sensors

Article
LoRa-Based IoT Network Assessment in Rural and
Urban Scenarios
Aikaterini I. Griva 1, * , Achilles D. Boursianis 1 , Shaohua Wan 2 , Panagiotis Sarigiannidis 3 ,
Konstantinos E. Psannis 4 , George Karagiannidis 5 and Sotirios K. Goudos 1, *

1 ELEDIA@AUTH, School of Physics, Aristotle University of Thessaloniki, 54124 Thessaloniki, Greece

2 Shenzhen Institute for Advanced Study, University of Electronic Science and Technology of China,
Shenzhen 518110, China
3 Department of Electrical and Computer Engineering, University of Western Macedonia, 50131 Kozani, Greece
4 Department of Applied Informatics, School of Information Sciences, University of Macedonia,
54636 Thessaloniki, Greece
5 School of Electrical and Computer Engineering, Aristotle University of Thessaloniki,
54124 Thessaloniki, Greece
* Correspondence: [email protected] (A.I.G.); [email protected] (S.K.G.)

Abstract: The implementation of smart networks has made great progress due to the development
of the Internet of Things (IoT). LoRa is one of the most prominent technologies in the Internet of
Things industry, primarily due to its ability to achieve long-distance transmission while consuming
less power. In this work, we modeled different environments and assessed the performances of
networks by observing the effects of various factors and network parameters. The path loss model,
the deployment area size, the transmission power, the spreading factor, the number of nodes and
gateways, and the antenna gain have a significant effect on the main performance metrics such
as the energy consumption and the data extraction rate of a LoRa network. In order to examine
these parameters, we performed simulations in OMNeT++ using the open source framework FLoRa.
The scenarios which were investigated in this work include the simulation of rural and urban
environments and a parking area model. The results indicate that the optimization of the key
parameters could have a huge impact on the deployment of smart networks.
Citation: Griva, A.I.; Boursianis,
A.D.; Wan, S.; Sarigiannidis, P.; Keywords: Internet of Things (IoT); long-range network; smart agriculture; smart city; low-power
Psannis, K.E.; Karagiannidis, G.; wide-area network (LPWAN); data extraction rate (DER); network energy consumption (NEC)
Goudos, S.K. LoRa-Based IoT
Network Assessment in Rural and
Urban Scenarios. Sensors 2023, 23,
1695. https://doi.org/10.3390/
1. Introduction
s23031695
In recent years, Internet of Things (IoT) technologies and techniques have been devel-
Academic Editor: Carles Gomez
oped to cope with modern requirements. Smart cities [1], smart homes and buildings [2],
Received: 23 December 2022 healthcare [3], manufacturing [4], and smart agriculture [5] are some of the most notable
Revised: 19 January 2023 areas where IoT technologies are being adopted to address many challenges and improve
Accepted: 1 February 2023 the way we live. Everyday life devices are being equipped with sensors that communi-
Published: 3 February 2023 cate through the Internet. Based on Cisco, it is predicted that there will be 500 billion
Internet-connected devices by 2030 [6].
The key requirement for data access between devices is to cover large distances and
consume less power. The low-power wide-area network (LPWAN) architecture is one
Copyright: © 2023 by the authors. of the most prominent in that field, and long-range (LoRa) technology supported by the
Licensee MDPI, Basel, Switzerland.
low-power wide-area networking (LoRaWAN) specification has already been adopted in
This article is an open access article
numerous systems in wireless communication. A LoRa network consists of two distinct
distributed under the terms and
parts: LoRa and LoRaWAN. Each component corresponds to a different layer of the
conditions of the Creative Commons
protocol stack.
Attribution (CC BY) license (https://
LoRa’s design favors long-range applications that require a low rate of transmission
creativecommons.org/licenses/by/
and low energy consumption [7]. LoRa operates in the industrial, scientific, and medical
4.0/).

Sensors 2023, 23, 1695. https://doi.org/10.3390/s23031695 https://www.mdpi.com/journal/sensors

Sensors 2023, 23, 1695 2 of 14

(ISM) bands (from 863 to 870 MHz in Europe, from 902 to 928 MHz in the USA, and from
470 to 510 MHz in Asia).
LoRaWAN is a communication protocol [8] that uses an ALOHA-based MAC protocol.
That way, the LoRa end-devices comply with the practical requirement for low complexity.
LoRa end-node devices and gateways communicate over the physical layer, but there is no
association between nodes and a specific gateway. Data from end nodes can be received
by any gateway within a definite communication range, and the messages are forwarded
toward the network server using the Internet protocol. Finally, the messages are delivered
to the application server.
In our preliminary work [9], we simulated an open-field cultivation scenario using
FLoRa and evaluated the performance of the network. The motivation for our work
stems from the above-mentioned discussion and from the fact that we want to thoroughly
investigate the most important parameters that affect the operation of LoRa networks in
various simulations using the common path loss models in different environments. As such,
we can evaluate the performance of LoRa networks under different conditions and draw a
generalized conclusion about the efficiency of the network in urban and rural environments.
In this work, we focus on the assessment of performance metrics in LoRa networks
under different propagation and environmental scenarios. We selected the two most
common path loss models that are used to simulate various LoRa networks in the rural
environment, and we explored how the selection of the technical parameters affects the
performance of each scenario network. Moreover, we decided to extend the research to an
urban area by modeling a wide variety of nodes. We also used the Oulu path loss model to
compare different propagation scenarios. Finally, we modeled a dense network of nodes
with the same technical characteristics to simulate a smart parking area that can be found in
every modern city, and we compared the performance of this network to the previous ones.
The remainder of the paper is structured as follows. Section 2 presents the work
directly related to this paper. The problem definition and the simulation scheme are
provided in Section 3. In Section 4, the authors analyzed the simulation scenarios and
evaluated the results. Finally, the conclusions of this study are summarized in Section 5.

2. Related Work
According to the relevant literature, there have been various studies on LoRa networks.
Some of these have focused on LoRa network scalability. In [10], the authors released the
LoRaSim simulator to study the scalability and performance of LoRa networks through
simulation. They developed models that describe LoRa communication behavior, and
they determined that LoRa networks can scale well by adding more gateways and/or by
selecting dynamic transmission parameters. This was further extended in [11], in which
the authors presented LoRaWANSim, a tool that employs bidirectional communication.
Georgiou and Raza in [12] proposed a stochastic geometry framework to evaluate the
behavior of a single gateway LoRa network. They found that, with an increasing number
of end devices, the coverage probability decreased exponentially due to interfering signals
that used the same spreading factor. In a similar study, the authors in [13] highlighted that
the network scalability is more precise under the combined impact of co-SF and inter-SF
interference. In [14], in the case of a single-gateway LoRaWAN deployment, the scalability
was examined in terms of the number of nodes per gateway. The authors developed a
simulation model that measures the impact of interference to determine the scalability of
a single gateway. Furthermore, the authors in [15] used the ns-3 module to analyze the
scalability. The results show that the allocation of the network parameters to nodes has a
huge impact on the performance of LoRaWAN networks. Moreover, this work examines the
capacity for various traffic types. In [16], the scalability of the network increases by using a
new medium access protocol in a MATLAB simulator. In [17], a literature overview was
presented, and various performance determinants were analyzed on LoRa-based networks.
Various experimental tests have been performed in real-world environments. LoRa net-
works have been studied in cases ranging from indoor [18,19] and urban/ suburban [20,21]
Sensors 2023, 23, 1695 3 of 14

scenarios to rural [22] and mountain [23,24] environments. In [25], the researchers provided
a comprehensive evaluation of LoRa networks in urban, suburban, and rural environments,
considering both static and dynamic conditions. The authors in [26] evaluated the coverage
and simulated the path loss model in urban, forest, and coastal environments. In [27], a
path loss model based on experimental scenarios was proposed to assess the efficiency of
LoRa networks in urban and rural environments in terms of coverage. The work presented
in [28] focused on the evaluation of the transmission performance and the link quality of a
network considering the deployment scenario and the parameter configuration. In [29], a
smart building scenario was implemented to appraise the communication performance of
LoRa networks without considering the power consumption. The authors in [30] presented
a theoretical study and an experimental evaluation of a LoRa network. Moreover, the
impact of the coding rate of the communication link is discussed in this work. In [31], the
authors developed the FLoRa simulation tool to implement and evaluate the adaptive data
rate (ADR) mechanism in LoRa networks. In [32], the authors evaluated the impact of
SF on performance. Another method to improve the performance based on SF network
clustering was presented in [33], and the authors in [34] developed an algorithm to further
improve the efficiency of the LoRa network compared to the ADR algorithm.

3. Problem Definition
3.1. Evaluation Metrics
Based on the related work that was presented, we decided to assess the effectiveness
of LoRa networks using two evaluation metrics:
• Data extraction rate (DER) is defined as the amount of messages which were received
correctly divided by the number of messages that were sent to the server. DER is
computed between 0 and 1. When the ratio is closer to 1, it means that the LoRa
network is working more efficiently.
• Network energy consumption (NEC) is defined as the energy consumed by the net-
work divided by the number of successfully received messages. A low value for NEC
implies a more efficient network.

3.2. Path Loss Models

To model radio wave propagation, most researchers are using the log-normal shad-
owing model, which depends on empirical data [10]. The Oulu city model [18] and the
Okumura–Hata (OH) model have been used to assess the coverage in LoRa networks
[25,35]. In our research, we chose to look more closely at these three different path-loss
models. A short summary of each model is given.

3.2.1. Log-Distance Path Loss Model with Shadowing

The log-distance path loss model with shadowing is widely used in wireless commu-
nication to model the obstruction on the propagation path between a base station (BS) and
a mobile station (MS). The equation is given as follows:

¯ (d0 ) + 10 × n × log d + Xσ
PL(d) = PL (1)
d0
where the PL(d) stands for the path loss, PL(d0 ) is the mean path loss measured in dB, n
is used for the path loss exponent, Xs is the loss due to shadow fading with a zero-mean
Gaussian distribution, and σ denotes the standard deviation. In this work, n was set to 2
and sigma was set to 5 dB in order to simulate the rural environment, whilst n was selected
equal to 2.08 and σ was set to 3.57 dB to simulate the urban environment.
Sensors 2023, 23, 1695 4 of 14

3.2.2. The Oulu Path Loss Equation

Based on experimental data, the authors in [36] provided the Oulu path loss model
using the following equation:

EPL = B + α × logR (2)

where EPL is the expected path loss, B is the path loss in dB, α describes the path loss
exponent, and R is the distance between the node and the base station divided by the 1 km
reference distance.
The standard deviation of shadow fading describes a deviation between the measured
path loss and expected path loss and is computed as follows

σSF = std( PL − EPL) (3)

The city of Oulu is a medium-sized city with high residential buildings in the center,
located one the seashore and with a mainly flat terrain. The approach provided by this
area can be used to model many similar urban environments all over the world. The
measurements were conducted using a mobile node on the roof of a car moving over the
ground and using a node on a boat over the water. The base station was on the roof of the
University of Oulu, 24 meters above sea level.

3.2.3. Okumura–Hata Path Loss Model

The Okumura–Hata empirical model is a widely used radio propagation model for
predicting path loss. The model is appropriate for linking distances from 1 to 20 km and a
frequency range of 150–1500 MHz. The user antenna heights range from 1 to 10 m and the
BS antenna height ranges from 30 to 200 m. The Okumura–Hata model is popular for its
accuracy and simplicity. This model has the advantage of being adjusted in many different
environments from open areas to large cities by selecting the appropriate variant, as is
apparent from the following equations.
The Okumura–Hata path loss equations are modified as follows:

PL = A + B × logR + C (4)

With regard to an urban environment, the two following Okumura–Hata variants

can exist

Lurban = 69.5 + 26.16 × log f c − 13.82 × loghb − a(hm )

(5)
+ (44.9 − 6.55 × loghb ) × logR

For small or medium-sized cities,

a(hm ) = (1.11 × log f c − 0.7) × hm − (1.56 × log f c − 0.8) (6)

and for large cities,

(
(8.29 × (log(1.54 × hm ))2 − 1.1, f c ≤ 300MHz
a( hm ) = (7)
(3.2 × (log(11.75 × hm ))2 − 4.97, f c > 300MHz

Under the sub-urban environment,

fC 2
Lsuburban = Lurban − 2 × (log( )) − 5.4 (8)
28
Under the free/open/rural environment,

Lopen = Lurban − 4.78 × (log f c )2 − 18.33 × log f c + 40.94 (9)

Sensors 2023, 23, 1695 5 of 14

where Lurban is the path loss in urban areas, Lsuburban is the path loss in suburban areas, and
Lopen is the path loss in open-rural areas in dB. hb is the height of the base station antenna,
and hm is the height of the mobile station antenna in m. Moreover, f c is the frequency of
the transmission in MHz and R is the distance between the station and the mobile stations
in km. Finally, the a(hm ) is the correction factor for mobile antenna height.

3.3. Simulation Tools

As LoRa technology is widely used in IoT applications, various simulation envi-
ronments were developed to study, evaluate, and optimize LoRa networks before their
implementation. A short overview of the most commonly used simulators that have been
developed over the years is presented in this section.
• LoRaSim is a discrete event simulator developed to study the scalability in LoRa
networks and to model collisions. This tool is written in Python and allows modeling
in a 2-dimensional grid. By selecting a number of parameters such as the number
of nodes and base stations, the radio settings, and the simulation time, we can have
information regarding the collisions, transmissions, and total energy spent [10].
• The Framework for LoRa (FLoRa) is an open source simulation tool based on the OM-
NeT++ (ver. 5.3, OpenSim Ltd., Budapest, Hungary) [37] discrete event simulator and
the INET framework [38]. FLoRa provides a precise model of the physical layer taking
into consideration the capture effect as well as the collisions and also implements
the MAC layer. Moreover, FLoRa includes the various modules of a network such as
nodes, gateway(s), and server(s), supports bi-directional communication and enables
end-to-end simulations. One or more gateways can be simulated in the model and can
receive transmissions from nodes on multiple channels. The gateways communicate
with the server over the IP protocol. The tool is written in C++ and NED language
and is used to simulate LoRa networks. Moreover, it provides the statistics for energy
consumption in every node considering the three states (transmitting, receiving, and
sleeping) of a LoRa radio [31].
• The NS-3 module is a discrete-event network simulator written in C++ and Python.
This module is used to assess the performance of a network in various cases because
it can simulate many aspects of a network such as the layers of the system and the
protocols that are used [15].
• LoRaWANSim is a LoRaWAN simulator that is written in MATLAB. This simulator
was developed to characterize the behavior of LoRaWAN networks. In LoRaWANSim,
we can model and also modify parameters related to the PHY layer and the LoRaWAN
protocol [11].
Considering the requirements of our study, the FLoRa open source simulation tool
was selected. As shown in Figure 1, the tool was used to model the LoRa physical layer and
the LoRaWAN MAC protocol, as well as the network elements such as network servers,
nodes, and gateways. The module describing the energy consumption was also used.

Figure 1. Modules in FLoRa and the corresponding protocol stack [31].

4. Numerical Results
In this part, we analyzed the various scenarios we implemented to assess the perfor-
mance of LoRa Networks. In every simulation, we assumed that the nodes were uniformly
distributed on a square deployment area, whereas the gateways were arbitrarily placed.
Sensors 2023, 23, 1695 6 of 14

Each scenario lasted 7 days, and the physical layer of the LoRa network was simulated in
each simulation environment by choosing the right European regional parameters.
Five configuration parameters were selected to determine the energy consumption,
the transmission range, the data rate, and the noise durability of a LoRa transmission [10].
• Transmission power (TP). TP can be set between -4 and 20 dBm. When the TP increases,
the energy consumption of the network and the signal-to-noise ratio (SNR) are also
increased.
• Bandwidth (BW). BW can be set to 125 kHz, 250 kHz, or 500 kHz. A bigger BW
provides a higher data rate but decreases the radio sensitivity.
• Spreading factor (SF). SF can be in the range of 7–12. A higher SF improves the
communication range but increases the energy consumption.
• Carrier frequency (CF). CF can be selected between 137 MHz and 1020 MHz according
to the ISM band in the region of operation.
• Coding rate (CR). CR can be configured to 4/5, 4/6, 4/7, or 4/8 to provide security
from interference. By choosing a higher CR, the network becomes more reliable, but
the air time increases.

4.1. Case Study 1—Rural Environment

The rural environment was simulated by using the log-distance path loss model and
the Okumura–Hata path loss model. In each scenario, the transmission power was set to
10 dBm, the bandwidth was fixed at 125 kHz, the code rate was selected as 4/8, the chosen
spreading factor was 7, and the carrier frequency was selected as 868 MHz (Table 1).

Table 1. Simulation parameters in the rural environment.

Parameter Value
Transmission power (TP) 10 dBm
Carrier frequency (CF) 868 MHz
Spreading factor (SF) 7
Code rate (CR) 4/8
Bandwidth (BW) 125 kHz

4.1.1. Log-Distance Path Loss Model with Shadowing

Scenario 1.1 In the first scenario, we assess the effectiveness of a LoRa network in
three squared areas. The first deployment area was set to 300 m × 300 m, the second
deployment area was set to 500 m × 500 m, and the third area was set to 1000 m × 1000 m.
The number of nodes was increased from 10 to 1000 in each area. Figure 2a,b show that,
with an increasing number of nodes, there is a decrease in the network performance, due to
the more frequent incidents of collisions between the packets that were sent from nodes
with identical characteristics. As we can obtain by increasing the nodes count from 10 to
1000 in the deployment area B, the collisions are increased by 64%. Moreover, the results
of the simulations indicate that the performance metrics (DER, NEC) we obtained are
affected by the dimensions of the deployment area. This occurs because, as the dimension
of the deployment space is increased, the exact number of nodes must be spread across a
larger space.
Scenario 1.2 In the second scenario, the impact of the effective isotropic radiated power
was explored. We modeled 100 uniformly distributed nodes in a 500 m × 500 m deployment
area, and a single gateway was located at the point (x = 0, y = 0). We simulated various
antennas, such as the isotropic antenna, the 0.173 m dipole antenna, and isotropic antennas
with constant gain from 1 dBi to 6 dBi. Figure 2c,d show that the obtained performance
metrics of the network are improved as the EIRP increases. Moreover, we can detect an
increase of 4.41% by replacing the antenna from dipole (EIRP = 12.15 dBm) to isotropic
antenna with a constant of gain 6 dBi (EIRP = 16 dBm).
Sensors 2023, 23, 1695 7 of 14

(a) (b)

(c) (d)

(e) (f)

Figure 2. Log-normal shadowing model in rural environment. (a) DER as a function of the number
of nodes; (b) NEC as a function of the number of nodes; (c) DER as a function of the EIRP; (d) NEC as
a function of the EIRP; (e) DER as a function of the number of gateways; and (f) NEC as a function of
the number of gateways.

Scenario 1.3 In this series of simulations, we appraise the performance of 100 LoRa
nodes deployed in the same 500 m × 500 m area using multiple gateways. Firstly, a
single GW0 was located at the point (x = 0, y = 0). Then, a second GW1 was placed at
(x = xmax /2, y = ymax /2). In the next set of simulations, we used the first GW0 , a GW2
located at (x = xmax /2, 0), and a GW3 located at (0, y = ymax /2). For the last set, we used
all of the previous gateways in the deployment area. This gateway layout was selected
for simplicity. The performance of the network is increased by using more gateways, as
shown in Figure 2e,f. When the number of gateways is increased from 1 to 2, we obtained
an increase of 12.75% in the network performance. This increase is affected by the location
of the second gateway. We selected to place it at the point (x = xmax /2, y = ymax /2) to
reduce the distance between the gateways and the nodes. The simulations show that, if
we place the second gateway at the other side of the deployment area, for example, at the
point (x = xmax , y = ymax ), the increase in the performance is changed to 11.2%.

4.1.2. Okumura–Hata Path Loss Model

Scenario 1.4 In the next set of simulations, six different deployment areas were selected.
The smallest deployment area was set to 5 km × 5 km, the largest deployment area
was set to 15 km ×15 km, and the rest of the deployment areas were selected between
those boundaries. A gateway was located at (x = 0, y = 0) and 100 nodes were placed
at each deployment area. Figure 3a,b show that the efficiency of the network decreases
by expanding the deployment area’s size. Regarding the percentage of the packets that
arrive at the gateway having a power level below the minimum sensitivity level, we
can divide the area into four zones, as shown in Figure 4, namely 0–7 km, 7–11 km,
Sensors 2023, 23, 1695 8 of 14

11–14 km, and more than 14 km. Within a 7 km range from the gateway, every node can
successfully communicate with the gateway. In the 7–11 km range, 2.89% of the nodes
cannot reach the gateway. In the third zone, 11–14 km, the figure rises from 2.89% to 24%.
Finally, above 14 km, the number of nodes that cannot communicate with the gateway has
significantly increased.

(a) (b)

(c) (d)

(e) (f)

Figure 3. Okumura–Hata model in the rural environment. (a) DER as a function of the size of the
deployment area; (b) NEC as a function of the size of the deployment area; (c) DER relative to the
transmission power; (d) NEC relative to the transmission power; (e) DER relative to the spreading
factor; and (f) NEC relative to the spreading factor.

Figure 4. The area divided into zones.

Sensors 2023, 23, 1695 9 of 14

Scenario 1.5 In this scenario, we analyze the influence of the TP on the main perfor-
mance metrics of a LoRa network. We placed a gateway at (x = 0, y = 0) and 100 nodes
uniformly distributed at the deployment space that was selected to be 8 km × 8 km.
Figure 3c,d show that with a growing value from -4 dBm to 10 dBm, the performance
of the network increases. For example, by increasing the TP from 2 dBm to 6 dBm, the
performance is increased by 45.3%. A threshold is obtained at 11 dBm and that means that
we can achieve the maximum performance of the network in the specific deployment area
that we have simulated, by choosing between a big range of transmission power values.
Scenario 1.6 In this scenario, we selected the same layout as in Scenario 1.6. The TP
was set to 10 dBm and we obtained the performance of the network by changing the value
of SF. Figure 3e,f show that, while we increase the value of the SF, the time on air is also
increased. Consequently, we observe an increase in the energy consumption of the network.
For example, by changing the SF from SF11 to SF12, the energy consumption increases
by 50.76%.

4.2. Case Study 2—Urban environment

The urban environment was simulated using the log-distance path loss model, the
Okumura–Hata model, and the Oulu city model. To simulate the urban environment, the
transmission power was chosen between 2 and 14 dBm, the code rate was set to 4/8, the
bandwidth was set to 125 kHz, the carrier frequency was selected to 868 MHz, and the
spreading factor was in the range of 7–12 (Table 2).

Table 2. Simulation parameters in urban environment.

Parameter Value
Transmission power (TP) 2–14 dBm
Carrier frequency (CF) 868 MHz
Spreading factor (SF) 7–12
Code rate (CR) 4/8
Bandwidth (BW) 125 kHz

4.2.1. Log-Distance Path Loss Model with Shadowing

Scenario 2.1 In this set of simulations, three different deployment areas were selected.
Deployment area A was set to 30 m × 30 m, B was set to 50 m × 50 m, and C was set
to 100 m × 100 m. The performance of a LoRa network was evaluated for an increasing
number of nodes. Figure 5a,b show that, while the number of nodes increases from 100 to
1000, the efficiency of the network is decreasing and comes as a result of the increase in
the number of collisions between the packets. The number of this decrease compared to
the rural environment is lower due to the range of the TP and SF values which were used
to simulate the nodes in an urban environment. In the urban environment, the number of
undetectable packages depends on the combination of the location, the TP, and the SF of
the nodes. For example, by simulating 100 nodes in deployment area C, the percentage of
the packages that reach the gateway with a power level lower than −100 dBm was almost
39%. With an increasing number of nodes, the percentage of undetectable packages was
slightly reduced. As reflected by the previous simulations, the size of the deployment space
affects the achieved performance.
Sensors 2023, 23, 1695 10 of 14

(a) (b)

Figure 5. Log-normal shadowing model in the urban environment: (a) DER as a function of the
number of nodes; and (b) NEC as a function of the number of nodes.

4.2.2. Okumura–Hata Path Loss Model

Scenario 2.2 Six deployment areas were modeled to evaluate the performance of the
network in this set of simulations. The smallest deployment area was set to 500 m × 500 m,
and the largest deployment area was set to 3 km × 3 km. A gateway was modeled at a
height of 30 m from the ground at (x = 0, y = 0), and 600 nodes were located uniformly
at each deployment area at a height of 1 m. Figure 6a,b show that the performance of the
network slowly decreases by increasing the deployment area’s dimensions.
Scenario 2.3 In this scenario, we changed the height of the gateway in deployment
area B. When we raise the height of the gateway, the performance of the network is
increased, as shown in Figure 6c,d. For example, by changing the height from 30 m to
35 m, the performance is increased by 5% which is very important because we can deploy
the same network in a wider area (1.5 km × 1.5 km) without changing any configuration
parameters.

(a) (b)

(c) (d)

Figure 6. Okumura–Hata path loss model in an urban environment: (a) DER as a function of the size
of the deployment area; (b) NEC as a function of the size of the deployment area; (c) DER relative to
the height of gateway; and (d) NEC relative to the height of gateway.

4.2.3. Oulu Path Loss Model

Scenario 2.4 At this setup, we used the Oulu city path loss model to evaluate the
performance of a LoRa network (Table 3). The deployment area was set to be 2 km × 2 km
and a gateway was located at the point (x = 0, y = 0) at 24 m from sea level. Figure 7a,b show
the results for a growing number of nodes by simulating the different propagation models.
Sensors 2023, 23, 1695 11 of 14

(a) (b)

Figure 7. Oulu city path loss model in an urban environment: (a) DER as a function of the number of
nodes; and (b) NEC as a function of the number of nodes.

Table 3. Simulation parameters in the city of Oulu [36].

Metric Car Boat Free Space

Path loss exponent (n) 2.32 1.76 2.00
Path loss intercept (B) 128.95 126.43 91.22
Shadow fading (σs f ) 7.8 dB 8.0 dB -

The performance of the network decreases when we increase the number of nodes due
to the frequent incidents of collisions between the packets. As expected, the best results are
obtained using the free space path loss model. On the other hand, when the car model was
used, we detected the lowest value in the performance of the network. The car model was
selected to simulate an area where buildings and other physical obstacles are blocking the
path between the nodes and the gateway.

4.3. Case Study 3—Parking Model Environment

In the last series of simulations, we attempted to evaluate the performance of a parking
model environment. The deployment area was set to 100 m × 100 m in every simulation
set. Firstly, to model 600 nodes in the urban environment, we used the European regional
parameters. Then, for the same number of nodes, we selected the TP to be fixed at 10 dBm
and the SF at 7 to simulate a parking area with a large number of identical nodes. Finally,
50 nodes were simulated using the appropriate parameters to model the LoRa physical
layer in the rural environment.
Figure 8a,b show the results for the three different cases. Comparing the urban model
to the parking area model, we can see that the data extraction rate of the network is
decreased by 38.79% when we select identical nodes. This decrease is caused due to a
large number of collisions between the packets. Regarding the energy consumption of the
network, we observe that it slightly decreased by 2% in the parking area model. Finally, we
can obtain that the performance in the rural environment is significantly high and the data
extraction rate is very close to the highest value.

(a) (b)

Figure 8. Parking model: (a) DER relative to the type of area; and (b) NEC relative to the type of area.
Sensors 2023, 23, 1695 12 of 14

5. Conclusions
In this work, an assessment of LoRa networks was presented in three different use
cases: the rural environment, the urban environment, and the parking model environment.
Firstly, we explored the way in which the network density impacts performance in LoRa
networks in both rural and urban environments. The results show that, while the amount
of nodes increases, the network’s performance decreases. This is more obvious in the
implementation of rural environments because we modeled nodes with identical technical
characteristics. Moreover, we explored the impact of EIRP in our model and highlighted
the importance of the number and the location of the gateways in the performance of LoRa
networks by selecting the log-distance path loss model with shadowing. The impact of the
transmission power and the spreading factor was explored using the Okumura–Hata path
loss model. By increasing the transmission power, the number of the delivered packages
was also increased until the stimulation upper threshold of 11 dBm was reached. While
we increased the spreading factor from SF 11 to SF 12, the energy consumption of the
network was increased by 50.76%. Furthermore, the results show that the selection of the
dimensions of the deployment area and the height of the gateway are very important to
implement more efficient networks. By choosing the Oulu path loss model, we managed
to record the impact of the propagation model on the network performance. Finally,
we simulated a parking model area to underline the network behavior by changing the
environmental parameters of the modeled area. As future work, the current project can be
extended in several directions, such as performing experimental evaluations in a real-world
environment to assess the simulation results. To develop a full picture of the behavior of a
LoRa network in specific environments, for example, in a smart agriculture scenario or/and
in an urban area in Greece, we are planning to simulate and experimentally implement
LoRa networks in order to define the numeric parameters of the path loss model taking into
consideration the different type of crops in the agriculture scenario and the specifications of
the urban area, respectively. Finally, further research should be undertaken to investigate
the improvement of the performance of LoRa networks using optimization algorithms.

Author Contributions: Conceptualization, A.I.G., A.D.B. and S.K.G.; methodology, A.I.G., A.D.B.
and S.K.G.; software, A.I.G. and A.D.B.; validation, A.I.G., A.D.B. and S.K.G.; formal analysis, A.I.G.,
A.D.B. and S.K.G.; investigation, S.K.G.; resources, P.S.; data curation, S.K.G.; writing—original draft
preparation, A.I.G.; writing—review and editing, A.D.B. and S.K.G.; visualization, A.I.G., A.D.B. and
S.K.G.; supervision, S.K.G., S.W., P.S., K.E.P. and G.K.; project administration, S.K.G. All authors have
read and agreed to the published version of the manuscript.
Funding: This work has received funding from the European Union’s Horizon 2020 research and
innovation program under grant agreement No. 957406 (TERMINET).
Institutional Review Board Statement: Not applicable.
Informed Consent Statement: Not applicable.
Data Availability Statement: Data sharing not applicable.
Conflicts of Interest: The authors declare no conflict of interest.

References
1. Shehab, M.J.; Kassem, I.; Kutty, A.A.; Kucukvar, M.; Onat, N.; Khattab, T. 5G Networks Towards Smart and Sustainable Cities: A
Review of Recent Developments, Applications and Future Perspectives. IEEE Access 2022, 10, 2987–3006. [CrossRef]
2. Minoli, D.; Sohraby, K.; Occhiogrosso, B. IoT Considerations, Requirements, and Architectures for Smart Buildings—Energy
Optimization and Next-Generation Building Management Systems. IEEE Internet Things J. 2017, 4, 269–283. . JIOT.2017.2647881.
[CrossRef]
3. Baker, S.B.; Xiang, W.; Atkinson, I. Internet of Things for Smart Healthcare: Technologies, Challenges, and Opportunities. IEEE
Access 2017, 5, 26521–26544. [CrossRef]
4. Gallardo, J.L.; Ahmed, M.A.; Jara, N. LoRa IoT-Based Architecture for Advanced Metering Infrastructure in Residential Smart
Grid. IEEE Access 2021, 9, 124295–124312. [CrossRef]
Sensors 2023, 23, 1695 13 of 14

5. Boursianis, A.D.; Papadopoulou, M.S.; Diamantoulakis, P.; Liopa-Tsakalidi, A.; Barouchas, P.; Salahas, G.; Karagiannidis, G.; Wan,
S.; Goudos, S.K. Internet of Things (IoT) and Agricultural Unmanned Aerial Vehicles (UAVs) in smart farming: A comprehensive
review. Internet Things 2022, 18, 100187. [CrossRef]
6. Zikria, Y.B.; Ali, R.; Afzal, M.K.; Kim, S.W. Next-Generation Internet of Things (IoT): Opportunities, Challenges, and Solutions.
Sensors 2021, 21, 1174. [CrossRef]
7. LoRa Alliance, March 2018. Available online: http://www.lora-alliance.org (accessed on 21 December 2022).
8. LoRa Alliance. A Technical Overview of LoRa and LoRaWAN, Nov. 2015. Available online: https://lora-alliance.org/wp-
content/uploads/2020/11/what-is-lorawan.pdf (accessed on 21 December 2022).
9. Griva, A.; Boursianis, A.D.; Wan, S.; Sarigiannidis, P.; Karagiannidis, G.; Goudos, S.K. Performance Evaluation of LoRa Networks
in an Open Field Cultivation Scenario. In Proceedings of the 10th International Conference on Modern Circuits and Systems
Technologies (MOCAST), Thessaloniki, Greece, 5–7 July 2021; pp. 1–5. [CrossRef]
10. Bor, M.C.; Roedig, U.; Voigt, T.; Alonso, J.M. Do LoRa Low-Power Wide-Area Networks Scale? Association for Computing Machinery:
New York, NY, USA, 2016. [CrossRef]
11. Marini, R.; Mikhaylov, K.; Pasolini, G.; Buratti, C. LoRaWANSim: A Flexible Simulator for LoRaWAN Networks. Sensors 2021,
21, 695. [CrossRef]
12. Georgiou, O.; Raza, U. Low Power Wide Area Network Analysis: Can LoRa Scale? IEEE Wirel. Commun. Lett. 2017, 6, 162–165.
[CrossRef]
13. Mahmood, A.; Sisinni, E.; Guntupalli, L.; Rondón, R.; Hassan, S.A.; Gidlund, M. Scalability Analysis of a LoRa Network Under
Imperfect Orthogonality. IEEE Trans. Ind. Inform. 2019, 15, 1425–1436. [CrossRef]
14. Haxhibeqiri, J.; Van den Abeele, F.; Moerman, I.; Hoebeke, J. LoRa Scalability: A Simulation Model Based on Interference
Measurements. Sensors 2017, 17, 1193. [CrossRef]
15. Van den Abeele, F.; Haxhibeqiri, J.; Moerman, I.; Hoebeke, J. Scalability Analysis of Large-Scale LoRaWAN Networks in ns-3.
IEEE Internet Things J. 2017, 4, 2186–2198. [CrossRef]
16. Fernandes, R.; Luís, M.; Sargento, S. Large-Scale LoRa Networks: A Mode Adaptive Protocol. IEEE Internet Things J. 2021,
8, 13487–13502. [CrossRef]
17. Gkotsiopoulos, P.; Zorbas, D.; Douligeris, C. Performance Determinants in LoRa Networks: A Literature Review. IEEE Commun.
Surv. Tutorials 2021, 23, 1721–1758. [CrossRef]
18. Petäjäjärvi, J.; Mikhaylov, K.; Hämäläinen, M.; Iinatti, J. Evaluation of LoRa LPWAN technology for remote health and wellbeing
monitoring. In Proceedings of the 10th International Symposium on Medical Information and Communication Technology
(ISMICT), Worcester, MA, USA, 20–23 March 2016; pp. 1–5. [CrossRef]
19. Fraile, L.P.; Tsampas, S.; Mylonas, G.; Amaxilatis, D. A Comparative Study of LoRa and IEEE 802.15.4-Based IoT Deployments
Inside School Buildings. IEEE Access 2020, 8, 160957–160981. [CrossRef]
20. Astrain, J.J.; Falcone, F.; Lopez-Martin, A.J.; Sanchis, P.; Villadangos, J.; Matias, I.R. Monitoring of Electric Buses Within an Urban
Smart City Environment. IEEE Sensors J. 2022, 22, 11364–11372. [CrossRef]
21. Hosseinzadeh, S.; Almoathen, M.; Larijani, H.; Curtis, K. A Neural Network Propagation Model for LoRaWAN and Critical
Analysis with Real-World Measurements. Big Data Cogn. Comput. 2017, 1, 7. [CrossRef]
22. Yim, D.; Chung, J.; Cho, Y.; Song, H.; Jin, D.; Kim, S.; Ko, S.; Smith, A.; Riegsecker, A. An experimental LoRa performance
evaluation in tree farm. In Proceedings of the 2018 IEEE Sensors Applications Symposium (SAS), Seoul, Republic of Korea, 12–14
March 2018; pp. 1–6. [CrossRef]
23. Iova, O.; Murphy, A.; Picco, G.; Ghiro, L.; Molteni, D.; Ossi, F.; Cagnacci, F. LoRa from the City to the Mountains: Exploration
of Hardware and Environmental Factors. In Proceedings of the International Conference on Embedded Wireless Systems and
Networks (EWSN) 2017, Uppsala, Sweden, 20–22 February 2017.
24. Bianco, G.M.; Giuliano, R.; Marrocco, G.; Mazzenga, F.; Mejia-Aguilar, A. LoRa System for Search and Rescue: Path-Loss Models
and Procedures in Mountain Scenarios. IEEE Internet Things J. 2021, 8, 1985–1999. [CrossRef]
25. Sanchez-Iborra, R.; Sanchez-Gomez, J.; Ballesta-Viñas, J.; Cano, M.D.; Skarmeta, A.F. Performance Evaluation of LoRa Considering
Scenario Conditions. Sensors 2018, 18, 772. [CrossRef]
26. Callebaut, G.; Van der Perre, L. Characterization of LoRa Point-to-Point Path Loss: Measurement Campaigns and Modeling
Considering Censored Data. IEEE Internet Things J. 2020, 7, 1910–1918. [CrossRef]
27. El Chall, R.; Lahoud, S.; El Helou, M. LoRaWAN Network: Radio Propagation Models and Performance Evaluation in Various
Environments in Lebanon. IEEE Internet Things J. 2019, 6, 2366–2378. [CrossRef]
28. Kulkarni, P.; Hakim, Q.O.A.; Lakas, A. Experimental Evaluation of a Campus-Deployed IoT Network Using LoRa. IEEE Sensors J.
2020, 20, 2803–2811. [CrossRef]
29. Liang, R.; Zhao, L.; Wang, P. Performance Evaluations of LoRa Wireless Communication in Building Environments. Sensors 2020,
20, 3828. [CrossRef] [PubMed]
30. Faber, M.J.; van der Zwaag, K.M.; dos Santos, W.G.V.; Rocha, H.R.d.O.; Segatto, M.E.V.; Silva, J.A.L. A Theoretical and
Experimental Evaluation on the Performance of LoRa Technology. IEEE Sensors J. 2020, 20, 9480–9489. [CrossRef]
31. Slabicki, M.; Premsankar, G.; Di Francesco, M. Adaptive configuration of lora networks for dense IoT deployments. In
Proceedings of the NOMS 2018—2018 IEEE/IFIP Network Operations and Management Symposium, Taipei, Taiwan, 23–27 April
2018; pp. 1–9. [CrossRef]
Sensors 2023, 23, 1695 14 of 14

32. Torres, A.P.A.; Silva, C.B.D.; Filho, H.T. An Experimental Study on the Use of LoRa Technology in Vehicle Communication. IEEE
Access 2021, 9, 26633–26640. [CrossRef]
33. Zhu, G.; Liao, C.H.; Sakdejayont, T.; Lai, I.W.; Narusue, Y.; Morikawa, H. Improving the Capacity of a Mesh LoRa Network by
Spreading-Factor-Based Network Clustering. IEEE Access 2019, 7, 21584–21596. [CrossRef]
34. Al-Gumaei, Y.A.; Aslam, N.; Chen, X.; Raza, M.; Ansari, R.I. Optimizing Power Allocation in LoRaWAN IoT Applications. IEEE
Internet Things J. 2022, 9, 3429–3442. [CrossRef]
35. Paredes, M.; Bertoldo, S.; Carosso, L.; Lucianaz, C.; Marchetta, E.; Allegretti, M.; Savi, P. Propagation measurements for a LoRa
network in an urban environment. J. Electromagn. Waves Appl. 2019, 33, 2022–2036. [CrossRef]
36. Petajajarvi, J.; Mikhaylov, K.; Roivainen, A.; Hanninen, T.; Pettissalo, M. On the coverage of LPWANs: Range evaluation and
channel attenuation model for LoRa technology. In Proceedings of the 14th International Conference on ITS Telecommunications
(ITST), Copenhagen, Denmark, 2–4 December 2015; pp. 55–59. [CrossRef]
37. OMNeT++ Discrete Event Simulator, Feb. 2020. Available online: https://omnetpp.org/ (accessed on 21 December 2022).
38. INET Framework—What Is INET Framework? 2020. Available online: https://inet.omnetpp.org/Introduction.html (accessed on
21 December 2022).

Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual
author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to
people or property resulting from any ideas, methods, instructions or products referred to in the content.