Professional Documents
Culture Documents
by
My Name
(July 2023)
ii
Abstract
Effective water networks monitoring is crucial. It ensures efficient and sustainable use of
water resources. Despite significant efforts to develop water monitoring systems, water
loss in distribution networks remains a major issue. An estimated 30% of water is lost
due to leaks and other inefficiencies. This is not only wasteful but can also have serious
consequences for communities that rely on the water supply. Monitoring the water network
can help detect leaks and other issues early on, allowing for timely repairs and preventing
further loss. Additionally, it can provide valuable data on water usage patterns, helping to
identify areas where conservation efforts may be needed.
Traditional water monitoring systems face challenges in rural areas. Due to the lim-
itation on network connectivity and power supply. Making traditional water monitoring
systems less effective. Because these systems often rely on continuous data transmission
and processing, which requires a stable and reliable connection to the network and a steady
power supply.
To address these challenges, this thesis proposes a novel approach to water monitoring
using mobile LoRaWAN gateways. The gateways collect data from sensors and transmit
it to a central server via a low-power wide-area network. The proposed system is designed
to be cost-effective, scalable, and adaptable to different network configurations. Further-
more, this thesis proposes the integration of fog computing at remote sites of the water
networks to enable real-time data processing and analysis. This approach reduces the need
for centralized computing and improving the overall efficiency of the system.
We study the effect of gateway mobility in LoRaWAN on the network communication
flow and transmission parameters. The performance of various gateway arrival scenarios is
analyzed, including synchronized transmission, semi-synchronized transmission, and non-
synchronized transmission. In each scenario, we explore the trade-off between freshness of
the data, data collection rate and end-nodes power consumption.
The limits for reliable communication at different moving speeds are analytically com-
puted. These limits are then validated by both numerical simulations and real experiments.
We validate our analysis by both real experiments and simulations. Additionally, we in-
vestigate the limits of reliable communication in terms of gateway speed, throughput, and
energy consumption. Also, we present a case study exploiting the proposed architecture to
provide coverage for Wadi El-Gemal national park in Egypt
iii
The numerical results show that reliable communication is achieved over all spreading
factors (SFs) for gateway speeds up to 150 km/h. Negligible performance degradation is
noticed at SFs = 11, 12 at speeds more than 100 km/h. The synchronized transmission
model ensures the best performance in terms of throughput and power consumption at
the expense of the freshness of data. Non-synchronized transmission allows time-flexible
data collection at the expense of increased power consumption. Using four gateways, the
same throughput as semi-synchronized transmission can be achieved with only five times
the energy consumption. However, a single gateway requires seventeen times more energy.
Using ten gateways increases the throughput to the level of the synchronized scenario, but
at eight times the energy consumption.
iv
Acknowledgments
This work is dedicated to Professor Atef Ghuniem, may Allah have mercy upon his soul.
I am deeply thankful to Dr. Ahmed Elzanaty for his useful insights and constant gaudiness
during working on the research papers based on this thesis. I would like to express my
sincere gratitude to my husband, Mahmoud for his unwavering support and assistance in
data collection. I also want to thank iTi for providing me with the time and opportunity
to conduct this research and for their help in setting up ns-3. Finally, I am deeply grateful
to my family and kids for their love and encouragement throughout this journey.
1
Contents
Abstract ii
Acknowledgments iv
Acronyms 3
List of Tables 5
List of Figures 6
1 Introduction 8
1.1 Monitoring Water Networks . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
1.2 Challenges in Monitoring Rural Areas . . . . . . . . . . . . . . . . . . . . . 10
1.2.1 Core Network Challenges . . . . . . . . . . . . . . . . . . . . . . . . 10
1.2.2 End-Devices Challenges . . . . . . . . . . . . . . . . . . . . . . . . . 11
1.2.3 Security of Communication . . . . . . . . . . . . . . . . . . . . . . . 11
1.2.4 Freshness of Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
1.3 Fog Computing for Rural Data Collection and Processing . . . . . . . . . . 13
1.4 LoRaWAN for Rural Connectivity . . . . . . . . . . . . . . . . . . . . . . . 13
1.5 The Proposed Data Collection Model . . . . . . . . . . . . . . . . . . . . . . 14
1.6 Thesis Organization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
3 System Architecture 33
3.1 System Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
3.1.1 LoRaWAN Mobility Limitations . . . . . . . . . . . . . . . . . . . . 36
3.1.2 Path Loss Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
Bibliography 71
3
Acronyms
BW Bandwidth. 24
SF Spreading Factor. 24
SNR Signal to Noise Ratio. 55
Tv Visibilty Time. 33
4
List of Tables
3.1 Upper bounds of transmission distance and visibility time for each SF . . . 39
4.1 Detailed time duration and current consumption values for the power con-
sumption model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
List of Figures
5.12 Average energy consumption at the end-node for different SF values and
multiple gateways at speed=80 Km/h and Ron = 8windows/hr . . . . . . . 65
5.13 Distribution of end-nodes in the Wadi El-Gemal area. . . . . . . . . . . . . 66
5.14 Average number of received packets per end-node within effective visibility
time of different speeds and SF values. . . . . . . . . . . . . . . . . . . . . . 67
8
Chapter 1
Introduction
Systems monitoring via the internet of things (IoT) has been utilized in various domains
and applications. Many of these systems are applied over a wide geographical area or in
rural sites. For example, water monitoring networks, where timely and accurate data is
crucial to ensure efficient management and conservation of water resources. This results in
some challenges such as network connectivity, data collection, and latency.
In this chapter we give an overview of the importance of water networks monitoring,
explain the challenges in-terms of data collection and connectivity and, propose a solution
for data collection in rural areas with limited or no communication networks. The research
will focus on developing a fog-enabled architecture for water leakage monitoring, leveraging
LoRaWAN technology for low-power, long-range connectivity between sensors and mobile
gateways.
a risk of water scarcity and compromised public health, making it a critical issue that needs
to be addressed.
Water loss in water networks occurs due to many reasons. As shown in Fig. 1.1 there
are two main types of leakage; leakages in user residencies and leakages through the dis-
tribution networks. Two common types of leaks shown in the figure are distribution pipe
leakage and burst events. These types of leaks can result from various factors. Some of
the factors include corrosion, water pressure, and soil movement. Water leaks can cause
significant water loss, which can be difficult to detect and repair, especially in remote or
inaccessible areas. Aging infrastructure and poor maintenance of household water lines can
also cause residential water loss. This can occur due to factors such as corrosion, material
degradation, and improper installation or maintenance, resulting in substantial water loss
at the household level.
The Internet of Things (IoT) plays a vital role in water monitoring networks, which
are essential for the efficient management and conservation of water resources. By col-
lecting real-time data on water quality, quantity, and usage, these networks help identify
potential issues, detect and prevent leaks, and optimize water distribution systems. In such
monitoring systems, smart devices are used to exchange data and information over the in-
ternet. These devices may include sensors, meters, and controllers, which measure various
water-related parameters such as pH, temperature, pressure, and flow rate.
However, water monitoring networks can be complex and heterogeneous, involving var-
ious devices that communicate via different communication protocols. This heterogeneity
10
can pose a challenge for data collection, integration, and analysis. Therefore, there is a need
for interoperability and standardization in IoT-based water monitoring systems to ensure
seamless communication and integration of different devices and data sources. To address
this challenge, various communication technologies and protocols have been proposed [5],
such as LoRaWAN, Sigfox, NB-IoT, and MQTT, which provide reliable, low-power, and
long-range connectivity for IoT devices.
To reduce water loss in distribution networks, the authors in[6] present a is a smart
water management system. It uses IoT sensors and cloud computing to monitor water flow.
It can detect leaks in real-time. The system was deployed in a water distribution network
in Dubai and resulted in a 37% reduction in water loss. Our work in[7] demonstrate that
monitoring of the city-wide water distribution networks of Hurgada, Egypt can reduce non-
revenue water losses that accounts for around 40% of the water pumped in water networks.
Another study by [8] proposed a system for real-time monitoring of water loss. This system
uses machine learning algorithms to analyze flow data. The data is collected from pressure
sensors. The system was evaluated in a water distribution network in China and achieved
a 70% reduction in water loss.
Data collection in rural scenarios is challenging due to the lack of infrastructure com-
pared to urban areas [9] in both end devices and core network domains. Those challenges
include congestion, service delay, data privacy and security and lack of interoperability.
The increasing number of connected devices is a major challenge for traditional cloud
based solutions. Sending all collected data simultaneously to the cloud is not feasible. The
exponential growth of connected devices create massive amounts of data. For example
connected cars and traffic cameras can generate tens of megabytes of data per second. This
would require prohibitively high network bandwidth to transmit all the data to the cloud.
Additionally, core network services struggle to provide uninterrupted services to devices with
intermittent network connectivity, which is common in rural areas with limited connectivity.
[10]. This shows that serving a large number of devices at network edges is costly. The costs
come in terms of bandwidth, storage, latency, and energy consumption of the network.[11].
11
To address the weaknesses inherited by this cloud computing paradigm, fog computing
has been proposed by Cisco in 2005. Fog Computing encourages a shift of handling every-
thing in the core of cloud computing, to handling at the edges of a network. This approach
also reduces the burden on the cloud computing infrastructure and lowers the overall net-
work bandwidth consumption. It also support more latency sensitive applications [12]. The
use of fog computing in water network monitoring can enable real-time monitoring of the
water distribution network and provide timely feedback, thus improving the efficiency and
reliability of the monitoring system.
IoT devices are small and equipped with multiple sensors, requiring a continuous source
of energy, which presents a significant challenge in terms of battery life and cost. Low
power consumption is essential in IoT-based smart cities to address these issues. Selecting
the connectivity protocol to provide data access to these devices with acceptable data
rates and minimal power consumption is challenging. Additionally, monitoring services like
traffic cameras, smart metering, and systems spread over wide geographical areas require
long-range capabilities.
Several standards such as ZigBee and Bluetooth, have been used to implement IoT
services. Mainly due to their low power consumption. However, their limited coverage
constitutes a major obstacle[13]. Whereas, using cellular communication is not always
available due to lack of infrastructure. Deep satellite communication was an initial solution
to support IoT in rural areas. However, it was found to be inefficient in terms of cost and
link budget. Recently, low power wide area networks (LPWAN) have been efficiently used
for data collection in many scenarios. It’s most beneficial where the network is sporadically
distributed over a wide geographical range and works with low data rates [14–16].
As the Internet of Things (IoT) continues to grow, so does the need for robust security
mechanisms to protect against emerging security threats. The current security paradigm
for IoT applications is inadequate in addressing these challenges. One major challenge is
maintaining security credentials and software updates on a large number of devices located
12
on the network edge. This can be a daunting task, particularly when devices are deployed
in remote or hard-to-reach locations in rural areas.
Another challenge is the need to protect resource-constrained devices such as sensors,
which typically have limited processing power and memory. In addition, security breaches
on such devices can result in intolerable disruption, making it essential to respond to security
compromises in a way that minimizes the impact on the system as a whole. Furthermore,
the use of lightweight IoT connectivity protocols with simple or non-security algorithms can
also exacerbate these challenges by providing limited protection against potential threats.
To address these challenges, researchers have proposed various security mechanisms that
leverage advanced technologies such as blockchain and machine learning. These approaches
aim to provide secure, efficient and scalable solutions for IoT applications. Furthermore,
the use of low power wide area networks such as LoRaWAN, which are designed to provide
long-range communication with low power consumption, can also enhance security in IoT
applications by reducing the number of devices needed and simplifying the deployment of
secure IoT networks.[17].
The timeliness of the collected data is crucial in ensuring the accuracy and reliability of
the information. In certain applications such as agriculture, water, and climate monitoring,
some degree of information latency can be tolerated. However, for critical applications like
monitoring toxic gas emissions or detecting fires, real-time information is crucial. Several
studies have highlighted the importance of timely data collection in rural areas [18, 19].
Traditionally, data processing is done at the core network, which may not meet the
latency requirement for applications in rural areas. The round trip delay of data can also
be an issue. This can lead to unsatisfactory performance for applications that require low
latency. Hence, fog computing has emerged as a promising alternative to the centralized ap-
proach. Fog computing enables moving some of the storage, processing, and decision-making
functionalities near the end-nodes, thereby reducing the back and forth traffic between the
end-nodes and the core network. This approach not only saves bandwidth and energy but
also significantly reduces response time for real-time applications.
The freshness of collected data cannot be overstated in applications that require real-
time information in rural areas. While traditional data processing at the core network may
13
not meet the latency requirements, fog computing provides a viable alternative by moving
some of the functionalities closer to the end-nodes. This reduces the round trip delay of
data, resulting in better performance for applications that require low latency.
Fog computing was initially proposed for sustainable smart cities in [20]. The authors
in [21] proposed a multi-tiered framework that relies on fog computing to save energy and
reduce delays. Also, in [21] a fog-based approach is suggested as a more suitable architecture
for IoT applications. Fog computing is a reliable alternative to the traditional centralized
approach. It moves some of the storage, processing, and decision-making functionalities
near the end-nodes. This approach involves deploying computational resources, such as
storage, processing, and decision-making functionalities, in proximity to the end-nodes, i.e.,
the devices that generate or consume data. By doing so, the data can be processed and
analyzed closer to its source, which can result in significant improvements in response time,
bandwidth usage, and energy consumption.[7].
Fog computing can be particularly useful for real-time applications that require low
latency and high bandwidth [12], such as video streaming, and industrial control systems.
In such applications, fog computing can enable real-time processing and decision-making
at the edge [22, 23], which can significantly reduce the response time and improve the
overall performance of the system. Additionally, fog computing can help to overcome the
challenges posed by limited network connectivity, by providing local processing and storage
capabilities that can operate in areas with limited or intermittent network coverage [24, 25].
LoRaWAN has also been successfully deployed in various smart city monitoring appli-
cations [26], especially in rural areas, as it was utilized in mobility assisted data collection
techniques introduced in Wireless sensor networks (WSNs), where the mobility of sensor
nodes [27] or sink node [28–30] using Unmanned aerial vehicles (UAVs) as LoRaWAN gate-
ways is leveraged in expanding the coverage, reducing the transmission power and; hence,
the battery life of nodes, also, improving the localization of mobile end-nodes [31]. On
the other hand using mobile gateways in data collection is very beneficial in environmen-
14
tal monitoring and pristine sites. The absence of the heavy infrastructure required for
communication towers leave the environment and the natural habitats unaffected.
LoRaWAN technology operates in the unlicensed industrial, scientific, and medical
(ISM) band, which means that there are no licensing fees required to deploy this tech-
nology. This significantly reduces the deployment cost and attracts both the industry and
academia’s attention. Due to its cost-effectiveness, LoRaWAN has gained popularity in
recent years, and many research studies have been conducted to analyze, evaluate, and
experiment with its performance.
Studies have shown that LoRaWAN has a good connectivity range, making it suitable
for use in both urban and rural environments. In urban areas, LoRaWAN can provide
connectivity within a 5 km range, while in open space, it can provide connectivity within a
15 km range. Additionally, these studies have shown that LoRaWAN has a low packet-loss
ratio of less than 30% [32, 33].Which makes it an ideal technology for applications that
require reliable and efficient data transfer.
In this work, we propose a novel architecture for IoT data collection in rural areas. The
system considers a vehicle-Mounted Fog-Node that connects through the LoRaWAN pro-
tocol to fixed sensors deployed near the road. The fog-node consists of a controller board,
which is used for storage, GPS localization, and acts as a controller of the LoRaWAN gate-
way collecting the data. The fog node also acts as a minor version of the LoRaWAN network
and application servers reducing latency and the need for continuous cloud connectivity.
We investigate different gateway arrival scenarios to compromise between freshness
of the data, communication reliability and power consumption. We consider three prac-
tical scenarios in which the transmission is synchronized, semi-synchronized, and non-
synchronized and, study the effect of each scenario on the communication.
We explain the effects of the mobility of LoRaWAN gateways and their arrival model on
the LoRa communication parameters and gateway visibility time. We also study the effect
of using multiple gateways on the aforementioned parameters. We derive the corresponding
number of received packets and power consumption. Finally, we validate our analysis with
data from real experiments, as well as extensive simulations in both MATLAB and ns-3 ®
network simulations.
15
We apply our minimally invasive system to monitor one of Egypt’s main natural reserves.
Wadi El-Gemal national park, A coastal area along the red sea in south Sinai (24.4747° N,
35.0987° E). Many researchers are concerned with monitoring the marine and land environ-
ment of the area [34, 35] as well as the effect of tourism on the natural ecosystem [36]. To
preserve the stone nature of the land, in addition to the huge mangrove forests and coral
reefs existing, cellular communication towers are located only outside the boundaries of the
natural reserve, with many blind spots inside (areas with no cellular coverage). Hence, the
advantage and feasibility of our proposed system for this location. The contributions of this
work can be summarized as follows:
1. We propose a novel architecture for a reliable data collection technique for water
monitoring networks in rural areas using a mobile vehicle-mounted LoRaWAN
gateway.
2. We analyze different gateway dispatching scenarios and investigate the trade-
off between power consumption, end-nodes battery life, freshness of data and
communication reliability.
3. We setup and implement real experiments to validate the derived results and
the MATLAB simulation results of all proposed scenarios.
4. A case study of the synchronized transmission scenario in Wadi El-Gemal using
ns-3® network simulator.
Chapter 2 gives an overview of fog computing and LoRaWAN, explaining each technol-
ogy and how it benefits our proposed model, also, presents a literature review of related
work. Chapter 3 describes our proposed model in details and investigates the effect of
gateway mobility on the LoRaWAN communication. Three gateway arrival scenarios are
presented in 4 and the trade-off between power consumption, freshness of data and com-
munication reliability in each scenario is highlighted. Chapter 5 shows results of a real
experiment of LoRaWAN gateway mobility with speeds up to 120Km/h, validation of the
analytical results, ns-3® simulated results and practical results, MATLAB simulations of
throughput and power consumption for all scenarios, and ns-3® simulated case study of
monitoring Wadi El-Gemal national park in Egypt. Finally, chapter 6 explains our conclu-
sions and future work.
16
Chapter 2
Combining the benefits of Fog computing where computing resources can be distributed
at edge of the network, closer to the IoT devices. With the use of low-power, wide-
area networks (LPWANs) such as LoRaWAN can further enhance the efficiency and cost-
effectiveness of IoT water monitoring networks by providing long-range, low-power connec-
tivity. In this chapter we give an overview of each technology, highlight its advantages for
our proposed model and, present literature review of the related work.
The system architectures of Fog systems adopt a three-tier architecture similar to other
large-scale distributed computing systems. An illustrative model of such architecture is
shown in figure 2.1. The three tiers are:
• Cloud Layer: includes components that are responsible for functionalities such
as storing the data received from the Fog layer as a backup.
17
• Fog Layer: divided into two types of Fog nodes.The Fog server and the Fog
edge nodes. The Fog server includes modules that are responsible for manage-
ment functionalities such as application deployment, network configuration, and
billing. The Fog edge nodes provide computing, storage, and communication
capabilities to smart objects, and communicate with the Fog server. Meanwhile,
the communications between the Fog edge nodes and the cloud nodes are routed
through the Fog servers. Application components deployed in the Fog layer are
responsible for filtering the collected data from the smart objects and ensuring
real-time interactions
• IoT Device/End-User: consists of smart objects such as sensors and laptops
as well as connectivity devices.
The geographical distribution of Fog servers provides better support for user mobility;
heterogeneity and interoperability as Fog nodes coming from different vendors and operate
under different protocols; and, supports for big data analytic as well as, protecting data
18
privacy and enhancing the overall security of the network. The main advantages of Fog
computing rises from four key features[1]:
• Cognition: the awareness of the client objectives. Due to the close proximity
of Fog to the end users, Fog devices could be fully aware of the nature of the
customer requirements and be able to adapt to these requirements.
• Efficiency: the interoperability between Fog and cloud, as Fog can distribute
computing, storage, and control functions anywhere between the cloud and the
endpoint to take full advantage of the resources available anywhere on the net-
work. Moreover, the closer proximity of the Fog to the endpoints enable it to
be more closely integrated with the end-user systems to enhance overall system
efficiency and performance.
• Agility: enables rapid innovation and affordable scaling as it is usually much
feasible to experiment with client edge devices rather than waiting for large
networks migration to a new or updated technology.
• Latency: is the main benefit of processing at the network edge. As Fog can
support time-sensitive functions such as alarming systems, vehicle to vehicle
communication and augmented reality applications.
Fog architecture offers many compelling benefits for high availability, computing, stor-
age, as well as for networking in rural areas. Multiple Fog devices participate to cooperate
in the processing which reduces processing time and processors complexity. Compared to
processing at cloud data centers, energy consumption and operation costs in the Fog com-
puting paradigm is low. The Fog is located closer to the user, so the distance between users
and Fog devices could be one or a few hops [37]. Some of the main benefits Fog computing
introduce into rural IoT are discussed below.
Fog nodes fill the gap between the cloud and the end-devices “things” to enable a
service continuum. Meaning, that if you rely on the cloud for a certain service and you lose
connectivity, your nearest Fog node could act as a backup server for this service. Fog and
Cloud are interdependent [38]. For example, cloud services may be used to manage the Fog.
19
Conversely, Fog can act as the proxy of the cloud to deliver cloud services to endpoints, and
act as the proxy of the endpoints to interact with the cloud.
In cloud computing paradigm, events or data are transmitted to the data center inside
the core network. Results are sent back to the end users after a series of processing, which
puts high load on the core network. By designing module-based data analytical components
that can be remotely pushed into Fog devices on demand, Fog can insure uninterrupted
services, by acting as a backup for the cloud or by splitting the traffic and processing
load of the cloud’s core network by handling acquisition, aggregation and pre-processing,
reducing the data transportation and storage and, balancing computation power on data
processing[1].
Being in the near proximity to the end users, Fog avoids the back-and-forth traffic
between cloud and mobile users. This not only saves the backbone bandwidth, but also
significantly reduces the energy consumption of both the connected things and the core
networks. This close proximity of Fog nodes to users reduces the latency effectively. In [2],
Shanhe Yi et al, constructed an experiment comparing between Fog and cloud performances
in terms of latency and bandwidth consumption.
The authors build a proof-of-concept Fog computing platform, consisting of two Fog
subsystems with a network link between them and both are connected to the Amazon EC2
cloud through WAN as shown in figure 2.2, all of which were equipped with access points
so that the mobile devices they used can access the Fog as well as the Amazon EC2 cloud.
Then they measured both uplink and downlink bandwidth while connecting to each system
showing that, the bandwidth consumption for connecting to the cloud is almost fifty times
more the consumption of Fog.
Figure 2.2: Experiment setup comparing Fog and cloud latency performance [2]
20
Real-time interaction is a critical requirement for many smart city applications. For
instance, in vehicle-to-vehicle communication, latency below a few tens of milliseconds is
necessary for the system to function correctly [39]. However, mainstream cloud services
struggle to meet such low latency requirements due to the distance between the cloud
servers and the end devices [40]. This issue can be addressed through Fog computing,
which allows processing to occur at the network edge near the end devices by the Fog
nodes. This approach enables processing at specific locations, reducing latency significantly
and enabling location-aware real-time services. By pushing the computation to the network
edge, the Fog computing paradigm ensures that data processing occurs closer to the source
of data generation, reducing the need to transmit data to the cloud for processing. This
can result in significant latency reduction and improved performance for latency-sensitive
applications [12].
The experiment conducted in [2] measured Fog and cloud based network performances
in terms of latency. They used use round trip time (RTT) as their metric and, implemented
a face recognition application running across a smartphone and a Fog or a cloud. The app
on the smartphone sends a picture of someone face to the remote server and, the remote
server will then try to recognize the face by matching it in the local face photo database
of size 1521. They compared the RTT for both systems showing that the latency of cloud
is twelve times more than Fog, while the processing time on both servers was almost the
same. The response time of the Fog server was almost eight times faster than that of a
cloud server.
Data privacy and security remains the main concern of most users which poses a na-
tional security threat for large scale smart city applications. According to HP, 70 percent
of IoT devices in a smart city were at risk of attack due to sufficient vulnerabilities such
as insufficient authorization, inadequate software protections, and weak encrypted commu-
nication protocols [41]. Moreover, the open nature of smart cities and its dependency on
wireless communication between the smart cities three elements; devices, Fog and, cloud
causes more vulnerabilities.
21
Three different types of security need to be ensured: network connection security, data
security, and user privacy [42]. Fog’s proximity to end users and locality on the edge enable
it to help address certain IoT security challenges such as updating security checks and
credentials on a large number of physical devices. If an update is available, it will be much
more feasible and reliable for the devices to get this update through its nearest Fog server
instead of connecting to the cloud all at once.
The Fog nodes also can work as an authentication layer between the end devices and the
cloud to apply some sort of access control over the services and data. On the other hand,
Fog layer could be used as an extra layer of protection to source limited devices which cannot
execute complex security algorithms. More importantly, using Fog ensure that users will
take possession of their own data locally, and eliminate the need of transmitting sensitive
data for processing. If remote processing is needed, these data could be always processed
at a secure local Fog server.
2.2 LoRaWAN
One of the widely utilized technologies in IoT applications is the low range wide area
network (LoRaWAN) protocol operating in the ISM band, by the LoRa Alliance. LoRaWAN
can be potential solutions for connectivity in rural areas due to their low power consumption,
long range, low complexity, and minimal cost [14–16]. In this section, we present some basic
background about the LoRaWAN protocol. Highlighting some of the design choices that
will be investigated in our system analysis.
22
LoRaWAN is the media access control (MAC) layer above LoRa. LoRaWAN MAC
layer defines three classes of end devices for different operation needs. Classes A, B, and C
as shown in figure 2.4; where the key difference is their trade-off between communication
latency and power consumption [3, 33]. Class A networks are mainly intended for monitoring
applications. In a Class A network, transmissions are always initiated by the end devices,
in a totally asynchronous manner. After each uplink transmission, the end device will open
two reception windows, waiting for any command or data packet returned by the server.
Class B is for battery powered actuators where uplink and downlink transmissions are
equally needed. Class B end devices synchronize with the server by beacon packets which
are broadcasted by Class B gateways and can hence receive downlink data or command
packets in specific time windows. Finally, Class C is defined for end devices with no energy
constraints. Which means that such devices can keep their receive window open at all
times. In this work, we use LoRaWAN to utilize its off-the-shelf devices and software
without needing any firmware modification.
24
Figure 2.4: LoRa end-devices reception slot timing in different classes [3]
This shows that the higher the spreading factor (SF) gets, the longer the chirp signal is,
hence the further the signal could reach. The choice of (SF) affects the symbol duration
and the supported data rate[43]. And, the bandwidth (BW) is related to the symbol length
ls as
SF
BW = fmax − fmin = (2.2)
ls
SF leverage the trade-off between transmission rate, reliability, and coverage, i.e., the higher
the SF is, the lower the data rate and wider coverage. The relation between the input bit
25
Rb = RS SF . (2.3)
TOA = ts Ns (2.5)
Ns = Np + Npl (2.6)
where Ns is the number of symbols, Np is the of preamble symbols and Npl is the number
of payload symbols [45].
ALOHA is used for medium access, and each end-node starts its transmission window
with a random time offset. If there is a collision and transmission fails, the transmitter
waits for a random amount of time before re-transmitting. Although collision would only
happen if two nodes transmitted simultaneously on the same frequency using the same
SF, this might cause a significant time delay which would be intolerable in case of limited
gateway availability due to movement and taking into account the standard requirements
of D airtime duty cycle for end-nodes.
LoRaWAN utilizes two layers of security: one for the network and one for the appli-
cation. The network security ensures authenticity of the node in the network while the
application layer of security ensures the network operator does not have access to the end
user’s application data. AES encryption is used, with the key exchange utilizing an IEEE
EUI64 identifier.
In this section, we explore the challenges and techniques for data collection in rural
areas, with a particular focus on the use of LoRaWAN and fog computing. The literature
review highlights the limitations of traditional data collection methods such as cellular
networks and satellite communications in rural areas, and the need for more efficient and
26
cost-effective solutions. The potential of mobility-assisted data collection in rural areas and
how it can improve the coverage and reliability of the data collected. The section also
discusses the potential of LoRaWAN, a low-power wide-area network technology, and its
ability to provide long-range and low-cost communication for IoT devices in rural areas.
And, emphasizes the effect of Fog computing on enhancing network performance.
Data collection in rural areas is a complex and challenging task due to several factors.
Firstly, the lack of proper infrastructure such as roads, electricity, and communication
networks makes it difficult to access remote areas. Secondly, the sparse population in these
areas leads to low demand for services, which limits the incentive for service providers to
invest in infrastructure. Thirdly, harsh environmental conditions such as extreme weather
and natural disasters pose significant risks to equipment and personnel. Ezeugwu et al.
[46] examined the challenges of data collection for agricultural research in Nigeria. They
identified the lack of infrastructure, inadequate funding, and inadequate technical expertise
as major challenges. They proposed the use of mobile data collection tools and capacity
building for researchers and farmers to overcome these challenges.
Data collection challenges in rural areas are reflected into the problem of network cover-
age, as the protocols commonly used in IoT such as RFID, NFC and BLE are not suitable
because of their short range capabilities. Whereas, using cellular communication is not
always available due to lack of infrastructure. One of the initial solutions to support IoT in
rural areas was using deep satellite communication, which was not very efficient in terms
of cost and link budget. Wireless sensor networks (WSN) were then introduced to connect
end-nodes via protocols such as WiFi or Zigbee [47, 48], and aggregate data over a wide
range till it reaches a point where internet connectivity is available. WSNs also, introduced
mobility-assisted data collection techniques, where the mobility of the sink node or gateway
is leveraged in expanding the coverage, reducing the transmission power and hence, the
battery life of the nodes and localization [49].
Several research studies have addressed the lack of communication infrastructure as a
major challenge to data collection. For instance, in their study on rural water management,
Wu et al. [50] proposed the use of LoRaWAN and fog computing to enable mobility-assisted
data collection in remote areas. Similarly, in their study on rural healthcare, Guo et al.
27
In recent years, LoRaWAN has emerged as a promising wireless technology for data
collection in rural areas due to its long-range and low power consumption capabilities. The
study in [58] evaluates the performance of LoRaWAN in a rural area, highlighting its poten-
tial to enable Internet of Things (IoT) applications such as precision agriculture and smart
28
water management. The authors demonstrate that LoRaWAN can provide reliable and
energy-efficient communication over long distances, even in harsh outdoor environments.
One of the major benefits of using LoRaWAN is its ability to cover a large distance
while consuming low power. LoRaWAN transmission performance in rural areas has been
analytically and imperially analyzed in the literature. In [59] the authors investigated in
the 868 MHz band. Extensive measurement campaigns were carried out at urban and rural
environments. Their campaigns’ results showed that the coverage in rural areas exceeded
15 km. A comparison between Long Term Evolution - Narrow Band (LTE-NB) and Lo-
RaWAN performances was carried out in [60–62]. (LTE-NB) achieved better coverage in
urban areas and better performance with applications that requires high network through-
put and availability. However, LoRaWAN achieved better coverage in rural areas and better
performance with applications that requires long battery life.
Network scalability is another benefit of using LoRaWAN. LoRaWAN technology allows
for the deployment of many sensors and gateways, which can be easily managed through
a central network server. This feature is particularly important in rural areas where data
collection may require a large number of sensors distributed over a vast geographical area.
The study presented in [63] reviews LoRaWAN scalability potential and issues for deploying
massive IoT. To manage scalability issues in LoRaWAN, different techniques can be used
such as slotted medium access [64], multi-hop communication [65, 66], adaptive data rate
optimization (ADR) [67] which adjusts the data transmission rate based on the signal quality
and the battery level of the end-node, and capacity planning [68, 69].
Data collection using LoRaWAN has been widely studied and implemented in various
fields. For instance, in [70–72], it has been used for soil moisture monitoring, temperature
sensing, and crop growth monitoring. LoRaWAN has also been used for environmental
monitoring such as air and water quality monitoring [73, 73–75], water management [76, 77],
and fire detection [78, 78, 79]. In smart cities, it has been used for traffic management [80, 81]
waste management [82, 83], and parking management [84, 85]. Furthermore, it has been
used in tracking, and communication in areas here traditional communication technologies
may not be able to provide coverage. In areas with rough terrain such as mines or pristine
areas [86, 87].
The deployment of LoRaWAN networks in rural areas poses several challenges. One
major challenge is the limited availability of LoRaWAN gateways, which may result in
coverage gaps and signal attenuation. To address this issue, mobility assisted data collection
29
is used. Mobile gateways or end-nodes that can be placed on moving vehicles or drones
to improve the coverage and reliability of data collection in remote and inaccessible rural
areas.
While, mobility assisted data collection significantly enhances the network connectivity.
Throughout the literature, this is commonly applied using UAVs where the gateway would
stationary hover over the end-nodes while collecting the data, with the altitude being the
main effective parameter[28, 29, 31]. Nevertheless, our work focuses on studying the effect of
gateway mobility with relatively high speeds on data transmission and the flow of network
operation in LoRaWAN. Also, we propose utilizing transportation vehicles instead of drones
to overcome the limitations of the high power consumption, low travel time and, the inability
of carrying heavy antennas of high gain [88].
Drones as mobile LoRaWAN gateways are used in [89] for fire detection in forests where
a network of ground Infrared cameras was installed to constantly observe any thermal
changes in the area. LoRaWAN gateways would fly over the targeted areas, collect the
readings of cameras, and head back to offload the data into a server for early fire detection.
The same approach was implemented in [90] to guarantee rapid detection of environmental
parameters in infrastructure-less areas.
Mobility in LoRaWAN networks is extensively studied in the fixed gateway and mobile
end-nodes scenarios. This is expected due to the wide range of IoT applications in smart
wearable devices, vehicle communication, and tracking devices that benefit from this sce-
nario. The authors in [16] reviewed various IoT systems for mobility support and concluded
that mobility in LoRaWAN can be easily achieved in uplink while operating in any of the
three classes, with the expense of some latency in downlink due to the time needed by the
network server to update its forwarding table as the end-nodes change their location.
In the opposite scenario, where the end-nodes are fixed and the gateway is mobile,
the work carried out is almost limited to unmanned aerial systems that use LoRaWAN gate-
ways as mobile data sinks for various applications such as localization [31, 91], the backup
communication system in case of environmental disasters or hostile locations such as mines
and forests [92, 93], and many environmental monitoring applications [94]. The perfor-
mance analysis of such systems is mainly concerned with the efficient scheduling of the
30
data collection and routing of the LoRaWAN gateway due to the limited power capacity in
UASs [95–97]. Unmanned aerial systems collect data while hovering over the end-nodes at
low speeds and stay in location until all transmission is complete. In comparison, our work
focuses on analyzing LoRaWAN communication over relatively high speeds and limited
gateway availability.
The mobility of LoRaWAN gateways requires an appropriate choice of spreading factor
and LoRa configuration. One option is to use higher spreading factors to ensure stable
communication through the vertical distance between the ground end-nodes and the drone
hovering over them in [90, 98]. However, using lower spreading factors with a shorter time
on air is more suitable for tracking applications and real-time communication when shorter
inter-transmission time is required [91, 93].
Experiments of the coverage and speed limits of mobile LoRaWAN gateways in the lit-
erature showed successful LoRaWAN communication on speeds of 10–50 km/h and vertical
distances of 30–300 m [29, 99–101], which is to be expected due to the limitations in power
in the unmanned aerial systems used in these experiments. Our previous work [102] presents
real experiment results and performance analysis of the mobility of LoRaWAN gateways
with relatively much higher speeds, showing that transmission can be successful with SFs
7 to 10 at speeds up to 160 km/h.
To ensure seamless connectivity in LoRaWAN mobile scenarios, several challenges need
to be addressed to ensure the reliability and efficiency of LoRaWAN in such scenarios. One
of the challenges is the Doppler effect, which causes a shift in the frequency of the signal as
the device moves, leading to a higher bit error rate and decreased signal quality. This issue
was studied in [103], where the authors proposed a new algorithm to mitigate the effects of
the Doppler shift on LoRaWAN signals.
In addition, the handover of an end-node from the coverage of one gateway to another
is also a critical issue in LoRaWAN mobile scenarios, as it affects the continuity of the con-
nection and the quality of service. Furthermore, (ADR) technique is crucial for optimizing
the network performance and prolonging the battery life. Finally, end-node localization is
also a critical factor in LoRaWAN mobile scenarios, as it enables tracking the location of
assets or vehicles and facilitating various applications such as fleet management and logis-
tics. These issues were addressed in [104], where the authors conducted an empirical study
on the performance of LoRaWAN in mobile scenarios and proposed several solutions to
improve the reliability and efficiency of the network.
31
Mobile fog computing is discussed in [111]. The primary advantages is the ability to
enable real-time processing and decision-making closer to the data source. Combining mo-
bile fog computing with LoRaWAN mobile gateways offers better scalability, as additional
gateways can be easily deployed to expand the coverage area and handle increasing data
traffic. This could enable efficient and effective IoT applications with improved performance
and reduced operational costs especially in rural areas.
33
Chapter 3
System Architecture
This section explains the proposed LoRaWAN mobility assisted data collection model.
We explain the effect of gateway mobility on LoRaWAN communication. And, we analyze
the system parameters in terms of range, maximum throughput, and power consumption.
We consider rural areas with minimal IoT infrastructure. The proposed system, de-
picted in Fig. 3.1, consists of a fog-vehicle equipped with a LoRaWAN gateway and reduced
capability application and network servers, end-nodes sparsely located by the water distri-
bution pipes alongside the highway, and, a cloud-based fully-functional LoRaWAN network
and application servers.
In our setup, the end-nodes are located remotely and may transmit data intermittently
during the day. To conserve energy, these nodes only open transmission windows at a
pre-configured rate per hour (Ron ) to check for the presence of a nearby gateway before
transmitting data. The total time at which the gateway and end-node are in range is
defined as the gateway visibility time (Tv ), and the data transmission time should not
exceed this duration. In this sparse deployments of end-nodes in remote areas, we consider
the nodes to be distanced far enough apart so that only one node is transmitting within
the gateway visibility time (Tv ). By adhering to these guidelines, the Lorawan network can
optimize energy consumption, maintain reliable communication, and ensure efficient data
transmission even in low-power, remote settings.
34
Figure 3.1: Proposed system model for moving fog-node data collection
35
The proposed model for data collection in rural areas using mobile gateways involves a
gateway that moves at a speed of vkm/h to collect data from the end-nodes. The collected
data is then transmitted to the network or application server when the gateway reaches a
location with an internet connection. The model considers three different gateway arrival
scenarios, each offering a different level of flexibility in the data collection schedule, which
affects the freshness of the data. However, there is a trade-off between power consumption
and communication reliability.
The first scenario is the synchronized transmission, where the gateway arrives at a fixed
time at each end-node location, and data is transmitted to the network or application server.
This scenario ensures high reliability in data transmission and the least power consumption
but requires complex synchronization with the gateway due to the fixed schedule. The
second scenario is semi-synchronized transmission, where the gateway arrives at each loca-
tion within a time window, allowing more flexibility in the data collection schedule. This
approach increases power consumption but maintains communication reliability.
The third scenario is non-synchronized transmission, where the gateway collects data at
each location, regardless of the time, and transmits the data to the network or application
server when it reaches a location with an internet connection. This scenario provides the
highest level of flexibility in the data collection schedule and the highest power consump-
tion, but may result in lower communication reliability due to the lack of synchronization.
This is enhanced by using multiple gateways to collect the data. These scenarios are an-
alyzed in detail in the following chapter, considering factors such as power consumption,
communication reliability, and data freshness.
The fog-vehicle has three main functionalities: i) storage sink to all collected data;
ii) a virtualized network server to coordinate LoRaWAN configurations (e.g., spreading
factor allocation and transmit time coordination for the nodes); iii) a virtualized application
server responsible for real-time decision making such as activating some actuator depending
on the measured data [112]. Each virtualized server module reports and synchronizes its
data with the corresponding main server when internet connectivity is available. More
advanced functionalities responsible for data verification, compression, or ciphering can be
implemented on the fog node according to application requirements.
The mobility of the gateways can have an impact on the communication between nodes
and the gateway. The speed of the gateway can affect the transmission time and cause
the Doppler effect, which can cause signal fading and limit the range of communication.
36
Coherence Time tc
60 ts : SF = 12
ts : SF = 11
ts : SF = 10
50 ts : SF = 9
ts : SF = 8
40 ts : SF = 7
Time (ms)
30
20
10
Figure 3.2: Symbol times for different spreading factors and coherence time with mobility
1
tc = . (3.1)
fD
When the symbol time of LoRa is larger than the coherence time, the amplitude and phase
changes imposed by the channel vary considerably during the transmission of one symbol.
To cope with the Doppler spread, the system should be designed such that ts < tc .
In this regard, we inspect the coherence time and the symbol duration for various
gateway speeds and spreading factors, respectively, as shown in Fig. 3.2. In this figure,
we consider a center frequency of 868 MHz and 125 kHz bandwidth. We conclude that
transmission with SFs up to 10 is less sensitive to Doppler shift if the gateway speed is
below 160 Km/h; however, higher SFs will be highly subject to Doppler shift at speeds
greater than 80 Km/h. On the other hand, higher SFs imply a longer communication range
that increases the visibility time for moving Gateway at higher speeds. Therefore, there is a
trade-off between the selection of lower SFs to mitigate the Doppler effect and the selection
of higher SFs to enhance the communication range.
38
The visibility time is defined as twice the maximum transmission time, assuming the
transmission starts as the gateway approaches the end-node from a distance, passes by the
node, and then moves away in the opposite direction, i.e.,
2 dmax
Tv = . (3.2)
v
in the case of moving gateway, the spreading factor used at the end-node should be optimized
taking into account the maximum distance between the end-node and gateway and the total
time on air (TOA) required to transmit all packets within the gateway visibility time (Tv ).
Selecting lower SF values might achieve higher throughput, but requires higher sensitivity
levels at the receiver, resulting in lower path loss endurance. On the other hand, higher
SFs endure higher path loss levels at the expense of longer transmission time which would
reduce the throughput due to the duty cycle limitations within the gateway visibility time,
in addition to the Doppler shift effect stated earlier.
To study the effect of these boundaries on the maximum throughput for each SF, we need
toi) calculate the maximum distance the signal could travel under the path loss conditions,
ii) calculate the maximum visibility time depending on the gateway moving speed and the
time at which, end-nodes’ transmission starts, iii) take into account the transmission duty
cycle. The bounds of transmission distance and visibility time for each SF are listed in
Table 3.1.
To further study the effect of mobility on the total number of received packets, we con-
sider the available TOA within the total visibility time determined by speed v of the moving
fog-node, while taking into account the 1% duty cycle requirements of the LoRaWAN pro-
tocol. Each transmission takes total time T = 100 T OA where, 0.01T is the actual T OA
calculated as Eq.5 in [113] and, 0.99T is the waiting period forced by the standard. In this
analysis, the visibility time Tv is defined as twice the time it takes the moving fog-node to
travel distance dmax , assuming the end-node is centered along the traveling path.
The distance dmax is computed as the maximum possible distance for a certain trans-
mitted power Pt to meet the receiver sensitivity Prs requirements shown in Table 3.1. We
calculate dmax using the Okumura-Hata propagation loss model for the channel.
In the rest of the paper, we consider the maximum speed of the gateway such that is
the Doppler effect is minimal. Now, we investigate the impact of the spreading factor on
the communication setup, i.e., gateway visibility, time, and throughput.
39
Table 3.1: Upper bounds of transmission distance and visibility time for each SF
SF SF7 SF8 SF9 SF10 SF11 SF12
Prs dBm -123 -126 -129 -132 -134.5 -137
2dmax km 7.4 8.4 9.4 10.5 11.5 12.6
T60Km/h min 7.41 8.40 9.42 10.50 11.52 12.61
T80Km/h min 5.55 6.30 7.05 7.88 8.63 9.45
T100Km/h min 4.44 5.04 5.64 6.30 6.90 7.56
The key difference between the A, B, and C classes in LoRaWAN, is the trade-off made
between latency and Energy consumption [114]. Class A devices are only awake during
scheduled transmissions, which requires for precise scheduling of the transmission windows
with the presence of the moving gateway. Class C devices will have a better chance of
catching the gateway, as the end-nodes are usually connected to a power source and are
always listening. Lastly, Class B is a hybrid profile where beacon signals can be used to
synchronize with the gateway and start the transmission only when the gateway is present,
but at the expense of power consumption.
In our proposed system, we utilize Class A assuming perfect synchronized transmission
windows. The TOA nearly doubles with each step in spreading factor, the energy con-
sumption and hence battery drainage increase due to transmission time. The daily energy
consumption can be calculated as:
where, Vin is the input voltage, Itr is the drained current for some transmission power Pt
and SF, and N is the number of transmissions during the day, neglecting the current drained
at sleep mode.
We notice that, for the same payload length, using higher spreading factors unnecessarily
within lower link budgets, will result in wasting doubles of the actual amount of energy
needed for transmission. On the other hand, if the link budget is not specified, it may be
beneficial to transmit at high SF to guarantee delivery at first attempt as the equivalent
power cost can be less than the total cost of re-transmission of failed packets at lower
spreading factors [115].
40
For the path loss model, we consider the Okumura-Hata propagation model. The
Okumura-Hata propagation model is widely used for rural scenarios due to its compat-
ibility with macrocell characteristics. It takes into account the frequency of the signal,
distance, antenna height, and other environmental factors such as terrain, land use, and
climate. The path loss is calculated using a formula that includes several parameters and
constants, such as the base station antenna height, mobile station antenna height, frequency,
and city size. The model is based on empirical measurements of signal propagation and has
been extensively validated in various scenarios.
The Okumura-Hata model can be used to predict the signal strength at different dis-
tances from the base station, which is crucial for designing and optimizing wireless commu-
nication systems. It is particularly useful for large-scale deployments in rural areas where
the terrain and environment can significantly impact signal propagation. By accurately
modeling the path loss, network planners can estimate the coverage area and capacity of a
wireless network, optimize the placement of base stations and antennas, and select appro-
priate transmission parameters such as power and modulation scheme.
The path loss for the Okumura-Hata [116] is written as
where fc is the carrier frequency, hb and hm are the antenna height at the base station
(gateway) and end node, respectively, and d is the link distance (all in kilometers). The
functions A(fc , hb , hm ), B(hb ), and C(fc ) depend on the environment, and they can be
represented for rural scenario as in [116].
The maximum communication distance, dmax , is computed as the maximum possible
distance such that the received power Pr is higher than the receiver sensitivity Prs , shown
in the first row of Table 3.1. We calculate dmax using the Okumura-Hata discussed above
as
dmax = argmax d s.t. Pr (d) ≜ Pt − Pl (d) ≥ Prs (SF) (3.5)
d∈R
Chapter 4
We present different gateway arrival scenarios to enhance the freshness of data, and we
analyze the optimum wake-up schedule for the end-node in order to guarantee to catch the
presence of the gateway in each scenario as well as to save power. We study the effect of
synchronization on power consumption and throughput.
This scenario assumes perfect synchronization between the start of end-node transmis-
sion and the time the gateway arrives into range. This setup necessitates precise gateway
arrival scheduling or some sort of synchronization method to wake up the end-nodes at
the time of gateway arrival. This maximizes the throughput and minimizes the power
consumption as the end-nodes are awake only during transmission time.
42
However, this synchronized setup can limit the freshness of data performance as data
can only be transmitted during pre-scheduled transmission windows. End-nodes may miss
the transmission if they are not awake at the exact time of transmission, which can result
in a delay in data delivery. To overcome this issue, precise gateway arrival scheduling or
synchronization methods are required to wake up the end-nodes at the time of the gateway
arrival. Despite this limitation, the synchronized scenario is beneficial for applications that
require high throughput and can tolerate some delay in data delivery.
In the case of perfect synchronization, the end-node is assumed to start transmission
once the gateway is within the end-node communication range as illustrated in Fig. 4.1.
The communication starts when the gateway is at a distance of d0 =dmax from the end-
node, as depicted in Fig. 3.1 in the system model. The maximum number of transmitted
packets of a certain payload Kpl is the total number of packets to be transmitted within
the visibility time Tv of a certain gateway speed v while applying the 1% duty cycle (D)
constrain regulated by LoRaWAN standard, i.e., for each transmission, the end-node must
wait time equal to (1 − D) TOA before the next transmission.[44].
The maximum number of transmitted packets in the case of synchronized transmission
(Kpl ST ) is calculated such that
TOA
(Kpl ST − 1) + TOA ≤ Tv (4.1)
D
notice that there is no need to account for the (1 − D) TOA waiting time for the last
transmitted packet. With simple rearrangement of variables and applying floor function,
the expression becomes
ST D(Tv − TOA)
Kpl = +1 (4.2)
TOA
this expression takes into account the total transmission time for each packet TOA as well
as the uplink transmission duty cycle D = 0.01. The value of TOA for a certain payload
length is calculated as in [113].
The left-hand side of the equation calculates the total time needed to transmit all the
packets, taking into account the TOA and the waiting time between packets. The waiting
time is not needed for the last packet, hence the (Kpl ST − 1) factor. If the total time needed
to transmit all the packets is less than or equal to the visibility time (Tv ), then all packets
can be transmitted within the synchronization window.
The equation is then rearranged to solve for (Kpl ST ), which represents the maximum
number of transmitted packets. The floor function is used to round down the result of the
44
division to the nearest integer. This is because the number of transmitted packets must be
a whole number, and cannot be a fractional value.
as:
SS D(Tv − tx − TOA)
Kpl = + 1. (4.3)
TOA
Next, we calculate the expectation of this expression. We find that the output of the
floor function is a discrete set of values rather than a continuous range of values. Since
the average of a discrete set of values is not well-defined, we cannot directly calculate the
average of a floor function. However, as an estimate of the worst case expected value,
we could take the lower bound. And, the average number of transmitted packets can be
expressed as
SS D(Tv − tx − TOA) D(Tv − 0.5Tmax − TOA)
E[Kpl ]≥E = (4.4)
TOA TOA
where E[tx ] = 0.5Tmax . Taking the lower bound of the floor function is necessary in this
case because it ensures that the average number of transmitted packets is at least as large
as the minimum number of packets that could be transmitted. Which means that the
actual number of packets transmitted could be slightly higher. The equation shows that
the average number of packets that can be transmitted increases with the visibility time,
but decreases with the time on air of a packet and the waiting time of each spreading factor
due to the duty cycle constrain.
case, but with Tmax = H × 3600 s. To detect the presence of a gateway, the end-node
continues to open receive windows with fixed size Ton as shown in 3.1 at a rate per hour
Ron along the arrival time frame. The actual transmission starts when the arrival time
resides in one of these slots with a probability. The purpose of this approach is to improve
the availability of gateways for end-nodes in a more realistic scenario where gateway arrivals
are unpredictable. The probability of gateway detection is expressed as:
Ton Ron H for ta ∈ [0, Tmax ]
Tmax
PTX = (4.5)
0
otherwise
with, Ton Ron H ≤ Tmax . The equation takes into account the arrival time ta of the gateway
which is assumed to be completely random and uniformly distributed over a number of
hours H. The probability of gateway detection is then calculated based on the size of the
receive windows, Ton , the rate per hour that the end-node opens the receive windows, Ron ,
and the total time period for gateway arrival.
To minimize the opening rate for power saving multiple gateways are considered. Using
multiple gateways NG increases the probability of gateway detection to
NG Ton Ron H for ta ∈ [0, Tmax ]
Tmax
PTX = (4.6)
0
otherwise
This equation is similar to the first one, but it includes the additional factor of NG in the
calculation. The increase in the number of gateways increases the probability of detec-
tion, which in turn leads to more reliable Communication. Throughput is influenced by
several factors, including the probability of gateway detection, the number of gateways,
and the number of transmission windows. The maximum throughput for non-synchronized
transmission with multiple gateways, denoted as Kpl NS , can be expressed as the product
of the maximum throughput for synchronized transmission (Kpl ST ) and the probability of
successful transmission (PTX ).
The maximum throughput for the non-synchronized case Kpl NS with multiple gateways
can be expressed as
Kpl NS = Kpl ST PTX (4.7)
However, when the presence of a gateway is detected, the transmission can only start from
the beginning of the next transmission window, which means that the actual position of the
47
It’s worth mentioning that this scenario can offer better performance regarding the
freshness of the data [117]. With the use of multiple gateways in a mobile gateway scenario,
the latency is reduced because the end node is constantly searching for the nearest gateway,
resulting in a faster transmission of data. This means that the data is fresher and more up-
to-date, which can be critical for real-time applications such as monitoring environmental
conditions or tracking the location of assets. In contrast to the other scenarios, where the
freshness of data is at its best only for the data generated in the limited time of the gateway
arrival.
48
Figure 4.2: End-node procedure Figure 4.3: End-node procedure in semi-synchronized and
in synchronized transmission non-synchronized transmission
49
The Flow diagram in figure 4.2 shows the transmission procedure at the end-nodes. In
a pre-configured time the end-node wakes up, start its radio preparations such as applying
the configured SF , transmission power level and select an uplink sub-channel. The collected
data is formatted into packets and transmitted to the present gateway. The end-node opens
a receive window and waits for downlink communication for a certain time specified by
LoRaWAN standard. When the time for this receive window experies, the end-node opens
a second one giving the gateway a chance for delayed response before ending transmission.
If a downlink message is received in any of the windows, the end-node immediately starts
its end of transmission sequence and goes into deep sleep mode till its time for the next
schedule transmission.
Figure 4.3 shows the flow diagram for both semi-synchronized and non-synchronized
scenarios. To compromise between having the end-nodes transmitting only in precise win-
dows and, the end-nodes being always active, the end-nodes wake up at a pre-configured
rate per hour to detect the gateway presence. If the gateway is detected, transmission pro-
cedure starts as explained above. If no gateway is detected the end-nodes goes to sleep till
its time for the next wake up window. Semi-synchronized end-nodes go into deep sleep as
soon as transmission is complete or the counter for ping slots expires, and, wake up again
in the scheduled time in which the gateway is expected again. While, non-synchronized
end-nodes continue waking up for ping slots through the day, as long as there are data to
be transmitted.
Table 4.1: Detailed time duration and current consumption values for the power consump-
tion model
Time Duration Current Consumption
State Variable Value (ms) Variable Value (mA)
1 Wake-up TW U 168.2 IW U 22.1
2 Radio Preparation TRP 83.8 IRP 13.3
3 Ping Slot TP S 160.0 IP S 10 × 10− 3
4 Transmission TX Kpl × T OA IX 83.0
5 1st wait window TW 1 983.3 IW 1 27.0
6 1st Rx window TRW SF Dependant IRW 38.1
7 Radio Off TRO 147.4 IRO 13.2
8 Post Processing TP P 268.0 IP P 21.0
9 Turn-Off sequence TT S 38.6 IT S 13.3
10 Sleep TS TON − TActive IS 45 × 10− 3
The equation given represents the calculation of daily energy consumption in a lorawan
network for data collection. The total daily energy consumption is calculated using Equation
4.9, which is the sum of the power consumed by four different states of operation: wake-up,
search, transmit, and receive. The power consumption for wake-up and sleep states is given
by EOn−Off .
The power consumption for wake-up and sleep states EOn−Off is repeated depending
on the frequency of data collection. It can be further broken down into four components:
TWU , IWU for wake-up , TRP , IRP for radio preparation , TTS , ITS for turn off sequence, and
TS , IS for sleep state.
N
X
EOn−Off = V (TWU IWU + TRP IRP + TTS ITS + TS IS ) (4.10)
n=1
where V is the input voltage usually equals 3.3 V, N is the number of data collection times
per day, and Ix is the drained current for a certain state for time duration Tx . Table 4.1
lists the detailed parameters of the energy consumption model used in this work, with more
details to be found in [118].
ESearch is the energy consumed by the end-node while searching for the gateway presence,
this state only occurs in the semi-synchronized and non-synchronized scenarios, and its
51
where, ESearch = 0 for the synchronized transmission scenario and, H = 1 for the semi-
synchronized scenario.
ETransmit is the energy consumed for data transmission, which is dependent on the used
SF and the number of transmitted packets Kpl . While EReceive is the energy consumed
while opening the two receive windows and processing downlink messages.
N
X
ETransmit = V (TX (Kpl , SF) IX + TRO IRO + TPP IPP ) (4.12)
n=1
The summation limit for each state depends on the transmission model and the number
of trails, as shown in Figure 4.2. The transmission procedure is straightforward, each state
occurs once, and the main factor in the energy consumption is the transmission state, which
depends on the used SF, and the number of packets transmitted. However, in Figure 4.3,
the process of gateway search consumes a significant amount of energy, which increases with
high Ron values or low probability of detection. Moreover, there is a worst-case scenario
where the gateway is not detected. This dramatically increases the energy consumption
values for semi-synchronized ESS and nons-ynchronized transmissions ENS .
52
Chapter 5
In this section, we validate our analysis by the means of three different methods: (i)
a theoretical performance analysis is conducted; (ii) extensive simulation for the proposed
system with different gateway arrival scenarios, i.e., synchronization conditions.The simu-
lation results are discussed in terms of throughput and power consumption; (iii) real-world
experiment is carried out to assess the effect of gateway mobility on throughput. Further-
more, to test the practicability of our proposed scheme, we propose a case study in which
we test the proposed system in the Wadi El-Gemal area by running ns-3® simulations.
In this experiment, we validate our analysis of the effect of gateway mobility on the
LoRaWAN communication link and the total number of correctly received packets. We
compare the number of received packets recorded in the actual experiment setup to the
numbers indicated by both analytical results calculated via (4.2) and simulation results via
ns-3 simulation. The number of correctly received packets is calculated as the number of
successfully decoded packets at the gateway in both the real and simulated experiments.
However, in the analytical results, the number of received packets considered is the num-
ber of received packets whose received signal strength (RSSI) is greater than the receiver
sensitivity for each SF.
53
Payload Preamble Coding Rate Low Data Rate Optimization Header Mode
25 Byte 8 Byte 4/5 DE=0 (disabled) H=0 (explicit)
Path loss Model Parameters
Fig. 5.1 shows the real experiment setup where the left image shows the gateway which
is placed on top of a car traveling down the road parallel to Suez Canal University campus in
Ismailia, Egypt presented in the middle image, while the image on the right shows a screen-
shot of the received packets as they appear on the Chirpstack server® . The gateway is built
using SX1301, Semtech™ transceiver. The SX1301 transceiver is controlled by Raspberry Pi
II, which also acts as a LoRaWAN network server and data storage. The Raspberry Pi is
later connected via IP protocol to the application server created on Chirpstack® . For the
end-node, module SX1872 is used, with 868 MHz antenna of gain 2.5 dBi. The end-node
is placed on the ground alongside an empty road with a clear line of sight to the gateway
at an initial distance of 2.5 km. The end-node transmits 25 packets using each SF to the
gateway. The gateway is moving inward with speed v ∈ [60 : 120] Km/h and continues
until the node is out of range.
The same setup used in the real experiment was applied in simulations. The simula-
tion setup parameters for the MATLAB and ns-3® network simulations, such as antenna
heights, gateway speed, initial distance, and packet configurations, are listed in Table 5.1.
The Okumura–Hata model was used as a propagation model in all simulations. For ns-3
simulations, C++ scripts were written to configure the LoRaWAN module and mobility
module. The total number of symbols in an LoRaWAN frame after coding and the corre-
sponding TOA are calculated as in [45].
where:
Here, Tsym is the symbol time, SF is the spreading factor, CRC is the cyclic redundancy
check, MHDR is the MAC header, length is the length of the payload, payload is the payload
size, and nPreamble is the number of preamble symbols.
Real experiment data, simulation data in ns-3 simulator, and analytical results are
compared in Figure 5.2. The figure depicts the maximum number of packets received in
real-world experiments and simulations, as well as the data from analytical expressions at
various spreading factor values and a gateway speed of 90 km/h. The results show that
the data from the experiment and simulation matched the expected analytical performance.
The analytical data act as a lower bound for the remaining two curves. We notice that the
level of received packets is constant for SFs 7 and 8 on 25 packets. This is due to the fact
that the upper bound for possible received packets using these spreading factors is much
higher than the 25 packets transmitted in this test. Furthermore, we find that for higher
spreading factors, the number of received packets is reduced due to the limited visibility
time and longer TOA values. We notice that for higher spreading factors, the data recorded
in the real experiment are less than the analytical calculations, as expected, because the
symbol time exceeds the channel coherence time.
55
Analytical (4.2)
25 Simulation
Experiment
15
10
0
7 8 9 10 11 12
Spreading Factor
Figure 5.2: Number of received packets at speed = 90 km/h for different SF values
Furthermore, we studied the effect of increasing the transmitted power, and gateway
height on the visibility time and the total throughput. Figure 5.3 shows the change in
throughput in Kbps against increasing the transmitted power, using antenna gain of 2dBi
and gateway speed of 100km/h. When transmitting at a higher power, the signal can travel
further and with less attenuation, which can result in a higher signal-to-noise ratio (SNR) at
the receiver. This improved SNR can help to reduce packet loss and increase the reliability
of the transmission. However, we find that increasing transmission parameters would not
linearly result in increasing of throughput. Thus, keeping the transmitted power at the
minimum value to achieve acceptable throughput is better for maintaining the battery life
of the end nodes.
56
SF =7
SF =8
SF =9
SF = 10
SF = 11
SF = 12
Throughput (Kbps)
101
100
14 15 16 17 18 19 20 21 22
Transmitted Power (dBm)
Figure 5.3: Throughput at speed 100 Km/h, antenna gain 2dBi and, different SF values
Similarly, increasing the height of the gateway can improve the coverage area and reduce
the effects of obstructions and multi-path fading. This can also result in a higher SNR and
improved reception of transmissions. Especially for higher spreading factors, which may
benefit more from improved signal quality. On the other hand, for lower spreading factors,
the impact may be less significant since these transmissions already have a higher data
rate and shorter time on air. Figure 5.4 shows the change in throughput against increasing
the antenna height with transmitted power of 14dBm, antenna gain of 2dBi and gateway
speed of 100km/h. We find that the gain of larger coverage and hence larger visibility time
are wasted by the transmission duty cycle constraint and the throughput isn’t effected as
required.
57
SF =7
SF =8
SF =9
SF = 10
SF = 11
SF = 12
Throughput (Kbps)
101
100
2 3 4 5 6 7 8 9 10
Antenna Height (m)
Figure 5.4: Throughput at speed 100 Km/h, Pt =14 dBm, Gain=2dBi and, different SF
values
We continue the analysis of the three gateway arrival scenarios by running MATLAB
simulation to investigate the effect of higher speeds on the visibility time, the resulting
throughput, and energy consumption at the end-node. The throughput is presented as the
number of received packets. For fairness of comparison, we test the performance for all
spreading factors with a payload length of 25 bytes. The setup for all experiments and
simulations is that the gateway height is 2.5m whereas the end-nodes are lying on the
ground level as depicted in 3.1.
Figure 5.5 depicts the number of received packets within the visibility time for each
spreading factor at various speeds. The graph shows that in the scenario of precise synchro-
nization, gateway mobility has no effect on LoRaWAN’s expected performance. The only
advantage of using higher SFs is increased coverage. Even though a longer coverage distance
58
means more visibility time, the transmission duty cycle constraint wastes this extra time.
The number of packets received at each speed demonstrated is considered the upper bound
for received packets at the same speed in later scenarios. For the remainder of this section,
we will conduct our analysis with a fixed speed of 80 km/h as an average vehicle speed.
80 SF =7
SF =8
SF =9
SF = 10
Number of Received Packets
SF = 11
60
SF = 12
40
20
0
60 70 80 90 100 110 120
Speed (Km/h)
Figure 5.5: Number of received packets within effective visibility time of different speeds
and SF values in the case of perfect synchronization
Figure 5.6 shows the energy consumption relative to the throughput shown in Fig-
ure 5.5. As expected, the higher SFs consume more energy even though they transmit
fewer packets due to their higher range and longer TOA, as explained earlier in Section ??.
The selection of SFs transmission value must be optimized to meet a certain coverage dis-
tance, throughput limits, and minimum power consumption level requirements, especially
for semi-synchronized and non-synchronized transmission, as discussed below.
Energy consumption values are normalized to a reference value, which is the energy
of transmitting six packets using SF = 12 with precise synchronization. The energy of
transmitting using SF = 12 with a gateway speed of 80 km/h is Eref = 1.674J for each
transmission window. This normalization will also be used for later scenarios for ease of
comparison. For better presentation, for the rest of this section, we concentrate on the
59
shortest SF = 7, the longest SF = 12, and the two in the middle. The missing spreading
factors behave similarly within their maximum received packet limits.
SF = 12
1
SF = 11
SF = 10
0.9
Normalized Energy Consumption
SF =9
SF =8
0.8 SF =7
0.7
0.6
0.5
0.4
0.3
Figure 5.6: Normalized energy consumption of packets transmission for different speeds and
SF values in the case of perfect synchronization
The end-node opens repeated ping slots, searching for the gateway at a rate Ron per
hour. We start with 4 windows per hour and increase up to 30 windows per hour. The num-
ber of windows are selected to guarantee gateway detection. The maximum synchronization
delay that could be endured to detect the gateway presence is around 15 minutes using
SF = 12, and assuming a gateway speed of 60 km/h or less. We compare the analytical
expression for the average number of received packets within effective visibility time (4.4)
against the simulation results of 10,000 instances in Figure 5.7.
Figure 5.7 shows the average number of received packets for different spreading factors
and different Ron values at a gateway speed of 80 km/h. In order to investigate the limits
on the system number of transmitted packets, we also investigate the worst-case scenario,
where the gateway is only detected at the last ping slot, in this case, the end-nodes transmit
60
the minimum number of packets. We find that the performance of the average throughput
values for each SF comply with the synchronized transmission behavior; however, the worst-
case scenario throughput of SF = 7 drops for an opening rate of fewer than 15 windows per
hour. We find that using SF = 9 with Ron = 15 windows per hour is optimal for coverage
range, guaranteed throughput, and power consumption.
30
Expected Value (4.4)
Simulated Value
25 Worst Case
Average Recieved Packets
20
15
10
4 8 12 16 20 24 28 30
Transmission Windows Rate Ron
Figure 5.7: Average and worst-case number of received packets of different SF and Ron
values at speed=80 Km/h
Figure 5.8 shows the average normalized energy consumption for each Ron for SFs =
7, 9, 10, and 12 corresponding to the average throughput values in Figure 5.7. We find
that the normalized energy consumption using SF = 7 and, Ron = 30 windows per hour is
0.48 for transmitting only 28 packets, which is fewer than half the number of transmitted
packets with almost the same energy consumption in the synchronized scenario. The same
performance is repeated for other spreading factors.
61
SF = 12
0.8
SF = 10
0.6
0.4
0.2
0
4 8 12 16 20 24 28 30
Transmission Windows Rate Ron
Figure 5.8: Average Normalized Energy consumption at the end-node for different SF and
Ron values at speed=80 Km/h
Similar to the previous case, the end-nodes open ping slots at the same rate per hour
Ron ; however, the search for the gateway is repeated over several hours. We propose using
multiple gateways for data collection to increase the gateway availability and enhance the
freshness of data. First, we investigate the transmission with a single gateway.
Figure 5.9 shows the average number of received packets for different SFs and, Ron
values. We compare the performance of semi-synchronized transmission to that of non-
synchronized transmission with a single gateway. We notice the effect of no synchronization
on the decrease in the number of received packets. To achieve the same values for average
received packets of the previous scenario using a single gateway, we need to operate at
Ron = 30 windows per hour or more, at the cost of energy consumption at the end-nodes
rising to up to 17 multiples for SF = 7 and 15 multiples for SF = 12, as shown in Figure 5.10.
62
30
SF = 7 (Kpl SS )
SF = 7 (Kpl NS )
SF =9
25 SF = 10
Average Received Packets SF = 12
20
15
10
4 8 12 16 20 24 28 30
Transmission Windows Rate per Hour Ron
Figure 5.9: Average and worst-case number of received packets of different SF and Ron
values at speed=80 Km/h and single gateway
63
SF = 12
12
SF = 10
0
4 6 8 10 12 14 16 18 20 22 24 26 28 30
Transmission Windows Rate per Hour Ron
Figure 5.10: Average normalized energy consumption for different SF and Ron values at
speed=80 Km/h and single gateway
Using multiple gateways can significantly enhance the system’s performance in terms of
throughput, energy consumption by end-nodes, and freshness of data, as the end-nodes will
detect the gateway presence more frequently. Figure 5.11 presents the effect of increasing
the number of gateways on the average number of received packets for different SFs at
gateway speed of 80 km/h and Ron = 8 windows per hour. The analytical values are
compared to the results of a MATLAB simulation of multiple gateway arrivals. Figure 5.12
shows the effect of increasing the number of gateways on the energy consumption of the
end-node. We observe an increase in energy consumption due to more transmitted packets
as the probability of gateway detection increases.
64
40
30
20
10
1 2 3 4 5 6 7 8 9 10
Number of Gateways
Figure 5.11: Average number of received packets for different SF values and multiple gate-
ways at speed=80 Km/h and Ron = 8windows/hr
Using four gateways increases the throughput achieved via SF = 7 to four times the
value achieved in the semi-synchronized scenario for the same parameters of gateway speed
and Ron value. The number of received packets increases from 7 to 28 packets, with only five
times the energy consumption when compared to using a single gateway. Data collection
using multiple gateways with no transmission synchronization between the gateway and end-
nodes can approach synchronized transmission performance. The synchronized transmission
throughput of SF = 7 is achieved by using ten gateways and only eight multiples of the
synchronized transmission power consumption, while it would require 40 multiples of power
consumption using a single gateway.
65
8 SF = 12
SF = 10
1 2 3 4 5 6 7 8 9 10
Number of Gateways
Figure 5.12: Average energy consumption at the end-node for different SF values and
multiple gateways at speed=80 Km/h and Ron = 8windows/hr
In this section, we present a case study of Wadi El-Gemal national park in Egypt,
a coastal area along the Red Sea in southeast Egypt (24.4747° N, 35.0987° E). We
simulate the area with the ns-3® simulator, distribute the end-nodes throughout the area,
and preconfigure their transmission parameters in terms of SF allocation and transmission
schedule to avoid collisions. We evaluate data collection performance using the synchronized
transmission model.
The area shown in Figure 5.13 is 58 × 64 km2 area. Based on our analysis, we conclude
that transmitting with SF = 7 is optimal for end-node throughput and power consumption.
Thus, we aim to use this spreading factor in most of the end-nodes. We intend to divide
the area in such a way that there is no overlap in transmission time between successive
nodes. The maximum communication distance using SF = 7 is 7.4 km, as stated earlier in
Table 3.1. As a result, we decided to divide the area into 10 × 10 km2 subareas to ensure
sufficient distance between the end-nodes operating with SF = 7.
66
Each subarea is thought of as a cluster of four end-nodes. To cover the entire area, we
use 146 end-nodes. The 146 distributed end-nodes and their associated spreading factor
are represented by colored dots. The allocation of SFs takes into account two factors: first,
the distance between the end-node and the gateway, and second, minimizing interference
between successive end-nodes. The simulation parameters are the same values listed in
Table 5.1.
The gateway collects data from one or two clusters as it passes by, depending on the
location of the road. For example, the gateway collects data from four end-nodes along the
circumferential road within the same visibility time. The two end-nodes closest to the road
have spreading factors of seven and eight, respectively, while the two farthest end-nodes
have spreading factors of 11 and 12. Whilst two clusters—one on each side—will also be
transmitting at the same time along the vertical unpaved routes running across the middle
area, spreading factors 7, 8, 9, and 10 were used alternately for the near end-nodes, while
67
11 and 12 were kept for the farthest end-nodes as the shorter ranges of lower spreading
factors will not be sufficient to reach the gateway.
The simulation implements the synchronized transmission scenario, in which all end-
nodes transmit continuously throughout the visibility time. The mobile gateway begins at
position zero, with an initial distance of 2.5 km from the first end-node. We assume having
multiple gateways, with only one gateway passing by each node at speeds of 80 km/h.
One gateway follows the circumferential road, while three others follow the middle routes,
and each end-node transmits six packets of 25 bytes payload. Simulation results showed
100% packet delivery rate for all circumferential end-nodes and all end-nodes operating in
SFs 10 and 11, while all end-nodes operating in SF11, as well as some of the end-nodes in
the middle area, achieved PDR 83.33% due to one packet loss.
SF = 7
80 SF = 8
SF = 9
Average Number of Received Packets
SF = 10
SF = 11
SF = 12
60 Synchronized Transmission
40
20
0
60 70 80 90 100 110 120
Speed (Km/h)
Figure 5.14: Average number of received packets per end-node within effective visibility
time of different speeds and SF values.
To further test our setup, we repeat the simulation with different gateway speeds while
setting the end-nodes to transmit continuously until the gateway is out of range. Fig-
ure 5.14 shows the average number of packets that the mobile gateway could receive from
each end-node operating in a certain SF. We notice that the results are close to those
68
of precise synchronized transmission shown in Figure 5.5, especially at lower speeds, and
as the gateway speed increases, the visibility time decreases and the number of collisions
increases.
69
Chapter 6
In this work, we proposed an architecture for IoT data collection in monitoring water
distribution networks via mobile LoRaWAN fog nodes. Our model simplifies the complexity
and cost of systems monitoring to sensor nodes only, as the mobility of the gateway will
easily extend the network coverage compared with adapting a fixed network infrastructure
approach. We analyzed the effects of gateway mobility and studied transmission limitations.
The proposed approach is evaluated through a series of experiments, including simulation
and real-world testing.
The system utilizes the long-range and low-power properties of LoRaWAN communica-
tion and gateway mobility to extend the coverage with as little effect on the ecosystem as
possible. The effects of gateway speed on the signal in terms of Doppler shift, range, and
visibility time are discussed. Performance analysis of the proposed system is presented, and
three gateway arrival models were investigated.
From this analysis we conclude that for perfect synchronization scenario and as long
as the distance between end-nodes and gateway is within 3.5 Km radius, it is advised to
use spreading factor 7 for achieving maximum throughput. As the gain of larger coverage
and hence larger visibility time are wasted by the transmission duty cycle constrain. Also,
increasing the transmitted power or investing in a more powerful antenna might not be fully
rewarding as the gained extra coverage mostly lies within the waiting period as well, and
keeping the transmitted power at the minimum value to achieve acceptable throughput is
better for maintaining the battery life of the end nodes.
The three models of gateway arrival emphasize the trade-off between energy consump-
tion and synchronization. Minimum energy consumption necessitates complete synchroniza-
70
tion between end-nodes and gateway, which may necessitate complex transmission schedul-
ing. While adopting a non-synchronized transmission model would allow for more flexible
transmission and a higher level of data freshness, it would also require more energy.
The results of our work show that the mobility of the LoRaWAN gateway limits the
throughput corresponding to each SF due to limited gateway availability and transmission
time. For synchronized transmission scenarios, using lower SFs provides better throughput;
however, for semi-synchronized and non-synchronized transmission, higher SFs could ensure
better performance as the longer visibility time allows for a higher probability of gateway
detection. Data collection with synchronized transmission achieves almost 50% less power
consumption compared to the semi-synchronized transmission for the same number of re-
ceived packets at gateway speed of 80 km/h.
Using multiple gateways with a non-synchronized transmission setup enhances the fresh-
ness of data and simplifies transmission scheduling at the end-nodes. Non-synchronized
transmission with a single gateway consumes 15 times more energy than the semi-synchronized
scenario for SF = 12 and 17 times more for SF = 7. However, it can achieve the same
throughput as that of a perfect synchronization scenario at the expense of eight times more
energy consumption using 10 gateways for SF = 7 and eight gateways for SF = 12.
The proposed system’s ability to adapt to different network configurations and its cost-
effectiveness make it a promising solution for water monitoring in various contexts. However,
further investigation of the optimum selection of LoRaWAN signal configuration is recom-
mended. For future work, we will consider a more general case study on the implementation
of our system and the optimal end-nodes placements, the routing of the mobile gateway,
as well as a data collection scheme for maximizing the sensing area and minimizing energy
consumption.
71
Bibliography
[2] S. Yi, Z. Hao, Z. Qin, and Q. Li, “Fog computing: Platform and applications,” in
2017 Third IEEE workshop on hot topics in web systems and technologies (HotWeb).
IEEE, 2017, pp. 73–78.
[3] L. Alliance, “LoRaWAN™ 1.0.4 specification,” LoRa Alliance San Ramon,CA, USA,
Tech. Rep., 2020.
[4] W. Bank, Water loss in distribution systems. World Bank Publications, 2019.
[7] S. Sobhi, M. A. Ali, and M. F. Abdelkader, “Combining fog computing and lorawan
technologies for smart cities applications,” 2019.
[8] Z. Liu, X. Zhao, T. Liu, F. Xue, X. Cheng, and T. Zhang, “Real-time water loss moni-
toring using machine learning and pressure sensors,” IEEE Transactions on Industrial
Informatics, vol. 16, no. 3, pp. 1757–1765, 2020.
72
[9] S. Dang, O. Amin, B. Shihada, and M.-S. Alouini, “What should 6g be?” Nature
Electronics, vol. 3, no. 1, pp. 20–29, 2020.
[10] A. Munir, P. Kansakar, and S. U. Khan, “Ifciot: Integrated fog cloud iot: A novel
architectural paradigm for the future internet of things.” IEEE Consumer Electronics
Magazine, vol. 6, no. 3, pp. 74–82, 2017.
[11] H. Yar, A. S. Imran, Z. A. Khan, M. Sajjad, and Z. Kastrati, “Towards smart home
automation using iot-enabled edge-computing paradigm,” Sensors, vol. 21, no. 14, p.
4932, 2021.
[14] F. Muteba, K. Djouani, and T. Olwal, “A comparative survey study on LPWAN IoT
technologies: Design, considerations, challenges and solutions,” Procedia Computer
Science, vol. 155, pp. 636–641, Jan 2019.
[15] L. Alliance, “A technical overview of LoRa and LoRaWAN; white paper,” The LoRa
Alliance, San Ramon,CA, USA, Tech. Rep., 2015.
[16] W. Ayoub, A. E. Samhat, F. Nouvel, M. Mroue, and J.-C. Prévotet, “Internet of mo-
bile things: Overview of lorawan, dash7, and nb-iot in lpwans standards and supported
mobility,” IEEE Communications Surveys & Tutorials, vol. 21, no. 2, pp. 1561–1581,
2018.
[22] Ç. Dilibal, “Development of edge-iomt computing architecture for smart healthcare
monitoring platform,” in 2020 4th International Symposium on Multidisciplinary
Studies and Innovative Technologies (ISMSIT). IEEE, 2020, pp. 1–4.
[23] S. Naveen and M. R. Kounte, “Key technologies and challenges in iot edge computing,”
in 2019 Third international conference on I-SMAC (IoT in social, mobile, analytics
and cloud)(I-SMAC). IEEE, 2019, pp. 61–65.
[26] N. Ahmed, D. De, and I. Hussain, “Internet of things (iot) for smart precision agri-
culture and farming in rural areas,” IEEE Internet of Things Journal, vol. 5, no. 6,
pp. 4890–4899, 2018.
74
[28] L.-Y. Chen, H.-S. Huang, C.-J. Wu, Y.-T. Tsai, and Y.-S. Chang, “A LoRa-based
air quality monitor on unmanned aerial vehicle for smart city,” in 2018 International
Conference on System Science and Engineering (ICSSE), 2018, pp. 1–5.
[32] M. Chiani and A. Elzanaty, “On the LoRa modulation for IoT: Waveform properties
and spectral analysis,” IEEE Internet of Things Journal, vol. 6, no. 5, pp. 8463–8470,
2019.
[36] R. Tawfik and M. Sarhan, “Ecotourism and protected areas sustainable financing:
A case study of wadi el gemal visitor center,” Journal of Spatial and Organizational
Dynamics, vol. 9, no. 2, pp. 156–172, 2021.
[38] H. Rahman and M. I. Hussain, “Fog-based semantic model for supporting interoper-
ability in iot,” IET Communications, vol. 13, no. 11, pp. 1651–1661, 2019.
[40] Y. Wu, “Cloud-edge orchestration for the internet of things: Architecture and ai-
powered data processing,” IEEE Internet of Things Journal, vol. 8, no. 16, pp. 12 792–
12 805, 2020.
[43] M. Chiani and A. Elzanaty, “On the LoRa modulation for IoT: Waveform properties
and spectral analysis,” IEEE Internet of Things Journal, vol. 6, no. 5, pp. 8463–8470,
2019.
[44] L. Alliance, “A technical overview of LoRa and LoRaWAN; white paper,” The LoRa
Alliance, San Ramon,CA, USA, Tech. Rep., 2015.
[46] C. I. Ezeugwu, J. Ogbu, and C. C. Ngwakwe, “Towards effective data collection for
agricultural research and development in nigeria,” International Journal of Agricul-
tural and Environmental Information Systems (IJAEIS), vol. 11, pp. 67–79, 2020.
[49] A. Mehto, S. Tapaswi, and K. Pattanaik, “A review on rendezvous based data ac-
quisition methods in wireless sensor networks with mobile sink,” Wireless Networks,
vol. 26, pp. 2639–2663, 2020.
[50] X. Wu, Y. Li, L. Guo, X. Liu, L. Wang, Z. Jiang, and S. Zhang, “Mobility-assisted
data collection for rural water management using lorawan and fog computing,” IEEE
Access, vol. 8, pp. 180 013–180 023, 2020.
[51] L. Guo, Y. Li, X. Wu, S. Zhang, and F. Li, “Mobile healthcare services for rural areas:
Opportunities and challenges,” International Journal of Environmental Research and
Public Health, vol. 18, no. 9, p. 4683, 2021.
[52] M. S. Khan, F. K. Khan, S. Ullah, Y. Liaqat, and M. Khalid, “An overview of satel-
lite communication in precision agriculture,” Wireless Communications and Mobile
Computing, vol. 2020, p. 8821517, 2020.
[53] B. Pourghebleh and N. Jafari Navimipour, “Towards efficient data collection mech-
anisms in the vehicular ad hoc networks,” International Journal of Communication
Systems, vol. 32, no. 5, p. e3893, 2019.
[54] A. Hassan, N. Javaid, and Z. A. Khan, “A comparative analysis of zigbee and blue-
tooth low energy for indoor building automation,” Journal of Network and Computer
Applications, vol. 122, pp. 1–13, 2019.
77
[56] A. Hidayati, M. I. Nashiruddin et al., “Lpwa-based iot technology selection for smart
metering deployment in urban and sub urban areas: A state electricity company
perspective,” Buletin Pos dan Telekomunikasi, vol. 18, no. 2, pp. 75–94, 2020.
[59] R. El Chall, S. Lahoud, and M. El Helou, “Lorawan network: Radio propagation mod-
els and performance evaluation in various environments in lebanon,” IEEE Internet
of Things Journal, vol. 6, no. 2, pp. 2366–2378, 2019.
[66] J. Dias and A. Grilo, “Multi-hop lorawan uplink extension: specification and proto-
type implementation,” Journal of Ambient Intelligence and Humanized Computing,
vol. 11, pp. 945–959, 2020.
[69] T. Attia, M. Heusse, and A. Duda, “Message in message for improved lorawan capac-
ity,” in 2021 International Conference on Computer Communications and Networks
(ICCCN). IEEE, 2021, pp. 1–9.
[73] H.-Y. Miao, C.-T. Yang, E. Kristiani, H. Fathoni, Y.-S. Lin, and C.-Y. Chen, “On
construction of a campus outdoor air and water quality monitoring system using
lorawan,” Applied Sciences, vol. 12, no. 10, p. 5018, 2022.
[75] Y. Lalle, M. Fourati, L. C. Fourati, and J. P. Barraca, “Routing strategies for lorawan
multi-hop networks: A survey and an sdn-based solution for smart water grid,” IEEE
Access, vol. 9, pp. 168 624–168 647, 2021.
[80] A. Aneiba, B. Nangle, J. Hayes, and M. Albaarini, “Real-time iot urban road traffic
data monitoring using lorawan,” in Proceedings of the 9th International Conference
on the Internet of Things, 2019, pp. 1–8.
80
[83] N. Cruz, N. Cota, and J. Tremoceiro, “Lorawan and urban waste management—a
trial,” Sensors, vol. 21, no. 6, p. 2142, 2021.
[85] S. Nasim, M. Oussalah, T. Outila, and J. Jutila, “Smart parking system with place-
pod, lorawan iot sensors and android app,” in 2022 IEEE 23rd International Confer-
ence on Information Reuse and Integration for Data Science (IRI). IEEE, 2022, pp.
278–284.
[88] S.-Y. Wang, J.-E. Chang, H. Fan, and Y.-H. Sun, “Performance comparisons of nb-
iot, lte cat-m1, sigfox, and lora moving at high speeds in the air,” in 2020 IEEE
Symposium on Computers and Communications (ISCC), 2020, pp. 1–6.
[89] G. Hristov, J. Raychev, D. Kinaneva, and P. Zahariev, “Emerging methods for early
detection of forest fires using unmanned aerial vehicles and lorawan sensor networks,”
in 2018 28th EAEEIE annual conference (EAEEIE). IEEE, 2018, pp. 1–9.
81
[90] M. Pan, C. Chen, X. Yin, and Z. Huang, “Uav-aided emergency environmental mon-
itoring in infrastructure-less areas: Lora mesh networking approach,” IEEE Internet
of Things Journal, vol. 9, no. 4, pp. 2918–2932, 2021.
[92] M. Stellin, S. Sabino, and A. Grilo, “Lorawan networking in mobile scenarios using a
wifi mesh of uav gateways,” Electronics, vol. 9, no. 4, p. 630, 2020.
[98] J. Liu, J. Wu, and M. Liu, “Uav monitoring and forecasting model in intelligent traffic
oriented applications,” Computer Communications, vol. 153, pp. 499–506, 2020.
[99] J.-M. Martinez-Caro and M.-D. Cano, “Iot system integrating unmanned aerial vehi-
cles and lora technology: A performance evaluation study,” Wireless Communications
and Mobile Computing, vol. 2019, 2019.
82
[100] Y. Chen, S. Qiao, G. Zhang, Y. J. Xu, L. Chen, and L. Wu, “Investigating the
potential use of sentinel-1 data for monitoring wetland water level changes in china’s
momoge national nature reserve,” PeerJ, vol. 8, p. e8616, 2020.
[104] K. Kousias, G. Caso, Ö. Alay, and F. Lemic, “Empirical analysis of lorawan adaptive
data rate for mobile internet of things applications,” in Proceedings of the 2019 on
Wireless of the Students, by the Students, and for the Students Workshop, 2019, pp.
9–11.
[105] J. Jalowiczor, J. Rozhon, and M. Voznak, “Study of the efficiency of fog computing
in an optimized lorawan cloud architecture,” Sensors, vol. 21, no. 9, p. 3159, 2021.
[108] N. Ahmed, D. De, and I. Hussain, “Internet of things (iot) for smart precision agri-
culture and farming in rural areas,” IEEE Internet of Things Journal, vol. 5, no. 6,
pp. 4890–4899, 2018.
83
[109] R. Litoriya, A. Gulati, M. Yadav, R. S. Ghosh, and P. Pandey, “Social, ethical, and
regulatory issues of fog computing in healthcare 4.0 applications,” in Fog computing
for healthcare 4.0 environments. Springer, 2021, pp. 593–609.
[110] R. Jain, M. Gupta, A. Nayyar, and N. Sharma, “Adoption of fog computing in health-
care 4.0,” in Fog computing for healthcare 4.0 environments. Springer, 2021, pp. 3–36.
[111] C. Chang, A. Hadachi, J. Mass, and S. N. Srirama, “Mobile fog computing,” Fog
Computing: theory and practice, pp. 1–41, 2020.
[114] L. Alliance, “LoRaWAN™ 1.0.4 specification,” LoRa Alliance San Ramon,CA, USA,
Tech. Rep., 2020.
[115] W. Oelers and H. Naumann, “Mobile Systems IV,” Triptec HL UG, Lübeck,Germany,
Tech. Rep., 2021.
[118] L. Casals, B. Mir, R. Vidal, and C. Gomez, “Modeling the energy performance of
lorawan,” Sensors, vol. 17, no. 10, p. 2364, 2017.