [1], the smallest node was Fig. 2. Block diagram WaterGrid-Sense.
achieved in [13], for green gas sensing as well as algorithm
for energy conservation. The sensor node utilized the CMOS
C. General Operations
technology since it is good for integrating diverse sensing
capabilities, signal processing on a single board [12], and WaterGrid-Sense is based on LoRa using 868 MHz. Data is
realized as a building block for energy efficient node design. transmitted by attaching an external LoRa compliant antenna
However, what we gathered from the literature is that LoRa through the onboard antenna interfacing, which establishes a
platforms are lacking, most manufacturers focus on producing link with the gateway. The device is powered by a Li-Ion
only LoRa modules, hence there is a need to develop a generic battery and harvests external energy through a solar panel.
monitoring, and control platform based on LoRa. The node is designed to save energy at all costs, increase the
battery life by employing LoRa. LoRa ensures that the battery
III. WATERGRID -S ENSE N ODE OVERVIEW lasts longer due to the deployed network setup, which uses a
star topology and allows inactive nodes to enter sleep mode to
This section presents an overview of a LoRa-based sensor
save battery energy. Solar energy charges the battery during the
node called the WaterGrid-Sense.
day when there is sunlight and node runs on the battery source
at nighttime. The battery voltage is 3.7 V, with a capacity of
A. WaterGrid-Sense 1000mAh and the maximum charge the battery can handle is
WaterGrid-Sense is a LoRa-based sensor node currently 4.2 V at 500 mA.
used in a smart water management system (SWMS). It is a Currently, WaterGrid-Sense has two sensor interfaces: a
smart interface platform with the ability to monitor and control pressure sensor and a pulse sensor. The Pressure sensor
in real-time the components involved in SWMS. WaterGrid- uses I2 C intra-board communication for data transmission
Sense comes as a full-stack node that includes a single from the pressure sensor, attached to the water pipe, to the
PCB, processing unit, power management unit, two transceiver WaterGrid-Sense interface. The pulse sensor is attached to
interfaces, and sensor interfaces, with a small size design (See a water Reed switch on the actual water meter, the pulse
Fig. 1). The node provides a wide range of usage for differ- magnetic field sends an analog signal to the pulse sensor
ent applications and WaterGrid-Sense supports a long-range interface on the WaterGrid-Sense. Both pressure value and
communication based on LoRa/LoRaWAN using 868 MHz. water meter reading embedded into one packet and send to
The network stack used for SWMS is a three-layered the back-end system through LoRa Gateway. Moreover, other
network stack: application, abstraction and the perception. supporting components, such as voltage regulator are in place
However, this paper will focus on the perception layer of the to protect the device and regulate the functionality of all other
stack, which consists of end-devices (EDs) or sensor nodes components.
and the communication medium use with abstraction layer
(GW). LoRa is employed in the physical layer, which uses D. Handlings
a LoRaWAN communication protocol between the GW and
the ED. To use the WaterGrid-Sense the following activities are
performed:
1) Interfacing the Mote With the Water Meter: The WaterGrid-
B. WaterGrid-Sense Components Sense interfaces with four components for the water meter
WaterGrid-Sense has an onboard LoRa module using application: the battery to power up the device; the solar panel,
LoRaWAN™Class A protocol stack, sensor interfacing, and the Reed switch which connects via cable to the meter.
onboard processor, memory and finally onboard battery and The device and the battery are packed inside a small package
solar interfacing. They form the main components of the LoRa and placed inside a metal enclosure while the solar panel is
device (ED) to function in a LoRa network. Fig. 2 shows attached to the enclosure to expose it to the sun. Moreover,
the block diagram featuring the major components of the the antenna is placed atop of the metal enclosure for better line
WaterGrid-Sense: of sight between the LoRa gateway and the device. This is to

Authorized licensed use limited to: VIT University. Downloaded on May 18,2022 at 11:47:02 UTC from IEEE Xplore. Restrictions apply.
KHUTSOANE et al.: WATERGRID-SENSE: LoRa-BASED SENSOR NODE FOR INDUSTRIAL IoT APPLICATIONS 2725

Fig. 3. WaterGrid-Sense configuration menu.

keep it safe and protect against damages from people or natural

disasters.
2) Device Configuration: To connect the device to the net-
work the first time, it must be configured via an USB inter-
face. This involves flashing the settings for water metering Fig. 4. Experimental design framework.
application to the device. That is, set the present water meter
reading, set the clock date and time, and set the water
•Evaluate the reliability and effectiveness of the commu-
counter multiplier. Once this is done, the device is plugged
to the laptop and then we start the configuration program nication link using RSSI, SNR, SF, PDR and PER.
by running the command, sudo minicom –s this command The experiment scenario implemented is the node located
starts a minicom serial port communication program [19] that indoors and the gateway located outdoors, hence indoor to
enables configuration of WaterGrid-Sense. Once the configura- outdoor communication. Based on the scenario, we tried to
tion program starts, then a menu with a list of options appears constrain the communication as much as possible to emulate
as shown in Fig. 3. available options are system statistics, worst case scenarios found in industrial environments.
the system log, view the mote settings, perform testing, run
the device by joining the network and so on. C. Power Consumption Rate
WSN has been constrained by energy consumption which is
IV. E MPIRICAL E VALUATION considered inefficient leading to several nodes’ failure. Thus,
This section presents investigation into the battery life and extending the lifetime of sensor nodes, power conservation,
battery usage of the device without the external solar energy and management techniques play a significant role in sen-
source as well as its behavior during the initial communication sor nodes design [1]. To address this menace, several tech-
routine when first connected to the network and also, while in niques have been proposed such as operating in synchronous
operation. mode [5] and so on. This study is conducted to measure how
long the node could last if operated only on battery without
A. Setting solar source.
Experiments were performed in a laboratory and conducted
for a short duration. While the SWMS deployed at the Council D. Reliability of Communication Links
for Scientific and Industrial Research (CSIR) campus and Deploying WSN in the industry comes with some chal-
nodes were attached to water meters and pressure sensors lenges that emanate from the environment. Industrial envi-
installed in the actual water grid around the campus. The ronments are harsh (such as noise, interference, etc.) and
deployment environment is considered harsh due to the pres- can adversely affect the deployed wireless communication
ence of trees, tall buildings, hills and so on. This allows testing links [1], [3], [5], [8], [10]. Thus, the deployed technologies
of different network aspects such as the link budget since in such environments should be able to handle such harsh
the network communication is wireless. Moreover, it can be conditions to ensure the reliability of the system at hand with
affected by many factors such as distance, obstacles, external respect to data transmission. Real-time data will require links
network interference and so on [3], [5], [8]–[10], [20]. Since that are more reliable and can provide wide bandwidth and
LoRa is a leading LPWAN amongst others, we anticipate good high data rates. In this case, WaterGrid-Sense generates and
sensitivity, low path loss, and good obstacle penetration [20]. logs data to the server and, therefore, its communication is
not immune to the harsh environment. Thus, the data logged
B. Tasks will be used to measure the reliability of the communication
In this paper, our goal is to evaluate the power consumption link when the device is located indoors and communicating
rate and the communication link reliability or behavior of the with the gateway located outdoors. The experimental design
WaterGrid-Sense deployed at the CSIR campus. Tasks are set framework is shown in Fig.4.
as follows:
• Study/observe initial communication with the gateway.
V. DATA C OLLECTION
• Study/observe communication behavior overtime. Two kinds of data are collected, the network data and
• Study/observe power consumption overtime, and present the power consumption data. The network data is collected
Current Voltage and Power results. through the backend server, as the node communicated with
• Evaluate how long the node can operate on battery the gateway. In addition, the experiment setup depicted in
without external energy harvester source. Fig. 5 was used to power up the node and measure its energy

Authorized licensed use limited to: VIT University. Downloaded on May 18,2022 at 11:47:02 UTC from IEEE Xplore. Restrictions apply.
2726 IEEE SENSORS JOURNAL, VOL. 20, NO. 5, MARCH 1, 2020

voltage (VL ) in mV, which is the voltage across the dummy

resistor, and battery voltage (VL ) in mV, which is the voltage
across the battery. (2) computes the current flowing through
from the battery, (3) computes the voltage through the node
while (4) measures the power consumed by the node
VL
IL = (2)
R
VN = V B + VL (3)
P = I2R (4)
Fig. 5. Experiment setup.
V B = VL + VN (5)
where I L is the current from the battery through the load
consumption over time while in operation. Upon the setup of to the circuit, VL is the voltage across the dummy resistor,
the experiment, the link budget was computed using an online R is the 1.4 ohms’ dummy resistor, VN is voltage at the
link calculator called LigoWave [21]. To calculate the link sensor node, V B is battery voltage calculated using (5) while
budget, (1) is used and parameters substituted as described P represents the power obtained through (3). The node under
below. The results of this computation are discussed in the test was configured together with other nodes connected to
next section. the network, to transmit data after every 10 minutes. The data
Pr x (d Bm) = Pt x (d Bm) + G syst em (d B) was logged over two days and used to measure the reliability
of the communication link using the received signal strength
−L syst em (d B) − L channel (d B) − M(d B) (1)
indicator (RSSI) and the signal to noise ratio (SNR).
where:
 Pr x = expected power loss occurrence at the receiver B. Gateway
 Pt x = the transmitter power The gateway utilized is the MultiConnect Conduit from
 G syst em = system gains such as those associated with Multitech. It is a flexible gateway that offers configurabil-
directional antennas, etc. ity, ease of management, and communication scalability for
 L syst em = losses associated with the system such as feed- industrial IoT. For our network, it is deployed outdoors, placed
lines, antennas (height of an antenna) etc. 10 meters above a hill on CSIR campus.
 L channel = losses due to the propagation channel, either
VI. R ESULTS AND A NALYSIS
calculated via a wide range of channel models or from
empirical data This section presents the results and their analysis from the
 M = fading margin, again either calculated or from empir- experiments performed.
ical data
A. Link Path
A. Set-up and Variables This section presents the results of the link budget computed
The purpose of this experiment is to observe the power by the online calculator as shown in Fig. 6. It shows path loss,
consumption of WaterGrid-Sense while in its normal operation receiver signal level, thermal fade margin, link available, and
over time and to assess its behavior during the initial com- the distance between the node and gateway [21]. In Fig. 6,
munication routine when first joining the network, in relation a limited link occurs when after all the link calculation the
to battery usage. The study presents an indoor to an outdoor incident power at the receiver is lower than that required,
communication scenario with the node placed indoors and to meet the SNR requirement of the receiver in order to be
gateway outdoors. As shown in Fig. 5, the sensor node was able to demodulate the received data. LoRa, on the other hand,
programmed, configured, and connected to the network. The can detect signals up to -134 dBm below the noise floor.
battery voltage was 3.7 V, capacity of 1000 mAh, and with a Moreover, normalization is a process of adjusting the height of
maximum charge of 4.2V at 500 mA was used to power the the antennas in reference to the line of sight (LOS) shown on
node. A dummy resistor of 1.4 ohms was attached between the LOS path after the calculation; this process is required to
the battery and the node to measure the current and voltage. improve the link. Therefore, the gateway is placed 10 meters
The 868 MHz antenna was attached to the node to connect above a hill.
to the LoRa GW and the experimental setup was left running
for two days while the node was in normal operation and the B. Power Consumption Rate
data was logged. For the power consumed by the WaterGrid-Sense device
Fig. 5 shows the whole setup, where a Delphin Expert deployed for two days with external power source, Fig. 7
Key 100L [22] data logger was used for logging the data shows the initial operation when the node performs join oper-
in mV for both the battery and the load created by the dummy ation with the LoRa network server. Multiple beacons were
resistor. The data logger was configured to record 10 samples sent to the server and after the joining operation, the sensor
per second and then display the corresponding node operation node enters the active state for some time until it enters sleep
with respect to time. From the logged data, we have load and transmission mode.

Authorized licensed use limited to: VIT University. Downloaded on May 18,2022 at 11:47:02 UTC from IEEE Xplore. Restrictions apply.
KHUTSOANE et al.: WATERGRID-SENSE: LoRa-BASED SENSOR NODE FOR INDUSTRIAL IoT APPLICATIONS 2727

Fig. 9. Current with respect to time.

Fig. 6. Link-budget estimation calculation results.

Fig. 10. Power consumption with respect to time.

states: sleep, sensor-reading, and radio-transmission. The sleep

state uses 0.1 mV, where no operation is active on the
device and the active state is divided into two states which
are the Sensor-reading state which uses about 23mV and
the Radio-transmission state, which uses 73mV. Accordingly,
more voltage was drawn from the battery as the sensor node
Fig. 7. LoRa join operation. transmits the data to the gateway. The advantage is that the
transmission only occurs after every 10 minutes, contributing
more to energy conservation.
Furthermore, the sensor node under observation was mon-
itored from the server side during its operational states.
We found that it sometimes initiates the linking procedure
with the network due to weak received signal strength from
the gateway side. However, the node has the capability to go
from the sleep state to radio-transmission state and transmit
data successfully as shown in Fig. 9. From around 15:40 pm
to 18:40 pm the sensor node was operating from the sleep
state and radio-transmission state, which resulted in a constant
voltage and consequently, a lot of energy conservation as well.
Power consumed in respect to the states of the sensor node
during transmission was 190 mW, sensing states or idle state
the node is 52 mW and 0.1 mW for sleep state as shown in
Fig. 8. Node operation over time. Fig. 10. In general, the battery consumed was about 0.067 V
per day, which resulted in the battery lasting about 2 months.
Fig. 8 shows the operation of the node against time. The In the same vein, after the stipulated period, an energy
blue line represents the battery voltage usage and the red line harvesting technique was introduced via a solar panel to charge
represents the different states of the sensor node over time. the battery, thus, extending the lifetime of each sensor node
The voltage dropped from 3.770V to about 3.703V over a on the network. To obtain the cut-off point required to drive
period of 1 day. The sensor node has 3 different operational sensor node voltage, the sensor node was powered with a

Authorized licensed use limited to: VIT University. Downloaded on May 18,2022 at 11:47:02 UTC from IEEE Xplore. Restrictions apply.
2728 IEEE SENSORS JOURNAL, VOL. 20, NO. 5, MARCH 1, 2020

Fig. 11. Test node RSSI over time.

Fig. 12. Test node SNR values over time.

TABLE I TABLE II
AVERAGE P OWER C ONSUMPTION AVERAGE C OMMUNICATION P ERFORMANCE

The RSSI indicated the received signal power level after a

combination of all possible loss along the propagation. The
voltage supplier and the voltage was reduced accordingly, and
higher the RSSI value the stronger the received signal strength.
the resultant voltage cut-off was 3.2V.
Accordingly, LoRa can detect signals up to -134 dBm below
The node while in operation has three states as observed
the noise floor. This makes it one of the robust LPWAN
in Fig. 7, Fig. 8, and Fig. 9, the transmission state (Tx),
wireless communication technology. Also, Fig. 12 presents
the idle/reading state, and the sleep state respectively. The
SNR with ranges between 9 dB down to -3.1 dB throughout.
results in Fig. 9 show that during Tx the node draws an average
current of 0.52 mA, during the idle state 0.16 mA and 0.1 mA
for sleep state. These results, in the context of this paper, are VII. C ONCLUSION
acceptable and correspond to works found in the literature.
IIoT literature on sensor nodes revealed that the design
of low powered nodes is still an ongoing research. In this
C. Communication Link Reliability paper, we conducted node experiments using a novel LoRa
In the above experimental setup, we tracked the network based full-stack sensor node for IWSN called WaterGrid-
data from the server side as it was transmitted from the test Sense. It integrated all the main components on a single board
sensor node shown in Fig. 5. The goal was to measure the and followed the SoC design for energy conservation. As the
reliability of the communication link of LoRa when the ED state-of-the-art node designs suggest that a node should have
deployed indoors is communicating with the gateway located an external or on-chip energy harvesting source, WaterGrid-
outdoors. Fig. 11 shows the RSSI measured in dBm. The best Sense employs a solar panel to harvest energy and recharges
value attained was -91 dBm and worst value was -119 dBm. the battery source that powers the sensor node.

Authorized licensed use limited to: VIT University. Downloaded on May 18,2022 at 11:47:02 UTC from IEEE Xplore. Restrictions apply.
KHUTSOANE et al.: WATERGRID-SENSE: LoRa-BASED SENSOR NODE FOR INDUSTRIAL IoT APPLICATIONS 2729

To assess the WaterGrid-Sense for deployment, we con- [18] M. Imran, K. Shahzad, N. Ahmad, M. O’Nils, N. Lawal, and
ducted two experiments, 1) power consumption of the sensor B. Oelmann, “Energy-efficient SRAM FPGA-based wireless vision sen-
sor node: SENTIOF-CAM,” IEEE Trans. Circuits Syst. Video Technol.,
node and, 2) communication reliability provided by the sensor vol. 24, no. 12, pp. 2132–2143, Dec. 2014.
node. The results obtained show that without the external [19] Minicom(1)—Linux Man Page. Accessed: Jun. 13, 2017. [Online].
energy harvester the sensor node can operate for more than Available: https://linux.die.net/man/1/minicom
[20] LoRa Community. Accessed: Jun. 14, 2017. https://semtech.force.
two months on a battery until cut-off voltage, which is 3.2 V. com/lora/LC_ArticleSearch?ArticleType=Data_Sheet_kav
In terms of communication reliability, the node produced [21] Log in—LigoWave Link Calc. Accessed: Jun. 14, 2017. [Online]. Avail-
outstanding results as summarized. able: https://linkcalc.ligowave.com/Account/login
[22] Models | Delphin Technology AG. Accessed: Sep. 13, 2017.
[Online]. Available: http://www.delphin.com/products/measuring-and-
ACKNOWLEDGMENT testing-devices/expert-key/models.html
The authors would like to acknowledge Dr. Hartmut IIgner
for providing them with the measurement equipment for the Oratile Khutsoane (S’18) received the B.Sc.
lab experiment. degree in computer science and electronics,
the B.Sc. (Hons.) degree in computer science,
and the M.Sc. degree in computer science from
R EFERENCES North-West University, South Africa, in 2015,
[1] V. C. Gungor and G. P. Hancke, “Industrial wireless sensor networks: 2016, and 2018, respectively. He is currently a
Challenges, design principles, and technical approaches,” IEEE Trans. Software Engineer focusing on the development
Ind. Electron., vol. 56, no. 10, pp. 4258–4265, Oct. 2009. of spectrum management tools with the Council
[2] D. V. Diego et al., “Survey and systematic mapping of indus- for Scientific and Industrial Research (CSIR),
trial Wireless Sensor Networks,” J. Netw. Comput. Appl., vol. 97, South Africa. His research interests are wireless
pp. 96–125, Nov. 2017. sensor networks, low power-wide area networks,
[3] R. Ibrahim, T. D. Chung, S. M. Hassan, K. Bingi, and and advanced Internet of Things.
S. K. B. Salahuddin, “Solar energy harvester for industrial wireless
sensor nodes,” Procedia Comput. Sci., vol. 105, pp. 111–118, Jun. 2017.
[4] O. Khutsoane, B. Isong, and A. M. Abu-Mahfouz, “IoT devices and Bassey Isong (M’09) received the B.Sc. degree
applications based on LoRa/LoRaWAN,” in Proc. IECON 43rd Annu. in computer science from the University of Cal-
Conf. IEEE Ind. Electron. Soc., Oct./Nov. 2017, pp. 6107–6112. abar, Nigeria, in 2004, the M.Sc. degrees in
[5] G. R. Delgado, A. M. Oliveira, L.-F. A. Cavalcante, M. A. Spohn, and computer science and software engineering from
B. F. Antônio, “On the Impact of local processing for motor monitoring the Blekinge Institute of Technology, Sweden
systems in industrial environments using Wireless Sensor Networks,” in 2008 and 2010, respectively, and the Ph.D.
Int. J. Distrib. Sens. Netw., vol. 9, no. 7, Jul. 2013, Art. no. 471917. degree in computer science from the North-West
[6] A. J. Wixted, P. Kinnaird, H. Larijani, A. Tait, A. Ahmadinia, and University in 2014. He is a Senior Lecturer with
N. Strachan, “Evaluation of LoRa and LoRaWAN for wireless sensor the Department of Computer Science and a
networks,” in Proc. IEEE Sensors, Oct./Nov. 2016, pp. 1–3. Faculty Member with the North-West University,
[7] J. Haxhibeqiri, I. Moerman, and J. Hoebeke, “Low overhead scheduling Mafikeng Campus, South Africa. His research
of LoRa transmissions for improved scalability,” IEEE Internet Things interests include and are not limited to software engineering, security,
J., vol. 6, no. 2, pp. 3097–3109, Apr. 2019. software defined wireless sensor network, low-power wide-area net-
[8] M. Cheffena, “Industrial wireless sensor networks: Channel model- works, cloud computing, Internet of Things, machine learning, and big
ing and performance evaluation,” J. Wireless Commun. Netw., vol. 1, data. He is also a member of IEEE Computer, Communication, and
pp. 1–8, Dec. 2012. Education Societies.
[9] H. Chen, X. Wu, G. Liu, and Y. Wang, “A novel multi-module sep-
arated linear UWSNs sensor node,” IEEE Sensors J., vol. 16, no. 11, Naison Gasela received the B.Sc. degree in
pp. 4119–4126, Jun. 1, 2016. economic data processing from Kharkov Engi-
[10] A. Ajith Kumar S, K. Ovsthus, and L. M. Kristensen. “An neering Economics Institute, USSR, in 1985,
industrial perspective on wireless sensor networks—A survey of the PGDip and M.Sc. degrees in software engi-
requirements, protocols, and challenges,” IEEE Commun. Surveys neering from the University of Abertay Dundee,
Tuts., vol. 16, no. 3, pp. 1391–1412, 3rd Quart., 2014, doi: 10. U.K., and the Ph.D. degree in computer science
1109/SURV.2014.012114.00058. from the National University of Science and Tech-
[11] M. Stoopman, K. Philips, and W. A. Serdijn, “An RF-powered DLL- nology, Zimbabwe, in 2001. He is an Associate
based 2.4-GHz transmitter for autonomous wireless sensor nodes,” IEEE Professor with the Department of Computer Sci-
Trans. Microw. Theory Techn., vol. 65, no. 7, pp. 2399–2408, Jul. 2017. ence, North-West University, Mafikeng Campus,
[12] X. Liu, M. Zhang, and J. Van der Spiegel, “A low-power multifunctional South Africa. His research interests include big
CMOS sensor node for an electronic facade,” IEEE Trans. Circuits Syst. data, machine learning and artificial intelligence, software engineering,
I, Reg. Papers, vol. 61, no. 9, pp. 2550–2559, Sep. 2014. software defined wireless sensor network, cloud computing, and the
[13] J. Lu, H. Okada, T. Itoh, T. Harada, and R. Maeda, “Toward the world Internet of Things. He is also a member of ACM and ORSSA.
smallest wireless sensor nodes with ultralow power consumption,” IEEE
Sensors J., vol. 14, no. 6, pp. 2035–2041, Jun. 2014.
[14] S. Paul et al., “A Sub-cm3 energy-harvesting stacked wireless sen- Adnan M. Abu-Mahfouz (M’12–SM’17) re-
sor node featuring a near-threshold voltage IA-32 microcontroller in ceived the M.Eng. and Ph.D. degrees in
14-nm tri-gate CMOS for always-on always-sensing applications,” IEEE computer engineering from the University of
J. Solid-State Circuits, vol. 52, no. 4, pp. 961–971, Apr. 2017. Pretoria. He is currently a Principal Researcher
[15] P. Cheong, K.-F. Chang, Y.-H. Lai, S.-K. Ho, I.-K. Sou, and K.-W. Tam, with the Council for Scientific and Industrial
“A zigbee-based wireless sensor network node for ultraviolet detection Research (CSIR), a Research and Innovation
of flame,” IEEE Trans. Ind. Electron., vol. 58, no. 11, pp. 5271–5277, Associate with the Tshwane University of Tech-
Nov. 2011. nology, and an Extraordinary Faculty Member
[16] Y.-C. Kan and C.-K. Chen, “A wearable inertial sensor node for with the University of Pretoria. He is an Associate
body motion analysis,” IEEE Sensors J., vol. 12, no. 3, pp. 651–657, Editor with IEEE ACCESS, the IEEE INTERNET
Mar. 2012. OF THINGS, and the IEEE TRANSACTION ON
[17] A. Somov, A. Baranov, D. Spirjakin, and R. Passerone, “Circuit Design INDUSTRIAL INFORMATICS, and a member of many IEEE technical commu-
and Power Consumption Analysis of Wireless Gas Sensor Nodes: One- nities. He is the Founder of the Smart Networks Collaboration initiative
Sensor Versus Two-Sensor Approach,” IEEE Sensors J., vol. 14, no. 6, that aims to develop efficient and secure networks for the future smart
pp. 2056–2063, Jun. 2014. systems, such as smart cities, smart grid, and smart water grid.

Authorized licensed use limited to: VIT University. Downloaded on May 18,2022 at 11:47:02 UTC from IEEE Xplore. Restrictions apply.