Received November 3, 2020, accepted November 8, 2020, date of publication November 16, 2020,

date of current version November 30, 2020.

Digital Object Identifier 10.1109/ACCESS.2020.3038031

An Energy-Efficient Mobility-Based
Cluster Head Selection for Lifetime
Enhancement of Wireless Sensor Networks
SEHAR UMBREEN, DANISH SHEHZAD, NUMAN SHAFI ,
BILAL KHAN, AND USMAN HABIB
Department of Computer Science, National University of Computer and Emerging Sciences, Islamabad, Chiniot-Faisalabad Campus, Chiniot 35400, Pakistan
Corresponding author: Numan Shafi ([email protected])

ABSTRACT Wireless based sensor networks contain sensors for environment monitoring but have restricted
resources. Many clustering protocols are designed to prolong network lifetime but have problems of
inadequate cluster head selection criteria, fixed clustering, and static rounds which consume more energy.
It is needed to develop an adaptive clustering strategy for better CH selection and load balancing. In this
article, we introduced an energy-efficient mobility based cluster head selection mechanism to overcome these
limitations. CH selection is based on dedicated parameters that have a huge impact on the sensor energy
consumption. The weightage of each node is calculated on the base of the node’s mobility level, residual
energy, distance to sink, and density of neighbors. Inter-cluster communication uses single-hop/multi-hop.
MATLAB is used to perform simulations. Results show that the proposed approach EEMCS performs better
as compared to the existing algorithms CRPD, LEACH, and MODLEACH in terms of load balancing,
network stability, energy depletion, and throughput. Energy utilization in the case of EEMCS is much less
and the network lifetime is greater than other existing protocols.

INDEX TERMS Wireless sensor network, cluster, LEACH, sensor nodes, mobility, cluster head, routing,
homogeneous.

I. INTRODUCTION the lifetime of the network; but in a dynamic environment,

Wireless sensor network (WSN) is composed of small, tiny nodes are free to move from one position to another. The
sensors. A sensor consists of a processing unit, energy unit, transmission of collected data to the base station from the sen-
and modules for wireless communication. Sensor nodes are sor nodes can be done directly or through other intermediate
responsible for monitoring the environment, record the phys- nodes. Mobility management here is quite a problematic task
ical condition, and send aggregated data to base stations as the movement of sensors is usually unpredictable. Sen-
[1]. WSN has become a vital component of the modern sor networks should be capable enough to reorganize them-
era because of its escalating deployment in almost every selves and synchronization is crucial to make communication
field]. There exist several types of WSNs such as terrestrial, error-free and consistent.
underground, underwater, multimedia, and mobile [2], [3]. Due to the limitation of resources, routing in WSNs is
Real-time applications are also taking advantage of WSNs. more challenging as compared to other wireless networks.
WSNs are used for battlefield monitoring, critical missions’ Therefore, new routing mechanisms are designed from time
surveillance, and enemy movement detection [4]. In the to time in the context of the challenges faced during topo-
health sector, there are many applications; sensors are being logical formation. Inappropriate topology formation results
used for monitoring, testing, disease detection, and diagnos- in more energy consumption and a decline in the network’s
tics [5]. In traffic control, monitoring and tracking of vehicles lifetime. To design a well-organized routing protocol, many
can be done with the help of WSN. influencing factors are important such as node’s energy, data
Nodes in a network can be static or dynamic [6]. In the storage, distance from a base station, time constraints, node’s
static case, all the nodes are fixed at their position throughout mobility, etc.
Routing protocols based on their network structure can be
The associate editor coordinating the review of this manuscript and categorized into flat and hierarchical protocols [2]. In flat
approving it for publication was Noor Zaman . networks, all the nodes have the same role and may use a

This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 License.

VOLUME 8, 2020 For more information, see https://creativecommons.org/licenses/by-nc-nd/4.0/ 207779
 S. Umbreen et al.: Energy-Efficient Mobility-Based CH Selection for Lifetime Enhancement of WSNs

TABLE 1. List of the abbreviation.

FIGURE 1. Clustered wireless sensor network.

single-hop for intra-cluster communication (within-cluster

flooding mechanism that causes high bandwidth consump- data transmission) and multi-hop for inter-cluster commu-
tion and excessive energy utilization. In the hierarchical nication (between clusters data transmission) based on the
protocol, nodes are clustered into groups based on some distance.
key criteria; and the optimum node among a specific group The hierarchical cluster-based protocol is used for routing
is selected as cluster head. Clustering is one of the most as it consumes less energy as compared to other similar
popular mechanisms used for topology formation to prolong protocols. The base architecture of the LEACH protocol is
the lifetime of WSN [7]. Figure 1, shows a typical clustered considered for the implementation of enhanced CH selection.
wireless sensor network [8]. A clustering technique orga- Simulations were performed in MATLAB to determine the
nizes the nodes in a well-structured form based on a set of performance of the proposed scheme and comparison is done
pre-defined criteria. The hierarchical clustering protocol is with other existing clustering protocols.
divided into different phases that include cluster head selec- This article is organized as follows: Section 2 covers the
tion, cluster creation, data collection, and data transmission. related work regarding the selection of the cluster head and
An appropriate clustering mechanism can ensure efficient dynamic clustering. The flow of the proposed methodology
utilization of resources which results in terms of prolonging is presented in Section 3. The architecture of the cluster
the network’s lifetime. Many clustering techniques including head selection mechanism and parameters that we consid-
LEACH, LEACH-M, CRPD, MBC, MODLEACH, etc. have ered for the development of improved algorithms are also
been designed to extend network lifetime [9], [10]. However, mentioned in this Section. Section 4 includes the evaluation
they have limitations such as fixed clustering, inadequate metrics that are used to make comparisons between the state
cluster head selection criteria, and static rounds, etc. It is of the art research work and the proposed methodology.
needed to develop an adaptive clustering strategy for better A detailed results comparison is provided to show the promi-
CH selection and load balancing. nence of the proposed algorithm. Finally, Section 5 provides
The main objective of the proposed research is to pro- the conclusion of this research and gives the future direction.
vide an energy-efficient hierarchical clustering mechanism
(EEMCS). This mechanism reduces energy consumption, II. RELATED WORK
prolong the lifetime of a wireless sensor network in a In WSNs, the limited battery power of sensors, severe envi-
dynamic environment. EEMCS provides an efficient clus- ronmental situations, heterogeneity, quality of service, pro-
ter head (CH) selection mechanism by considering dif- tection of the nodes, and fully functional deployment are a
ferent key parameters. Moreover, EEMCS can handle the few of the challenges [11]. To retain the network alive and
mobility of sensors to make an adaptive routing protocol to prolong the lifetime of WSNs, multi-constrained routing
for enhancing the lifetime of the network. This research mechanisms have been designed in the literature. Based on
is particularly focused on the best cluster head selection the network structure, the routing mechanism is classified
by considering major factors that directly affect the energy into two categorize. 1. Flat networks 2. Hieratical networks.
consumption of sensor nodes. For data transmission, we use In flat networks, no distinct role is assigned to the nodes

207780 VOLUME 8, 2020

S. Umbreen et al.: Energy-Efficient Mobility-Based CH Selection for Lifetime Enhancement of WSNs

and the flooding mechanism causes more energy consump- the next round and no data receive during the marked time
tion. While in hierarchical routing, nodes have distinct roles slot then this node is considered as moved to another cluster.
based on several specified rules. Based on predefined criteria CH removes this node from its cluster list and reschedules the
best node is selected as CH in the cluster [9]. Cluster-based slots for other nodes. A node can join a new cluster based on
routing is the most widely used mechanism to attain effec- the Receive signal strength indicator (RSSI) of the broadcast
tive communication through efficient utilization of resources. message. CH adds the new coming node and updates the clus-
Better node management, load balancing, and scalability can ter members list, TDMA schedule, and broadcast an updated
also be achieved by using a clustering approach [12], [13]. schedule in its cluster.
By considering the diverse challenges of WSNs the main aim LEACH Mobile Extended [21] is an extension of Mobile
of researchers is to reduce energy consumption and ensure LEACH and suitable for a large area. In this protocol, CH is
energy-efficient data transmission or communication. Several selected on the base of node mobility. A node that has
hierarchical cluster-based routing protocols were developed the minimum movement and has a low reduction of signal
to make the network energy efficiency and prolong the life- strength during data transmission is selected as CH. Then the
time of the network by considering different parameters such cluster formation is the same as discussed in previous LEACH
as energy, distance, coverage, mobility, density, etc. In the lit- variants. During the transmission phase if the node or CH
erature, many algorithms were proposed to deal with energy moves then the node sends a disjoint request to its CH and join
efficiency problems in specific domains [14]–[18]. Still, request to the new cluster head. TDMA schedule is updated
researches are conducted for combining different domains after joining or disjoining a node.
and develop a hybrid model that is more energy-efficient and LEACH-MAC [22] (Medium Access Control) is designed
prolong the network lifetime. Proposed mechanism results to restrict the number of CH advertisement messages and
are compared with the current models that are best for solving calculate an optimal number K of cluster heads using the
such problems. A few of these protocols are discussed in this following equation. s
article. N Efs M
LEACH [19] is an efficient and simple round-based low K= − 2
(2.2)
25 Emp dtoBS
energy adaptive clustering protocol for WSN. LEACH uses
a probability Equation for calculating cluster head. The where N is sensor nodes, M is the sensing area, the average
Equation is as follows: distance between the base station and cluster heads is d2toBS .
popt This protocol uses a random but uniform time interval for

   if
1 all nodes to listen to CH broadcast message. If any CH broad-
T (s) = 1 − popt r mod popt (2.1)
cast message is received in this interval of time and cluster
/G

0 if s ∈

headcount is less than the optimal number then this node
where Popt is the pre-determined percentage probability of becomes cluster head itself otherwise it becomes a normal
cluster head and r is the present round number. G is the set of node. After the cluster head election, the whole process is the
all the nodes that have already been selected as cluster heads. same as LEACH.
This protocol operates in the form of several rounds that are CRPD [23] is a cluster-based routing protocol for dynamic
divided into two phases. wireless sensor networks to reduce energy consumption and
1. Set-up phase improve energy efficiency by update the network topology.
In the set-up phase cluster head (CH) is selected randomly. It includes four phases:
Each node gets a chance to become CH by using a probability 1. Neighbor discovery
equation. After the CH election, a broadcast message is sent In this phase neighbor of every node is calculated. Within
to all nodes to make clusters. In each cluster CH of that par- the transmission range of nodes, every node broadcasts a
ticular cluster make a schedule using Time-Division Multi- detect message to calculate its neighbor. Each sensor node
Access (TDMA) and this schedule to all its cluster members has two sets to store the information of neighbor nodes: Nbr
to avoid collision during data transmission. store Id of neighbor nodes and Nbr INFO store Id, coordinate,
2. Steady phase and residual energy of its neighbors.
In this phase sensed data transmission is done 2. Cluster head selection and cluster formation
within-cluster (Nodes send data to CH) and outside of cluster CH is selected on the base of residual energy and the num-
(CH transmit data to Base Station) using TDMA Schedule. ber of neighbor degree. The energy threshold is considered
LEACH-M [20] protocol improved data transmission rate 40% of the initial energy. A node with the largest neighbor
as compare to LEACH but consume more energy. It is formed degree and residual energy greater than the energy threshold
specifically for mobile nodes. The set-up phase is the same is selected as a cluster head. Then CH broadcast I-am-CH
as LEACH but the steady phase is different from the old message and nodes join their cluster head according to their
protocol. In a steady phase after every round, CH checks the signal strength.
list of received data and match it with its TDMA schedule, 3. Data aggregation and Route construction
if any node does not send data in its allocated time then After the cluster head, the formation data aggregation
that time slot is marked. If the same node remains absent in phase is started. Data can be sent to the base station using

VOLUME 8, 2020 207781

 S. Umbreen et al.: Energy-Efficient Mobility-Based CH Selection for Lifetime Enhancement of WSNs

a single-hop if the distance is less or can use multi-hop if TABLE 2. Summary of literature review.
the base station is far away. It also set time value time1 with
data transmission to prevent from getting duplication of
data.
4. Re-clustering and Re-routing
It can be done whenever a node dead, moves to a new
location, or join a cluster. First, check whether it is CH or nor-
mal node and then re-clustering is done according to node
type and condition.
MBC [24] is a mobility-based clustering protocol. In which
cluster head selection is based on the residual energy and
mobility of the node. We can modify the probability Equa-
tion for electing the cluster head by introducing the velocity
factor in it. Here Ecurr is the current energy of the node
and Emax is the maximum energy of the node. Vmax repre-
sents the maximum velocity of the sensor node and Vcurr
is the current velocity of the node. T(s) is the probability
of cluster head selection [15]. The modified Equation is as
following [26]:
Ecurr Vmax −Vcurr
Tnew = T(s) ∗( ∗ ) (2.3)
Emax Vmax
MODLEACH [25] is a variant of the LEACH protocol. It is
based on the energy threshold for an efficient cluster head
formation scheme. If the energy of an existing CH is still
greater than the threshold then it will not change its status
and maintain its CH state for the next round. If the energy of
CH does not satisfy the threshold then it is replaced according
to the pre-existing algorithm LEACH.
We proposed EEMCS an Energy-Efficient Mobility based
Cluster head selection for lifetime enhancement of wireless
sensor networks, for better utilization of energy and increas-
ing the lifetime of the network using an improved cluster
head selection scheme. It is which is suitable for the dynamic
nature of nodes and mitigates the problem of fixed round
time for changing the cluster head. The cluster head selec-
tion process of the proposed method includes the residual
energy of sensor nodes, their mobility level, number of nodes
associated, and node’s distance from the base station. The
cluster formation is designed in such a way that whenever
a node is a move in the network field, they can easily join
a new cluster according to its new position and leave the
previous cluster without disturbing the whole cluster. If the
moving node is cluster head then it selects a new cluster
head from its cluster and then leaves that cluster. In this
way, we can save the cluster from destroying when it’s clus-
ter head move. Table 2 shows a summary of this literature
review.
III. METHODOLOGY
The novel EEMCS protocol can be divided into several
phases that include cluster head selection, cluster creation,
data collection, and data transmission. However, the main A. ARCHITECTURE
focus of this work is on the efficient cluster head selection The main architecture of the EEMCS mechanism is based on
which results in the improvement of the overall life span of the Network Development Model and Radio Energy Model.
WSN. In this section proposed methodology is described in For the energy dissipation of nodes, the radio energy model
detail. is used. The network model is used to construct the routing

207782 VOLUME 8, 2020

S. Umbreen et al.: Energy-Efficient Mobility-Based CH Selection for Lifetime Enhancement of WSNs

FIGURE 2. First order radio energy depletion model for Wireless Sensor
Network.
mechanism to collect sensed data and transmit it to the base
station.
1) RADIO ENERGY MODEL
As we already discussed that all the sensor nodes have a
restricted amount of resources and we aim to increase the net-
work life through efficient energy consumption techniques.
Figure 2 shows the first order radio model. In every WSNs
data communication consume most of the energy of sensors. FIGURE 3. Sensor nodes random deployment.
Therefore to calculate the utilization of energy in a network
we need to estimate the cost of energy both for data transmis- 2) NETWORK DEVELOPMENT MODEL
sion and data reception. So, in this research first order radio In this research, we have made some assumptions regarding
energy model is used to calculate the energy consumption of the network model and sensor nodes, which are as
transmitting and receiving data energy during the communi- following:
cation of sensor nodes and base stations. If the distance is • The network contains n number of nodes.
short between sensor nodes and cluster head then we use a • There is one base station in the network.
free space model otherwise the multi-path model is used for • The deployment of nodes is random.
communication. Energy consumption while transmitting ‘k’ • All the sensor nodes are mobile.
number of bits to a node placed at the distance ‘d’ meters is • Sensor nodes know their energy and position coordinates.
written as [21]: • The base station is static and placed outside the network.
ETx (k, d) = ETx_ele (k) + ETx_mp (k, d) (3.1) • The base station has an unlimited battery, storage, and
(
Eele ∗ k + Efs ∗ k ∗ d , d ≤ do
2 calculating ability.
ETx (k, d) = (3.2) • All sensor nodes have the same data transmission range.
Eele ∗ k + Emp ∗ k ∗ d 4 , d > do
• The initial energy of all the nodes is the same and nodes
ERx (k) = ERx_ele (k) + kEele = Eele ∗ k (3.3)
are homogeneous.
where: • The energy threshold is defined to diminish the uneven

Eele is the transmitter or receiver per bit energy deple- energy reduction effect on a node’s premature death
tion [33]. The threshold used in this article is 40 percent of
Emp is for multi-path fading the maximum energy of nodes in the network.
Efs is for the free-space • Nodes always have data to transmit according to the

ETx is the energy dissipation at the transmitter side TDMA schedule.

ERx is the per-bit energy dissipation at the receiver side
a: SENSOR NODES DEPLOYMENT
k is data in bits
First, the network area of 200 m x 200 m is created and
d is the distance between a transmitter and receiver
100 sensor nodes are arranged in the related field randomly.
node
All the nodes are free to move in any direction because the
It is also assumed that the ideal MAC layer is working for network is dynamic. Nodes deployment is shown in Figure 3.
collision-free transmission. When the transmission is done by We can change the network field and the number of sensor
single-hop, there is no relay node between sensors and base nodes as per requirement. The base station is set in the middle
stations. To calculate whether the node is using single-hop of the network area. Nodes are homogeneous and a unique
or multi-hop, we use ‘do ’. The threshold value for ‘do ’ is identity ‘ID’ is assigned to each sensor. When all nodes are
calculated as: deployed randomly, the BS broadcasts a ‘Hello’ message
s packet in the entire network. Each sensor after receiving the
Efs start message packet, send back their locations information,
do = (3.4)
Emp energy status to the base station.

VOLUME 8, 2020 207783

 S. Umbreen et al.: Energy-Efficient Mobility-Based CH Selection for Lifetime Enhancement of WSNs

b: RESIDUAL ENERGY CALCULATION extra energy. Therefore the mobility factor of sensor nodes
In each round, every node calculates its remaining energy. was considered and for that, mobility in the existing network
As nodes start transmitting and receiving data, their energy was calculated by changing the position of random nodes
starts decreasing and each node energy consumption is in the network. The node’s new position is also selected
different from other nodes. The change in energy deple- randomly. The mobility level of a node can be calculated
tion is according to their inter-communication or intra- by measuring the difference between the previous and new
communication distance. Residual energy is calculated using position of the node. The following Equation is used to find
the following formula [23]: out the level of mobility [34]:
Ecurrent q
Er = (3.5) ML = (xnew − xcurr )2 + (ynew − ycurr )2 (3.7)
Einitial
where (xnew , ynew ) are the coordinates of the sensor node at
c: DISTANCE CALCULATION the new position and (xcurr , ycurr ) are the coordinates points of
Distance plays a significant role in the energy dissipation nodes at the last calculated position. There is an inverse rela-
process of nodes. The nodes closer to the base station need tionship between a node’s mobility and chances for becoming
less energy to transmit the data as compared to the distant a cluster head; therefore a node having high mobility level has
nodes. In the EEMCS model, Euclidean distance is used to fewer chances to be selected as a cluster head.
calculate the distance between nodes. The distance between
3) CLUSTER HEAD SELECTION AND FORMATION
the member nodes and cluster head is represented by DtoCH
and the distance between the cluster head and the base station To select optimal cluster head many methods were proposed
is denoted as DtoBS. To calculate the distance between the sink in the literature, some of them have purely based on proba-
node, cluster head, and member nodes the following formula bility and some variants on defined threshold values for the
is used. selection of cluster head [18]–[20]. The optimal number of
cluster heads in the network is also important; better practice
q
Distance = (Xj − Xi )2 + (Yj − Yi )2 (3.6) is to choose cluster head based on parameters that directly
affect energy utilization as the main aim of CH selection is
d: SUITABLE NUMBER OF CLUSTERS
to reduce energy consumption and achieve optimal resource
In a sensor network, the communication cost between nodes utilization. In EEMCS, cluster heads are selected by consid-
is high. To provide a solution aimed at reducing communi- ering the significant parameters including nodes remaining
cation costs, nodes are arranged into groups. The challenge energy, mobility, distance to the base station, and neighbors
is how to divide the network into groups and how to calcu- count. After calculating the defined criterion for cluster head
late the optimal number of clusters. We observed from the selection, the energy of each node is checked against the
literature that the surplus number of clusters consumes more threshold value because a certain energy level is the first
energy and cause extra overhead. In EEMCS methodology, requirement to become a candidate for cluster head selection.
we utilized different ranges and checked suitable different In EEMCS energy threshold value is Threshold 40 percent of
ranges that can help to increase the lifetime of the network. the maximum energy of nodes in the network [23].
e: NEIGHBORHOOD DISCOVERY All the nodes that satisfy the energy threshold are eligible
After calculating the distance of nodes from the base station, for cluster head selection. The weightage of nodes is calcu-
the neighbor discovery phase starts. In this phase, all nodes lated based on their mobility level, remaining energy, their
broadcast a message to detect their neighbors within their distance to the base station, and neighbor degree (number of
radio transmission range. This detection message contains the nodes that are in their transmission range), and weights are
id of each node. All those nodes which receive the detect assigned to each parameter discussed above. The flow of clus-
message; stores the message as the sender node’s id in its ter head selection is shown in Figure 4. Let us consider that
neighbor list, and increase the neighbor count [23]. If a node node i have the maximum weightage among its neighbor and
is within the range of two cluster heads then the node joins satisfy the energy threshold; then node i broadcast ‘‘IMCH’’
that cluster head which is close as compared to other cluster message in its transmission range. For control packets, a new
heads. In this way, all nodes get information about their ‘1 bit’ field was introduced, which we call ‘‘IMCH’’. The
neighbors and neighbor count within their range. node sets its value to 1 if it wants to say ‘‘IMCH’’ otherwise
it is not set. Adding 1-bit field to the packet does not carry
f: MOBILITY significant overhead. All the nodes that receive the ‘‘IMCH’’
The mobility factor plays an important role in any network. message, store the ID of the sender node in a list. Those nodes
Nodes in WSNs can move at any time, in any direction. which are not able to become a CH, wait to receive ‘‘IMCH’’
A high level of mobility can destroy network structure message from their neighbors. This process is repeated until
and can be the reason for the unexpected behavior of net- all the nodes either receive ‘‘IMCH’’ message or change
work communication. If a cluster head changes its position their type to CH. After CH selection the next step is to join
rapidly then the whole cluster can die which can cause the appropriate CH. All nodes that receive CH messages,
resource wastage; moreover, frequent re-clustering consumes calculate their distance with each CH and send a join request

207784 VOLUME 8, 2020

S. Umbreen et al.: Energy-Efficient Mobility-Based CH Selection for Lifetime Enhancement of WSNs

FIGURE 4. Cluster head selection and formation flow diagram.

TABLE 3. Weights of different parameters for CH selection.

FIGURE 5. Cluster head formed in sensor network.

FIGURE 6. Weight calculation for cluster head selection.

to that CH which closer to the node as compared to other
cluster heads. Each node only joins one cluster head at a get the most stable, balanced, and energy-efficient cluster
time and two clusters cannot join each other. Cluster head head.
formation in a network environment is shown in Figure 5. The node with the highest weightage become a cluster
All nodes in the network join the nodes according to their head first. The cluster head then set a TDMA schedule for
distance and transmission range. Weights assigned to each its member node so they can send data when their allocated
parameter are given in Table 3. We selected that weightage time starts. The cluster head collects data from its cluster
which gives less energy utilization and greater network life- and sets the data to the base station using single hope or
time. Figure 6 shows the effect of different weights on energy multi hope based on the distance between the cluster head
utilization and network lifetime. These parameters help to and base station. The following Equation is used to calculate

VOLUME 8, 2020 207785

 S. Umbreen et al.: Energy-Efficient Mobility-Based CH Selection for Lifetime Enhancement of WSNs

the weightage of each node.

Er ∗ w1 + Degree ∗ w2
Weightage = (3.8)
ML ∗ w3 + DtoBS ∗ w4

4) RE-CLUSTERING CASES
As the network is based on dynamic nodes so we have to
face different scenarios because of node mobility. If a node
moves from one cluster to another cluster or a cluster head
is about to die then we need to change the network topology
and re-clustering is needed.
FIGURE 7. New node added into existing network.
Nodes failure due to energy consumption or other problems
can also become a reason for re-routing or re-clustering.
There could also a possibility that for some reason or net-
work requirement we need to add new sensor nodes in the
network. To resolve all these issues and make the network
keep working in real-time, we have to provide a suitable
algorithm that can deal with all these problems. Re-routing
and re-clustering situations are categorized into three cases
which are as follows.
a: NODE DEAD
This situation occurs when a network is running and sensor
nodes are kept transmitting data then their energy keeps
decreasing. There comes a time when the residual energy FIGURE 8. Node moved and join same cluster head.

is reduced enough so it is about to die. At this time when

the residual energy of sensor nodes is equal or less than c: NODE MOVE
the energy death threshold, the node needs to check its type As the network consist of mobile nodes and nodes are free
whether the node is a normal node or a cluster head. If it is to move in any direction. When a node changes its position,
a normal node, then it simply broadcast a dead message to we must check first its type. If the moving node is normal then
its neighbors with its ID only but if the dying node is CH it simply broadcasts a leave message which contains the id of
then before dying CH select another node from its cluster a moving node in its current position so its neighbors remove
to become a new CH. New CH selection is based on the the moving node from their neighbor list. And after changing
weightage. The dying cluster headfirst selects a node that position, the node broadcast joins message on a new position
has the highest weightage among its neighbors and its energy to neighbors within its transmission range. Here the new
also greater than the energy threshold. After that previous CH neighbors are considered and it joins the CH according to the
broadcast a dead message which contains the id of both the previous case that we have discussed above. But if the moved
new CH and dying cluster head. Now all nodes that receive node is CH then we need to find out a new CH before the
this dead message check the id, whether it is its CH id or a previous CH move from its cluster. Moving CH first selects
normal node id. If it is a normal node then the receiving a node from its cluster that satisfies the energy threshold
node deletes this node from its neighbor list. If the dead and has the highest weightage among its neighbors. Then the
message contains the id of the CH node then the receiving moving CH broadcast a leave message which contains the id
node replaces its CH with the new CH id and removes the of new CH and id of leaving CH.
dying CH from its list. All nodes that receive leave messages first check whether
the sender node is a normal node or CH. If it is CH then
b: NODE JOINING they replace the moving node id with the new CH id and
Due to some application requirements or other reasons, remove the moved CH from their CH list. But if the leaving
we need to add new sensor nodes to the existing network. node is normal then they just remove the moving node from
So, whenever a new node joins the network it broadcast a their neighbor list. The moved node then joins the cluster
join message within its transmission range. All the nodes according to its new position using the join mechanism dis-
that receive node join message, add the new node into their cussed above. Nodes movement and joining the cluster head
neighbor list, and replay back with their energy, coordinates, according to their new position is shown in Figure 8 and 9.
id, and CH id. In this way, the newly added node get the
information of its neighbors and store their information. Now 5) COMMUNICATION PHASE
the new node checks all the neighbours CH id and joins that After the selection of the cluster head and cluster formation
CH which is closer to the newly added node. A new node phase, the data communication phase starts. Now each node
added to the existing network is shown in Figure 7. sends data to their respective cluster head directly according

207786 VOLUME 8, 2020

S. Umbreen et al.: Energy-Efficient Mobility-Based CH Selection for Lifetime Enhancement of WSNs

FIGURE 9. Node moved and join other cluster heads.

to the TDMA schedule. When every node transmits data to

their cluster head, it’s time to send that data to the base station.
For this purpose, EEMCS uses acknowledgment messages to
ensure reliable data delivery and to avoid message repetition;
we used a time counter that how many times the data is sent.
After receiving data from cluster members, each cluster head
chose a path based on the minimum distance to the base
station [41]. CH check its distance from the base station,
if the distance is the minimum as compared to other nodes
that are in transmission range (TR) of CH then CH sends data
directly to BS; otherwise, it selects a node from its neighbor
list whose distance to BS is less as compared to the CH. Then
the neighbor node again checks from its neighbor that does
any node from its neighbor list exists with less distance as
compared to the node itself or not. If yes then pass the data
to that node otherwise send data to the base station. There
could be a chance that some nodes are deployed far away
and not in the range of all the cluster heads. For such nodes,
we use a multi-hop path for data transmission. Intra cluster
communication is done by using a single hop mechanism and
inter-cluster communication uses a multi-hop transmission
path for data delivery. Figure 10 shows the complete flow
chart of the proposed algorithm.
FIGURE 10. Flow diagram of proposed mechanism.
B. TIME COMPLEXITY ANALYSIS
Time complexity is used to estimate the total time required
all functions that are used in them. For cluster head selec-
to execute the algorithm and to evaluates the algorithm effi-
tion (CH) we need to calculate the neighbor degree which
ciency [35]. Each phase of the proposed method runs for
takes O(1) time, where 1 is network max degree. O(1) time
a particular time. For every node sending, receiving, and
is required for CH selection. Cluster formation requires O(1)
processing control messages is called a cycle. To calculate
time within one cycle each node can find out its role in the
the time complexity of the proposed scheme we need to find
network. We assume that the occurrence of re-clustering cases
out the complexity of each phase.
is represented as α (0 ≤ α ≤ n), where n is the total number
In CRPD [23] the time complexity of the complete
of sensor nodes in the network. Therefore, the worst-case
algorithm is the sum of the complexity of each phase.
time complexity of this phase is O (n). O(1) time is needed
In CRPD, the time required for neighbor discovery is O(1),
to aggregate the data from cluster members and O(d/TR) is
whereas in cluster head selection plus the cluster formation
the route construction time, where TR is the transmission
phase needs O(1) time for neighbor discovery, and O(1 +
range of each node, and d is cluster heads farthest distance
dd/Rce) time for route construction, and data aggregation
from BS.
required, respectively, where Rc is the communication range
and d is cluster heads farthest distance from BS. Similarly, C. EVALUATION METRICS
O(n) time is required for re-clustering. MATLAB platform is used to perform the different simula-
The time complexity of LEACH and MODLEACH is tions and evaluate the performance of the proposed method-
O(n) [44]. In contrast, our proposed algorithm has four phases ology. In EEMCS, evaluation is done on the basis of energy
and different functions are performed in every phase. So, efficiency, network lifetime, and performance of the proposed
the time complexity of each phase is calculated based on work and comparison is done with the existing mechanisms.

VOLUME 8, 2020 207787

 S. Umbreen et al.: Energy-Efficient Mobility-Based CH Selection for Lifetime Enhancement of WSNs

Key parameters are considered to be evaluation metrics and TABLE 4. Parameters used for simulations.
are discussed as following:
1) NETWORK LIFETIME
Network lifetime means that how many rounds a node
survives in the network. If any node survives for a longer
period in a network then it means network’s lifetime is
elongated.
2) ENERGY CONSUMPTION
These metrics give the estimation that how much energy is
consumed in one round, by all the nodes that are in the
network. One round is completed when all the nodes send
data to the base station.
3) AVERAGE ENERGY
The average energy is the valuation of the remaining energy
of all the nodes after the simulation.
4) THROUGHPUT
Throughput gives the total number of data packets transmitted
to the base station during each round.
IV. SIMULATIONS AND RESULTS COMPARISON
This section consists of all the simulation results and
performance comparison of EEMCS with existing and
well-known algorithms including LEACH [19], MODLEACH
[25], and CRPD [23]. We performed several simulations with
different initial energy and changed the number of nodes to
check the performance of our work. MATLAB is used for the
implementation of the proposed protocol.
A. PARAMETERS FOR SIMULATION
Simulations are performed by varying the number of nodes
between 50-250; to evaluate the effect of node density on
the entire network. Nodes are randomly distributed in the FIGURE 11. Cluster formation of 100-250 nodes.
network. The network area varies from 50 meters square to
200 m2 to check the effect of network scalability. The base of data packets sent to the base station in each round. Another
station is placed in the left upper corner of the network. metrics is the average energy, the sum of the remaining
Simulation parameters are listed in the following table. energy of all the nodes left collectively in the network to
Where Eo represents the initial energy of nodes, Eele , Efs , and keep the communication between nodes and base stations.
Eamp are energy depletion parameters of the first-order radio In Figure 11 (a), (b), (c) and (d) are the cluster formation of
model, and values given to these parameters are constantly 100, 150, 200 and 250 nodes respectively.
based on radio model standard. We use the same values of 1) NODE DENSITY EFFECT
node join and node move as in [23]. Node Join is the proba- The simulation was started initially with 50 nodes and
bility of new nodes to join the network and node movement within 50 x 50-meter square area to check the network per-
probability is represented by Node Move. formance of all algorithms. We kept increasing the number of
B. EXPERIMENT RESULTS nodes in the same region to analyze the effect of nodes density
Secondly, evaluation of the EEMCS protocol was done based on the network lifetime, energy utilization, and throughput.
on the stability and lifetime of network topology. For that, Figure 12 shows the network lifetime comparison using 100,
we examined that for how long a node can survive in the net- 150, 200, and 250 nodes.
work. We recorded the number of rounds when the first node Tests were also performed using the different number of
exhausted all of its energy. Performance is also evaluated on nodes in the same network area to check the node density
the base of the sum of energy consumption of all the sensor effect on the network. It was observed from simulations that
nodes for sending data packets to the base station in one with an increase in the number of nodes in the same region,
round. Energy overhead for sending control packets (adver- the first node of CRPD and EEMCS dies very early as com-
tisement message packets, acknowledgment packet, Hello pare to LEACH and MODLEACH in the case of 100 nodes
packets) are also included in the overall energy consumption. but the overall stability of the network of the EEMCS scheme
We estimated the throughput by calculating the total number is much better.
207788 VOLUME 8, 2020
S. Umbreen et al.: Energy-Efficient Mobility-Based CH Selection for Lifetime Enhancement of WSNs

TABLE 5. Network lifetime comparison statistics using 100, 150, 200,

250 nodes.

FIGURE 12. Nodes density effect within the area of 50 × 50.

FIGURE 13. Percentage of network life performance comparison.

Dense networks result in the decline of a lifetime of the

network in case of existing protocols; while it did not have
much effect on the performance of EEMCS. The percentage
analysis is shown in Figure 13. Statistical results comparison
is also given in Table 5. FIGURE 14. Control packets overhead comparison analysis.
The control packets overhead on the network are shown
in figure 14. The results show that overhead due to control at 29 rounds. The whole network in the case of EEMCS
packets increases for all the schemes along with the increase survives for a longer period. In the CRPD algorithm, all
in the number of nodes in the network. The overhead specif- nodes exhausted their energy after 278 rounds. The energy
ically for the LEACH is noteworthy as the network size consumption of the proposed algorithm after each round is
increases followed by MODLEACH and CRPD. It is worth quite less and the average remaining energy of the whole
noticing here that EEMCS, on the other hand, controls the network is high as compared to the existing schemes.
overhead with the increase in the size of the network. 3) TEST CASE 2
2) TEST CASE 1 In the second case, we increased the area of the network 100 x
In this case, we simulate EEMCS and other existing algo- 100 m2 but the initial energy level is the same 0.1 Joule.
rithms by using initial energy 0.1 Joule. At the start of the Figure 17 and 18 represents the result comparison of the case
simulation, the number of alive nodes were 50 and deployed 2 simulation. Here the results depict that after increasing the
randomly in a region of 50 x 50 m2 . Figure 15 and 16 shows area of the network, the lifetime of the network decreases a
the comparison of the result of case 1. bit in the case of EEMCS and CRPD but it has a great impact
It can be seen from results that the number of alive nodes on the lifetime of LEACH. Now the first node of the proposed
for EEMCS remains the same till 85 rounds; but on the work is dead after 81 rounds but EEMCS still performs better
other hand first node of the existing protocol, CRPD dies than the existing one.

VOLUME 8, 2020 207789

 S. Umbreen et al.: Energy-Efficient Mobility-Based CH Selection for Lifetime Enhancement of WSNs

FIGURE 18. Network survival comparison of case 2.

FIGURE 15. Network performance using 50 nodes in the 50∗ 50 TABLE 6. Simulation results statistics using 50 sensors, 0.1 J energy and
m2 network area. in 50∗ 50 network area.

TABLE 7. Simulation results statistics using 50 sensors, 0.1 J energy, and

in 100∗ 100 network area.

TABLE 8. Simulation results statistics using 50 sensors, 0.1 J energy, and

in the 150∗ 150 network area.
FIGURE 16. Network survival comparison of case 1.

remarkable difference between the energy consumption of the

proposed algorithm and the existing protocol. As we increase
the number of node in the network, the network performance
of the proposed algorithm increases gradually on the other
hand the performance of existing algorithms start decreasing.
4) TEST CASE 3
In this case experiments were performed using 50 nodes in the
region of 150 x 150 m2 network. We observe that in EEMCS,
the first node energy exhausted after 66 rounds as in the
FIGURE 17. Network performance using 50 nodes in the 100∗ 100 m2
network area.
previous case first node dead at 81 rounds and the first node
of CRPD and LEACH exhausted fast. In this case, the energy
The number of data packets delivery after increasing the consumption of both models is almost the same, but there is a
area of the network decreases but still the throughput of notable increase in throughput and average remaining energy
EEMCS is higher than the existing protocols. There is also a as compared to CRPD, LEACH, and MODLEACH. The

207790 VOLUME 8, 2020

S. Umbreen et al.: Energy-Efficient Mobility-Based CH Selection for Lifetime Enhancement of WSNs

TABLE 9. Simulation results statistics using 50 sensors, 0.1 J energy, and

in 200∗ 200 network area.

TABLE 10. Simulation results statistics using 50 sensors, 0.1 J energy, and
in 250∗ 250 network area.

FIGURE 20. Network survival comparison of case 3.

FIGURE 21. Network survival comparison of case 4.

FIGURE 19. Network performance using 50 nodes in the 150∗ 150 m2

network area.

throughput of existing protocols decreases with the increase

in the network area. Figures 19 and 20 show the comparison
of the results of case 3.

5) TEST CASE 4
In this simulation, we increase the network area to 200 x FIGURE 22. Network performance using 50 nodes in the
200 m2 and examine its impact on the lifetime of the network, 200∗ 200 m2 network area.
throughput, and energy consumption. Simulation runs on
50 nodes with an initial energy of 0.1 Joules. The network 6) TEST CASE 5
area is the same throughout the experiments. Figures 21 and In the last case, the network area was increased to 250 x 250
22 show the comparison of the results of case 4. Results m2 . Figure 23 and 24 show the comparison of the results of
show that after increasing the area four times, the lifetime case 5. We analyzed that the life of the EEMCS protocol is
of the network in the case of CRPD, LEACH, and MOD- greater than the other algorithms. There is more stability in
LEACH reduce to more than 50 percent of their initial life- the network of the proposed algorithm whereas in LEACH
time but in EEMCS protocol lifetime is reduce around 20- and CRPD first node dies within the first 10 rounds. EEMCS
30 percent. Energy consumption is almost the same and also performed better in the case of energy consumption
throughput is also increased as compared to other existing of sensor nodes and average energy utilization of nodes. In
algorithms. terms of throughput, the performance of LEACH is reduced.

VOLUME 8, 2020 207791

 S. Umbreen et al.: Energy-Efficient Mobility-Based CH Selection for Lifetime Enhancement of WSNs

because, at the end of each round, the energy consumed by

the nodes is less as compared to the CRPD, LEACH, and
MODLEACH. This is due to the selection of parameters
for CH selection which have a high impact on energy con-
sumption. However, in CRPD cluster head selection is based
on the number of neighbours degree and residual energy.
A CH with a maximum number of neighbours consumes
more energy as compared to a stable cluster head. Unneces-
sary re-clustering or cluster head selection in each round also
consumes additional energy. These problems were handled
in EEMCS technique by using a mechanism of adaptive
re-clustering where re-clustering is done only when it is
necessary. Results show that the throughput of EEMCS is far
better than CRPD. Whereas the throughput of LEACH and
MODLEACH is very less. This is due to the proper routing
mechanism and efficient clustering in EEMCS.
FIGURE 23. Network performance using 50 nodes in 250∗ 250 network
area.
V. CONCLUSION
Appropriate resource utilization in WSN is very important to
keep the network alive for a longer period, especially where
the network environment is composed of mobile nodes. In this
work, an efficient routing mechanism EEMCS is developed
for WSNs to prolong the network lifetime and reduce energy
consumption in a dynamic environment. This mechanism
overcomes the overhead of cluster formation by performing
re-clustering only when it is essential. EEMCS makes the
cluster formation adaptive and saves energy as compared to
other existing mechanisms where re-clustering is done after
a fixed round. Another advantage of the proposed scheme
is that it maintains a load balance over nodes during data
transmission. CH uses a relay node from its neighbor with
a minimum distance to the base station. Through EEMCS
FIGURE 24. Network survival comparison of case 5.
a stable and balanced network environment with efficient
energy utilization was achieved.
MODLEACH is better than LEACH but sent fewer packets
MATLAB has been used for the implementation of the
as compare to CRPD. EEMCS throughput is far better than
proposed mechanism and the results are compared with exist-
all three existing comparative protocols.
ing protocols CRPD, LEACH, and MODLEACH. The use of
C. COMPARATIVE EVALUATION EEMCS has resulted in an improvement of energy utilization
We took on average ten (10) simulations while changing by 20-30% as compared to CRPD and is more stable than
parameters to compare the performance of our work with the LEACH and MODLEACH in large scale networks. More-
results of the existing protocols. In every situation, the first over, the network survival rate of EEMCS on 150 nodes in
node of EEMCS survives more rounds as compared to CRPD a 50 × 50 m2 area is 77% whereas the network lifetime of
and the overall network lifetime is better as well in EEMCS. CRPD, MODLEACH, and LEACH is 62%, 58%, and 46%
In CRPD the energy of the node closest to the base station respectively. The energy consumption in EEMCS due to its
is consumed rapidly, this is because in CRPD when all the dynamic nature is quite less as a result nodes can sustain in
clusters collect data from their member nodes, they select the network for a longer period. The throughput of EEMCS is
a node that has the minimum distance to the base station, comparatively higher than the existing protocols in all cases.
from their neighbours. As we increased the network area In the future, we aim to extend the proposed work for
we see the network lifetime of LEACH and MOD-LEACH heterogeneous nodes as well. The mobility prediction model
decreases rapidly as compared to CRPD and EEMCS. In the can be used to predict the movement of sensor nodes that
case of LEACH and MOD-LEACH the reason for the fast will help to choose the most steady cluster head in terms of
decline of the energy of nodes is that each node that does not mobility.
belong to any cluster sends data directly to the base station
without considering any energy consumption factor. CRPD REFERENCES
and EEMCS are more stable than LEACH and MODLEACH
[1] M. Sedighimanesh, J. Baqeri, and A. Sedighimanesh, ‘‘Increasing wireless
but the performance of EEMCS is 20-30 % better than CRPD. sensor networks lifetime with new method,’’ Int. J. Wireless Mobile Netw.,
Moreover, EEMCS is better in terms of energy consumption vol. 8, no. 4, pp. 65–80, Aug. 2016, doi: 10.5121/ijwmn.2016.8404.
207792 VOLUME 8, 2020
S. Umbreen et al.: Energy-Efficient Mobility-Based CH Selection for Lifetime Enhancement of WSNs

[2] K. A. Darabkh and L. Al-Jdayeh, ‘‘A new fixed clustering based [23] S. Wang, J. Yu, M. Atiquzzaman, H. Chen, and L. Ni, ‘‘CRPD:
algorithm for wireless sensor networks,’’ in Proc. 14th Int. Wireless A novel clustering routing protocol for dynamic wireless sensor net-
Commun. Mobile Comput. Conf., Jun. 2018, pp. 71–76, doi: 10.1109/ works,’’ Pers. Ubiquitous Comput., vol. 22, no. 3, pp. 545–559, Feb. 2018,
IWCMC.2018.8450485. doi: 10.1007/s00779-018-1117-6.
[3] M. Elhoseny and A. E. Hassanien, ‘‘Hierarchical and clustering WSN mod- [24] S. Deng, L. Shen, and J. Li, ‘‘Mobility-based clustering protocol for
els: Their requirements for complex applications,’’ in Dynamic Wireless wireless sensor networks with mobile nodes,’’ IET Wireless Sensor Syst.,
Sensor Networks (Studies in Systems, Decision and Control), vol 165. vol. 1, no. 1, pp. 39–47, Mar. 2011, doi: 10.1049/iet-wss.2010.0084.
Cham, Switzerland: Springer, 2019, doi: 10.1007/978-3-319-92807-4_3. [25] D. Mahmood, N. Javaid, S. Mahmood, S. Qureshi, A. M. Memon, and
[4] T. Bala, V. Bhatia, S. Kumawat, and V. Jaglan, ‘‘A survey: Issues and T. Zaman, ‘‘MODLEACH: A variant of LEACH for WSNs,’’ in Proc.
challenges in wireless sensir network,’’ Int. J. Eng. Technol., vol. 7, no. 2, 8th Int. Conf. Broadband Wireless Comput., Commun. Appl., Oct. 2013,
pp. 53–55, Mar. 2018, doi: 10.14419/ijet.v7i2.4.10041. pp. 158–163, doi: 10.1109/BWCCA.2013.34.
[5] D. D. K. Rathinam, D. Surendran, A. Shilpa, A. S. Grace, and [26] Z. Zhao, K. Xu, G. Hui, and L. Hu, ‘‘An energy-efficient clustering routing
J. Sherin, ‘‘Modern agriculture using wireless sensor network (WSN),’’ protocol for wireless sensor networks based on AGNES with balanced
in Proc. 5th Int. Conf. Adv. Comput. Commun. Syst. (ICACCS), Mar. 2019, energy consumption optimization,’’ Sensors, vol. 18, no. 11, p. 3938,
pp. 515–519, doi: 10.1109/ICACCS.2019.8728284. Nov. 2018, doi: 10.3390/s18113938.
[6] K. Guleria and A. K. Verma, ‘‘Comprehensive review for energy effi- [27] A. Al-Baz and A. El-Sayed, ‘‘A new algorithm for cluster head selection
cient hierarchical routing protocols on wireless sensor networks,’’ Wireless in LEACH protocol for wireless sensor networks,’’ Int. J. Commun. Syst.,
Netw., vol. 25, no. 3, pp. 1159–1183, Apr. 2019, doi: 10.1007/s11276-018- vol. 31, no. 1, p. e3407, Jan. 2018, doi: 10.1002/dac.3407.
1696-1. [28] K. A. Darabkh and J. N. Zomot, ‘‘An improved cluster head selec-
[7] M. Tokala and R. Nallamekala, ‘‘Secured algorithm for routing the mil- tion algorithm for wireless sensor networks,’’ in Proc. 14th Int.
itary field data using dynamic sink: WSN,’’ in Proc. 2nd Int. Conf. Wireless Commun. Mobile Comput. Conf., Jun. 2018, pp. 65–70,
Inventive Commun. Comput. Technol. (ICICCT), Apr. 2018, pp. 471–476, doi: 10.1109/IWCMC.2018.8450446.
doi: 10.1109/ICICCT.2018.8473343. [29] N. Mazumdar and H. Om, ‘‘Distributed fuzzy approach to unequal cluster-
[8] S. Yahiaoui, M. Omar, A. Bouabdallah, E. Natalizio, and Y. Challal,
ing and routing algorithm for wireless sensor networks,’’ Int. J. Commun.
‘‘An energy efficient and QoS aware routing protocol for wireless sen-
Syst., vol. 31, no. 12, p. e3709, May 2018, doi: 10.1002/dac.3709.
sor and actuator networks,’’ AEU—Int. J. Electron. Commun., vol. 83,
[30] K. Cengiz and T. Dag, ‘‘Energy aware multi-hop routing protocol
pp. 193–203, Jan. 2018, doi: 10.1016/j.aeue.2017.08.045.
for WSNs,’’ IEEE Access, vol. 6, pp. 2622–2633, 2018, doi: 10.1109/
[9] H. El Alami and A. Najid, ‘‘ECH: An enhanced clustering hierarchy
ACCESS.2017.2784542.
approach to maximize lifetime of wireless sensor networks,’’ IEEE Access,
[31] M. Mirzaie and S. M. Mazinani, ‘‘Adaptive MCFL: An adaptive multi-
vol. 7, pp. 107142–107153, 2019, doi: 10.1109/ACCESS.2019.2933052.
[10] K. Haseeb, K. M. Almustafa, Z. Jan, T. Saba, and U. Tariq, ‘‘Secure clustering algorithm using fuzzy logic in wireless sensor network,’’ Com-
and energy-aware heuristic routing protocol for wireless sensor network,’’ put. Commun., vol. 111, pp. 56–67, Oct. 2017, doi: 10.1016/j.comcom.
IEEE Access, vol. 8, pp. 163962–163974, 2020, doi: 10.1109/access. 2017.07.005.
2020.3022285. [32] J.-Y. Lee, K.-D. Jung, S.-J. Moon, and H.-Y. Jeong, ‘‘Improvement on
[11] H. Mostafaei and M. S. Obaidat, ‘‘Learning automaton-based self- LEACH protocol of a wide-area wireless sensor network,’’ Multimedia
protection algorithm for wireless sensor networks,’’ IET Netw., vol. 7, no. 5, Tools Appl., vol. 76, no. 19, pp. 19843–19860, Oct. 2017, doi: 10.1007/
pp. 353–361, Sep. 2018, doi: 10.1049/iet-net.2018.0005. s11042-016-3732-4.
[12] Y. H. Robinson, E. G. Julie, R. Kumar, and L. H. Son, ‘‘Probability- [33] B. Kundaliya and S. K. Hadia, ‘‘Enhancing network lifetime with an
based cluster head selection and fuzzy multipath routing for prolonging improved MOD-LEACH,’’ Int. J. Electr. Comput. Eng., vol. 9, no. 5,
lifetime of wireless sensor networks,’’ Peer Netw. Appl., vol. 12, no. 5, pp. 3615–3622, Oct. 2019, doi: 10.11591/ijece.v9i5.pp3615-3622.
pp. 1061–1075, May 2019, doi: 10.1007/s12083-019-00758-8. [34] F. Belabed and R. Boouallegue, ‘‘Clustering approach using node mobil-
[13] B. Jan, H. Farman, M. Khan, and S. H. Ahmad, ‘‘An adaptive energy ity in wireless sensor networks,’’ in Proc. 13th Int. Wireless Com-
efficient scheme for energy constrained wireless sensor networks,’’ in mun. Mobile Comput. Conf., Jun. 2017, pp. 987–992, doi: 10.1109/
Proc. 34th ACM/SIGAPP Symp. Appl. Comput., Apr. 2019, pp. 2391–2398, IWCMC.2017.7986420.
doi: 10.1145/3297280.3297515. [35] Song, Liu, He, and Jiang, ‘‘Research on data fusion scheme for wireless
[14] H. Mostafaei, ‘‘Energy-efficient algorithm for reliable routing of wire- sensor networks with combined improved LEACH and compressed sens-
less sensor networks,’’ IEEE Trans. Ind. Electron., vol. 66, no. 7, ing,’’ Sensors, vol. 19, no. 21, p. 4704, Oct. 2019, doi: 10.3390/s19214704.
pp. 5567–5575, Jul. 2019, doi: 10.1109/TIE.2018.2869345.
[15] K. Cengiz and T. Dag, ‘‘Low energy fixed clustering algorithm (LEFCA)
for wireless sensor networks,’’ in Proc. Int. Conf. Comput. Netw. Commun. SEHAR UMBREEN, photograph and biography not available at the time of
(CoCoNet), Dec. 2015, pp. 79–84, doi: 10.1109/CoCoNet.2015.7411170. publication.
[16] K. Cengiz and T. Dag, ‘‘Multi-hop low energy fixed clustering algorithm
(M-LEFCA) for WSNs,’’ in Proc. IEEE 3rd Int. Symp. Telecommun.
Technol. (ISTT), Nov. 2016, pp. 31–34, doi: 10.1109/ISTT.2016.7918080. DANISH SHEHZAD, photograph and biography not available at the time of
[17] T. M. Behera, S. K. Mohapatra, U. C. Samal, M. S. Khan, M. Danesh- publication.
mand, and A. H. Gandomi, ‘‘Residual energy-based cluster-head selection
in WSNs for IoT application,’’ IEEE Internet Things J., vol. 6, no. 3,
pp. 5132–5139, Jun. 2019, doi: 10.1109/JIOT.2019.2897119. NUMAN SHAFI, photograph and biography not available at the time of
[18] P. Gou, F. Li, Z. Li, and X. Jia, ‘‘Improved LEACH protocol publication.
based on efficient clustering in wireless sensor networks,’’ J. Com-
put. Methods Sci. Eng., vol. 19, no. 3, pp. 827–838, Jul. 2019,
doi: 10.3233/JCM-190006. BILAL KHAN received the B.S. degree in computer science from the Depart-
[19] W. R. Heinzelman, A. Chandrakasan, and H. Balakrishnan, ‘‘Energy- ment of Computer Science, University of Peshawar, Pakistan, in 2005, the
efficient communication protocol for wireless microsensor networks,’’ M.S. degree in computer security from the University of Bradford, U.K., in
in Proc. 33rd Annu. Hawaii Int. Conf. Syst. Sci., 2000, p. 223,
2008, and the Ph.D. degree in computer networks from Dongguk University.
doi: 10.1109/hicss.2000.926982.
He worked as a Senior Researcher with the Center of Excellence in Infor-
[20] D.-S. Kim and Y.-J. Chung, ‘‘Self-organization routing protocol support-
ing mobile nodes for wireless sensor network,’’ in Proc. 1st Int. Multi- mation Assurance (CoEIA), King Saud University. He has been teaching in
Symposiums Comput. Comput. Sci. (IMSCCS), Jun. 2006, pp. 622–626, FAST-NUCES, Pakistan, since 2017. His research interests include devising
doi: 10.1109/imsccs.2006.265. performance efficient and energy-efficient MAC layer protocols for wireless
[21] G. S. Kumar, P. M. V. Vinu, and K. P. Jacob, ‘‘Mobility metric based local and personal area networks.
LEACH-mobile protocol,’’ in Proc. 16th Int. Conf. Adv. Comput. Com-
mun., Dec. 2008, pp. 248–253, doi: 10.1109/ADCOM.2008.4760456.
[22] P. K. Batra and K. Kant, ‘‘LEACH-MAC: A new cluster head selection USMAN HABIB, photograph and biography not available at the time of
algorithm for wireless sensor networks,’’ Wireless Netw., vol. 22, no. 1, publication.
pp. 49–60, Jan. 2016, doi: 10.1007/s11276-015-0951-y.

VOLUME 8, 2020 207793