Download as pdf or txt
Download as pdf or txt
You are on page 1of 8

LoRaWAN-based Smart Parking Service: Deployment and

Performance Evaluation
Juan Ramón Santana Pablo Sotres Jesús Pérez
Network Planning and Mobile Network Planning and Mobile Network Planning and Mobile
Communications Laboratory Communications Laboratory Communications Laboratory
Universidad de Cantabria Universidad de Cantabria Universidad de Cantabria
Santander Spain Santander Spain Santander Spain
jrsantana@tlmat.unican.es psotres@tlmat.unican.es jesus@tlmat.unican.es

Luis Sánchez Jorge Lanza Luis Muñoz


Network Planning and Mobile Network Planning and Mobile Network Planning and Mobile
Communications Laboratory Communications Laboratory Communications Laboratory
Universidad de Cantabria Universidad de Cantabria Universidad de Cantabria
Santander Spain Santander Spain Santander Spain
lsanchez@tlmat.unican.es jlanza@tlmat.unican.es luis@tlmat.unican.es

ABSTRACT Through the continuous monitoring of two parking sensor


deployments in the city, we derive some conclusions and lessons
The dramatic increase of the urban population has put a lot of learnt that could benefit the planning and deployment of city-
pressure in modern urban transportation systems. This not only scale IoT infrastructures employing LoRaWAN networks as their
implies noteworthy air pollution, and waste in time and energy, wireless access technology.
but also has led to the critical issue of the parking spots scarcity.
Finding a parking spot has become one of the most common
CCS CONCEPTS
problems mentioned by drivers due to the day-by-day increase in
• Networks → Network performance evaluation →
number of vehicles. To overcome this problem, smart parking
services based on sensors to detect parking spot occupancy status Network performance analysis.
and inform drivers about their availability have been proposed.
With the successful deployment of this kind of solutions, costs KEYWORDS
associated with traffic jams or wasted gas can be considerably IoT, LPWAN, LoRa, LoRaWAN, smart cities, smart parking
reduced. However, one of the main challenges to be addressed
when planning and deploying this kind of services is the ACM Reference Format:
communication technology employed by parking sensors, which Juan Ramón Santana, Pablo Sotres, Jesús Pérez, Luis Sánchez, Jorge Lanza,
are typically buried under the asphalt, and has to be able to cover Luis Muñoz. 2022. LoRaWAN-based Smart Parking Service: Deployment
large areas of the city in the most cost-efficient way. In this paper, and Performance Evaluation. In Proceedings of the 19th ACM International
Symposium on Performance Evaluation of Wireless Ad Hoc, Sensor, &
we analyze the behavior and performance of a LoRaWAN
Ubiquitous Networks (PE-WASUN '22). ACM, Montreal, QC, Canada, 8
network employed for supporting a Smart Parking service in the pages. https://10.1145/3551663.3558685
city of Santander. The sensors and network deployment are
described in the paper and the thorough experimental assessment
and evaluation of the network behavior is presented. The goal of 1 INTRODUCTION
this evaluation is to provide a better understanding of the key
The demographic explosion that started in the 1950 has caused
factors affecting the communication of outdoor parking spots.
an uncontrolled growth of urban areas, causing pollution,
Permission to make digital or hard copies of all or part of this work for personal or overexploitation and poverty, among others. Besides, this is a
classroom use is granted without fee provided that copies are not made or distributed growing concern, as the urban population is increasing over time
for profit or commercial advantage and that copies bear this notice and the full
citation on the first page. Copyrights for components of this work owned by oth ers and, according to the United Nations (UN), up to the 68% of the
than ACM must be honored. Abstracting with credit is permitted. To copy otherwise, total world population will live in cities by 2050 [1]. As a result,
or republish, to post on servers or to redistribute to lists, requires prior specific
permission and/or a fee. Request permissions from Permissions@acm.org.
city livability and citizens’ quality of life are being deteriorated
PE-WASUN’22, October 24–28, 2022, Montreal, QC, Canada over time.
© 2022 Association for Computing Machinery. In this scenario, the Smart City concept has been proposed as
ACM ISBN 978-1-4503-9483-3/22/10…$15.00
https://doi.org/10.1145/3551663.3558685 a solution to improve the city livability employing novel
technologies as tools to overcome new challenges that have

107
PE-WASUN '22, October 24–28, 2022, Montreal, QC, Canada Juan Ramón Santana et al.

emerged from overpopulation. Among these technologies, 2 LoRa AND LoRaWAN OVERVIEW
Internet of Things (IoT) becomes of utmost importance as a mean
LoRa and LoRaWAN terms are often used interchangeably, but
to monitor the city and ease the decision-making process.
they refer to different communication technology concepts. While
Moreover, the IoT market keeps growing at a remarkable pace,
LoRa, a proprietary radio modulation technique developed by
and forecasts predict up to 27 billion of connections by 2025 [2],
Semtech [18], corresponds to the PHY layer, LoRaWAN is the
partly due to the impact of Smart City projects [3] and, in
specification proposed for the Medium Access Control (MAC)
particular, Smart Parking ones [4].
layer operations, a LPWAN protocol built upon LoRa recognized
Smart Parking has become one of the key services within the
by the International Telecommunication Union (ITU). In this
Smart City, as a mean to reduce the road occupancy and the
section we introduce both communication technologies,
pollution. Therefore, IoT-based parking services are already a
highlighting their main features.
reality, leveraging low-power communication technologies as
their deployment environment, usually in the asphalt, lacks of
2.1 LoRa Modulation
power supply and their access for battery replacement is a non-
trivial problem [5]. In this context, Low Power Wide Area LoRa enables long-rage communication using a modulation
Networks (LPWAN) technologies are well-suited for this type of based on Chirp Spread Spectrum (CSS), a spread spectrum
services, and used not only in Smart Cities, but also in rural areas technique that encodes information through wideband linear
[6], [7] or other mobility-related applications [8], which can even frequency modulated pulses, in three different bandwidths (125,
provide satellite navigation-independent positioning thanks to 250 and 500 KHz). This technique allows demodulation even
their features [9]. The communication technologies covered under below the noise level. Besides, LoRa tolerates small frequency
LPWANs include licensed technologies, such as Narrow Band IoT offset caused by the Doppler effect, and is resilient to fading and
(NB-IoT) or LTE-M, as well as unlicensed ones, from which we multipath environments [19]. In fact, this resilience is provided by
can highlight Sigfox and LoRaWAN [10]. the support of different Spreading Factors (SF), one of the main
Among existing LPWAN communication technologies, features of LoRa.
LoRaWAN presents several advantages that makes it suitable for SFs are directly related to chirps, a signal characterized by a
Smart Parking services based on sensor devices deployed outdoors linear frequency excursion all over the channel bandwidth in a
[11]. The use of LoRaWAN is actually growing thanks to its low- time period. Chirps decrease from the highest frequency to the
power consumption and long-range communication, sacrificing lowest (down-chirps) or vice versa (up-chirps). SFs range between
the throughput, which is not a critical feature in parking 7 and 12, and as a function of time, they are defined depending on
monitoring. the time taken by each chirp. The higher the SF, the lower Signal
Although some Smart Parking solutions based on LoRaWAN to Noise Ratio (SNR) is required in the receiver for demodulation.
can already be found in the literature [12], [13], either they do not In CSS, the number of bits per symbol are defined by the SF used,
delve deep into the network performance evaluation [14]–[16], or being 2SF bits of information per symbol. LoRa typically uses 8
they do not present their results based on a real long-term urban preamble symbols and 2 synchronization symbols, in addition to
deployment [17]. In this paper, we describe a LoRaWAN-based the header and the payload. In summary, a higher SF means a
Smart Parking deployment in detail, and evaluate its performance, more resilient communication against noise and a higher payload
putting special emphasis on the characteristics of these types of per frame, at the expense of power consumption, as the time-on-
deployments and extracting some conclusions about the air of chirps is longer, and data rate. A summary of these
challenges and key aspects that should be considered for the characteristics can be seen in Table 1, where the airtime is
deployment of this kind of networks. calculated considering the LoRaWAN protocol. LoRa SFs are
Thus, the key contributions of the paper are both, the thorough inherently orthogonal, thus they behave as noise to each other.
analysis of the behavior and performance of a real-world Moreover, LoRa implements a variable correction scheme,
LoRaWAN network deployed in a Smart City scenario for the
support of a Smart Parking service, as well as the derivation of
Table 1. LoRa modulation features (125 KHz bandwidth)
practical aspects that should lead to optimal planning and
deployment of this kind of networks for supporting smart services
in urban scenarios. 10-byte
Bits
Spreading SNR limit payload Bitrate
The remainder of the paper is structured as follows. Section II (Chips /
Factor (dB) airtime (bps)
presents an overview of both, the LoRa and LoRaWAN Symbol)
(ms)
communication technologies. Section III describes in detail the
7 128 -7.5 61.7 5470
LoRaWAN-based Smart Parking service and the deployment
carried out in Santander, as well as the system that enables the 8 256 -10 113.2 3125
LoRaWAN data collection. Then, in Section IV we present the 9 512 -12.5 205.8 1760
performance evaluation of the Smart Parking service and discuss 10 1024 -15 370.7 980
the results emerged from the use of LoRaWAN. Finally, Section V 11 2048 -17.5 823.3 440
summarizes the lessons learned and the future work. 12 4096 -20 1482.8 250

108
LoRaWAN-based Smart Parking Service PE-WASUN’22, October 24–28, 2022, Montreal, QC, Canada

typically with a 4/5 coding rate. LoRa is meant to be used in the communication, while increasing the device power
sub-GHz spectrum. In Europe, LoRa transmits in the ISM band consumption.
within the 863-870 MHz frequency range, regulated by the
LoRaWAN protocol defines the Adaptative Data Rate (ADR)
European Telecommunications Standards Institute (ETSI), with a
feature, from which devices set the SF depending on the SNR at
bandwidth of 0.3 MHz per channel.
gateway level, aiming to use the lowest possible SF and reduce the
power consumption. Besides, LoRaWAN restrictions apply to
2.2 LoRaWAN Specification
avoid collisions due to the use of unlicensed bands. In particular,
LoRaWAN specification, currently in its 1.1 version [20], is regulatory authorities impose a restrictive use of the channel,
proposed and maintained by the LoRa Alliance [21], aiming at limiting transmissions to a 1% duty cycle per node. Finally, it is
optimizing the data transmission rate and the energy worth highlighting the two existing security schemas for device
consumption, supplying the needs of IoT devices. It features end- activation: Activation By Personalization (ABP) and Over-The-Air
to-end protected bidirectional communication through a star-of- Activation (OTAA). In the first case, session keys used for the
stars network topology in which gateways relay messages authentication are hardcoded into the ED before the deployment,
between end-devices and a network server. These are the three while in the second case they are derived from a root key and
fundamental components of the LoRaWAN architecture, which assigned through a join procedure once the ED connects to a
are defined below: LoRaWAN network.
• End-devices (ED): are sensors or actuators wirelessly In summary, thanks to the characteristics of LoRa and
connected to a LoRaWAN network. They are able to LoRaWAN, battery-powered devices are meant to offer [22]:
send (uplink) or receive (downlink) traffic to/from a • Low power consumption: depending on the
network server through LoRaWAN gateways. conditions, devices can last more than 10 years before
• Gateways (GW): connect the EDs and the Network running out of battery.
Server, demodulating the uplink messages received • Long-range communication: up to 5 km in urban
from EDs, and modulating downlink messages the other scenarios, and up to 15 km in rural or open areas, being
way around. In contrast to EDs, GWs are not usually even longer under Line of Sight (LOS) propagation.
power-constrained devices and are connected to the
Internet through traditional high-speed network
technologies (e.g. Wi-Fi or Ethernet). Connection 3 SMART PARKING SERVICE
between EDs and GWs follows the N-to-N
The LoRaWAN Smart Parking service presented in this paper
communication pattern, thus EDs uplinks are
has been deployed in the city of Santander, and is based on the
broadcasted, and can be received by any number of
use of parking sensors buried under the asphalt to monitor the
GWs.
occupancy status of parking spots in a 1-to-1 basis (i.e. one
• Network Server (NS): is the core component in the parking sensor per parking spot). Deployed parking sensors are
LoRaWAN architecture backend, which receives all the manufactured by Byda [23] and shown in Figure 1.
uplink messages from EDs and expose them to the Parking sensors are equipped with a radar that detect any
application layer. GWs and Network Server follows a N- object placed on top of it at a certain distance, thus recognizing
to-1 communication pattern. Besides, downlink when the cars are parked, while diminishing other irrelevant
messages are only sent back to EDs through the GW objects (e.g. tree leaves). Parking sensors send their status in two
with best SNR for the received uplinks. different ways. Firstly, upon an asynchronous event, triggered
once the parking sensors find a status change. Secondly, on a
LoRaWAN specification defines three device classes, regular basis, as parking sensors send a beacon every hour, which
depending on the transmitting and receiving windows: includes the parking status as well, regardless if it has been
• Class A: all devices must at least implement this class, changed or not.
meant to reduce the energy consumption to the
minimum. Devices can communicate to the gateway at
any time, but they can only open two receiving
windows after each transmission. Thus, downlinks can
be only received after an uplink has been sent.

• Class B: devices open a periodic receiving window


defined by a synchronization beacon sent by the
gateway, thus the server can send downlinks to devices
at specific time intervals.

• Class C: devices are always listening, but for the Figure 1. Byda parking sensors
transmission periods, reducing the latency for downlink

109
PE-WASUN '22, October 24–28, 2022, Montreal, QC, Canada Juan Ramón Santana et al.

coverage. They take advantage of the preexisting LoRaWAN


GW
network infrastructure in Santander, which consists of 10
LIGHTHOUSE
uc-tlmat-kerlink-09 SMART PARKING gateways distributed within the city, as shown in Figure 2.
DEPLOYMENT

3.1 Downtown Deployment


The first deployment scenario is located in “Paseo Menéndez
Pelayo”, in the city downtown, which is an area characterized for
GW uc-tlmat-rak-01
having high building density, and also medium to high traffic
density caused by vehicles moving around and searching for a
parking spot. It is a two-way street with parking spots located at
uc-tlmat-kerlink-01 both sides of the road, as shown in Figure 3.
GW
uc-tlmat-kerlink-02 The deployment consists of 45 parking sensors, with a distance
uc-tlmat-kerlink-08
GW to the closest LoRaWAN gateways from 250 to 385 meters.
Furthermore, the number of buildings and the lack of LOS
uc-tlmat-kerlink-04 uc-tlmat-kerlink-06 GW
DOWNTOWN between gateways and parking sensors makes this scenario far
GW SMART PARKING
GW
DEPLOYMENT from ideal in terms of LoRa coverage.
uc-tlmat-kerlink-05 GW
uc-tlmat-kerlink-03
3.2 Lighthouse Deployment
The second scenario consist on a parking area near
Santander’s lighthouse, located in the “Avenida del Faro” avenue.
There are several major differences with the previous scenario, as
Figure 2. Gateway distribution and deployment areas this is a very touristic and leisure area, with low traffic density
and only crowded in particular periods depending mainly on the
weather and the vacation periods. Besides, this area is located in
Parking sensors are equipped with Class A LoRaWAN
one of the highest points in the city, thus with LOS to several
transceivers, activated under the OTAA schema, and
LoRaWAN gateways in the city with distances up to 3 Km.
implementing the ADR mechanism, so as to optimize the energy
A total of 24 parking sensors have been deployed in this area,
consumption and reducing the airtime as much as possible. as shown in Figure 4. It is worth highlighting the LoRaWAN
Besides, the application-level protocol specifies that all the gateway deployed in the same area. Therefore, the distance
transmissions are confirmed. Thus, parking sensors wait for an
between the parking sensors and the closest gateway is only about
explicit acknowledgement and force a retry if they do not receive
25 to 60 meters, which makes this deployment an almost ideal
it. After a certain number of failed retransmissions, parking
scenario in terms of communication.
sensors assumes that it has lost connection and triggers a reboot
Thanks to the different characteristics of both scenarios, we
that restarts the join procedure. can carry out a more realistic performance evaluation of a
The deployment areas have been chosen to create two
LoRaWAN parking service under real conditions in an urban
heterogeneous scenarios in terms of use and communication
scenario, as described in Section 4.

Figure 3. Parking sensors in the Downtown deployment

110
LoRaWAN-based Smart Parking Service PE-WASUN’22, October 24–28, 2022, Montreal, QC, Canada

LoRaWAN uplinks received by any gateway, including network-


related metadata (e.g. Signal to Noise Ratio (SNR), Received Signal
Strength Indicator (RSSI), Frame counter).
The Parking Data Extractor component is an in-house
development created to obtain and process all the uplinks received
at TTN level. This component retrieves and store in a local
database the packets received by the MQTT subscription. Finally,
it sends the parking service information to the SmartSantander
platform [26], so as to make it available for the city and other
aggregated value services. The whole system is depicted in Figure
5.

4 PERFORMANCE EVALUATION
The performance evaluation of the deployed parking sensors
in both scenarios has been carried out through the analysis of a
Figure 4. Parking sensors in the Lighthouse deployment
dataset consisting of 76833 packets sent by the parking sensors,
where 39676 of them correspond to the Downtown deployment.
3.3 Smart Parking System The dataset comprises the period between June 1st 2022 and July
7th 2022.
The Smart Parking system has two central components beyond The distribution of the packets received by each gateway and
the IoT infrastructure deployed in the city of Santander: i) the the corresponding SF is shown in Figure 6. As expected, the
LoRaWAN Network Server and Application Server, where we distribution of packets clearly depends on the gateways located
have leveraged the collaborative solution provided by The Things around the deployment areas, where “uc-tlmat-kerlink-03” and
Network (TTN) [24]; and ii), an ad-hoc developed component “uc-tlmat-kerlink-04” are the closest to the Downtown
named Parking Data Extractor. deployment, while the “uc-tlmat-kerlink-09” is the one located in
TTN is an open collaborative platform that provides the means the Lighthouse deployment. It is worth highlighting the number
to manage a LoRaWAN gateway infrastructure through a of packets received by the “uc-tlmat-kerlink-08” gateway, which
community shared back-end. TTN is based on The Things Stack is 3 kilometers away from the Lighthouse deployment. In this
(TTS), an open source software stack that implement, among sense, even for the lowest SF, some packets are received, thus, it
others, a LoRaWAN Network Server as well as an Application can be assessed, as expected, that the presence and density of
Server. From a user perspective, TTN provides a secured MQTT obstacles on the signal propagation and the received SNR have an
[25] broker, offering a subscription-based mechanism to receive impact that is much higher than the actual distance for LoRaWAN
deployments.

GW-B

GW-C

LoRaWAN based
GW-A
Smart Parking Sensors
(buried under the asphalt)

PARKING
DATA EXTRACTOR

SMARTSANTANDER
SMART CITY TESTBED

Figure 6. Number of packets received per LoRaWAN


Figure 5. Smart Parking Service backend system gateway and the corresponding SF

111
PE-WASUN '22, October 24–28, 2022, Montreal, QC, Canada Juan Ramón Santana et al.

Figure 7. Downtown deployment RSSI CDF for Free Figure 9. Downtown deployment RSSI PDF for Free
parking status (left) and Occupied parking status (right) parking status (left) and Occupied parking status (right)

Figure 8. Lighthouse deployment RSSI CDF for Free Figure 10. Lighthouse deployment RSSI PDF for Free
parking status (left) and Occupied parking status (right) parking status (left) and Occupied parking status (right)

So as to evaluate the behavior of the network, it is important other obstacles in the LOS does not mask the impact of parked
to consider the unique characteristics of Smart Parking services, vehicles, as it is the case in the Downtown deployment.
as the signal propagation will be affected by the parking status. Finally, the impact of the distance on the reporting of the
Therefore, we have studied the impact on the RSSI for both, the sensors’ observations have been also studied and represented in
free and occupied status for the different deployments. In this Figure 11 and Figure 12. To this end, we have only chosen the
sense, Figure 7 and Figure 8 show the Cumulative Distribution nearest gateway to the different deployments, so as to have more
Function (CDF) for each one of them, taking the maximum RSSI granularity in regards to distance. Therefore, we have considered
value from all the gateways receiving the same packet. As the distance between the gateway that received the packet and the
expected, the impact on the RSSI is significantly higher when the corresponding parking sensor. Both figures show as well the
parking spot is occupied, with a difference of 5 and 3 dBm between packet distribution in the axis, and the linear regression in red
the highest RSSIs for 80% of the received packets in the Lighthouse from the received packets.
and Downtown deployments, respectively. On the other hand, the As shown in the figures, RSSI decreases with the distance,
measured RSSI in the Lighthouse deployment is much higher than being steeper in the case of the Lighthouse, as the gateway is
in the Downtown deployment, as the gateway is closer, and the much closer. The concentration of the received packets around -
obstacles are almost negligible compared to the Downtown, thus 120 dBm and the prevalence of SF12 show that the Downtown
reducing the multipath losses. This can be appreciated in detail in deployment is already in the LoRaWAN coverage border, mainly
Figure 9 and Figure 10, where the Probability Density Function for the furthest parking sensors. In fact, although the LoRa
(PDF) for both scenarios and states are depicted. In the case of the receiver has a higher sensitivity, received packets are barely
downtown, most of the packets are received with a RSSI of -120 below -125 dBm, also due to the noise at the gateways.
dBm, being also the average RSSI for the occupied status -120.22
dBm, while in the free status the average RSSI rises to -117.53
dBm. On the contrary, in the Lighthouse deployment, packets are 5 CONCLUSIONS AND FUTURE WORK
more uniformly distributed along all the values of the RSSI, with Within this paper we have presented the assessment of
an average RSSI of -89.17 dBm and -82.86 dBm for the occupied LoRaWAN technology behavior when employed for supporting
and free parking status, respectively. As can be seen in the figures, the wireless communication of a Smart City service. In particular,
the average attenuation impact of parked vehicles is higher in the a real-world deployment of a Smart Parking service has been used
Lighthouse deployment. The reason behind this is related to the as the scenario for this thorough study.
better coverage conditions of this deployment, where the lack of

112
LoRaWAN-based Smart Parking Service PE-WASUN’22, October 24–28, 2022, Montreal, QC, Canada

Figure 11. Downtown deployment RSSI and distance distribution for Free parking status (left)
and Occupied parking status (right)

Figure 12. Lighthouse deployment RSSI and distance distribution for Free parking status (left)
and Occupied parking status (right)

Several conclusions can be derived from the results that have understanding of the radio propagation conditions (e.g. avoiding
been presented in Section 4. Firstly, the performance of LoRaWAN the use of simple propagation models but employing modelling
networks, in terms of coverage and communication robustness, is tools that incorporates urban scenario features such as density
far from the theorical values when devices are deployed in dense and type of obstacles). In the case of the typical propagation
urban areas, and they can be even worse in presence of the special environments that can be found in downtown urban areas the
characteristics of Smart City services, where, for example in the behavior observed in the analysis that has been performed
Smart Parking case, cars parked over the sensors significantly recommends to deploy gateways in a picocell distribution,
degrade the signal strength. limiting the range covered per gateway to 400 meters.
However, it has been experimentally demonstrated that Nonetheless, in situations where LOS can be guaranteed (or, at
LoRaWAN can still be suitable for this type of services, as long as least, obstacles density is low), even considering the presence of
a careful deployment plan is carried out in advance. In this sense, large obstacles (i.e. cars) just over the IoT devices, LoRaWAN has
for realizing robust and resilient large-scale LoRaWAN-based demonstrated large coverage areas. Thus, identifying key
communications in urban scenarios, the results obtained during locations with good overview of the city (i.e. typically high points
the evaluation described in the paper encourages the careful in the surroundings of the city) and placing there LoRa GWs,

113
PE-WASUN '22, October 24–28, 2022, Montreal, QC, Canada Juan Ramón Santana et al.

provides a good and efficient coverage to the underlying smart [8] W. Ayoub, A. E. Samhat, F. Nouvel, M. Mroue, and J.-C. Prévotet, “Internet of
Mobile Things: Overview of LoRaWAN, DASH7, and NB-IoT in LPWANs
city services. Standards and Supported Mobility,” IEEE Commun. Surv. Tutor., vol. 21, no. 2,
As the future steps, we plan to evaluate in detail the coverage pp. 1561–1581, 2019, doi: 10.1109/COMST.2018.2877382.
[9] C. Sanchez, B. Arpi, A. Vazquez-Rodas, F. Astudillo-Salinas, and L. I. Minchala,
in the whole city, and compare our results with other datasets for “Performance Evaluation of RSSI-based Positioning System with Low-cost
LoRaWAN networks, such as the LoED dataset [27]. Besides, we LoRa Devices,” in Proceedings of the 16th ACM International Symposium on
also plan to gather more information within the different SFs with Performance Evaluation of Wireless Ad Hoc, Sensor, & Ubiquitous Networks,
New York, NY, USA, Nov. 2019, pp. 37–44. doi: 10.1145/3345860.3361512.
a longer dataset, forcing the use of lower SFs to compare the [10] A. Ikpehai et al., “Low-Power Wide Area Network Technologies for Internet-
number of packets demodulated by the furthest gateways, of-Things: A Comparative Review,” IEEE Internet of Things J., vol. 6, no. 2, pp.
2225–2240, Apr. 2019, doi: 10.1109/JIOT.2018.2883728.
exploring the limits of LoRaWAN in urban areas. On the other [11] H. Wang et al., “LP-INDEX: Explore the Best Practice of LPWAN Technologies
hand, we are also working on the use of the network performance in Smart City,” in 2020 IEEE International Smart Cities Conference (ISC2), Sep.
information to infer service data (e.g. obtain the parking spot 2020, pp. 1–5. doi: 10.1109/ISC251055.2020.9239030.
[12] T. Lin, H. Rivano, and F. Le Mouël, “A Survey of Smart Parking Solutions,”
status based on the received RSSI) and machine learning IEEE Trans. Intell. Transp. Syst., vol. 18, no. 12, pp. 3229–3253, Dec. 2017, doi:
techniques to interpolate service information for deployments 10.1109/TITS.2017.2685143.
[13] J. J. Barriga et al., “Smart Parking: A Literature Review from the Technological
under critical coverage conditions. Perspective,” Appl. Sci., vol. 9, no. 21, Art. no. 21, Jan. 2019, doi:
10.3390/app9214569.
S. A. A’ssri, F. H. K. Zaman, and S. Mubdi, “The efficient parking bay allocation
ACKNOWLEDGMENTS
[14]
and management system using LoRaWAN,” in 2017 IEEE 8th Control and
This work has been partially supported by the Korea Agency for System Graduate Research Colloquium (ICSGRC), Aug. 2017, pp. 127–131. doi:
10.1109/ICSGRC.2017.8070581.
Infrastructure Advancement (KAIA) for the development of the [15] X. Meng, H. Wan, and T. Qin, “Design of Urban Parking Space Monitoring
action “K-City Network: Global Cooperation Program 2021. System Based on LPWAN,” in 2020 IEEE 20th International Conference on
Communication Technology (ICCT), Oct. 2020, pp. 1029–1034. doi:
Solution Demonstration Project”; and the Spanish State Research 10.1109/ICCT50939.2020.9295856.
Agency (AEI) by means of the projects FIERCE (Future Internet [16] P. Sotres, J. Lanza, L. Sánchez, J. R. Santana, C. López, and L. Muñoz, “Breaking
Enabled Resilient CitiEs) under Grant Agreement No. RTI2018- Vendors and City Locks through a Semantic-enabled Global Interoperable
Internet-of-Things System: A Smart Parking Case,” Sensors, vol. 19, no. 2, Art.
093475-A-I00, and SITED (Semantically-enabled Interoperable no. 2, Jan. 2019, doi: 10.3390/s19020229.
Trustworthy Enriched Data-spaces) under Grant Agreement No. [17] M. O. Farooq and D. Pesch, “Analyzing LoRa: A use case perspective,” in 2018
IEEE 4th World Forum on Internet of Things (WF-IoT), Feb. 2018, pp. 355–360.
PID2021-125725OB-I00. Authors would also like to acknowledge doi: 10.1109/WF-IoT.2018.8355224.
Dtonic Corporation for his work and support during the project. [18] “Semtech LoRa Technology Overview | Semtech.”
https://www.semtech.com/lora (accessed Jul. 01, 2022).
[19] C. Demeslay, P. Rostaing, and R. Gautier, “Theoretical Performance of LoRa
REFERENCES System in Multipath and Interference Channels,” IEEE Internet of Things J.,
[1] Department of Economic and Social Affairs of the United Nations, “World vol. 9, no. 9, pp. 6830–6843, May 2022, doi: 10.1109/JIOT.2021.3114439.
Urbanization Prospects (2018 revision).” https://population.un.org/wup/ [20] “LoRaWAN® Specification v1.1,” LoRa Alliance®. https://lora-
(accessed Jul. 01, 2022). alliance.org/resource_hub/lorawan-specification-v1-1/ (accessed Jul. 01,
[2] “State of IoT 2022: Number of connected IoT devices growing 18% to 14.4 2022).
billion globally,” IoT Analytics, May 18, 2022. https://iot- [21] “LoRa Alliance,” LoRa Alliance®. https://lora-alliance.org/ (accessed Jul. 01,
analytics.com/number-connected-iot-devices/ (accessed Jul. 01, 2022). 2022).
[3] “New Research on 1,600 Enterprise IoT Projects: Upsurge in Smart City and [22] K. Staniec and M. Kowal, “LoRa Performance under Variable Interference and
Connected Building Related IoT Projects,” IoT Analytics, Jan. 25, 2018. Heavy-Multipath Conditions,” Wirel. Commun. Mob. Comput., vol. 2018, p.
https://iot-analytics.com/global-overview-1600-enterprise-iot-projects/ e6931083, Apr. 2018, doi: 10.1155/2018/6931083.
(accessed Jul. 01, 2022). [23] “BYDA Sensor & System.” http://www.bydasensor.com/index.jsp (accessed
[4] “11% of global public parking spaces are now ‘smart’ – New report finds,” IoT Jul. 01, 2022).
Analytics, Dec. 19, 2018. https://iot-analytics.com/smart-parking-market- [24] T. T. Network, “The Things Network,” The Things Network.
report-2019-2023/ (accessed Jul. 01, 2022). https://www.thethingsnetwork.org/ (accessed Jul. 01, 2022).
[5] P. Sotres, J. R. Santana, L. Sánchez, J. Lanza, and L. Muñoz, “Practical Lessons [25] “MQTT Version 5.0.” https://docs.oasis-open.org/mqtt/mqtt/v5.0/mqtt-
From the Deployment and Management of a Smart City Internet-of-Things v5.0.html (accessed Jul. 01, 2022).
Infrastructure: The SmartSantander Testbed Case,” IEEE Access, vol. 5, pp. [26] L. Sanchez et al., “SmartSantander: IoT experimentation over a smart city
14309–14322, 2017, doi: 10.1109/ACCESS.2017.2723659. testbed,” Comput. Netw., vol. 61, pp. 217–238, Mar. 2014, doi:
[6] M. L. Liya and D. Arjun, “A Survey of LPWAN Technology in Agricultural 10.1016/j.bjp.2013.12.020.
Field,” in 2020 Fourth International Conference on I-SMAC (IoT in Social, Mobile, [27] L. Bhatia, M. Breza, R. Marfievici, and J. A. McCann, “LoED: The LoRaWAN at
Analytics and Cloud) (I-SMAC), Oct. 2020, pp. 313–317. doi: 10.1109/I- the Edge Dataset.” Zenodo, Sep. 25, 2020. doi: 10.5281/zenodo.4121430.
SMAC49090.2020.9243410.
[7] A. Grunwald, M. Schaarschmidt, and C. Westerkamp, “LoRaWAN in a rural
context: Use cases and opportunities for agricultural businesses,” in Mobile
Communication - Technologies and Applications; 24. ITG-Symposium, May
2019, pp. 1–6.

114

You might also like