sensors

Article
An Integrated IoT Architecture for Smart Metering
Using Next Generation Sensor for Water Management
Based on LoRaWAN Technology: A Pilot Study
Vlastimil Slaný 1, * , Adam Lučanský 1 , Petr Koudelka 1 , Jan Mareček 1 , Eva Krčálová 1
and Radek Martínek 2
1 Department of Agricultural, Food and Environmental Engineering, Mendel Univerzity in Brno,
61300 Brno, Czech Republic; [email protected] (A.L.); [email protected] (P.K.);
[email protected] (J.M.); [email protected] (E.K.)
2 Department of Cybernetics and Biomedical Engineering, VSB—Technical University of Ostrava,
70800 Ostrava-Poruba, Czech Republic; [email protected]
* Correspondence: [email protected]; Tel.: +420-545-132-371




Received: 7 July 2020; Accepted: 6 August 2020; Published: 20 August 2020


Abstract: This pilot study focuses on the design, implementation, optimization and verification of
a novel solution of smart measuring of water consumption and crisis detection leading to a smart
water management platform. The system implemented consists of a modular IoT platform based
on a PCB (Printed Circuit Board) design using the M2.COM standard, a LoraWAN modem and
a LoraWAN gateway based on the Raspberry Pi platform. The prototype is modular, low-cost,
low-power, low-complex and it fully reflects the requirements of strategic technological concepts
of Smart City and Industry 4.0, i.e., data integration, interoperability, (I)IoT, etc. The study was
produced in cooperation with M.I.S Protivanov and VODARENSKA AKCIOVA SPOLECNOST,
a.s. (industry partners distributing drinking water in the Olomouc and South-Moravian regions)
to depict the current situation in the Czech Republic, characterized by extreme weather fluctuations
and increasingly frequent periods of drought. These drinking water distributors are also constantly
placing new demands on these smart solutions. These requirements include, above all, reliability of
data transmission, modularity and, last but not least, low cost. However, smart water management
(water consumption, distribution, system identification, equipment maintenance, etc.) is becoming an
important topic worldwide. The functionality of the system was first verified in laboratory conditions
and, then, in real operation. The study also includes checking signal propagation in the municipal area
of the village of Zdarna, where the radius of the proposed measuring system was tested. A laboratory
test with simulation of water leakage is also part of this work. Subsequently, the system was tested in
a residential unit by means of water leakage detection using the MNF method (minimum night flow);
the detection success rate was 95%.

Keywords: smart metering; LoRaWAN; smart water; modularity; water management

1. Introduction
The topic of water management has gained prominence recently because of extreme weather in
the Czech Republic [1], which is currently mapped by the INTERSUCHO project [2]. The Aqueduct
prediction model [3] expects two or three times higher water stress until 2030 in almost half of the
territory of the Czech Republic. Figure 1 shows the progression of annual precipitation and the average
temperature in the Czech Republic.

Sensors 2020, 20, 4712; doi:10.3390/s20174712 www.mdpi.com/journal/sensors

Sensors 2020, 20, 4712 2 of 23

Territorial precipitation in Czechia

1200
Total percipitation per year
Linear: y = - 0.00197*x + 678 Linear regresion (Total percipitation per year)
1000 Long-term percipitation normal 1981 - 2010 [mm]
Total Precipitation [mm]

800

600

400

200

0
1980 1985 1990 1995 2000 2005 2010 2015 2020

12
Average air temperature in [°C]
Linear: y = 0.048*x - 87 Linear regresion (Average air temperature)
10 Long-term air temperature normal 1981-2010
Temperature [° C]

0
1980 1985 1990 1995 2000 2005 2010 2015 2020

Figure 1. Territorial precipitation and territorial temperature in Czechia [4]

However, Ref. [5–7] proved that extreme weather fluctuations and drought are not problems of
only the Czech Republic. They are becoming a global issue with no promise of improvement in the
future [8] along with the associated intelligent water management (water consumption, distribution,
water quality monitoring, system identification, equipment maintenance, etc.) [9–11]. The analysis
conducted in the last few years proved that these intelligent measuring systems are highly beneficial
and effective, not only in the case of involving citizens in reducing energy consumption, but also when
identifying possible waste of resources [12,13]. For example, the average distribution loss in drinking
water in the EU countries is 23% [14].
In the Czech Republic, a volumetric flow meter is currently most often used to measure the
consumption, mainly due to its low price and good reliability. Its function is based on measuring the
speed of water flow through a pipe when it is rotated by a piston or turbine. The volumetric water
flow is directly proportional to the speed of rotation of the blades. Mechanical water flow meters may
become clogged if the water is dirty or contains larger particles (leading to increased maintenance costs),
which is their disadvantage. At low flow rates, the error rate of these water meters also increases.
Recently, these mechanical flow meters have been equipped with additional electronics allowing
them to read remotely. These water meters are indicated as AMR (Automatic meter reading) [15].
Sensors 2020, 20, 4712 3 of 23

The number of these water meters in the Czech Republic is still low, mainly due to the purchase
price paid by the end customer. Another option may be fully electronic water meters employing new
measuring methods, such as electromagnetic [12], fluidic and ultrasonic meters [15]. According to
Act 274/2001 Coll. each building that wants to be connected to the water supply network must be
equipped with one of the aforementioned types of water meter (the water meter must be certified for
the required use) [16]. Table 1 shows an overview of households connected in the Czech Republic by
region. The total number of buildings connected is 94.7% with a total consumption of 609.7 million m3
of drinking water, when the losses reached 95 million m3 , which is 15.8% of the total consumption
(the losses are broken down by individual entities connected to the distribution network) [17].
The system monitoring consumption of drinking water has not yet been used on a larger scale in the
Czech Republic. Mostly, they are just big cities where pilot projects are taking place. For example, in
Prague, remote readings of water meters have been being implemented since 2013. In 2019, the number
of water meters installed reached 8244, which is approximately 7.2% of the total number of 113,848
water meters (operated by Pražské vodovody a kanalizace, a.s) [18]. It is also important to state that
sewage (waste water) pricing in the Czech Republic is broken down by individual entities on the basis
of drinking water consumption. In the event of an accident, the entities are, therefore, charged not
only for the drinking water consumed, but also for the waste water.
As for foreign companies, we can mention, for example, Thames Water supplying drinking water
to the Thames Valley (in the south-east of England) and most of London. In 2013, this company came
up with a program to install smart water meters. The goal of this program is to provide each of the
9 million customers with their own water meter by the end of 2030, thus allowing remote readings
(except for problematic locations). This is the first smart metering project in the water industry in the
United Kingdom. In 2019/20, 449,000 water meters, installed primarily in the area with the highest
water leakage, were managed by this company. Thanks to this step, the year-on-year reduction in
leakage is 13.8%. [19] When reducing water consumption and using this technology, we will also
experience the following benefits:

• The load on the pumps will be reduced, which will be reflected primarily in longer service life
and lower maintenance costs
• Better sustainability and management of current water resources (in certain parts of the year,
for example, restrictions on drinking water consumption are announced when, for example, it is
forbidden to water gardens, etc., with tap water)
• Early detection of unusual conditions—such as unauthorized entry into the shaft, drop in ambient
temperature below the allowable value or detection of zero water consumption at a certain time
(used, for example, to check senior citizens, wherein zero water consumption may mean a health
issue of this person)
• Fault detection—for example, water meter failure, backflow or attempted unauthorized
manipulation of the water meter.

The development of IoT (Internet of Things) technologies enables us to use smart technologies
in areas yet unimaginable. Their continuous integration into daily life is associated with the broad
support of Smart City projects, which use the data collected (consumption of water, gas, electricity,
or street lighting) for improving the quality of life in cities [20–23]. The Smart City concept envisages
the deployment of sensors at important nodes in the distribution network, which will allow us to
detect the large leaks quickly [24,25].
The basic methods of water leakage detection include the AMR method (which is also included in
our solution). Leakage detection is evaluated on the basis of the data read, which is then processed
and evaluated [26]. Hamilton and Charalambous [27] list several other basic methods for detecting
water leaks:
Sensors 2020, 20, 4712 4 of 23

• Gas injection method: Is very effective in cases where acoustic methods cannot be used due to the
absence of sound signals. Tracer gas, which spreads well through the pipeline and rises to the
surface at the point of the defect, where it can be located with a suitable measuring device with
sensitivity to this gas, is fed into the circulation system.
• Manual listening stick: Is used to listen to the leaks on fittings and to pinpoint the location of the
leak. The listening stick may be made of metal, wood, or plastic.
• Hydrophones: This is a sensitive microphone that is in direct contact with water and can,
therefore, capture sounds transmitted by water and not just by the pipes. Based on the analysis of
the noise, a possible water leakage can be detected [28,29].
• Noise logger corelation: These devices are placed in accessible places on the pipeline, where they
try to detect and evaluate the noise. The leakage is always located between two noise loggers,
where the position of the fault can be calculated [30].
• Thermal imaging: They can detect temperature differences that reveal the presence of water.
They are mainly used to detect cracked pipes in buildings [31,32].

However, the aforementioned methods are relatively expensive and technically demanding.
In addition, they are mainly used to detect water leakage on the supplier’s side. In contrast, our solution
focuses on end users.
Many types of water meters without the necessary support for remote reading are available on
the market. They are designed primarily for the visual representation of water consumption when
a human operator’s presence is still required. However, the Directive of the European Parliament
and the Council (EU) 2018/2002 [33] sets out a new rule of obligation to install only devices that
allow remote reading from 25 October 2020 [33] in order to increase energy efficiency throughout
the energy chain and to provide consumers with accurate, truthful, and reliable information of their
energy consumption (especially concerning the billing of heat and hot water). Devices installed before
this date must be replaced no later than 1 January 2027 [33]. Nevertheless, the market does not offer
an open and widely reputable communication standard for data transmission. Thus, most producers
develop their own solutions with the implementation of novel communication protocols, such as
LoraWAN, Sigfox, or NarrowBand IoT [13].
The study focuses on the implementation of IoT remote system for monitoring water consumption
(including possible leaks) using the LoraWAN technology. The current development in the IoT area
allows us to meet the ever-increasing demands on the economical use of water resources, caused by
the current climate situation in the Czech republic. Since most producers must rely on signal coverage
by the local providers, who charge a monthly fee for each device connected to their network, we also
proposed a new low-cost modular LoraWAN gateway. The modularity, where other peripherals can
be connected to our device via the integrated MikroBUS bus, whether it is another water meter or a
flood sensor, is also another advantage of our system. A temperature sensor, which can be used to
control the temperature in water supply shafts, is also directly integrated into the device. The system
is also ready for possible replacement of the communication module) with another type (e.g., SigFox).
However, this version has not been tested yet.

Related Works and Solutions

The use of smart technologies in measuring systems is a frequent topic in the research, and
many various solutions in the area of energy reading (water, gas, or electricity) are described.
Jin, G. et al. [34] and Li et al. [35] proposed a system for measuring water consumption based on
image recognition technology, using MCU (Microcontroller unit) STM32F103ZET6 and camera module
OV7725. The original image of the water meter counter is preprocessed using greying, edge extraction,
Otsu’s binarization, and tilt correction. The digits extracted in this way are then recognized by
CNN (Convolutional Neural Network) using the structure of a classical two-stage target detection
network [36]. The data is then transmitted to the server using the NB-IoT (NarrowBand-Internet
Sensors 2020, 20, 4712 5 of 23

of Things) technology. Kelnar, M. et al. [37] described the implementation of a low-cost, two-way,
and fibreless optics communication system for water meter reading, which is implemented again using
image recognition. Al-Ali et al. [38] proposed an IoT device determined to measure the consumption
of electricity, water, and gas, providing remote access, and making the desired configuration. The data
is read by the pulse output and displayed to the user with the goal to increase the outline of the
consumption. The sensors are connected to the central control unit (Raspberry Pi), collecting the data
required and, then, sending it to the server. Similar issues are dealt with in [13,39,40].
Besides neuron networks, ultrasound sensors provide more precise reading, especially in a low flow
rate zone [41]. In [42], the measuring system based on a control unit (a centralizer), which can communicate
with ultrasound sensors via Mbus or wirelessly using the LoRaWAN technology, is described. The data is
available for the users in cloud storage, where it is derived from the control unit by means of a mobile
network. The shortcomings of this method are discussed in [43]. Short sound path, weak echo signal,
and different temperature gradient result in the decline of accuracy and stability.
Another reading method uses the pulse output. The pulse is usually a mechanical contact
switching with each countdown wheel turning in the water meter. Other water meters are equipped
with a magnetic needle [44], and the sensor is placed on the water meter glass. The sensor detects the
needle’s magnetic field and indicates the result as the number of its rotation cycles. The pulse signal
obtained can be transferred to the PLC (Programmable logic controller) and then processed.
There are several different solutions to data transmission. The most common technologies in the
area of smart measuring systems and long-distance data transmission include the LoraWAN [45–47],
Sigfox [48,49] or NB-IoT technologies [34,50]. These technologies are compared in [51], where their
benefits and shortcomings are discussed. In particular, the safety of IoT networks is an important
and frequently discussed issue. For example, Lin et al. [52] described the architecture, the useful
technologies, and the IoT security, discussed the potential safety issues and privacy protection,
and proposed possible solutions. Navarro-Ortiz et al. [53] focuses on the safety of the LoRaWAN
technology and describes the possibility of its improvement on the HW level. The analysis of LoraWAN,
Sigfox, and NB-IoT vulnerability is conducted in [54]. This work [55] describes the concept of Small
Cell Cloud (SCC) composed of multiple Cloud-enabled Small Cells (CeSCs), which provide radio
connection for mobile User Equipment (UE), such as smart-phones or wearables.
First, analysis of the available solutions of smart water management and comparison of the
technologies used for remote data transmission were conducted in Section 2. Based on this review
and the requirements on the system, we selected the LoraWAN technology. Section 3 describes the
development of the motherboard with the M.2 module. Finally, the results of system testing under
laboratory conditions with the controlled simulation of water leakage and in an apartment building,
and gateway testing in real operation are shown in Section 4.

2. Matherial and Methods

The entire process of the prototype design is described in Figure 2.

Smart water meter Gateway platform

Signal coverage test
solution analysis solution

Next generation
Data transfer Advanced
Optimalization smart mettering
solution analysis measurement
platform

Smart sensor
Laboratory testing
platform solution

Figure 2. Smart water meter concept.

Sensors 2020, 20, 4712 6 of 23

The proposed platform is modular, as shown in Figure 3, and consists of two modules: (1) Carrier
board and (2) M2.COM processor module with the necessary communication protocol. Both modules
are described below. Due to the use of the device in water supply shafts, it is necessary to protect the
device especially from high humidity. For this purpose, standardized junction boxes with IP protection
index 67 are used.

IP network

Raspberry Pi based gateway The things network

LoraWAN MQTT

Server > InfluxDB

Motherboard and M.2 module > Grafana

MicroBUS
QWIIC

Flowmeters Users

Figure 3. Concept of the proposed smart metering system.

When proposing the solution, the following requirements were set on the basis of cooperation
with the companies MIS s.r.o. and VAS a.s.:

• Modularity between the modules connected for reading the data from the sensors and the rest of
the system
• Use of exclusively standardized interfaces
• Possibility to connect other external sensors (e.g., a flood sensor) via these standardized interfaces
• DC-DC converter to the battery power supply
• Minimum battery life of two years
• Integration of the LoraWAN gateway into the data transmission system
• Data transmission interval from the device to the gateway, at least once an hour.
• Low production costs for surface deployment

Another part of the measuring chain comprises a gateway that collects the data from the individual
sensors. The gateway used is based on the Raspberry Pi platform using an iC880A-LoRaWAN®
Concentrator 868 MHz board by IMST (Germany).
The modular system of the proposed device comprises a front-end module with two open-collector
inputs, so a magnetic reed switch and any commercially available readout attachment for flow meters
can be used. The over-voltage protection and the RC filter of the inputs are set to filter pulses less than
20 ms and noise.
The general diagram of the proposed measuring system is described in Figure 4.
Sensors 2020, 20, 4712 7 of 23

RF

RN2483 LoraWAN modem

UART

MikroBUS MikroBUS M.2 module

external modul external modul ( processor unit )

SPI SPI
I2C I2C M2.COM
UART UART
GPIO GPIO

2 - 15 V
3.3 V
power Carrier board
source
power supply

Figure 4. Diagram of the proposed measuring sensor.

2.1. Carrier Board

As shown in Figure 4, the carrier board consists of a power connector (JST-PH2) with a power
range from 2 to 15 V, an M2.COM connector, a DC-DC converter, and connectors selected depending
on the interfaces. The complete design is presented in Figure 5. The power supply of the entire
device is designed with regard to the secured functionality, both in a low voltage range of the LiFePO4
battery (2.6 V) and in a high range of 12 V of the accumulator (maximum 13 V). We assume that the
maximum current on the 3.3 V branch does not exceed 200 mA. A buck-boost LTC3129 chip was
used as integrated switch-mode power supply [56]. The proposed power supply circuit enables the
processor module to shut down, if it is supplied externally, to prevent overcharging between these
sources. This is a preparation for the M2.COM module with an external connector (USB), which will
be used to power the entire device during programming.

Figure 5. Printed Circuit Board (PCB) with M2.COM module and microBUS module for reading from
pulse water meter.
Sensors 2020, 20, 4712 8 of 23

2.2. M2.COM Processor Unit

The main part of the processor unit comprises a module that includes a processor, a modem,
and selected peripheries (F-RAM memory, Hall probe, and two LEDs). A block diagram of this module
is shown in Figure 6. The peripheries of the microprocessor are connected to the side of the M2.COM
connector, which is also used as an input for the power supply and other signals (resetting or switching
off external buses). The antenna can be connected through the SMA connector with 50 Ω conduction
to the LoRaWAN module. The basic properties of the STM32L431RB processor are summarized in
Table 1 [57].

M2.COM processor unit

Antenna output
UART
RN2483

M2. COM LoraWAN modem

I2C FRAM
FM24CL16-DG

Processor I2C
Thermometer
STM32L431RB humidity meter

GPIO
Hall effect sensor
Si7201

GPIO
2 x LED

SWD debug

Figure 6. Block diagram of the processor module.

Table 1. Basic properties of the STM32L431RB processor.

Chip type STM32L431RB

SRAM memory size 64 kB
FLASH memory size 128 kB
Operating voltage 1.71–3.6 V
Core frequency 80 MHz
Pin count 64 pins
Temperature range −40 ◦ C–105 ◦ C
I2C 3
UART 3
SPI 3

The I2C (Inter-Integrated Circuit) temperature and humidity sensor (Si7006-A20-IM) is placed
on the top of the board. It serves for testing the communication and functionality after the assembly
(the data successfully sent indicates a functional critical path). Its accuracy is determined by
the producer ±1 °C (−10 °C to 85 °C) a ±5% RH (0–90% RH) [58].
State and configuration preservation is secured by F-RAM memory, which is characterized by
practically an infinite number of transcripts (1014 ), thus enabling saving the current state of the water
meter counter without destroying the memory cells. We choose the FM24CL16B−DG chip with the I2C
interface and a capacity of 16 kbit [59]. Wireless communication and data transmission are performed
via an RN2483 modem made by Microchip. This module provides the LoRaWAN protocol using a
simple UART (Universal asynchronous receiver-transmitter) interface (communication with module
passes on the serial port using the ASCII protocol) and enables us to use the LoRaWAN CLASS A
Sensors 2020, 20, 4712 9 of 23

protocol, designed for low-energy equipment [60]. The connector, 6-wire JST-PH (1mm spacing) in the
SMD (Surface Mount Device) case, was used for program tuning and recording purposes according to
the DroneCode standard. Besides the power supply wire and the negative pole, there are also two
SWD wires (Serial Wire Debug) and the UART.

2.3. Reading Module

The proposed module shown in Figure 7 can be used to sense any switching mechanism,
optionally with a phototransistor. The module consists of an RC filter connected with a diode providing
time settings to stabilize output oscillation during the switching on and off process. Moreover,
overvoltage protection and dual Schmitt trigger NC7WZ14P6 [61] are used. The proposed module is
actually shown in Figure 5, top view, right side.

Figure 7. Scheme of data reading module from the selected water meters.

2.4. Firmware
The firmware for the selected STM32 microcontroller on the proposed IoT platform uses the
ecosystem of Ivory and Ivory-Tower made by Galois, which is written in the Haskell functional
programming language [62,63]. The advantage of this solution is the ability to quickly change the
functionality based on the current requirements. The Tower framework’s main concept is a description
of behaviour by Hoare monitors with synchronous typed channels that resemble a Petri net. Such a
description of behaviour is then transformed into the C language along with the auxiliary code (FreeRTOS).

2.5. Transmission Frequency

The device contains internal F-RAM memory, where the current state of the pulse counter is
stored at regular intervals to maintain its state after a power outage. This data is subsequently sent in
a predefined time interval using the LoraWAN module. In our case, the time interval for sending the
data is set to 1 h. The minimum interval for sending the data is 3 min.

2.6. Energy Consumption

For the validation and testing purpose, the device was powered from a wall outlet adapter, but the
theoretical lifespan for operation using a battery (3xAA LR6 battery) is a minimum of two years with a
1 h interval of sending the data obtained (pulse counter, humidity and temperature) (calculation based
on STM32CubeMX) [64]. The general consumption at 132 µA is 3.0 V.
Sensors 2020, 20, 4712 10 of 23

2.7. LoraWAN Gateway

The gateway is based on Raspberry PI 3 B + a platform (see Figure 8), to which an iC880A
backplane is connected through an SPI (Serial Peripheral Interface) bus. The backplane is a board for
connecting the iC880A-SPI to the Raspberry Pi and serves for supplying the Raspberry PI and the
iC880A-SPI board itself (see Figure 6) [65,66].

Gateway schema
U.FL to SMA
Outdoor antena
iC880A - SPI
868 Mhz

SPI Power supply 5 V

Power supply
iC880A backplane

RJ45
POE SPI Power supply 5 V
splitter
Ethernet
Raspberry PI 3 B
with NixOS

Figure 8. Gateway schema based on Raspberry PI and ICC880A.

The iC880A module can be used for many applications, such as Smart Metering, IoT, or M2M
network building. It can receive packets from several different endpoints simultaneously, when they
are sent with different SF settings, through up to 8 channels concurrently. Hence, it provides the
possibility of robust communication between the gateway and a large number of endpoints at different
distances. The proper function of the iC880A needs a host system (in our case Raspberry Pi 3 B
combined with NixOS). This module is also connected with an external antenna using a U.FL−N
converter. The technical parameters of the antenna are summarized in Table 2.

Table 2. Technical specifications of outdoor 868 MHz antenna.

Frequency Range (MHz) 824 MHz–960 MHz

Bandwidth (MHz) 70
Gain (dBi) 10
VSWR kHz <=1.5
Input Impedance (Ω) 50
Polarization Vertical
Maximum input power (W) 50
Lightning protection DC Ground
Input connector type N Female
Operating temperature (°C) −40–60

As soon as the data is received by the gateway, it is immediately sent, via the IP Network, to the
server, where the data is further processed and evaluated.

LoraWAN Transmission Technology

The gateway, based on the LoraWAN technology, is a low-power consuming wireless network
protocol proposed for low-cost and safety communication in the IoT area. LoRa is a physical layer used to
establish a remote communication link, based on Chirp-Spread-Spectrum modulation, which preserves the
same properties of low-power consumption as FSK (Frequency-Shift Keying) modulation, but significantly
increases the possible communication range. Chirp-Spread-Spectrum is used mainly in military and space
Sensors 2020, 20, 4712 11 of 23

communications, due to the possible long communication distances, and is resistant to interference. LoRa
is the first low-cost implementation of this technology for the commercial use [67].
Another advantage of LoRa is the possibility of flexible configuration, which enables changing
the signal range by adjusting the bandwidth (BW), the spreading factor (SF), and the coding rate (CR).
According to EU863−870 Industrial, Scientific, and Medical (ISM) specifications, the BW can reach
values of 125 kHz, 250 kHz, and 500 kHz. The SF can be set in the range from 7 to 12, when it reaches
the value of 7 for short-distance communication (up to 4 km depending on the environment) and the
value of 12 for long-distance communication. The CR improves the robustness of code receiving of
4-bit data with redundancies for error correction in variants 4/5, 4/6, 4/7, or 4/8. The bit rate of LoRa
is calculated according to the following equation [68,69]:

BW 4
Rb = SF · · [bit s−1 ] (1)
2S F 4 + CR
where SF is the spreading faktor; BW is the band width; CR is the code rate.
Equation (1) shows that the SF is one of the dominant factors for the computation of the bit
rate, which defines the physical speed of data transmission (data rate (DR), DT), as shown in Table 3.
The DR in the range from 6 to 14 is not included because it is not implemented in ISM EU863−870 [69].

Table 3. LoRaWAN EU863-870 TX Data rate [69].

Data Rate Configuration Approximate Physical Bit Rate (bit/s)

0 SF12/125 kHz 250
1 SF11/125 kHz 440
2 SF10/125 kHz 980
3 SF9/125 kHz 1760
4 SF8/125 kHz 3125
5 SF7/125 kHz 5470

The nodes in the LoraWAN network are asynchronous, and the communication passes off after
the data is ready to be sent, no matter whether this condition is event-driven or scheduled. The nodes
often have to “wake up” in a mesh or synchronous network, such as a cellular one, synchronize with
the network and check the message. This synchronization consumes a considerable amount of energy
and is the main reason for shortening the battery life.
Terminal equipment is also used for different applications and has completely different
requirements. Terminal equipment of class A provides two−way communication when each uplink
transmission of the terminal equipment is followed by two short receiving windows for the downlink.
The emitting slot of the terminal equipment is based on its own communication needs (ALOHA
protocol). Hence, the terminal equipment requires downlink communication from the server shortly
after it sent the uplink to the server. Any other downlink communication has to wait before the
next sensor starts emitting. Terminal equipment of class B opens other receiving windows, such as
the reaction on a trigger from the gateway indicating the “listening” of the terminal equipment.
Terminal equipment of class C has almost continuously open receiving windows, closing only during
the data transmission to the server [67,70].
The concept of the device meets the requirements for classification in Class A, especially because
of the intended use in water supply shafts, where the battery is the only source of electric energy.

3. Results
The experimental workplace was built for primary testing, as shown in Figure 7. The closed
testing circuit consisted of a water tank, a pump, and one Meinecke WS-50 water meter with an output
for reed contact (when K = 1 and K = 10) and a MEDER KSK1A66-1020 reed switch. This reed contact
switches when a magnetic field is present. The reed switch was soldered to the end of the two-core
cable (originally one-channel video cable with a CINCH connector), and the JST-PH2 connector was
Sensors 2020, 20, 4712 12 of 23

added to its other side. The switch could be connected directly to the microprocessor output, but the
accurate and reliable measurement requires “front—end” (see Section 2.3), which avoids interference
and oscillations, and creates overvoltage protection.
Then, two water meters, SENSUS 420 [71], equipped with pulse module output, are connected in
parallel. An HRI-A (High-Resolution Interface) pulse module by SENSUS is connected to the interface
described above (see Section 2.3). Moreover, two ball valves are placed into the circuit (see Figure 9).
The first ball valve serves primarily for simulating a pipe rupture. If this ball valve is opened, part of
the water starts passing through the second branch and the third water meter. The second ball valve
regulates the pressure in the testing circuit to achieve the recommended operating conditions of the
individual water meters.

Figure 9. Laboratory testing circuit.

Figure 10 shows a time course of a laboratory experiment detecting a possible water leak during
a failure. The blue line indicates measurement with no water leak detected, which means that the
flow through water meter no. 1 is the same as through water meter no. 2. The red line represents a
simulation of failure and pipe rupture in the time of 60:00. The overall simulated water leak is sensed
using water meter no. 3, indicated by green colour. The water leak simulation was stopped in the time
of 1:49, and the total water leak reached 960 L. The threshold ability to detect water leaks was stated as
100 L, which can be lowered to 10 L by using another type of pulse output. In real operation, in the
Czech Republic, this method is used for transporting water over long distances, wherein there can
be up to 4 pumping stations on the route (a water meter is installed for each station). Subsequently,
during the distribution of drinking water, a so-called foot water meter is installed in each municipality,
which measures the water consumption of the entire municipality. Of course, the possibilities of water
leak detection in this manner also have their limits (for example, in the case of circular water mains,
where the water can be pumped back).
Sensors 2020, 20, 4712 13 of 23

Laboratory water leak detection experiment

8000
No water leak (Flow meter No. 1)
7000 With water leak (Flow meter No. 2)

Cumulative water flow in litres

Measured water leak size (Flow meter No. 3)
6000 Calculated water leak

5000

4000

3000

2000

1000

0
0 20 40 60 80 100 120 140 160 180
Time of measurement in minutes

Figure 10. Water leak detection experiment.

The green line in the graph represents the total measured size of the water leak (water meter no. 3)
and the black curve represents the calculated size of the water leak (based on the difference between
the values measured on water meter no. 1 and water meter no. 2) . Until no leakage is detected,
its value is still 0. If a leak is detected, the estimated water leak size is displayed. If the leak is stopped,
the total amount of water loss is still displayed, but the curve does not continue to grow (until another
water leak is detected). Ideally, when there is no water leakage, the red (water meter no. 2) and blue
(water meter no. 1) curves overlap. If a leak is detected, the red curve begins to decrease and the
difference between the red and blue curves is equal to the magnitude of the water leak detected.

3.1. Signal Coverage Test

The measurement took place in the village of Zdarna (area of 1.038 ha, 768 inhabitants,
305 buildings in the village, GPS (Global Positioning System) coordinates of 49.4875122 N, 16.6599706 E).
The aim was to compare the signal propagation and its radius in the village of Zdarna. A basic version
of the proposed Printed Circuit Board (PCB) with an integrated temperature and humidity sensor
(Si7006—A20—IM) was used, and the ADATA–P20000D 2000 mAh power bank ensured the power
supply. The entire device is shown in Figure 11.

Figure 11. Mobile measuring system composed of modular sensor and power bank.
Sensors 2020, 20, 4712 14 of 23

The measurement was conducted using the TTN Mapper application. The first step is a
registration of the gateway on www.thethingsnetwork.org, where the GPS coordinates, frequency plan,
and antenna type are assigned. The second step requires registration into the network defined by the
measuring device (in our case MONSTICK no. 5) with the assignment of the EUI identifier. Then,
the measurement can be started using the TTN Mapper mobile application. The measurement record
collects all data (temperature and humidity) that are captured by the registered gateway, see Table 4.

Table 4. Example of measurement record using TTN mapper.

Date and Time 2020–04–21 @ 09:28

Node MONSTICK5
Gateway ID 40D63CFFFE1F4309
Location accuracy 10.00
Packet ID 14508882
Distance 212.0 m
Data rate SF12BW125
Frequency 867.7 MHz
RSSI (Received Signal Strength Indication) −76.0
SNR ( Signal-to-noise ratio) 9.8
Altitude 641.4 m

Measurements in Zdarna Village

Over 350 measurements were made in the village of Zdarna (see Figure 12), both in a built-up
area and at places with potentially new building plots. We found out that dense vegetation can lead to
strong signal attenuation, but no locations without signal coverage were identified.

Figure 12. Measurement of the signal strength in Zdarna village (scale 1:6000).
Sensors 2020, 20, 4712 15 of 23

A weak signal was measured only on the village outskirts, in the southern and south-eastern
directions. Here, the signal strength ranged from 115 to 120 dBm due to the elevation and relatively
high density of buildings. However, these signal values do not prevent a successful data transmission.
The associated HEAT MAP, describing the signal transmission in this area, is shown in Figure 13.
Coverage in the non-completely measured area is calculated using interpolation.
The cloud icon in the middle of the image is used to indicate the position of the LoRaWAN
gateway (GPS coordinates of 49.4704200 N, 16.7590344 E). The maximum distance of the measured
point (the building in the village, which is the furthest from our gate) for data transmission is 870 m,
when the signal strength at this location reached −113 dBm with the SRN value of 10 dB.

Figure 13. Heat map of the signal coverage in scale 1:6000.

3.2. Measurement in Real Conditions

The functionality of the system was verified by reading the cold and hot water in a low-rise
apartment building. The results were written down manually to validate the accuracy of the
measurement and are shown in Figure 14 (hot water) and Figure 15 (cold water).
The frequency of the automatic and manual data collection was set to 1 h. Subsequently, the data
was manually paired, according to the timestamp. This timestamp contained every outgoing message
in the following format: YYYY—MM—DD HH:MM:SS. The method of detecting water leaks that has
been used is the so-called MNF method (minimum night flow), widely used in water supply. This is
the flow that runs into a certain zone/measuring district during the night, when most legitimate water
consumption is at its minimum. The period of the minimum night flow usually occurs between two
and four o’clock in the morning. In our case (measurement of water consumption in an apartment in a
low-rise building), the MNF period was set from 23:59 to 6:00, when the threshold value was set to
10 L. If the flow in the system exceeds this value, a water leak is assumed.
Sensors 2020, 20, 4712 16 of 23

The quantile-quantile or q-q plot is an exploratory graphical device used to check the validity
of a distributional assumption for a data set. If the data really follow the assumed distribution, then
the points on the q-q plot will fall approximately on a straight line. The q-q plot provides a visual
comparison of the sample quantiles to the corresponding theoretical quantiles. In general, if the points
in a q-q plot depart from a straight line, then the assumed distribution is called into question. The q-q
plots may be thought of as being “probability graph paper” that makes a plot of the ordered data
values into a straight line [72]. As can be seen in Figures 16 and 17, the resulting linear regression lines
overlap each other. Thus, both measurement methods have the same probability densities. The method
of measurement by means of an IoT sensor is thus relevant for practical application.

Hot water consumption measurement

1200 120

1000 100
Cumulative Water Consumption (l)

800 80

Number of pulses
600 60

400 40

200 20

0 0
3.4.2020 4.4.2020 5.4.2020 6.4.2020 7.4.2020 8.4.2020 9.4.2020 10.4.2020
Date of measurement (DD:MM:YYYY)

Real water consumption Pulse measurement by IoT module

Figure 14. Comparison between remote and manual reading—hot water.

Cold water consumption measurement

1400 140

1200 120
Cumulative Water Consumption (l)

1000 100
Number of pulses

800 80

600 60

400 40

200 20

0 0
3.4.2020 4.4.2020 5.4.2020 6.4.2020 7.4.2020 8.4.2020 9.4.2020 10.4.2020
Date of measurement (DD:MM:YYYY)

Real water consumption Pulse measurement by IoT module

Figure 15. Comparison between remote and manual reading—cold water.

Sensors 2020, 20, 4712 17 of 23

HOT WATER
120
Water consumption - number of pulses
100

80

60

40

20

0
-3,0 -2,5 -2,0 -1,5 -1,0 -0,5 0,0 0,5 1,0 1,5 2,0 2,5 3,0

-20

-40
Normal Theoretical Quantiles

Pulse measurement by IoT module Real water consumption

Figure 16. Q-Q plot—hot water.

COLD WATER
140

120
Water consumption - number of pulses

100

80

60

40

20

0
-3,0 -2,5 -2,0 -1,5 -1,0 -0,5 0,0 0,5 1,0 1,5 2,0 2,5 3,0
-20

-40
Normal Theoretical Quantiles

Pulse measurement by IoT module Real water consumption

Figure 17. Q-Q plot—cold water.

4. Discussion
Since the weather conditions constantly worsen and the Smart City project develops, we can expect
an increasing emphasis on sustainable water management, smart measuring of consumption, and the
detection of water leaks. The development of these technologies could positively affect the behaviour
of the individual consumers because the final user can have a detailed outline in their household about
their consumption. In case of the failure (pipe rupture, leak, running toilets or water taps), a warning
message is generated, which could reduce not only the possible leaks, but also the resulting financial
burden. This fact is supported by the EU thanks to Directive 2018/2002 of the European Parliament and
the Council, introducing mandatory monthly billing for thermal energy costs.
Sensors 2020, 20, 4712 18 of 23

Three low-power technologies for data transmission from water meters to the users are nowadays
suitable for use in longer distances; LoraWAN, Sigfox, and NarrowBand IoT. We can expect increasingly
massive deployment of these smart technologies due to the major development of IoT technologies
and the succession of 5G networks. Our platform is ready to integrate these technologies thanks to the
modular solution, which does not require the use of a particular water meter type and allows water
companies to choose different techniques and producers.
Two approaches to remote water meter readings are available; machine vision and pulse output.
The former is not suitable for wet-running (flood) water meters because of probable fogging or watering
the glass, which prevents image recognition. On the contrary, the latter can be enabled under any
conditions, and is less energy demanding, so it can be used without electricity.
Our solution enables water companies to connect multiple types of sensors for monitoring the
ambient temperature, flood sensors, or to secure the access to the shaft using a magnetic detector,
providing an informative message about the access to the area. Our system also includes the LoraWAN
gateway, which allows us to cover the wide area. Therefore, considerable funds could be saved. Local
operators charge fees for connecting each device to the network when the price is in the range from 0.5
to 1.5 USD, so using a higher number of sensors could be very expensive. Moreover, coverage by the
desired technology may not be available in the selected location. Deployment of the low-cost gateway
can be considered as an ideal solution. However, it still requires basic infrastructure in the form of
access to the Ethernet network and ensuring POE power.
The main advantages of our solution include modularity, where multiple sensors can be connected
to one device (for example, a flood sensor, a hatch opening or/and another water meter). This is
useful, for example, in the case of housing units that have water meters for both hot and cold water.
Thanks to the integrated temperature sensor, we can also detect too low a temperature, which could
lead to freezing of the pipeline and the subsequent failure. Thanks to this modularity, our device is also
ready for the use in other areas, such as waste management (smart bin) or readings of other types of
energy (such as electricity). Thanks to the MNF method, we are also able to detect possible water leaks.
However, nowadays there are also solutions that use artificial intelligence based on neural networks to
detect water leaks [73,74]. The device is initially in the learning mode, trying to learn the patterns of
the users’ behaviour. Nevertheless, at this time, the users are responsible for the reliability and proper
operation of the system. More advanced devices are able not to only report an error condition, but also
to stop the water supply to the building to prevent further possible leaks. The development of this
artificial intelligence seems to be another suitable step to improve our system. It is also important to
look at the economic aspects. The price of our equipment is around 40 USD per piece (in the case of
buying 100 or more pieces). It can also be assumed that companies will want to pass on these costs to
the consumers through the increased water and sewerage charges. However, it should also be noted
that water companies will have lower labour costs, since they will not need the staff who now have to
conduct the readings manually. There are also other problems associated with this fact. As described
above, there is increasing pressure to provide consumers with the most up-to-date information possible
and to bill energy consumption more frequently, which would mean hiring new staff to make these
readings. Another burning issue is privacy, where the end customers have water meters installed
directly in their houses or apartments and the company’s employees must be allowed to enter these
buildings. For these reasons, the interest of water companies in these smart solutions is still growing
in the Czech Republic.

5. Conclusions
The paper deals with the development and deployment of the IoT platform for smart measuring
of water consumption and the detection of failures using water leaks. The entire system was designed
as modular, wherein the solutions can be found according to the current needs of the desired area
and conditions. The LoraWAN technology is used for data transmission, including our LoraWAN
gateway. The first part of the paper describes the design and HW properties of the proposed system
Sensors 2020, 20, 4712 19 of 23

in detail and focuses on the review of the already existing solutions. The second part describes
the actual measurement and the testing of the proposed solution under both laboratory and real
conditions. The study also includes a heat map of signal propagation in the rural and urban areas.
Then, the measurement in the apartment unit was conducted using hand-operated reading as well as
automatic reading by means of the proposed IoT module. The results showed that the average relative
error, when measuring using the IoT module, was under 3%. The weaknesses of the solution were
identified and discussed. The discussion also outlined new opportunities that future research can take
due to the continuous innovation in IoT technologies.

Author Contributions: Conceptualization, V.S., R.M. and P.K.; methodology, R.M. and V.S.; software and
validation, A.L. and V.S.; formal analysis, P.K., R.M. and E.K.; investigation and resources, J.M. and E.K.; data
curation, A.L.; writing—original draft preparation, V.S., R.M., A.L. and P.K.; writing—review and editing, E.K.,
V.S. and A.L.; visualization, V.S., P.K. and A.L.; supervision, R.M. and J.M.; project administration, R.M. and V.S.;
funding acquisition, R.M. All authors have read and agreed to the published version of the manuscript.
Funding: This research was funded by the Ministry of Education of the Czech Republic, Project No. SP2020/156.
Acknowledgments: This work was supported by the European Regional Development Fund in Research
Platform focused on Industry 4.0 and Robotics in the Ostrava project, CZ.02.1.01/0.0/0.0/17_049/0008425
within Operational Programme Research, Development and Education.
Conflicts of Interest: The authors declare no conflict of interest.

Abbreviations

AMR Automatic meter reading

BW Bandwidth
CNN Convolutional Neural Network
CR Coding Rate
FSK Frequency-Shift Keying
GPS Global Positioning System
IoT Internet of Things
I2C Inter-Integrated Circuit
LPWAN Low-Power Wide-Area Network
MCU Microcontroller unit
MNF Minimal Night Flow
NB-IoT NarrowBand-Internet of Things
PCB Printed circuit board
PLC Programmable logic controller
RSSI Received Signal Strength Indication
SF Spreading Factor
SMD Surface Mount Device
SPI Serial Peripheral Interface
SNR Signal-to-noise ratio
SWD Serial Wire Debug
UART Universal asynchronous receiver-transmitter

References
1. Khan, S.J.; Deere, D.; Leusch, F.D.; Humpage, A.; Jenkins, M.; Cunliffe, D. Extreme weather events: Should
drinking water quality management systems adapt to changing risk profiles? Water Res. 2015, 85, 124–136.
[CrossRef] [PubMed]
2. Intersucho. Intersucho.cz [Online]. Available online: https://www.intersucho.cz/en/ (accessed on
6 July 2020).
3. World Resources Institute—Aqueduct Water Risk Atlas. Aqueduct [Online]. Available online: https:
//www.wri.org/aqueduct (accessed on 6 July 2020).
Sensors 2020, 20, 4712 20 of 23

4. Czech Hydrometeorological Institute. Historical Data—Meteorology and Climatology. Czech

Hydrometeorological Institute [Online]. Available online: http://portal.chmi.cz/historicka-data/pocasi/
zakladni-informace?l=en (accessed on 6 July 2020).
5. Grillakis, M.G. Increase in severe and extreme soil moisture droughts for Europe under climate change.
Sci. Total. Environ. 2019, 660, 1245–1255. [CrossRef] [CrossRef] [PubMed]
6. Easterling, D.R. Climate Extremes: Observations, Modeling, and Impacts. Science 2000, 289, 2068–2074.
[CrossRef] [CrossRef]
7. Ionita, M.; Tallaksen, L.M.; Kingston, D.G.; Stagge, J.H.; Laaha, G.; Van Lanen, H.A.J.; Scholz, P.; Chelcea, S.M.;
Haslinger, K. The European 2015 drought from a climatological perspective. Hydrol. Earth Syst. Sci. 2017, 21,
1397–1419. [CrossRef] [CrossRef]
8. Stagge, J.H.; Kingston, D.G.; Tallaksen, L.M.; Hannah, D.M. Observed drought indices show increasing
divergence across Europe. Sci. Rep. 2017, 7, 14045, doi:10.1038/s41598-017-14283-2. [CrossRef] [CrossRef]
9. Rathore, M.M.; Ahmad, A.; Paul, A.; Rho, S.; Rho, S. Urban planning and building smart cities based on the
Internet of Things using Big Data analytics. Comput. Netw. 2016, 101, 63–80. [CrossRef] [CrossRef]
10. Saab, C.; Shahrour, I.; Chehade, F.H. Risk Assessment of Water Accidental Contamination Using Smart Water
Quality Monitoring. Expo. Health 2020, 12, 281–293. [CrossRef] [CrossRef]
11. Gong, J.; Lambert, M.F.; Stephens, M.L.; Cazzolato, B.; Zhang, C. Detection of Emerging through-Wall Cracks
for Pipe Break Early Warning in Water Distribution Systems Using Permanent Acoustic Monitoring and
Acoustic Wave Analysis. Water Resour. Manag. 2020, 34, 2419–2432. [CrossRef] [CrossRef]
12. Thomas, B.; Damien, G.; Pierre, M.; Ariane, L.; Candice, M.; Stuart, W.; Rodney, S. Inteligent Metering for
Urban Water: A Review. Water 2013, 5, 1052–1081. [CrossRef]
13. Alvisi, S.; Casellato, F.; Franchini, M.; Govoni, M.; Luciani, C.; Poltronieri, F.; Riberto, G.; Stefanelli, C.;
Tortonesi, M. Wireless Middleware Solutions for Smart Water Metering. Sensors 2019, 8, 1853. [CrossRef]
[CrossRef]
14. EurEau—Europe’s Water in Figures. An Overview of the European Drinking Water and Waste Water
Sectors. 2017. Available online: https://www.danva.dk/media/3645/eureau_water_in_figures.pdf#page=
15 (accessed on 5 June 2020).
15. Li, X.J.; Chong, P.H.J. Design and Implementation of a Self-Powered Smart Water Meter. Sensors 2019, 19, 4177 .
[CrossRef] [CrossRef] [PubMed]
16. Act No. 274/2001 Coll—The Water Supply and Sewerage System Act. Available online: https://public.psp.
cz/en/sqw/sbirka.sqw?cz=274&r=2001 (accessed on 30 June 2020).
17. Report on Water Management in the Czech Republic in 2018. 2019. Available online: http://eagri.cz/public/
web/file/640180/Modra_zprava_2018_ENG_web.pdf (accessed on 15 July 2020).
18. PVK a. s. in 2019—Annual Report. 2019. Available online: https://www.pvk.cz/res/archive/1566/230634.
pdf?seek=1594712230 (accessed on 15 July 2020).
19. Combined Report—Annual Report, Annual Performance Report and Sustainability Report 2019/20. 2019.
Available online: https://www.thameswater.co.uk/about-us/investors/our-results (accessed on 15 July 2020).
20. Jamie, L.; Jesus, T.; Canovas, A.; Parra, L. An Integrated IoT Architecture for Smart Metering.
IEEE Commun. Mag. 2016, 54, 50–57. [CrossRef]
21. Kabalci, Y.; Kabalci, E.; Padmanaban, S.; Bo, H.N.J.; Blaabjerg, F. Internet of things applications as energy
internet in smart grids and smart environments. Electronics 2019, 8, 972. [CrossRef] [CrossRef]
22. Vavrinsky, E.; Kotradyova, V.; Svobodova, H.; Kopani, M.; Donoval, M.; Sedlak, S.; Subjak, J.; Zavodnik, T.
Detection of Advanced Wireless Sensors Used to Monitor the Impact of Environment Design on Human
Physiology. Adv. Electr. Electron. Eng. 2019, 17, 320–329. [CrossRef]
23. Nedoma, J.; Fajkus, M.; Bednarek, L.; Vasinek, V. Non-destructive Fiber-optic Sensor System for the
Measurement of Speed in Road Traffic. Adv. Electr. Electron. Eng. 2016, 14, 602–608. [CrossRef] [CrossRef]
24. Hsia, S.C.; Hsu, S.W.; Chang, Y. Remote monitoring and smart sensing for water meter system and leakage
detection. IET Wirel. Sens. Syst. 2012, 2, 402–408. [CrossRef] [CrossRef]
25. Rodney, A. S.; Rachelle, W.; Damien, G.; Kriengsak, P.; Guillermo, C. Web-based knowledge management
system: Linking smart metering to the future of urban water planning. Aust. Plan. 2010, 47, 66–74. [CrossRef]
26. Farah, E.; Shahrour, I. Smart water for leakage detection: Feedback about the use of automated meter reading
technology. In Proceedings of the 2017 Sensors Networks Smart Emerging Technologies (SENSET), Beirut,
Lebanon, 12–14 September 2017; pp. 1–4, [CrossRef]
Sensors 2020, 20, 4712 21 of 23

27. Hamilton, S.; Charalambous, B. Leak Detection: Technology and Implementation; IWA Publishing: London,
UK, 2013; ISBN 9781780404707. Available online: https://library.oapen.org/handle/20.500.12657/33035
(accessed on 15 July 2020).
28. Xu, J.; Chai, K.T.C.; Wu, G.; Han, B.; Wai, E.L.C.; Li, W.; Yeo J.; Nijhof, E.; Gu, Y. Low-cost, tiny-sized
MEMS hydrophone sensor for water pipeline leak detection. IEEE Trans. Ind. Electron. 2019, 66, 6374–6382.
[CrossRef] [CrossRef]
29. Martini, A.; Rivola, A.; Troncossi, M. Autocorrelation Analysis of Vibro-Acoustic Signals Measured in a test
field for water leak detection. Appl. Sci. 2018, 8, 2450. [CrossRef] [CrossRef]
30. Li, S.; Song, Y.; Zhou, G. Leak detection of water distribution pipeline subject to failure of socket joint based
on acoustic emission and pattern recognition. Measurement 2018, 115, 39–44. [CrossRef] [CrossRef]
31. Fahmy, M.; Moselhi, O. Automated detection and location of leaks in water mains using infrared photography.
J. Perform. Constr. Facil. 2010, 24, 242–248. [CrossRef] [CrossRef]
32. Pauline, E.; Alvarado, C.; Meza, G.-R. Water leak detection by termographic image analysis, in laboratory
tests. Proceedings 2020, 48, 15. [CrossRef] [CrossRef]
33. Directive (EU) 2018/2002 of the European Parliament and of the Council of 11 December 2018, Amending
Directive 2012/27/EU on Energy Efficiency. Available online: https://eur-lex.europa.eu/legal-content/EN/
TXT/HTML/?uri=CELEX:32018L2002 (accessed on 5 June 2020).
34. Jin, G.; Bai, K.; Zhang, Y.; He, H. A smart water metering system based on image recognition and narrowband
internet of things. Rev. D’Intell. Artif. 2019, 33, 293–298. [CrossRef] [CrossRef]
35. Li, C.; Su, Y.; Yuan, R.; Chu, D.; Zu, A.J. Light-weight spliced convolution network-based automatic water
meter reading in smart city. IEEE Access 2019, 7, 174359–174360. [CrossRef] [CrossRef]
36. Ding, X.Y.; Zhang, J.P.; Yang, J. A robust two-stage algorithm for local community detection. Knowl. Based Syst.
2018, 152, 188–199. [CrossRef] [CrossRef]
37. Kelnar, M.; Martinek, R.; Machacek, Z.; Vanus, J.; Bilik, P.; Zidek, J. Realization of prototype of a low-cost
bidirectional communication system through fibreless optics. Wirel. Pers. Commun. 2016, 91, 137–150.
[CrossRef] [CrossRef]
38. Al-Ali, A.R.; Landolsi, T.; Hassan, M.H.; Ezzeddine, M.; Abdelsalam, M.; Baseet, M. An IoT-Based
Smart Utility Meter. In Proceeding of the 2nd International Conference on Smart Grid and Smart Cities,
Kuala Lumpur, Malaysia, 12–14 August 2018; pp. 80–83. [CrossRef]
39. Anjana, S.; Sahana, M.N.; Ankith, S.; Natarajan, K.; Shobha, M. An IoT based 6LoWPAN enabled experiment
for water managementr. In Proceeding of the 2015 IEEE International Conference on Advanced Networks
and Telecommuncations Systems (ANTS), Kolkata, India, 15–18 December 2015; pp. 1–6. [CrossRef]
40. Taufik, M.; Rakhmania, A.E.; Afnani, Y.N. Prepaid water meter card based on internet of things.
In Proccedind of the 1st Annual Technology, Applied Science and Engineering Conference, East Java,
Indonesia, 29–30 August 2019. [CrossRef]
41. Lee, C.-H.; Jeon, H.-K.; Hong, Y.-S. An implementation of ultrasonic water meter using dToF measurement.
Cogent Eng. 2017, 4, 1371577. [CrossRef] [CrossRef]
42. Wang, J.-X.; Liu, Y.; Lei, Z.-B.; Wu, K.-H.; Zhao, X.-Y.; Feng, C.; Liu, H.-W.; Shuai, X.-H.; Tang, Z.-M.; Wu, L.-Y.;
et al. Smart water lora IoT system. In Proceeding of the 2018 International Conference on Communication
Engineering and Technology, Singapore, 24–26 February 2018; pp. 48–51. [CrossRef]
43. Luo, A.; Jiang, Y.; Li, Z.; Zhao, W. A new ultrasonic water meter with window time adjustment algorithm.
IOP Conf. Ser. Mater. Sci. Eng. 2019, 569, 052070. [CrossRef] [CrossRef]
44. Liu, J.; Li, H.; Lv, M.; Dong, S.; He, Y.; Wu, S. Investigation on water recording system for large user of water
supply network based on PLC. Procedia Environ. Sci. 2011, 11, 1016–1022. [CrossRef]
45. Khutsoane, O.; Isong, B.; Gasela, N.; Abu-Mahfouz, M. Water grid-gense: A loRa-based sensor node for
industrial IoT applications. IEEE Sens. J. 2020, 20, 2722–2729 [CrossRef] [CrossRef]
46. Codeluppi, G.; Cilfone, A.; Ferrarnn G. LoRaFarM: A LoRaWAN-based smart farming modular IoT
architecture. Sensors 2020, 20, 2028. [CrossRef] [CrossRef] [PubMed]
47. Haxhibeqiri, J.; De Poorter, E.; Moerman, I.; Hoebeke, J. A survey of LoRaWAN for IoT: From technology to
application. Sensors 2018, 18, 3995. [CrossRef] [CrossRef] [PubMed]
48. Joris, L.; Dupont, F.; Laurent, P.; Bellier, P.; Stoukatch, S.; Redoute, J. An autonomous sigfox wireless sensor
node for environmental monitoring. IEEE Sens. Lett. 2019, 3, 1–4. [CrossRef] [CrossRef]
Sensors 2020, 20, 4712 22 of 23

49. Rubio-Aparicio, J.; Cerdan-Cartagena, F.; Suardiaz-Muro, J.; Ybarra-Moreno, J. Design and implementation
of a mixed IoT LPWAN network architecture. Sensors 2019, 19, 675. [CrossRef]
50. Shi, Z.; Wang, M.; Lin, H.; Lin, H.; Gao, Z.; Huang, L. NB-IOT pipeline water leakage automatic monitoring
system based on cloud platform. In Proceedings of the 2019 IEEE 13th International Conference on
Anti-counterfeiting, Security, and Identification (ASID), Xiamen, China, 25–27 October 2019; pp. 272–276.
[CrossRef]
51. Borrero, J.D.; Zabalo, A. An autonomous wireless device for real-time monitoring of water needs. Sensors
2020, 20, 2078. [CrossRef] [CrossRef]
52. Lin, J.; Yu, W.; Zhang, N.; Yang, X.; Zhang, H.; Zhao, W. A survey on internet of things: Architecture,
enabling technologies, security and privacy, and applications. IEEE Internet Things J. 2017, 4, 1125–1142.
[CrossRef] [CrossRef]
53. Navarro-Ortiz, J.; Chinchilla-Romero, N.; Ramos-Munoz, J.J.; Munoz-Luengo, P. Improving hardware
security for LoRaWAN. In Proceeding of the 2019 IEEE Conference on Standards for Communications and
Networking, CSCN 2019, Granada, Spain, 28–30 October 2019. [CrossRef]
54. Coman, F.L.; Malarski, K.M.; Petersen, M.N.; Ruepp, S. Security issues in internet of things: Vulnerability
analysis of LoRaWAN, sigfox and NB-IoT. In Proceeding of the 2019 Global IoT Summit (GIoTS), Aarhus,
Denmark, 17–21 June 2019. [CrossRef]
55. Dolezal J.; Zeman T. Introduction to the computation offloading from mobile devices to the edge of mobile
network. Adv. Electr. Electron. Eng. 2019, 17, 413–422. [CrossRef] [CrossRef]
56. Linear Technology Corporation. [Datasheet]. LTC3129. ©2013 . Available online: https://www.analog.com/
media/en/technical-documentation/data-sheets/3129fc.pdf (accessed on 24 May2020).
57. STMicroelectronics. [Datasheet]. STM32L431RB. ©2018 . Available online: https://www.st.com/resource/
en/datasheet/stm32l431rb.pdf (accessed on 24 May 2020).
58. Silicon Laboratories Inc. [Datasheet]. Si7006-A20-IM. ©2016 . Available online: https://www.silabs.com/
documents/public/data-sheets/Si7006-A20.pdf (accessed on 24 May 2020).
59. Cypress Semiconductor Corporation. [Datasheet]. FM24CL16B-DG. ©2018. Available online: https://www.
cypress.com/file/136491/download (accessed on 24 May 2020).
60. Microchip Technology Inc. [Datasheet]. RN2483. ©2015 . Available online: https://ww1.microchip.com/
downloads/en/DeviceDoc/40001784B.pdf (accessed on 24 May 2020).
61. ON Semiconductor. [Datasheet]. NC7WZ14. ©2019. Available online: https://www.onsemi.com/pub/
Collateral/NC7WZ14-D.PDF (accessed on 24 May 2020).
62. Elliott, T.; Pike, L.; Winwood, S. Hickey, P.; Bielman, J.; Sharp, J.; Seidel, E.; Launchbury, J. Guilt Free Ivory.
ACM Sigplan Not. 2015, 10, 189–200. [CrossRef]
63. Jaloyan, G.-A.; Pike, L. Lock optimization for hoare monitors in real-Time systems. In Proceeding of the
2017 17th International Conference on Application of Concurrency to System Design (ACSD), Zaragoza,
Spain, 25–30 June 2017; pp. 49–53. [CrossRef]
64. STM32Cube Initialization Code Generator, STMicroelectronic [Software], Version 5.6.1. Available online:
https://www.st.com/en/development-tools/stm32cubemx.html#get-software (accessed on 17 July 2020).
65. Raspberry Pi Foundantion. [Datasheet]. RaspberryPi 3 B+. ©2019. Available online: https://www.
raspberrypi.org/products/raspberry-pi-3-model-b-plus/ (accessed on 30 May 2020).
66. Gonzalo Casas. [GitHub Project]. IC880A Backplane. ©2016. Available online: https://github.com/
gonzalocasas/ic880a-backplane/ (accessed on 30 May 2020).
67. LoraAliance. LoRaWAN™ What Is It? [Online]. ©2015. Available online: https://lora-alliance.org/sites/
default/files/2018-04/what-is-lorawan.pdf (accessed on 26 May 2020).
68. Mroue H.; Nasser A.; Parrein B.; Hamrioui S.; Mona-Cruz E.; Rouyer G. Analytical and simulation study for
LoRa modulation. In Proceeding of the 2018 25th International Conference on Telecommunications (ICT),
St. Malo, France, 26–28 June 2018; pp. 655–659. [CrossRef]
69. LoraAliance. LoRaWAN™ 1.0.2 Regional Parameters. [Online]. ©2017. Available online: https://lora-
alliance.org/sites/default/files/2018-05/lorawan_regional_parameters_v1.0.2_final_1944_1.pdf (accessed
on 26 May 2020).
70. Bezzera, N.; Åhlund, C.; Saguna, S.; Sousa, V., Jr. Temperature Impact in LoRaWAN—A Case Study in
Northern Sweden. Sensors 2019, 19, 4414. [CrossRef] [CrossRef]
Sensors 2020, 20, 4712 23 of 23

71. SENSUS. [Datasheet]. 420 WATER METER. ©2010. Available online: https://sensus.webdamdb.com/bp/#/
search/?q=%22420%20Water%20Meter%22&filters=%257B%257D (accessed on 30 May 2020).
72. Online Statistics Education: A Multimedia Course of Study. Project Leader: David M. Lane, Rice University.
Available online: http://onlinestatbook.com/2/advanced_graphs/q-q_plots.html (accessed on 2 July 2020).
73. Barradas, I.; Garza, L.E.; Morales-Menendez, R.; Vargas-martínez, A. Leaks Detection in a Pipeline Using
Artificial Neural Networks. In Proceeding of the Progress in Pattern Recognition, Image Analysis, Computer
Vision, and Applications, CIARP 2009, Jalisco, Mexico, 15–18 November 2009. [CrossRef]
74. Abdulla, M.B.; Herzallah, R.O.; Hammad, M.A. Pipeline leak detection using artificial neural network:
Experimental study. In Procceding of the 2013 5th International Conference on Modelling, Identification and
Control (ICMIC), Cairo, Egypt, 31 August–2 September 2013; pp. 328–332. [CrossRef]

© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access
article distributed under the terms and conditions of the Creative Commons Attribution
(CC BY) license (http://creativecommons.org/licenses/by/4.0/).