sensors

Article
Multiple Mobile Sinks for Quality of Service Improvement
in Large-Scale Wireless Sensor Networks
Abdelbari Ben Yagouta 1, * , Bechir Ben Gouissem 1 , Sami Mnasri 2,3, * , Mansoor Alghamdi 2 ,
Malek Alrashidi 2 , Majed Abdullah Alrowaily 4 , Ibrahim Alkhazi 2 , Rahma Gantassi 5
and Salem Hasnaoui 1

1 Communication System Laboratory (Sys’Com), National Engineering School of Tunis (ENIT), University of
Tunis El Manar (UTM), Tunis 1002, Tunisia; bechir.gouissem@enit.rnu.tn (B.B.G.);
salem.hasnaoui@enit.rnu.tn (S.H.)
2 Computer Science Department, Applied College, University of Tabuk, Tabuk 71491, Saudi Arabia;
malghamdi@ut.edu.sa (M.A.); mqalrashidi@ut.edu.sa (M.A.); i.alkhazi@ut.edu.sa (I.A.)
3 IRIT (RMESS), University of Toulouse II, 31058 Toulouse, France
4 Department of Computer Science, College of Computer and Information Sciences, Jouf University,
Sakaka 72341, Saudi Arabia; malrowaily@ju.edu.sa
5 Department of Electrical Engineering, Chonnam National University, Gwangju 61186, Republic of Korea;
rahmag@jnu.ac.kr
* Correspondence: abdelbari.benyagouta@enit.utm.tn (A.B.Y.); smnasri@ut.edu.sa (S.M.)

Abstract: The involvement of wireless sensor networks in large-scale real-time applications is ex-
ponentially growing. These applications can range from hazardous area supervision to military
applications. In such critical contexts, the simultaneous improvement of the quality of service and the
network lifetime represents a big challenge. To meet these requirements, using multiple mobile sinks
can be a key solution to accommodate the variations that may affect the network. Recent studies
were based on predefined mobility models for sinks and relied on multi-hop routing techniques.
Besides, most of these studies focused only on improving energy consumption without considering
Citation: Ben Yagouta, A.; Ben QoS metrics. In this paper, multiple mobile sinks with random mobile models are used to establish a
Gouissem, B.; Mnasri, S.; Alghamdi, tradeoff between power consumption and the quality of service. The simulation results show that
M.; Alrashidi, M.; Alrowaily, M.A.;
using hierarchical data routing with random mobile sinks represents an efficient method to balance
Alkhazi, I.; Gantassi, R.; Hasnaoui, S.
the distribution of the energy levels of nodes and to reduce the overall power consumption. Moreover,
Multiple Mobile Sinks for Quality of
it is proven that the proposed routing methods allow for minimizing the latency of the transmitted
Service Improvement in Large-Scale
data, increasing the reliability, and improving the throughput of the received data compared to recent
Wireless Sensor Networks. Sensors
2023, 23, 8534. https://doi.org/
works, which are based on predefined trajectories of mobile sinks and multi-hop architectures.
10.3390/s23208534
Keywords: cluster-based routing protocol; energy consumption; quality of service; multiple mobile
Academic Editors: Omprakash
sinks; large scale wireless sensor network
Kaiwartya, Sushil Kumar and
Upasana Dohare

Received: 4 September 2023

Revised: 27 September 2023 1. Introduction
Accepted: 29 September 2023 WSN (Wireless Sensor Networks) is a special case of Ad hoc networks [1], broadly
Published: 18 October 2023
used in various applications such as environment monitoring, object tracking, military
surveillance, traffic control, healthcare, etc. A WSN is a collection of large numbers of
sensor nodes (SN) distributed over a geographic area to monitor certain phenomena. Each
Copyright: © 2023 by the authors.
sensor node is limited in processing capability, wireless bandwidth, battery, and memory
Licensee MDPI, Basel, Switzerland. capacity. Mostly, it is difficult, even impossible, to recharge or change the battery, making
This article is an open access article energy consumption a significant constraint of WSNs lifetime [1,2].
distributed under the terms and WSNs have many advantages, and they are widely used due to their low cost, wireless
conditions of the Creative Commons communication capability, energy efficiency, and scalability, and they are suitable for
Attribution (CC BY) license (https:// Real-time monitoring applications.
creativecommons.org/licenses/by/ The SNs can sense, process, and transmit data either via multi-hop transmission or
4.0/). directly to a base station (BS). The BS sends the collected data to a remote-control station

Sensors 2023, 23, 8534. https://doi.org/10.3390/s23208534 https://www.mdpi.com/journal/sensors

Sensors 2023, 23, 8534 2 of 23

through radio networks or satellite connections. WSNs have unique features like autonomy,
self-organization, and Ad-hoc infrastructure, which makes them ideal for healthcare, smart
cities, and environmental surveillance [2,3].
Since wireless communication requires significantly more power than other tasks,
energy conservation is important while designing routing protocols for WSNs. The cluster-
ing approach is one of the best techniques for reducing the energy consumption of nodes.
Therefore, instead of each node sending its collected data individually, first, sensor nodes
organize themselves into clusters, and then an elected cluster head (CH) sends all aggregate
data to the sink.
Clustering is used in WSNs for several important reasons, as it offers several benefits
that contribute to the efficient and effective operation of these networks, such as:
• Energy Efficiency: As sensor nodes, WSNs are usually powered by batteries with
energy resources clustering, which plays a role in evenly distributing energy-intensive
tasks like data transmission and aggregation among the nodes. By assigning nodes
as Cluster Heads (CHs) for collecting and aggregating data, non-CH nodes can con-
serve energy and extend the network lifespan by operating in low-power modes for
longer periods.
• Reduced Communication Overhead: In clustered WSNs, the sensor nodes within a
cluster typically transmit collected data to their respective CH. The CH, then. Forwards
the data to the base station or sink node. This approach reduces communication
distances within the network since data does not need to be transmitted to the base
station. Consequently, reduced communication distances lead to energy consumption
and alleviate network congestion.
• Scalability: With clustered WSNs, new nodes can easily join existing clusters as the
network expands while CHs efficiently route data towards the Base Station (BS). This
allows for network expansion without impacting its performance.
• Load Balancing: Cluster Heads are vital in distributing data collection tasks among
sensor nodes within their cluster. This ensures that no single node becomes over-
whelmed with the responsibilities of gathering data. This load-balancing technique
plays a role in avoiding failures of nodes caused by excessive energy usage. Addition-
ally, clustering enhances the fusion of data, allowing for aggregation at the CH level.
As a result, redundant information collected by nodes is minimized, leading to the
transmission of precise and concise data to the base station.
• Prolonged Network Lifetime: The combination of reduced energy consumption, effi-
cient communication, and optimized data routing achieved through clustering signifi-
cantly extends the overall network lifetime.
In large-scale WSNs, coverage is one of the most important QoS metrics, and it
refers to how well the SNs in the network can monitor or sense the region of interest.
Coverage directly impacts the ability of a WSN to fulfill its intended purpose, which could
be environmental monitoring, surveillance, or any other sensitive application. In such a
context, the battery replacement of large amounts of nodes is a labor-consuming work.
Although the life of WSNs can be prolonged through energy-harvesting (EH) technology,
it is necessary to design an energy-efficient routing protocol for energy harvesting, as an
important part of nodes would be unavailable in the energy harvesting phase. In this phase,
a certain number of unavailable nodes would cause a coverage hole, affecting the WSN’s
monitoring function of the target environment.
In [4], authors propose an adaptive hierarchical clustering-based routing protocol for
EH-WSNs (HCEH-UC) to achieve uninterrupted coverage of the target region through
the distributed adjustment of the data transmission. The proposal balances the energy
consumption of nodes. Then, a distributed alternation of working modes is proposed
to adaptively control the number of nodes in the energy-harvesting mode, which could
lead to uninterrupted target coverage. The simulation results show that the proposed
HCEH-UC protocol can prolong the maximal lifetime coverage of WSNs compared with
Sensors 2023, 23, 8534 3 of 23

the conventional routing protocol and achieve uninterrupted target coverage using energy-
harvesting technology.
Despite this, numerous challenges such as Quality of Services (QoS), efficiency of used
energy, mobility, and lifetime restrict the use of WSN. The QoS and energy consumption
are relevant metrics used to assess the quality of paths in any designed routing protocol
in WSNs.
The Quality of Service (QoS) is defined by the International Telecommunication Union
regulations (ITU-T Supp. 9 of E.800 Series) [5] as the totality of characteristics of a telecom-
munications service that bear on its ability to satisfy stated and implied needs of the user
of the service. Also, the QoE (Quality of Experience) is defined as the degree of delight or
annoyance of the user of an application or service [6].
QoS in WSN refers to the ability of the network to provide certain guarantees regarding
latency, throughput, packet loss, and reliability for different types of traffic. Since WSNs
are typically deployed in harsh environments where resources are limited, providing QoS
is a hard task to resolve. However, it is essential to meet the application requirements, such
as monitoring critical infrastructure, conserving energy, and collecting data.
QoS is a very challenging subject, one of the big defis is how to guarantee QoS. In [7],
a new method for QoE parameters prediction in an overall telecommunication system
consisting of users and a telecommunication network, based on QoS indicators’ values
prediction, are overviewed. The presented results show the advantages of the proposed
overall model normalization techniques towards adequate prediction and presentation of
QoE in conjunction with QoS in the overall telecommunication systems.
Besides, the small battery energy is a major constraint for WSNs. As these nodes are
typically deployed in remote and inaccessible locations, recharging them is not feasible.
Therefore, the energy resources of the sensors must be used efficiently to prolong the
network’s lifespan [8,9]. If the network’s topology is not variable and the sink remains fixed,
the energy distribution will be increasingly uneven over time. The network’s longevity is
a crucial evaluation standard used to assess its performance, and it is typically measured
by determining the period when the first node dies. Over the years, numerous routing
protocols and algorithms have been suggested for energy-efficient WSNs, but many of
these works suppose that the sink is static [10–13]. In routing protocols based on multi-hop
communications, nodes close to the sink play a crucial role in transmitting data to other
sensors. As a result, their energy resources tend to deplete faster, resulting in the hot spots
issue [14].
Routing protocols using clustering help sensors sense and reassemble data from the
environment and transmit it to the sink with minimum costs. By grouping nodes, clustering
algorithms enhanced the performance of nodes and their ability to send data. In cluster-
based routing protocols, even cluster heads (CHs) located far away from the sink are more
likely to exhaust their battery reserves than those nearby since the needed hops for sending
data increase with the square of the distance [15,16]. Node deaths can disrupt the network
topology, reduce sensing coverage, and potentially result in a network partition, isolation
of nodes, and loss of data. Additionally, in real-time WSN applications, such as military
zone monitoring, enemy surveillance, natural disaster tracking (e.g., seismic activities),
and exploration of inaccessible areas, stringent quality-of-service constraints are essential.
These constraints include high data reliability, throughput network, low data delivery
latency, and high communication efficiency, apart from efficient energy usage.
Clustering technology is crucial in reducing the consumed power by attributing
sensors to clusters based on specific rules. The cluster features a set of CHs that act as relay
nodes for other members within the group. Clustering simplifies the network topology
and mitigates the need for sensor-sink communications. Moreover, CHs can leverage
data fusion techniques to eliminate repetitive data, thereby lessening the CHs burden. A
prominent example of a routing protocol which uses clustering is the “low-energy adaptive
clustering algorithm” (LEACH). However, selecting CH in the LEACH is not optimal, and
research work is needed to refine the protocol.
Sensors 2023, 23, 8534 4 of 23

Moreover, mobile WSN (MWSN) is a novel variant of networks used in dynamic and
mobile environments due to its capacity for self-configuration. In large WSNs, the network
can be logically portioned into sub-networks. Each one has its mobile sink. Using mobile
sinks is a highly effective approach for managing the imbalanced energy of WSNs. WSNs
supporting mobile sinks typically deploy intelligent vehicles or robots to carry the sink,
which can be moved freely around the sensing field. The implementation of mobile sinks
was suggested and evaluated to address the imbalanced energy problem in WSNs [17–23].
Mobile sinks controlling region of interest (RoI) gather information from static sensors
in one or multiple hops. A significant advantage of multiple mobile sinks is that they
can distribute the communication and computation load across the network, reducing the
burden on individual sensor nodes. This can significantly enhance the network lifetime
by mitigating the effects of energy depletion in specific nodes. Moreover, multiple mobile
sinks can help decrease data collection and transmission latency. By deploying sinks in
different parts of the network, the time taken to collect and transmit data from distant nodes
can be significantly reduced. However, there are some challenges in deploying mobile and
multiple sinks. One such challenge is finding an optimal placement of sinks to cover the
entire network efficiently. Also, the synchronization of mobile and multiple sinks can be
complex. In conclusion, deploying multiple mobile sinks in a WSN is a promising approach
for achieving better efficiency, network lifetime, and data collection, but it requires careful
deployment and synchronization to realize its full benefits.
The current research aims to investigate how energy consumption and QoS metrics
are enhanced by multiple mobile sinks that use a cluster-based routing protocol. We
will examine four different mobility models on these metrics, focusing on identifying the
number and cost of deploying mobile sinks. Our study aims to provide insights into how
network performance can be enhanced while balancing its investment costs.
The next sections are as follows: Section 2 introduces previous works and compares
them with our contribution. Section 3 presents the cluster-based routing protocols used in
this study. Section 4 discusses the suggested sink mobility models. Section 5 highlights the
findings of the simulations, along with analysis and discussions. Section 6 summarizes the
paper and its perspectives.

2. Review of the Literature

This section discusses the studies utilizing stationary and movable WSN sinks to
reduce power consumption and extend the network’s operating life. We will also discuss
the QoS challenges in WSNs applications in the literature.
In [24], a scheme for maximizing the longevity of WSNs utilizing a movable sink was
suggested to manage the delays in delivering data. Each node has a range of tolerance
for delay, within which it does not need to instantly transmit data when it is available.
Instead, the node can keep data in storage for a while and transfer it at the appropriate
time, i.e., when the mobile sink is at an optimal location to lengthen the network’s useful
functioning duration.
Moving sink nodes is among the viable ways used to extend the lifetime of the
network. As pointed out in [25], this technique can significantly improve network durability.
In [26,27], the authors delve further into using numerous mobile sinks to enhance energy
efficiency and network longevity. In another study [28], a joint optimization assessment
to optimize the network lifetime using mobile sinks is performed by determining K-
optimal trajectories and scheduling of sojourn time per position while abiding by the given
constraints by sensors and mobile sinks.
Hence, mobility is a prevalent approach for mitigating hotspots’ issues and extending
the lifetime of multi-hop WSN routing, as highlighted in [29]. Other studies, such as [30],
highlight the impact of using mobile sinks on power usage and longevity by selecting
optimal sink node numbers and parking positions. In [31], a network restructuring process
is proposed by modifying the adjacent nodes of a sink to optimize the lifetime and balance
the power usage among sinks.
Sensors 2023, 23, 8534 5 of 23

By ensuring that the total energy of the sink is below a specific threshold, only a set
of selected nodes are connected, which enhances the network lifespan. Research in [32]
highlights the benefits of using mobile sinks to prolong network life by randomly deploying
nodes in a square area or pre-defined rectangular or hexagonal grids. The hexagonal grid
deployment strategy is particularly effective since it maintains coverage and connectivity.
In the previously discussed studies, authors highlight only the issues of energy con-
sumption with multi-hop routing, considering some fixed or mobile sinks to reduce energy
and improve network lifetime. However, do not give importance to ever-increasing QoS
criteria, especially in real-time constraint applications. Since such routing already suffers
from an exhilarating node energy consumption and a huge data delivery delay due to the
transfer of data between nodes until reaching the BS in multi-hop traveling.
Our study investigates a more critical problem in the real-time WSN context; we use
multiple mobile sinks to improve energy consumption and QoS metrics. Existing cluster-
based routing already guarantees better energy conservation and fast data delivery [15–19].
We are also trying to find the optimal number of mobile wells that maintain good
power and QoS performance while considering the extra costs of mobile sinks deploying.
The weaknesses of the work based on multi-hop routing are mainly the border nodes
ensuring routing data of their affiliated sensors. These border nodes quickly lose their
energies and die, creating network partitioning and a huge loss of relevant data, especially
in a military context or vital monitoring.
For that, we decided to work with hierarchical routing, which showed its performance
concerning power usage and latency, representing the main weakness of mobility models
relying on predetermined and non-adaptive trajectories.
This type of mobility lacks the flexibility of adaptation, especially in a variable, stochas-
tic and unpredictable military context. Suppose a final node loses energy, dies, and goes
out of service. In that case, the mobile sink continues its regular trajectory, and all the nodes
that transmit their data through it become unable to deliver their data to the destination.
Therefore, we easily fall into the phenomenon of black holes, and the network becomes
partitioned, which is unacceptable in critical applications. While random mobility models
do not follow a trajectory, remain flexible, and adapt quickly to any change of context.
Recently, there has been a significant interest in the development of cluster-based
and power-efficient mobile protocols for routing. One such protocol, the “Energy-efficient
Cluster-based Dynamic Routing Algorithm” (ECDRA) [33], involves deploying a mobile
sink attached to a sensor that rotates circularly to dynamically change the topology of
the network in response to the sink’s position. However, LEACH [14] is a well-known
hierarchical routing protocol differentiating CHs from normal nodes (ONs). ONs transfer
their data to the appropriate CHs, which collectively transmit the data to BS. While LEACH
is more effective than classical routing protocols at increasing network lifetime, the random
selection of CHs can result in uneven distribution and flow between the BS and the CHs,
leading to higher energy consumption.
In [34], a framework that enhances energy efficiency and the QoS for WSN is presented.
The introduced hybrid technique utilizes a fitness function that considers key performance
indicators like the number of neighbors, the set of sensors for each cluster, and how long
each node remains the CH. This fitness is integrated with a probability threshold function
to influence the procedure of selecting CHs. Compared to previous homogeneous proto-
cols such as LEACH, the proposed method maintains optimal CH selection more stably
throughout network operation. Furthermore, compared with heterogeneous protocols like
“Developed Distributive Energy-Efficient Clustering and Enhanced Developed Distribu-
tive Energy-Efficient Clustering”, the proposed protocol displays superior performance
regarding the WSN lifetime, power usage, and throughput. However, this suggested
routing algorithm should offer more privacy and security features. Moreover, to prove its
consistency, this paradigm should be tested in a real-world context.
The authors of [35] introduced an evaluation technique to compare the “Secure Mobile
Sink Node Location Dynamic Routing Protocol” (SMSNDRP) with another algorithm
Sensors 2023, 23, 8534 6 of 23

named “routing protocol with K-means for forming Data Gathered Path” (KM-DGP). The
application of these two algorithms was on networks with Mobile Sinks of various sizes.
QoS and power usage are used to assess the quality of routes and energy consumption
patterns of both routing protocols on small (with single and multiple mobile Sinks) and
large networks. The proposed evaluation technique is implemented on NS3 using five
different scenarios. The findings suggest that compared with KM-DGP, SMSNDRP shows
improved network energy consumption on small, single networks. In contrast, for larger
networks with sixteen mobile Sink nodes or more, KM-DGP displays comparatively better
network energy consumption than SMSNDRP with four mobile sink nodes.
The study in [36] introduces a new high-performance communication protocol for
routing packets using multiple mobile nodes. The protocol relies on four main features:
assessment of packet delays, independent control of link quality and choosing active
neighbors of the nodes. Simulation studies on this protocol show that the latter improves
the packet forwarding rates, reduces power usage, and shortens average delays.
The authors in [37] presented a clustering paradigm for MWSN. Their technique in-
volves introducing super cluster heads (SCH) that are static and efficient sensors within the
MWSN to gather CH data from CHs. Combining SCHs with the “Minimum Transmission
Energy” (MTE) protocol reduces the distance required to transfer data from CH to BS,
ultimately improving energy efficiency. Under this approach, data is first transmitted from
CH to SCH and then forwarded to BS. This new technique promises to enhance the network
performance further.
Another study in [38] introduces a new energy-efficient routing system that employs
clustering and sink mobility techniques. The authors propose a two-step approach that
involves classifying the region of interest (RoI) into sectors and selecting a CH for each
sector based on the weight of each node member. Afterwards, each member calculates
the power usage of numerous routing paths and selects the most energy-efficient option.
Finally, CHs are linked in a chain via a greedy strategy for inter-cluster connectivity. The
findings show, as demonstrated through simulations that this new routing strategy is
better than similar approaches, like “Cluster-Chain Mobile Agent Routing” (CCMAR) and
“Energy-efficient Cluster-based Dynamic Routing Algorithm” (ECDRA).
According to a recent study [39], an auto-schedule routing algorithm relying on IoT
connections was introduced to enhance the power usage of Software-defined networking
(SDN) controlled embedded networks. The algorithm starts by constructing the “Neighbor
Distance Discovery Protocol”, which identifies the “minimum depletion path” by locating
the closest node to the BS. Next, the algorithm executes the “Multipath Cooperative Self-
Scheduling Protocol” to establish a non-traffic route. Additionally, the algorithm involves
the routing communications of each IoT object in building the routing medium. It computes
the average packet loss rate, node response rate, energy consumption, sensor absorption
rate, and transmission delay. Finally, the algorithm employs the “Lifetime Duty Cycled
Energy Efficient Protocol” to determine the network threshold latency and energy limits.
The research discussed in [40] explores the latest routing algorithms used in sensor
networks and proposes strategies for their development. This study highlights recent
advances in the strategy used to reduce the energy required for information transmission.
One key concern for IoT, which has gained much attention, is the energy requirements to
extend the lifespan of IoT networks. One of the approaches that has gained traction is the
design of routing protocols that minimize energy consumption during data transmission.
In recent studies, optimization paradigms have been utilized to address the energy
issues in WSNs by means of an energy-efficient multi-objective criterion as follows:
The proposed clustering and optimization-based routing approach in [41] is used to
improve the power efficiency and prolong the lifespan. The selection of CH is achieved in
parallel with the minimization of power usage, which effectively reduces dead sensor nodes.
The use of the “Sailfish optimizer algorithm” for optimal path selection also enhances the
energy efficiency of data transmission between CH and BS. However, the study does not
consider the node mobility in the proposed approach. In WSNs, nodes can move frequently
Sensors 2023, 23, 8534 7 of 23

due to environmental conditions or other factors. Hence, the network’s topology varies,
which may affect the performance of routing algorithms. Future research could address
this limitation by incorporating mobility models into the proposed approach to improve its
adaptability to dynamic network conditions.
Moreover, some studies have attempted to enhance the network power usage and its
lifespan via various optimization algorithms such as the PSO algorithm [42], bio-inspired
ant colony [43], etc. Another research [44] introduced a hybrid ACO-PSO routing paradigm
that employs mobile sinks to reduce overall power usage.
Hence, the research gap can be summarized as follows:
Despite the potential benefits of using multiple mobile sinks in WSNs, research gaps
should be addressed in this domain. One of the significant gaps is the establishment of
performant and robust routing paradigms for multiple mobile sinks since routing protocols
determine the efficiency of WSNs. The challenge of multiple mobile sinks is to design
a routing protocol to handle the changing positions of sinks and ensure efficient data
delivery. Current routing protocols used for multiple mobile sinks are based on centralized
approaches, which can lead to scalability issues and network congestion. Another research
gap is related to the synchronization of mobile multiple sinks. When multiple sinks move
in the RoI, it can be challenging to ensure that they are synchronized in terms of their
locations and data collection schedules. Synchronization is essential to avoid collisions and
ensure efficient data collection.
Furthermore, the deployment methodology of multiple mobile sinks is another area
where research is needed. Identifying an optimal number of mobile sinks, their place-
ment, and their trajectories requires sophisticated algorithms and optimization techniques.
Overall, the research gaps regarding using multiple mobile sinks in WSNs include es-
tablishing scalable and high-performance routing protocols, synchronization techniques,
and optimal deployment methods that can enable efficient and reliable data collection in
large-scale WSNs.
QoS in WSNs has been an interesting research topic in recent years. Many WSN
real-time-based applications require the support of QoS. However, the development of
sensor networks needs to consider various factors such as fault Tolerance, resource alloca-
tion, adaptive routing, data reliability, Real-time communication, scalability, and energy
efficiency [45,46].
Addressing these QoS challenges in WSNs often involves a combination of hardware
and software solutions, including efficient protocols, energy-efficient algorithms, and
adaptive strategies suitable to the specific application requirements. Researchers continue to
develop innovative approaches to overcome these challenges and enhance the performance
of WSNs in various domains [47,48].
Data aggregation is a method to effectively reduce the data transmission volume and
improve network lifetime. However, the data waiting for processing in the queue are
subject to an extra delay. In this paper [47], the authors propose an Adaptive Aggregation
Routing (AAR) scheme to avoid this problem by dynamically changing the forwarding
node according to the length of the data queue and balancing the aggregating and data-
sending load. Simulation results demonstrate that compared with the existing schemes,
the proposed scheme reduces the delay by 14.91%, improves the lifetime by 30.91%, and
increases energy efficiency by 76.40%.
Coverage is a fundamental QoS metric in WSNs that assesses the ability of the network
to adequately monitor a target area. It involves careful node deployment sensing range
configuration and may require adaptation strategies to maintain coverage over time. In [48],
authors propose an energy-efficient clustering routing protocol based on a high-QoS node
deployment with an inter-cluster routing mechanism (EECRP-HQSND-ICRM) in WSNs.
The new protocol introduces a node deployment strategy based on twofold coverage.
The proposed strategy divides the monitoring area into four small areas centered on the
base station (BS), and the CHs are selected in the respective cells to satisfy the uniformity
of the CHs distribution. The simulation results show that, compared with the general
Sensors 2023, 23, 8534 8 of 23

node deployment strategies, the deployment strategy of the proposed protocol has higher
information integrity and validity and lower redundancy.
One of the important challenges is the uncertainty of the service of requests. Recently,
intuitionistic fuzzy estimations of the QoS have been proposed, such as in this work [49],
where three intuitionistic fuzzy characterizations of virtual service devices are specified:
intuitionistic fuzzy traffic estimation, intuitionistic fuzzy flow estimation and intuitionistic
fuzzy estimation about probability. Six intuitionistic fuzzy estimations of the uncertainty
of comprise service devices are proposed. The proposed uncertainty estimations allow
for the definition of new Quality of Service (QoS) indicators. They can determine the
quality-of-service compositions across a wide range of service systems.

3. Methodology
3.1. Simulation Setup
The current research aims to assess the sensors using cluster-based routing protocols.
The evaluation will investigate throughput, reliability, packet latency time, and energy
consumption with four mobility models. The study compares the results of different sink
positions, ranging from one to eight static and mobile sinks. To ensure accuracy and
reliability, the simulation will be repeated 100 times for each scenario in different topologies.
The Castalia/OMNET++ simulator [50] will be utilized for the simulation process. The
“Throughput Test” application is implemented and used for this purpose.
Castalia is a discrete-event simulator specifically designed for WSNs and is built on
the OMNeT++ simulation framework. To evaluate the energy consumption and the QoS in
a large-scale WSN using Castalia, we follow these steps to set up the simulation, define
parameters, and select appropriate metrics:

3.1.1. Step 1: Install Castalia

• Download and install the Castalia simulator v3.2 and OMNeT++ framework v5.0
according to the installation instructions provided on the Castalia website [50].

3.1.2. Step 2: Create the Simulation Scenario

• Define the geographical area or environment where the WSN will be deployed.
• Define the number and initial positions of SNs and the static sink in the network.
• Define the random or deterministic deployment strategy.
• Define the mobility patterns of sinks.

3.1.3. Step 3: Configure Simulation Parameters

• Edit the Castalia configuration file (Config.ini) for each simulation scenario shown in
Appendix A.
• Configure various parameters, including but not limited to:
# Communication protocols (MAC and routing protocols).
# Radio models and channel characteristics.
# Node properties (battery capacity, transmission power, data rate).
# Simulation time and warm-up period.
# QoS-related parameters like latency, packet delivery ratio, and throughput
requirements.
# Energy models.

3.1.4. Step 4: Define QoS Metrics

• Select the specific QoS metrics you want to evaluate based on our research goals.
Common QoS metrics in WSN simulations include Packet Delivery Ratio (Reliability),
End-to-End Delay (Latency), Throughput, Network Lifetime and Coverage.
Sensors 2023, 23, 8534 9 of 23

3.1.5. Step 5: Run the Simulation

• Build and run the simulation using the OMNeT++ IDE or command-line tools as per
the Castalia documentation [50].
• Monitor and collect simulation results, which include the QoS metrics you defined in
step 4.

3.1.6. Step 6: Analyze and Interpret Results

• Use the collected data to analyze the QoS performance of the WSN.
• Generate graphs, plots, and statistics to visualize and interpret the results.
• Draw conclusions based on the evaluation of QoS metrics and how they relate to our
research objectives.

3.1.7. Step 7: Iterate and Refine

• Depending on our findings, we repeat and refine the simulation to further investigate
or optimize QoS in our WSN.

3.2. Cluster-Based Routing Protocols

Many recent articles have treated the impact of sink mobility in the WSN with multi-
hop routing mechanisms [15,16]. In such a type of routing, the nodes closest to the sink
dissipate their energies rapidly since they retransmit the collected data from distinct sensors
to the sink, which divides the network, isolates the sink, and creates energy holes.
The use of mobile sinks has considerably alleviated these concerns in terms of reli-
ability, throughput, and consumed energy. However, the data delivery delay was still
modest due to the accumulated delay for each hop. Furthermore, this routing technique
is not suitable for larger networks since the required number of hops is influenced by the
number of deployed nodes. In such case, other than the number of hops, the delivery delay
increases, and interferences between packets also increases, which rapidly and significantly
degrades the throughput [15,16].
The cluster-based routing protocols have proven good energy conservation results
and low data delivery latency [11–14]. For these reasons, in this study, we investigate
the effect of using multiple mobile sinks with a cluster routing paradigm in large-scale
WSNs. We will introduce the technique of such a routing protocol with the LEACH [14],
P-LEACH [51], and EA-CRP [52]. The Table 1 provides a brief comparison between the key
features of the three routing protocols to be studied.

3.3. Sink Mobility Patterns

To ensure unbiased results towards any mobility model, the authors of this study
chose to compare statics sinks outside the RoI with four random mobility models:
1. Random WayPoint Mobility Model (RWP) [53,54]: A model that includes pause times
between changes in directions and speed.
2. Random Walk Mobility Model (RW) [53,54]: A simple mobility model based on
random directions and speeds.
3. Random Direction Mobility Model (RD) [53,54]: A model that forces MNs to travel to
the edge of the simulation area before changing direction and speed.
4. Gauss Markov Mobility Model (GM) [53,54]: A memory model that uses one tuning
parameter to vary the degree of randomness in the mobility pattern.
The basis for selecting these models for evaluating network Quality of Service (QoS)
metrics is that they support unpredictable and random changes, like real-time scenarios.
Subsequently, we will delve into how each of these models operates. The functioning of
each mobility model is described in the article [21].
Figure 1a shows an example of a traveling path of a sink, which begins in the center
of the RoI, using the RW Mobility Model. At each point, the sink randomly chooses a
direction between 0 and 2 and a speed between 0 and 10m/s. At every 60, the sink changes
Sensors 2023, 23, 8534 10 of 23

direction and speed. This Model is a memoryless mobility pattern because it retains no
knowledge concerning its past locations and speed values [14].

Table 1. Brief Comparison between cluster-based routing protocols.

Protocol LEACH (2000) P-LEACH (2014) EA-CRP (2018)

The network is divided into The sensing area is alienated into
A uniform CH selection process in
partition clusters and uses the various layers, and certain clusters are
Key features which every node gets an equal
prediction techniques with mobile created inside each layer to minimize
chance for a CH job
sink tracking the communication cost.
Reducing the registration time for
Reducing the energy expenditure in
Maximizing the lifetime of new nodes, improving stability,
Aims the network and minimizing the cost
the network and reducing the energy
of communication between nodes
expenditure of the network.
Reduces the cluster setup overhead by
dividing the sensing area into layers
The prediction technique helps in
and sub-layers.
minimizing the energy
An equal chance is given to every The leader heads in different layers
expenditure.
node to become the CH, which reduce the load of CH by performing
Strengths The task of the cluster center is
divides the workload among all the data collection and aggregation
handled by the four gateway
the nodes within the layer.
nodes, which improve the
Minimizes the communication cost
network lifetime
amid the nodes as the width of the
layer decreases towards the BS
The multilayered structure in the
CH selection is random and does Sink mobility increases the
Weakness network can cause delays in
not consider the energy of nodes. message overhead and complexity
data transfer.
The weight function of energy and
Select the random number between The node with the highest battery
distance is calculated for each node.
CH selection method zero and one and compare it with capacity becomes the cluster
The node with the highest weight
the threshold to select the CH center CH
function value becomes the CH
Data transmission Single Hop Multi-Hop Multi Hop
CH rotation YES NO YES
Sensor node mobility Static Static Static
Sensors 2023, 23, x FOR PEER
topology REVIEW
Distributed Distributed Distributed 11
Deployment policy Random Random Random

Figure
Figure 1. Traveling 1. Traveling
patterns patterns
of a Mobile Sinkof a Mobile
using Sink
(a) the using (a)
Random WalktheMM
Random
(RW),Walk MM
(b) the (RW), (b) the Ra
Random
WayPoint
WayPoint MM (RWP), MM
(c) the (RWP), Direction
Random (c) the Random
MM (RD)Direction
and (d)MM
the (RD) and
Gauss (d) theMM
Markov Gauss Markov MM (G
(GM).

Figure 1b showsFigure 1b shows

an example of aan examplepath
traveling of aoftraveling path of
a sink, which a sink,
begins in which begins
the center of in the c
the RoI, using of
thethe RoI, using the RWP Mobility Model. The movement pattern of a sink using the
RWP Mobility Model. The movement pattern of a sink using the RWP
Mobility Model is similar to the RW Mobility Model if pause time is zero and [min-s
max-speed] = [speed-min, speed-max].
Figure 1c shows an example path of a sink, which begins in the center of th
using the RD Mobility Model. In this model, the sink chooses a random direction to t
Sensors 2023, 23, 8534 11 of 23

Mobility Model is similar to the RW Mobility Model if pause time is zero and [min-speed,
max-speed] = [speed-min, speed-max].
Figure 1c shows an example path of a sink, which begins in the center of the RoI, using
the RD Mobility Model. In this model, the sink chooses a random direction to travel, similar
to the RW Mobility Model. The sink then travels to the border of the simulation area in that
direction. Once the RoI boundary is reached (represented by dots in the figure), the sink
pauses for a specified time, chooses another angular direction (between 0 and 180 degrees)
and continues the process.
Figure 1d illustrates an example traveling pattern of a sink using the GM Mobility
Model; the sink begins its movement in the center of the RoI and moves for 1000 s. The
Gauss–Markov Mobility Model was designed to adapt to different levels of randomness via
one tuning parameter. Initially, the sink is assigned a current speed and direction. At fixed
intervals, n, movement occurs by updating the speed and direction. Specifically, the value
of speed and direction at the nth instance is calculated based on the value of speed and
direction at the n-1 instance and a random variable. This model can eliminate the sudden
stops and sharp turns encountered in the RW Mobility Model by allowing past velocities
(and directions) to influence future velocities (and directions).

4. Results and Discussion

4.1. Simulation Scenarios and Evaluation
In WSNs with cluster-based routing [55–57], the shorter the distance separating the
sink from the CH is, the more the power is conserved, and the lower the packet collection
latency is. Using numerous sinks to replace a single sink can significantly decrease these
distances. Using multiple sinks, every cluster head can communicate with the nearest sink.
Therefore, it is possible to enhance the QoS performances by deploying multiple sinks [58]
or relay nodes [59,60] to gather sub-regional data. This technique has been proven effective
in reducing distances and improving overall performance. Hence, it is a preferred solution
for achieving a better quality of service performance.
As a result, the primary sensor network is partitioned into smaller networks with a
low diameter. These sub-networks consist of sensors and a static or mobile sink, forming
a cluster. Cluster heads transfer information to the respective sink of the corresponding
sub-region.
The study’s initial scenario will focus on a simulation of a field measuring 400 m × 400 m,
hosting eight hundred nodes with random positions. This used static sink is situated be-
yond the RoI. Further, the simulation will be repeated four times using a mobile sink,
which follows one of the four designated mobility models in Figure 2a. Another sce-
nario implies the same RoI and number of nodes, but the sensing field is split into two
400 m × 200 m sections.
Each half will be controlled, once with a static sink positioned beyond the supervised
RoI, and four times with a mobile sink that follows one of the four mobility models, so
we will have two fixed sinks and two mobile sinks, as illustrated in Figure 2b in the third
scenario, it is the same principle as the second scenario except that we divide the main field
into four subfields sized of 200 m × 100 m. In this case, we will have four fixed and four
mobile sinks, as illustrated in Figure 2c. Finally, for the fourth scenario, we also keep the
same principle as the other scenarios, but this time, we divide the initial field into eight
subfields of 200 m × 100 m. In this case, we will have eight static and eight mobile sinks, as
shown in Figure 2d.
The assessment analysis of the introduced system relies on various assumptions, in-
cluding the supposed stationary state of all deployed sensor nodes, coupled with their
location awareness. The utilization of a cluster-based routing protocol has also been consid-
ered, where only CHs are authorized to transfer gathered information to the designated
sink. The latter changes its position through the network following a designated mobility
model that facilitates data collection from the respective cluster heads. Each sensor node
 low diameter. These sub-networks consist of sensors and a static or mobile sink, formi
a cluster. Cluster heads transfer information to the respective sink of the correspondi
sub-region.
The study’s initial scenario will focus on a simulation of a field measuring 400 m
Sensors 2023, 23, 8534 400 m, hosting eight hundred nodes with random positions. This used static
12 of 23sink is si
ated beyond the RoI. Further, the simulation will be repeated four times using a mob
sink, which follows one of the four designated mobility models in Figure 2a. Another s
nario implies the same RoI and number of nodes, but the sensing field is split into two 4
uniformly issues a standard amount of data per unit of time (i.e., one packet per second)
m × 200 m sections.
with an equivalent data length of 100 bytes.

Figure 2. Deployment of static and mobile sinks around the RoI, (a) Single Sink for the entire network,
Figure
(b) 2 Sinks for two 2. Deployment of
sub-networks, (c) static and mobile
four Sinks sinks
for four around the RoI,
sub-networks, (a) Single
(d) eight SinksSink
for for the entire n
eight
work, (b) 2 Sinks for two sub-networks, (c) four Sinks for four sub-networks, (d) eight Sinks for ei
sub-networks.
sub-networks.
The energy of transmission of each sensor is adapted to the adjacent nodes’ relative
Each half
distances. The mobile sink will be controlled,
is supposed onceenough
to hold with a static
power sink positioned
reserves beyond the supervis
to communicate
RoI, and four times with a mobile sink that follows one of the
and relocate at any point within the network. The static base stations in each scenario are four mobility models, so
positioned will
20 mhave
beyond twothe fixed sinks
field and two
margins. mobile sinks,
According to the as illustrated
operating in Figure
principles of 2b
thein the th
three hierarchical protocols described above, an election phase is planned each period (i.e., the m
scenario, it is the same principle as the second scenario except that we divide
field into
20 s for LEACH) four subfields
to choose a new CH. sized of 200 m × 100 m. In this case, we will have four fixed a
four mobile sinks, as illustrated
Additionally, all sensor nodes have identical in Figure 2c. Finally, for
communication the fourth
capacity and scenario,
computing we also ke
the same principle as the other scenarios,
resources. Table 2 highlights the relevant, considered parameters.but this time, we divide the initial field into eig
subfields of 200 m × 100 m. In this case, we will have eight static and eight mobile sin
as shown
Table 2. Parameters in Figure 2d.
of simulations.
The assessment analysis of the introduced system relies on various assumptions,
cluding the supposed stationary state of all deployed
Parameter Valuesensor nodes, coupled with th
location awareness.
RoI (Region of Interest) The utilization of a cluster-based
400 m × 400 mprotocol has also been co
routing
sidered, where only CHs are authorized to transfer gathered
Number of nodes 800 information to the des
nated sink. The latter changes its position through therounds
Number of rounds 1000 network following a designat
Round Time 20 s
mobility model that facilitates data collection from the respective cluster heads. Each se
Node deployment (topology) Random
sor node uniformly
Packet size issues a standard amount of data 100per unit of time (i.e., one packet p
bytes
second) with
Packet ratean equivalent data length of 100 bytes. 1 packet/s
Thenode
Initial energy of transmission of each sensor is adapted
energy 100 J to the adjacent nodes’ relat
distances.
Cluster RoutingTheProtocols
mobile sink is supposed to hold enough
LEACH, P-LEACH power reserves to communic
and EA-CRP
and
Sinkrelocate
Mobilityatmodels
any point within the network. The
Static, static
RWP, RW, base
RD stations
and GM in each scenario
Interval of mobility (speed) [1 m/s–10 m/s]
Time of move 10 s
Time of pause for the RWP model 5s

The metrics chosen for evaluation are as follows:

• Energy consumption (The consumed energy by all the sensor nodes)
• Throughput (Total data collected by sink)
• Data delivery rate (Reliability)
Sensors 2023, 23, 8534 13 of 23

• Delay or Packet Latency

To achieve accurate simulation results, we will use four random mobility models of
the sink (GM, RW, RWP and RD) and a static model (a static sink located outside the region
of interest; Fixed Sink) with three cluster-based routing protocols (LEACH, P-LEACH, and
EA-CRP).

4.2. Energy Consumption Evaluation

In the WSN, one can never talk about network performance evaluation without
studying the major concern of energy consumption. For that, in this simulation scenario,
we compared the energy consumed by all sensor nodes (Network Energy) by varying the
number of mobile and static sinks that monitor the network to study the impact of using
multiple mobile sinks on energy consumption in large-scale WSNs (LS-WS).
The simulation results illustrated in Figure 3a show that the use of a single mobile
sink, regardless of the model, decreases the average energy consumed compared to a fixed
Sensors 2023, 23, x FOR PEER REVIEW
sink by −12.5% for RW up to −19% for RD, and the best result obtained with14theof use
25 of the
RD mobility model and the P-LEACH routing protocol with a power gain of −20%.

(a) Network Energy Consumption for a Single (b) Network Energy Consumption for 2 Sinks
Sink 600,000
700,000 550,000
650,000
500,000
600,000
450,000
550,000
500,000 400,000
450,000 350,000
GM RW RWP RD GM RW RWP RD
Fixed Fixed
Mobile Mobile Mobile Mobile Mobile Mobile Mobile Mobile
Sink Sink
Sink Sink Sink Sink Sink Sink Sink Sink
LEACH 658,000 513,240 552,720 532,980 493,500 LEACH 592,200 461,916 497,448 479,682 444,150
P-LEACH 559,302 486,920 500,080 493,510 480,340 P-LEACH 447,440 389,536 400,064 394,800 384,272
EA-CRP 592,200 500,080 526,400 513,241 486,920 EA-CRP 503,370 425,068 447,440 436,254 413,882

LEACH P-LEACH EA-CRP LEACH P-LEACH EA-CRP

(c) Network Energy Consumption for 4 Sinks (d) Network Energy Consumption for 8 Sinks
500,000 450,000
450,000 400,000
400,000 350,000
350,000 300,000
300,000 250,000
250,000 200,000
GM RW RWP RD GM RW RWP RD
Fixed Fixed
Mobile Mobile Mobile Mobile Mobile Mobile Mobile Mobile
Sink Sink
Sink Sink Sink Sink Sink Sink Sink Sink
LEACH 473,760 369,532 397,958 383,745 355,320 LEACH 449,902 351,055 378,060 364,557 337,554
P-LEACH 313,208 272,675 280,044 276,360 268,990 P-LEACH 291,283 253,587 260,440 257,014 250,160
EA-CRP 377,527 318,801 335,580 327,190 310,411 EA-CRP 339,774 286,920 302,022 294,471 279,369

LEACH P-LEACH EA-CRP LEACH P-LEACH EA-CRP

Figure 3. Nodes Energy Consumption with different routing protocols by using multiple
statics3.and
Figure mobile
Nodes sinks,
Energy (a) Single Sink
Consumption withfor the entire
different network,
routing (b) 2by
protocols Sinks formultiple
using 2 sub-net-
statics and
works,
mobile (c) 4(a)
sinks, Sinks for Sink
Single four for
sub-networks, (d) 8 Sinks
the entire network, (b) 2for 8 sub-networks.
Sinks for 2 sub-networks, (c) 4 Sinks for four
sub-networks, (d) 8 Sinks for 8 sub-networks.
The simulation results illustrated in the other figures show that the use of multiple
mobile sinks decreases the average energy consumption compared to a single static sink,
as illustrated in Figure 3a:
 By using two mobile sinks (Figure 3b) from −25.6% for RW up to −31.3% for RD and
the best result obtained with the RD model and the P-LEACH routing protocol
−36.1%.
 By using four mobile sinks (Figure 3c) from −44% for RW to −48.3% for RD, the best
Sensors 2023, 23, 8534 14 of 23

The simulation results illustrated in the other figures show that the use of multiple
mobile sinks decreases the average energy consumption compared to a single static sink, as
illustrated in Figure 3a:
• By using two mobile sinks (Figure 3b) from −25.6% for RW up to −31.3% for RD
and the best result obtained with the RD model and the P-LEACH routing protocol
−36.1%.
• By using four mobile sinks (Figure 3c) from −44% for RW to −48.3% for RD, the best
result was obtained with the RD model and the P-LEACH routing protocol −57.5%.
• Using eight mobile sinks (Figure 3d) from −48% for RW to −52% for RD, the best
result was obtained with the RD model and the P-LEACH routing protocol −59.5%.
It can be drawn from the previous results that the best mobility model is RD, which
offers better energy conservation, around 60% less, with eight mobile sinks, compared
to a single fixed sink. More precisely, the sink with RD mobility model moves towards
the clusters of sensor nodes to reduce the distance between the sink and the CH and
consequently reduce the transmission energy consumption.
On the other hand, by comparing the different energy scenarios, we notice that using
four mobile sinks gives results very close to that of eight mobile sinks.
So, we can conclude that in terms of profitability, the use of four mobile sinks estab-
lishes a good compromise between energy conservation and investment cost (budget of
mobile sinks), and we can obtain better energy conservation of 57.5% less with only four
mobile sinks using the RD model and the P-LEACH routing protocol. Since consumed
power is expected to increase exponentially as the communication range increases, utilizing
a shorter transmission distance can significantly optimize the power consumed by mobile
sinks. This implies that the power conservation of the network will be higher, and its
lifetime will be extended as the subnet area becomes smaller. Nonetheless, it will incur a
higher cost of deploying adequate mobile sinks to cover the area.

4.3. Throughput Evaluation

When deploying WSN applications with service quality constraints, the amount of
collected data by the sink becomes important to consider.
To address this, in the second phase of the assessment, we analyzed the packets
received by the sink (throughput) of the network with multiple mobile and static sinks.
In this simulation scenario, we compared the amount of data collected (Throughput)
by varying the number of mobile and static sinks monitoring the network to study the
impact of multiple mobile sinks on the throughput in the LS-WSNs.
The simulation results illustrated in Figure 4a show that the use of a single mobile sink,
whatever the model, increases the average flow rate compared to a fixed sink by +8.6% for
RWP up to +14% for RD, and the best result obtained with the RD model and the EA-CRP
routing protocol with a throughput gain of +15.5%.
The simulation results illustrated in the other figures show that the use of multiple
mobile sinks increases the average flow compared to that of a single static sink, as illustrated
in Figure 4a respectively:
• Using two mobile sinks (Figure 4b) from +25% for RWP up to +31% for RD, the best
result is obtained with the RD model and the EA-CRP routing protocol + 38%.
• Using four mobile sinks (Figure 4c) from +49% for RWP to +57% for RD, the best result
is obtained with the RD model and the EA-CRP routing protocol + 73%.
• By using eight mobile sinks (Figure 4d) from 60.5% for RWP to 68.5% for RD, the best
result is obtained with the RD model and the EA-CRP routing protocol +90.2%.
It can be deduced from the previous results that the best mobility model is RD, which
offers almost double the throughput compared to a single fixed, using eight mobile sinks.
More precisely, the RD moves in the network by realistically approaching the clusters
of SN to reduce the distance between the sink and the CH, consequently reducing the
phenomena of collisions and packet retransmission and increasing the throughput.
Sensors 2023, 23, 8534 15 of 23
Sensors 2023, 23, x FOR PEER REVIEW 16 of 25

(a) Total Data Collected by a Single Sink (b) Total Data Collected by 2 Sinks
24,000 29,000
23,000 28,000
27,000
22,000 26,000
21,000 25,000
24,000
20,000 23,000
19,000 22,000
21,000
18,000 20,000
GM RW RWP RD GM RW RWP RD
Fixed Fixed
Mobile Mobile Mobile Mobile Mobile Mobile Mobile Mobile
Sink Sink
Sink Sink Sink Sink Sink Sink Sink Sink
LEACH 19,720 22,875 22,283 21,692 23,466 LEACH 21,692 25,162 24,511 23,861 25,812
P-LEACH 20,903 23,269 22,875 22,480 23,664 P-LEACH 24,038 26,759 26,306 25,852 27,213
EA-CRP 21,495 23,664 23,466 23,269 23,861 EA-CRP 25,794 28,396 28,159 27,922 28,633

LEACH P-LEACH EA-CRP LEACH P-LEACH EA-CRP

(c) Total Data Collected by 4 Sinks (d) Total Data Collected by 8 Sinks
36,000 40,000
34,000 38,000
32,000 36,000
30,000 34,000
32,000
28,000 30,000
26,000 28,000
24,000 26,000
22,000 24,000
22,000
20,000 20,000
GM RW RWP RD GM RW RWP RD
Fixed Fixed
Mobile Mobile Mobile Mobile Mobile Mobile Mobile Mobile
Sink Sink
Sink Sink Sink Sink Sink Sink Sink Sink
LEACH 24,295 28,181 27,452 26,724 28,910 LEACH 25,509 29,590 28,824 28,060 30,355
P-LEACH 28,845 32,110 31,567 31,022 32,655 P-LEACH 30,864 34,357 33,776 33,193 34,940
EA-CRP 32,242 35,495 35,198 34,902 35,791 EA-CRP 35,466 39,044 38,717 38,392 39,370

LEACH P-LEACH EA-CRP LEACH P-LEACH EA-CRP

Figure 4. Total data collected by multiple statics and mobiles sinks with different routing protocols,
Figure 4. Total
(a) Single Sinkdata collected
for the by multiple
entire network, statics
(b) two Sinksand
for mobiles sinks with(c)different
two sub-networks, routing
four Sinks protocols,
for four
(a)sub-networks,
Single Sink (d)
foreight Sinks for
the entire eight sub-networks.
network, (b) two Sinks for two sub-networks, (c) four Sinks for four
sub-networks, (d) eight Sinks for eight sub-networks.
4.4. Reliability Evaluation
OnIn this simulation
the other hand, scenario, we compared
by comparing the network's
the flow scenarios,reliability
we noteby varying
that the of four
the use
number of mobile and static sinks monitoring the network to study the impact of using
mobile sinks gives results very close to that of eight mobile sinks. So, we can conclude that
inmultiple
terms ofmobile sinks onthe
profitability, reliability in LS-WSN.
use of four mobile sinks establishes a good compromise between
the flowTherate
simulation
and theresults illustrated
investment cost,inand
Figure 5a show
we can obtain thata the usethroughput
better of a single mobile
of 73% more
sink,only
with whatever the model,
four mobile sinksincreases
by usingthe RD
average reliability
mobility model compared to a fixed
and EA-CRP sink from
routing protocol.
+17% for RWP up to + 20% for RD, and the best result is obtained with the RD model and
theReliability
4.4. EA-CRP routing protocol with a gain of +22%.
Evaluation
The simulation results illustrated in the other figures show that the use of multiple
In this simulation scenario, we compared the network’s reliability by varying the
mobile sinks increases the average reliability of the network compared to that of a single
number of mobile and static sinks monitoring the network to study the impact of using
static sink (Figure 5a):
multiple mobile sinks on reliability in LS-WSN.
 By using two mobile sinks (Figure 5b) from +43% for RWP up to +52% for RD, the
The simulation results illustrated in Figure 5a show that the use of a single mobile
best result is obtained with the RD model and the EA-CRP routing protocol + 65%.
sink, whatever the model, increases the average reliability compared to a fixed sink from
+17% for RWP up to + 20% for RD, and the best result is obtained with the RD model and
the EA-CRP routing protocol with a gain of +22%.
The simulation results illustrated in the other figures show that the use of multiple
mobile sinks increases the average reliability of the network compared to that of a single
static sink (Figure 5a):
Sensors 2023, 23, 8534 16 of 23

Sensors 2023, 23, x FOR PEER REVIEW 17 of 25

• By using two mobile sinks (Figure 5b) from +43% for RWP up to +52% for RD, the best
result is obtained with the RD model and the EA-CRP routing protocol + 65%.
•  By using four mobile sinks (Figure 5c) from +73% for RWK up to +88% for RD, the
By using four mobile sinks (Figure 5c) from +73% for RWK up to +88% for RD, the
best
bestresult
result is
is obtained withthe
obtained with the
RDRD model
model andand
the the EA-CRP
EA-CRP routing
routing protocol
protocol + 105%.+ 105%.
•  By
By using eight mobile sinks (Figure 5d) from 91.5% for RWK up to 110% for RD,for
using eight mobile sinks (Figure 5d) from 91.5% for RWK up to 110% theRD, the
best result is obtained with the RD model and the EA-CRP routing protocol + 140%.+ 140%.
best result is obtained with the RD model and the EA-CRP routing protocol

(a) Network Reliability (%) for a Single Sink (b) Network Reliability (%) for 2 Sinks

21 27
20 26
25
19
24
18
23
17 22
16 21
15 20
GM RW RWP RD GM RW RWP RD
Fixed Fixed
Mobile Mobile Mobile Mobile Mobile Mobile Mobile Mobile
Sink Sink
Sink Sink Sink Sink Sink Sink Sink Sink
LEACH 16 19.52 18.9 18.4 19.85 LEACH 19.2 23.4 22.68 22.08 23.82
P-LEACH 17.1 19.68 19.52 19.2 20 P-LEACH 21.375 24.6 24.4 24 25
EA-CRP 17.7 20 19.85 19.69 20.32 EA-CRP 23.01 26 25.805 25.597 26.416

LEACH P-LEACH EA-CRP LEACH P-LEACH EA-CRP

(c) Network Reliability (%) for 4 Sinks (d) Network Reliability (%) for 8 Sinks

34 40
38
32
36
30 34
28 32
30
26 28
24 26
24
22 22
20 20
GM RW RWP RD GM RW RWP RD
Fixed Fixed
Mobile Mobile Mobile Mobile Mobile Mobile Mobile Mobile
Sink Sink
Sink Sink Sink Sink Sink Sink Sink Sink
LEACH 23 28 22.7 27.2 28.6 LEACH 24.19 29.4 23.8 28.5 30
P-LEACH 26.7 30.75 30.5 30 31.2 P-LEACH 29.38 33.9 33.6 33 35
EA-CRP 29.9 33.8 33.5 33.2 34.3 EA-CRP 34.39 39 38.5 38.3 40

LEACH P-LEACH EA-CRP LEACH P-LEACH EA-CRP

Figure 5. Network Reliability with different routing protocols by using multiple statics and mobiles
Figure (a)Network
sinks, 5. Single SinkReliability withnetwork,
for the entire different(b)routing protocols
two Sinks for two by using multiple
sub-networks, statics
(c) four Sinksand
for mobiles
four sub-networks,
sinks, (a) Single Sink (d)for
eight
theSinks
entirefornetwork,
eight sub-networks.
(b) two Sinks for two sub-networks, (c) four Sinks for
four sub-networks, (d) eight Sinks for eight sub-networks.
It can be deduced from the previous results that the best mobility model is RD, which
offers more
It can bethan twice from
deduced more the
throughput
previouscompared to athe
results that single
bestfixed sink for
mobility modelthe whole
is RD, which
network, using eight mobile sinks. More precisely, the RD moves
offers more than twice more throughput compared to a single fixed sink for the in the network by ap- whole
proaching in a realistic way towards the SN clusters to reduce
network, using eight mobile sinks. More precisely, the RD moves in the network the distance between the by
sink and the CH and consequently reduce the phenomena of collisions
approaching in a realistic way towards the SN clusters to reduce the distance between and packet retrans-
mission,
the sink andconsequently
the CH and increasing the network
consequently reliability,
reduce especially with
the phenomena ofthe EA-CRP rout-
collisions and packet
ing protocol which uses the combination of two routing techniques,
retransmission, consequently increasing the network reliability, especially clustering, and with
multi-the EA-
hop.
CRP routing protocol which uses the combination of two routing techniques, clustering,
and multi-hop.
On the other hand, by comparing the flow scenarios, we note that the use of four
mobile sinks gives results very close to that of eight mobile sinks. So, we can conclude that
 Sensors 2023, 23, x FOR PEER REVIEW 18 of 25

Sensors 2023, 23, 8534 On the other hand, by comparing the flow scenarios, we note that the use of 17 four
of 23
mobile sinks gives results very close to that of eight mobile sinks. So, we can conclude that
in terms of profitability, the use of four mobile sinks establishes a good tradeoff between
inthe network
terms reliability and
of profitability, the the
use cost of investment,
of four mobile sinksand we can obtain
establishes better
a good reliability
tradeoff of
between
+105% more with only four mobile sinks by using the RD mobility model and the EA-CRP
the network reliability and the cost of investment, and we can obtain better reliability of
routing
+105% protocol.
more with only four mobile sinks by using the RD mobility model and the EA-CRP
routing protocol.
4.5. Packets Latency Time (End-to-End Delay) Evaluation
4.5. Packets
In thisLatency Time scenario,
simulation (End-to-End
we Delay)
comparedEvaluation
packet latency (the percentage of packets
thatInarrive at the sink scenario,
this simulation with less we
thancompared
1ms delay) by varying
packet latencythe
(thenumber of mobile
percentage and
of packets
static
that sinksatmonitoring
arrive theless
the sink with network to study
than 1ms theby
delay) impact of the
varying theuse of multiple
number mobile
of mobile andsinks
static
on packet
sinks latencythe
monitoring in LS-WSNs.
network to study the impact of the use of multiple mobile sinks on
packetThe simulation
latency results illustrated in Figure 6a show that the use of a mobile sink,
in LS-WSNs.
whatever the model,results
The simulation increases the mean
illustrated inpercentage of fastthat
Figure 6a show packets (packets
the use with delay
of a mobile sink,
less thenthe
whatever 1ms) compared
model, to that
increases theof a single
mean fixed sink
percentage frompackets
of fast +8% for RWP up
(packets to +28%
with delayfor
less
RD, 1ms)
then and the best result
compared is obtained
to that withfixed
of a single the RD
sinkmodel
from and
+8%theforLEACH
RWP uprouting
to +28% protocol
for RD,
+73%.
and the best result is obtained with the RD model and the LEACH routing protocol +73%.

Figure 6. Packets latency with different routing protocols by using multiple statics and mobiles sinks,
(a) Single Sink for the entire network, (b) two Sinks for two sub-networks, (c) four Sinks for four
sub-networks, (d) eight Sinks for eight sub-networks.
Sensors 2023, 23, 8534 18 of 23

The simulation results shown in the other figures show that the use of multiple mobile
sinks increases the mean percentage of fast packets compared to the use of a single static
sink (Figure 6a):
• By using two mobile sinks (Figure 6b) from +21% for RWP up to +43% for RD, the best
result is obtained with the RD model and the LEACH routing protocol +99.5%.
• By using four mobile sinks (Figure 6c) from +43% for RWP up to +69% for RD, the
best result is obtained with the RD model and the LEACH routing protocol +153%.
• Using eight mobile sinks (Figure 6d) from 51% for RWP to 78% for RD, the best result
is obtained with the RD model and the LEACH routing protocol +175%.
We can draw from the previous results that the best mobility model is RD, which offers
almost triple results and 2.73 times faster packets compared to a single fixed sink by using
eight mobile sinks. Therefore, we conclude that using multiple mobile sinks enhances the
packet latency, particularly in smaller areas.
More precisely, the RD moves in the network by approaching them in a realistic way
towards the SN clusters to avoid packet retransmission, decreases the distance between the
sink and the CH, and consequently increases the number of packets that reach sinks with
low delays due to the single hop routing technique used by LEACH.
On the other hand, by comparing the latency scenarios, we note that the use of four
mobile sinks gives results very close to that of eight mobile sinks. So, we can conclude that
in terms of profitability, using four mobile sinks establishes a good tradeoff between the
latency time and the cost of multiple mobile sinks. So, we can obtain a better latency of
+153% for fast packets with only four mobile sinks by using the RD model and the LEACH
routing protocol.
In conclusion, when the set of static or mobile sinks increases their proximity to sensor
nodes, fewer nodes are associated with each sink. This results in the reduction of the packet
interference and the un-saturation of the sink buffer. This avoids packet retransmission and
subsequently improves the number of packets that hold their sinks without additive delay.
Comparing the effect of static and mobile sinks, we can deduce that mobile sinks offer
gains in energy conservation of −59.5%, throughput of +90.2%, reliability of +140% and
latency of 175% with eight mobile sinks with the best RD the best mobility model for our
scenarios. The use of eight mobile sinks offers a non-significant gain compared to the use
of four mobile sinks. However, supplementary expenses were incurred due to the use of
a high number of mobile sinks. So, we can just be satisfied with four mobile sinks with
the RD mobility model, seeing the high costs of eight mobile sinks. Otherwise, it will be
a waste.

5. Limitations and Potential Solutions

Conducting a simulation study to improve QoS in large-scale WSNs using multiple
mobile sinks and random mobility models is a valuable approach. However, like any
research methodology, it has limitations. The Table 3 below resumes the approach and the
potential solutions or areas for future research:

Table 3. Limitations and Potential Solutions.

Criteria Limitation Potential Solutions/Future Research

The memoryless Random mobility models
used in simulations may not accurately
Investigate the use of mobility traces
represent real-world scenarios where sink
collected from real-world deployments to
Random Mobility mobility can be influenced by various factors
create more accurate random
such as environmental obstacles (Trees and
mobility models.
mountains) or climatic perturbations (wind,
rain, and snow)...
Sensors 2023, 23, 8534 19 of 23

Table 3. Cont.

Criteria Limitation Potential Solutions/Future Research

Consider a hierarchical routing protocol with
random mobility awareness of the sinks. The
The use of several mobile sinks in simulation
protocol allows dynamic and adaptive
can improve the QoS easily and give good
coordination between sinks and CHs to
Multiple mobile Sinks Cost results.
optimize routes and reduce the number of
However, in the real world, there is a
mobile sinks and associated costs while
significant investment cost behind it.
respecting important QoS metrics such as
delay and coverage.
Conduct empirical studies to optimize the
Random mobility models might not
mobility models with machine learning to
Optimization of adequately represent the movement patterns
ensure they accurately reflect the movement
Mobility Patterns of mobile sinks. Optimization of the
of sinks in realistic WSN applications with
behaviors of these models can be challenging.
QoS constraints.

6. Conclusions
This paper investigated the impact of using multiple mobile sinks on network energy
efficiency and QoS metrics using a cluster-based routing approach and random mobil-
ity patterns.
More specifically, this type of protocol uses hierarchical routing, which offers good
energy conservation and latency time results. Moreover, random mobility models run
through the network in the context of a simulated battlefield observation where fast and
unexpected events reoccur by getting closer to CHs to reduce power consumption during
the transmission phase, reduce delays during data collection and increase the number of
packets collected.
The simulation results obtained demonstrate that the Random Direction mobility
model with four sinks has a significant impact on power consumption and QoS metrics.
In particular, the EA-CRP and P-LEACH protocols achieve a significant improvement in
terms of energy consumption, throughput, and reliability, while the latency time gains
better with the LEACH protocol.
In addition, the simulation results show that RD is more suitable for LSWSN because
it maintains good performance in terms of power consumption and all the QoS criteria
studied for the large supervised RoI and that one can optimize the number of mobile sinks
according to the real-time constraints of the RCSF application and the allocated budget.
After an in-depth discussion of the current approach limits regarding certain uncer-
tainties in the operation of the mobility models used. We planned for future work to
first, investigate the use of mobility traces collected from real-world deployments to create
more accurate random mobility models. Secondly, consider a hierarchical routing protocol
with random mobility awareness of the sinks to reduce the number of mobile sinks and
associated costs while respecting QoS. Finally, optimize the mobility models to ensure
they accurately reflect the movement of sinks in realistic WSN applications with QoS
constraints. The proposed model may also be applied to numerous real-world engineering
problems [61,62].

Author Contributions: Conceptualization, A.B.Y.; Methodology, A.B.Y. and B.B.G.; Software, A.B.Y.
and R.G.; Validation, A.B.Y., S.M., M.A.A. and I.A.; Formal analysis, M.A. (Mansoor Alghamdi); Data
curation, A.B.Y. and M.A. (Malek Alrashidi); Writing—original draft, A.B.Y. and S.M.; Supervision,
S.H. All authors have read and agreed to the published version of the manuscript.
Funding: This research received no external funding.
Institutional Review Board Statement: Not applicable.
Informed Consent Statement: Not applicable.
Data Availability Statement: Data and codes are available by open request.
Sensors 2023, 23, 8534 20 of 23

Conflicts of Interest: The authors declare no conflict of interest.

Appendix A
The Config.ini file contains the Castalia simulation parameters:
• [General] #Common parameters
• include ../Parameters/Castalia.ini
• include ../Parameters/MAC/CSMA.ini
• SN.node[*].Communication.Radio.RadioParametersFile=“../Parameters/Radio/
CC2420.txt”
• sim-time-limit = {SimTime=1000}s
• simtime-scale = −10
• SN.field_x = 400
• SN.field_y = 400
• SN.numNodes = 401
• SN.deployment = “[1..401]->uniform” #random uniform deployment
• SN.node [0].ResourceManager.initialEnergy = 10000 # Joules
• SN.node [1..101].ResourceManager.initialEnergy = 100 # Joules
• SN.node [1..101].MobilityManagerName = “NoMobilityManager”
• SN.node[*].Communication.Radio.collisionModel = 2
• SN.node[*].Application.latencyHistogramMax = 10
• SN.node[*].Application.latencyHistogramBuckets = 20
• SN.node[*].ApplicationName = “ThroughputTest”
• SN.node[*].Communication.Radio.TxOutputPower = “−5dBm”
• SN.node[*].Communication.Routing.netBufferSize = 102400 #Bytes
• SN.node[*].Communication.MAC.macBufferSize = 102400 #Bytes
• SN.node[*].Communication.Radio.bufferSize = 102400 #Bytes
• SN.node[*].Application.packet_rate = 1 #Packet/s
• SN.node[*].Application.packetSize = 100 #Bytes
• SN.node[*].Communication.Radio.collectTraceInfo = false
• SN.node[*].Communication.MAC.collectTraceInfo = false
• SN.node[*].Communication.Routing.collectTraceInfo = false
• SN.node[*].Application.collectTraceInfo = false
• SN.node[*].SensorManager.collectTraceInfo = false
• SN.node[*].ResourceManager.collectTraceInfo = false
• SN.wirelessChannel.collectTraceInfo = false
• SN.wirelessChannel.sigma = 0
• SN.wirelessChannel.bidirectionalSigma = 0
• SN.wirelessChannel.pathLossExponent = 2.0
• SN.wirelessChannel.collectTraceInfo = false
• SN.wirelessChannel.onlyStaticNodes = false
• SN.node[*].Communication.RoutingProtocolName = “leach” #“p-leach”#”ea-crp”
• SN.node [0].Communication.Routing.isSink = true
• SN.node[*].Communication.Routing.slotLength = 0.2
• SN.node[*].Communication.Routing.roundLength = 20s
• SN.node[*].Communication.Routing.percentage = 0.05
• SN.node[*].Communication.Routing.powersConfig = xmldoc(“powersConfig.xml”)
• [Config Static] #Specific parameters for this scenario
• SN.node [0].MobilityManagerName = “NoMobilityManager”
• SN.node [0].xCoor = 200
• SN.node [0].yCoor = 450
• SN.node [0].zCoor = 0
• [Config GM] #Specific parameters for this scenario
• SN.node [0].MobilityManagerName = “GaussMobility”
Sensors 2023, 23, 8534 21 of 23

• SN.node [0].MobilityManager.updateInterval = 1000

• SN.node [0].MobilityManager.max_speed = 10
• [Config RWP] #Specific parameters for this scenario
• SN.node [0].MobilityManagerName = “WaypointMobility”
• SN.node [0].MobilityManager.updateInterval = 1000
• SN.node [0].MobilityManager.max_speed = 10
• SN.node [0].MobilityManager.pausetime = 5
• [Config RWK] #Specific parameters for this scenario
• SN.node [0].MobilityManagerName = “WalkMobility”
• SN.node [0].MobilityManager.updateInterval = 1000
• SN.node [0].MobilityManager.max_speed = 10
• SN.node [0].MobilityManager.movetime = 10

References
1. Cordeiro, C.D.M.; Agrawal, P. Ad Hoc and Sensor Networks: Theory and Applications; World Scientific Publication: Singapore, 2006;
ISBN 981-256-681-3/981-256-682-1.
2. Akyildiz, I.F.; Su, W.; Sankarasubramaniam, Y.; Cayirci, E. Wireless sensor networks: A survey. Comput. Netw. 2002, 38, 393–422.
[CrossRef]
3. Akkaya, K.; Younis, M. A survey on routing protocols in wireless sensor networks. A survey on routing protocols for wireless
sensor networks. Ad. Hoc. Netw. 2005, 3, 325–349. [CrossRef]
4. Han, B.; Ran, F.; Li, J.; Yan, L.; Shen, H.; Li, A. A Novel Adaptive Cluster Based Routing Protocol for Energy-Harvesting Wireless
Sensor Networks. Sensors 2022, 22, 1564. [CrossRef] [PubMed]
5. E.800SerSup9 (10/21) Guidelines on Regulatory Aspects of Quality of Service: Recommendation of the International Tele-
communication Union. Available online: https://www.itu.int/rec/T-REC-E.800SerSup9-202110-I (accessed on 20 August 2023).
6. P.10/G.100 (11/17) Vocabulary for Performance, Quality of Service and Quality of Experience: Recommendation of the Inter-
national Telecommunication Union. Available online: https://www.itu.int/rec/T-REC-P.10-201711-I/en (accessed on 20 Au-
gust 2023).
7. Poryazov, S.A.; Saranova, E.T.; Andonov, V.S. Overall Model Normalization towards Adequate Prediction and Presentation of
QoE in Overall Telecommunication Systems. In Proceedings of the 2019 14th International Conference on Advanced Technologies,
Systems and Services in Tele-communications (TELSIKS), Nis, Serbia, 23–25 October 2019; pp. 360–363. [CrossRef]
8. Quang, P.T.A.; Kim, D.-S. Enhancing Real-Time Delivery of Gradient Routing for Industrial Wireless Sensor Networks. IEEE
Trans. Ind. Informatics 2012, 8, 61–68. [CrossRef]
9. Lindsey, S.; Raghavendra, C.; Sivalingam, K.M. Data gathering algorithms in sensor networks using energy metrics. IEEE Trans.
Parallel Distrib. Syst. 2002, 13, 924–935.
10. Chen, G.; Li, C.; Ye, M.; Wu, J. An unequal cluster-based routing protocol in wireless sensor networks. Wirel. Networks 2009, 15,
193–207. [CrossRef]
11. Wei, D.; Jin, Y.; Vural, S.; Moessner, K.; Tafazolli, R. An Energy-Efficient Clustering Solution for Wireless Sensor Networks. IEEE
Trans. Wirel. Commun. 2011, 10, 3973–3983. [CrossRef]
12. Jumira, O.; Wolhuter, R.; Zeadally, S. Energy-efficient beaconless geographic routing in energy harvested wireless sensor networks.
Concurr. Comput. Pract. Exp. 2013, 25, 58–84. [CrossRef]
13. Yahya, B.; Ben-Othman, J. Energy efficient and QoS aware medium access control for wireless sensor networks. Concurr. Comput.
Pract. Exp. 2010, 22, 1252–1266. [CrossRef]
14. Heinzelman, W.R.; Chandrakasan, A.; Balakishnan, H. Energy-Efficient Communication Protocol For Wireless Micro-Sensor Net-
works. In Proceedings of the 33rd Annual Hawaii International Conference on System Sciences, Maui, HI, USA, 7 January 2000.
15. Yagouta, A.B.; Gouissem, B.B. Quality of Service and Energy Efficient Evaluation of Hierarchical and Flat Routing Protocols
for Wireless Sensor Networks. In Proceedings of the 10th International Conference on Sensor Technologies and Applications
(SENSORCOMM 2016), Nice, France, 24–28 July 2016.
16. Yagouta, A.B.; Gouissem, B.B. Compromises Between Quality of Service Metrics and Energy Consumption of Hierarchical and
Flat Routing Protocols for Wireless Sensors Network. Sens. Transducers 2016, 206, 15.
17. Vincze, Z.; Vida, R. Multi-hop Wireless Sensor Networks with Mobile Sink. In Proceedings of the 2005 ACM Conference on
Emerging Network Experiment and Technology, Toulouse, France, 24–27 October 2005; pp. 302–303.
18. Akkaya, K.; Younis, M.; Bangad, M. Sink Repositioning for Enhanced Performance in Wireless Sensor Networks. Elsevier Comput.
Netw. J. 2005, 49, 512–534. [CrossRef]
19. Mnasri, S.; Alrashidi, M. Energy-efficient IoT routing based on a new optimizer. Simul. Model. Pract. Theory 2022, 119, 102591.
[CrossRef]
20. Chatzigiannakis, I.; Kinalis, A.; Nikoletseas, S. Sink Mobility Protocols for Data Collection in Wireless Sensor Networks. In
Proceedings of the International Workshop on Mobility Management and Wireless Access, MobiWac ’06, Terromolinos, Spain,
2 October 2006; pp. 52–59.
Sensors 2023, 23, 8534 22 of 23

21. Yagouta, A.B.; Jabberi, M.; Gouissem, B.B. Impact of Sink Mobility on Quality of Service Performance and Energy Consumption in
Wireless Sensor Network with Cluster Based Routing Protocols. In Proceedings of the IEEE/ACS 14th International Conference
on Computer Systems and Applications, (AICSSA2017), Hammamet, Tunisia, 30 October–3 November 2017.
22. Hassanat, A.B.; Alkafaween, E.A.; Al-Nawaiseh, N.A.; Abbadi, M.A.; Alkasassbeh, M.; Alhasanat, M.B. Enhancing Genetic
Algorithms using Multi Mutations: Experimental Results on the Travelling Salesman Problem. Int. J. Comput. Sci. Inf. Secur. 2016,
14, 785.
23. Yagouta, A.B.; Gantassi, R.; Gouissem, B.B. Compromises between Energy Consumption and Quality of Service Metrics in
Wireless Sensor Networks with Mobile Sink and Cluster based Routing Protocols. In Proceedings of the 2017 International
Conference on Internet of Things, Embedded Systems and Communication (IINTEC 2107), Gafsa, Tunisia, 20–22 October 2017.
24. Yun, Y.; Xia, Y. Maximizing the Lifetime of Wireless Sensor Networks with Mobile Sink in Delay-Tolerant Applications. IEEE
Trans. Mob. Comput. 2010, 9, 1308–1318. [CrossRef]
25. Silva, R.; Silva, J.S.; Boavida, F. Mobility in wireless sensor networks—Survey and proposal. Comput. Commun. 2014, 52, 1–20.
[CrossRef]
26. Shrivastava, P.; Pokle, S.B. A multiple sink repositioning technique to improve the energy efficiency of wireless sensor networks.
In Proceedings of the 15th International Conference on Advanced Computing Technologies (ICACT ’13), Rajampet, India, 21–22
September 2013; p. 1.
27. Lee, E.; Park, S.; Yu, F.; Kim, S.-H. Exploiting mobility for efficient data dissemination in wireless sensor networks. J. Commun.
Networks 2009, 11, 337–349. [CrossRef]
28. Liang, W.; Luo, J. Network Lifetime Maximization in Sensor Networks With Multiple Mobile Sinks. In Proceedings of the IEEE
36th Conference on Local, Computer Networks, Bonn, Germany, 4–7 October 2011; pp. 350–357.
29. Wang, J.; Zuo, L.; Shen, J.; Li, B.; Lee, S. Multiple mobile sink-based routing algorithm for data dissemination in wireless sensor
networks. Concurr. Comput. Pract. Exp. 2015, 27, 2656–2667. [CrossRef]
30. Wang, J.; Li, B.; Xia, F.; Kim, C.-S.; Kim, J.-U. An Energy Efficient Distance-Aware Routing Algorithm with Multiple Mobile Sinks
for Wireless Sensor Networks. Sensors 2014, 14, 15163–15181. [CrossRef]
31. Jain, T.K.; Saini, D.S.; Bhooshan, S.V. Lifetime Optimization of a Multiple Sink Wireless Sensor Network through Energy Balancing.
J. Sensors 2015, 2015, 1–6. [CrossRef]
32. Anand, V.; Agrawal, D.; Tirkey, P.; Pandey, S. An energy efficient approach to extend network lifetime of wireless sensor networks.
Procedia Comput. Sci. 2016, 92, 425–430. [CrossRef]
33. Wang, J.; Cao, J.; Ji, S.; Park, J.H. Energy-efficient cluster-based dynamic routes adjustment approach for wireless sensor networks
with mobile sinks. J. Supercomput. 2017, 73, 3277–3290. [CrossRef]
34. Singh, P.; Singh, R. Energy-Efficient QoS-Aware Intelligent Hybrid Clustered Routing Protocol for Wireless Sensor Networks.
J. Sens. 2019, 2019, 8691878. [CrossRef]
35. Shiltagh, N.A.; Abdullah, M.Z.; Zarzoor, A.R. Evaluation of routing protocol with multi-mobile sinks in WSNs using QoS and
energy consumption parameters. Int. J. Electr. Comput. Eng. 2019, 9, 2880–2892. [CrossRef]
36. Li, Y.P.; Li, Y.; Wang, Y. High Performance Multi-Mobile Node Routing Communication Protocol based on Reliable Active
Node. In Proceedings of the 2020 6th International Symposium on System and Software Reliability (ISSSR), Chengdu, China,
24–25 October 2020; pp. 104–108. [CrossRef]
37. Kadiravan, G.; Sariga, A.; Sujatha, P. A Novel Energy Efficient Clustering Technique for Mobile Wireless Sensor Networks.
In Proceedings of the 2019 IEEE International Conference on System, Computation, Automation and Networking (ICSCAN),
Pondicherry, India, 29–30 March 2019; pp. 1–6. [CrossRef]
38. Wang, J.; Gao, Y.; Liu, W.; Sangaiah, A.K.; Kim, H.-J. Energy Efficient Routing Algorithm with Mobile Sink Support for Wireless
Sensor Networks. Sensors 2019, 19, 1494. [CrossRef] [PubMed]
39. Dhandapani, A.; Venkateswari, P.; Sivakumar, T.; Ramesh, C.; Vanitha, P. Cooperative self-scheduling routing protocol based
IOT communication for improving life time duty cycled energy efficient protocol in SDN controlled embedded network. Meas.
Sensors 2022, 24, 100475. [CrossRef]
40. Kaur, L.; Kaur, R. A survey on energy efficient routing techniques in WSNs focusing IoT applications and enhancing fog
computing paradigm. Glob. Transit. Proc. 2021, 2, 520–529. [CrossRef]
41. Mehta, D.; Saxena, S. MCH-EOR: Multi-objective Cluster Head Based Energy-aware Optimized Routing algorithm in Wireless
Sensor Networks. Sustain. Comput. Inform. Syst. 2020, 28, 100406. [CrossRef]
42. Wang, J.; Gao, Y.; Liu, W.; Sangaiah, A.K.; Kim, H.J. An Improved Routing Schema with Special Clustering using PSO Algorithm
for Heterogeneous Wireless Sensor Network. Sensors 2019, 19, 671. [CrossRef]
43. Wang, J.; Cao, J.; Sherratt, R.S.; Park, J.H. An improved ant colony optimization-based approach with mobile sink for wireless
sensor networks. J. Supercomput. 2018, 74, 6633–6645. [CrossRef]
44. Gao, Y.; Wang, J.; Wu, W.; Sangaiah, A.K.; Lim, S. A Hybrid Method for Mobile Agent Moving Trajectory Scheduling using ACO
and PSO in WSNs. Sensors 2019, 19, 575. [CrossRef]
45. Pundir, M.; Sandhu, J.K. A Systematic Review of Quality of Service in Wireless Sensor Networks using Machine Learning: Recent
Trend and Future Vision. J. Netw. Comput. Appl. 2021, 188, 103084. [CrossRef]
46. Osamy, W.; Khedr, A.M.; Salim, A.; El-Sawy, A.A.; Alreshoodi, M.; Alsukayti, I. Recent Advances and Future Prospects of Using
AI Solutions for Security, Fault Tolerance, and QoS Challenges in WSNs. Electronics 2022, 11, 4122. [CrossRef]
Sensors 2023, 23, 8534 23 of 23

47. Li, X.; Liu, A.; Xie, M.; Xiong, N.N.; Zeng, Z.; Cai, Z. Adaptive Aggregation Routing to Reduce Delay for Multi-Layer Wireless
Sensor Networks. Sensors 2018, 18, 1216. [CrossRef] [PubMed]
48. Xu, K.; Zhao, Z.; Luo, Y.; Hui, G.; Hu, L. An Energy-Efficient Clustering Routing Protocol Based on a High-QoS Node Deployment
with an Inter-Cluster Routing Mechanism in WSNs. Sensors 2019, 19, 2752. [CrossRef]
49. Poryazov, S.; Andonov, V.; Saranova, E.; Atanassov, K. Two Approaches to the Traffic Quality Intuitionistic Fuzzy Estimation of
Service Compositions. Mathematics 2022, 10, 4439. [CrossRef]
50. Castalia Wireless Sensor Networks Simulator. Available online: https://castalia.forge.nicta.com.au/index.php/en/ (accessed on
20 August 2023).
51. Cho, S.; Han, L.; Joo, B.; Han, S. P-LEACH: An Efficient Cluster-Based Technique to Track Mobile Sinks in Wireless Sensor
Networks. Int. J. Distrib. Sens. Networks 2014, 10, 803656. [CrossRef]
52. Darabkh, K.A.; Al-Maaitah, N.J.; Jafar, I.F.; Khalifeh, A.F. EA-CRP: A Novel Energy-aware Clustering and Routing Protocol in
Wireless Sensor Networks. Comput. Electr. Eng. 2018, 72, 702–718. [CrossRef]
53. Geng, F.; Xue, S. A comparative study of mobility models in the performance evaluation of MCL. In Proceedings of the 22nd
Wireless and Optical Communication Conference, IEEE, Chongqing, China, 16–18 May 2013. [CrossRef]
54. Camp, T.; Boleng, J.; Davies, V. A Survey of Mobility Models for Ad Hoc Network Research. Wireless Communications and
Mobile Computing (WCMC): Special issue on Mobile Ad Hoc Network Research: Research, Trends and Applications. Wirel.
Commun. Mob. Comput. 2002, 2, 483–502. [CrossRef]
55. Royer, E.M.; Melliar-Smith, P.M.; Moser, L.E. An analysis of the optimum node density for ad hoc mobile networks. In Proceedings
of the IEEE International Conference on Communications (ICC), Helsinki, Finland, 11–14 June 2001.
56. Tlili, S.; Mnasri, S.; Val, T. A multi-objective Gray Wolf algorithm for routing in IoT Collection Networks with real experiments.
In Proceedings of the 2021 National Computing Colleges Conference (NCCC), Taif, Saudi Arabia, 27–28 March 2021; pp. 1–5.
[CrossRef]
57. Mnasri, S.; Van Den Bossche, A.; Nasri, N.; Val, T. The 3D Redeployment of Nodes in Wireless Sensor Networks with Real Testbed
Prototyping. In Ad-hoc, Mobile, and Wireless Networks; ADHOC-NOW 2017. Lecture Notes in Computer Science; Puliafito, A.,
Bruneo, D., Distefano, S., Longo, F., Eds.; Springer: Cham, Switzerland, 2017; Volume 10517. [CrossRef]
58. Kalantari, M.; Shayman, M. Design Optimization of Multi-Sink Sensor Networks by Analogy to Electrostatic Theory. In
Proceedings of the IEEE Wireless Communications and Networking Conference (WCNC), Las Vegas, NV, USA, 3–6 April 2006;
pp. 431–438.
59. Ergen, S.C.; Varaiya, P. Optimal Placement of Relay Nodes for Energy Efficiency in Sensor Networks. In Proceedings of the IEEE
Inter-national Conference on Communications (ICC), Istanbul, Turkey, 11–15 June 2006.
60. Sheldon, M.; Chen, D.; Nixon, M.; Mok, A.K. A Practical Approach to Deploy Large Scale Wireless Sensor Networks. In
Proceedings of the IEEE MASS, Washington, DC, USA, 7 November 2005.
61. Tarawneh, A.S.; Hassanat, A.B.; Alkafaween, E.A.; Sarayrah, B.; Mnasri, S.; Altarawneh, G.A.; Alrashidi, M.; Alghamdi, M.;
Almuhaimeed, A. DeepKnuckle: Deep Learning for Finger Knuckle Print Recognition. Electronics 2022, 11, 513. [CrossRef]
62. Alkasassbeh, M.; Altarawneh, G.A.; Hassanat, A.B.A. On Enhancing the Performance of Nearest Neighbor Classifiers Using
Hassanat Distance Metric. arXiv 2015, arXiv:1501.00687. [CrossRef]

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