s

Available online at http://www.journalcra.com

INTERNATIONAL JOURNAL
OF CURRENT RESEARCH
International Journal of Current Research
Vol. 12, Issue, 04, pp.10176-10181, April, 2020

DOI: https://doi.org/10.24941/ijcr.38545.04.2020
ISSN: 0975-833X
RESEARCH ARTICLE

ENERGY EFFICIENT ALGORITHM FOR WIRELESS SENSOR NETWORKS BASED ON CLUSTER HEAD
SELECTION

Dr. Arunkumar1,*,*, J.R., Dr. Sundar Rajan M. 2 and Dr. Anusuya, R.3
13Assistant
Professor, Faculty of Computing and Software Engineering
2Associate
Professor, Faculty of Electrical and Computer Engineering
123Arbaminch Institute of Technology, Arbaminch University, Ethiopia

ARTICLE INFO ABSTRACT

Article History: This paper presents on key issue in WSN is to develop an energyenergy-efficient clustering Protocol with
Received 28th January, 2020 lower energy consumption for transferring information from channel heads to BS (Base Station). A
Received in revised form multi-level
level hierarchical structure performs in effectively route
route and collect data in WSN. Hierarchical
05th February, 2020 clustering algorithms are having the increased network lifetime. The primary point in these algorithms
Accepted 28th March, 2020 is the cluster head selection using midpoint algorithm to improve initial centroid selection procedure.
Published online 30th April, 2020 The proposed
proposed protocol works on selecting the best Cluster Head (CH) by considering some
parameters such as residual energy, distance between cluster heads, nodes and base station to work on
Key Words: long lifetime of network. Overall simulation performance is compared wit with standard algorithms and
Pulmonary Function Tests, Spirometry, achieved efficient energy consumption.
Swimming, Children.

Copyright © 2020, Arunkumar et al. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use,
distribution, and reproduction in any medium, provided the original work is properly cited.

Citation: Dr. Arunkumar,, J.R., Dr. SundarRajan M. and Dr. Anusuya, R.. R. 2020. “Energy
Energy Efficient Algorithm for Wireless Sensor Networks Based on Cluster Head
Selection”, International Journal of Current Research, 12, (4), xxx-xx.

INTRODUCTION the overall network energy consumption, balance the energy

consumption among the sensors and extend the lifetime of the
WIRELESS Sensor Network (WSN) (Lin-Feng Feng et al., 2014) is network. To reduce the amount of information that must be
collected with hundreds of sensor nodes that communicate transmitted to base station data fusion into routing protocol is
with low energy consuming routing protocols. The main issue followed. LEACH assumes each node in base station is
in WSN is the energy to be utilized in efficient way in order to transmitted directly with high power. LEACH K-MEANS
extend the lifetime in sensor networks. By considering the chooses k nodes as K clusters with calculated nodes cost. Node
model of sensor nodes (Celebi, 2013; Park et al., 2013) and having the smallest cost is selected as cluster head. Th
This cycle
their constraint lead to design energy efficient routing stops after fixing K-cluster heads with random distribution.
protocols. It should be low cost and low power sensors are able Following, channel heads gets overloaded and dies faster. The
to communicate in a limited range and work together to form a proposed Midpoint Leach K-means
means (MPLK
(MPLK-means) algorithm
sensor network for gathering data from a field. These sensors is to overcome the above issues through midpoint algorithm
carry out data processing and communication within
withi the nodes. (Krishnasamy, 2014) by accomplishing the selection process
It is enable to monitor and collect data in open medium and the with primary centroid and limited resource using kk-means. It
data is transmitted to electronic signals. Those signals are performs CH selection using residual parameter and provides
carried out to the base station (BS). The energy consumption is equal amount of sensor nodes to each cluster. Through this CH
the problem in energy efficient routing protocol. So should be gets balanced and gives better lifetime of nodes can be
simple and light weighted. According to the topology of achieved. It reduces energy consumption of CHs for data
network Wireless sensors networks routing protocols, LEACH communication. The threshold residual energy is defined to
(Low Energy Adaptive Clustering Hierarchy) (Chen, 2006) is calculate the amount of energy required to receive, combine
the simplest hierarchical protocol carry out clustering approach and transmit the average number of sensor nodes iin the cluster.
with low energy consumption in a wireless sensor network. Energy consumption is reduced by keeping CHs and the BS
Clustering is the main factor in LEACH algorithm to reduce closer for communication. If the distance between selected CH
node and the BS is greater than some threshold distance, the
CH will not communicate to the BS directly. In this case,
*Corresponding author: Dr. Arunkumar, multi-hop
hop communication will occur via other CHs. As a
Assistant Professor, Faculty of Computing and Software Engineering.
Engineering result, it provides enhanced network lifetime in WSN.
10177 Arunkumar et al. Energy efficient algorithm for wireless sensor networks based on cluster head selection

The main goal of this paper is to enhance a distributed The given set of data with fixed knumber of values into
clustering protocol in wireless sensor networks. There are four disjoint clusters is designed. The algorithm processed with two
parameters considered as residual energy of node distance separate sections the first stage is to define k centroids for each
between node and base station and its neighboring nodes, cluster. The second stage to the given data set is fixed with a
along with selection of cluster heads. A combined function of point to the nearest centroid. Euclidean distance is generally
parameters with different weights carried out in each node to considered to determine the distance between data points and
receive a score. This is used to choose a Cluster heads CHP. the centroids. The first step is completed after all the points are
The parameters are weighed based on their effects on energy- included in clusters, and an early grouping is done and to the
consumption decrement in different rounds of clustering. new centroids are calculated again, as the inclusion of new
Aggarwal, N., et.al. proposed the midpoint algorithm (Jenq- points may lead to a change in the cluster centroids. Once k
Shiou Leu, 2015) which has been used for initial CH selection new centroids are found, a loop is generated to create a new
assuming that the data points contain only positive values binding between the same data points and the nearest new
Ashish Upadhyay, et. al. (2006) introduced a hierarchical centroid. As a result of this loop, the k centroids may change
clustering algorithm for sensor networks, called Low-Energy their position in a step by step manner with fixed centroids
Adaptive Clustering Hierarchy (LEACH) is a classical positions results in coverage demand. The k means algorithm
clustering routing in wireless sensor networks .However the is a popular method for automatically classifying vector-based
cluster-head selection in LEACH protocol is lack of balancing data. In k-means algorithm (Kumar, 2014; Periyasamy, 2016),
the whole network energy consumption, with the result that the initial centroids are randomly selected from input data set
low energy nodes run out of energy prematurely and decline provides the algorithm with best outcome. From a given set of
the network life. This paper analyses the effectiveness of data set compute the k means algorithm by considering first 3
LEACH protocol in cluster-head selection, and proposes an points as initial centroids. If suppose the last 3 points as the
improved energy balanced clustering algorithm. Prernaet al. initial centroids is selected and third time selects any 3
(Periyasamy, 2016; Kuila, 2014) authors proposed Energy arbitrary data points as initial centroids and finally compute the
Efficient Clustering Algorithm (EEAC)for WSN. k-means algorithm. Each time the end clustering results will be
different by analyzing which one is the most suitable result.
This algorithm is based on static clustering concept and This is the limitation which needs to be dealt with in order to
dynamic cluster heads selection technique, which divides the make the k-means algorithm more efficient.
entire network area into a number of fixed regions. Cluster
Proposed Clustering Algorithm: Inorder to produce the
head (CH) is selected that it reduces Communicating distance
initial centroids (Park, 2013) efficient mid-point algorithm is
between nodes, for this reason energy consumption reduces
proposed. The Midpoint Leach K-means (MPLK-means)
while transmitting the data from one node to another and
algorithm gives a regular method to establish the initial
increases network efficiency as energy consumed. The
centroids. It also produces good clusters using k-mean method,
proposed (EECA) algorithm gives better result than the
compared with random selection of initial centroids. In
LEACH, LEACH-C and DR scheme with respect to
proposed algorithm, if the input data set contains the negative
throughput and stability. Jenq-ShiouLeu et al. (2014; Chamam,
value attributes, then all the attributes are transformed to
2010) proposed Energy Efficient Clustering Scheme for
positive space by subtracting each data point attribute with the
Prolonging the Lifetime of Wireless Sensor Network with
minimum attribute value in the data set. This transformation is
Isolated Nodes. A new regional energy aware clustering
required because in the algorithm the distance is calculated
method using isolated nodes for WSNs, called Regional
from origin to each data point. Then for different data points
Energy Aware Clustering with Isolated Nodes (REAC-IN). In
same Euclidean distance will be obtained to avoid incorrect
this algorithm cluster heads (CHs) are selected based on
selection of initial centroids. The next step is to calculate the
weight which is determined according to the residual energy of
distance of each point from the origin. After Then the original
each sensor and the regional average energy of all sensors in
data points are sorted into k equal sets. The mid-point is
each cluster and improves the cluster head selection process
calculated in all over the process. All the mid-points are taken
and solves the problem of node isolation, lifetime and stability
as the initial centroids. The following section explains about
of a network is more favorable. The rest of paper is structured
the algorithm carried out in balanced cluster formation and the
as follows: Section II discusses about the issues in existing
evaluation of clustering parameters in node.
algorithms to be overcome. Section III Provides complete
understanding of our proposed hierarchical clustering
Selection of Cluster Head: The midpoint algorithm
algorithm with the ability of cluster head Selection. Section IV
(Heinzelman et al., 2002) which has been used for initial CH
provides the simulation set up and result analysis respectively.
selection assuming that the data points contain only positive
Section V gives concluding remarks.
values. Here obtained from (1). Fig. 2 shows an example of a
Problem Statement: There is significant research effort for particular cluster of ten nodes where initial CHs have been
the development of energy efficient routing protocols in selected through the midpoint algorithm. Here the centroid of
Wireless Sensor Networks. several factors that should be taken desired number of clusters kept is cluster is a virtual node
into account when designing energy efficient routing protocols locating at the center position of the cluster. In this figure,
for WSNs. Energy efficiency is the most important factor, as it initial CH is denoted by encircled sensor node. To maintain the
directly affects the lifetime of the network.. Some of the connectivity of the network, residual energy of the CH is
existing approaches for clustering based on K-means algorithm checked every round. If the energy of the CH is smaller than
are discussed as follows: the threshold energy, the node having the next ID number is
selected as a new CH as in (Kumar, 2014). The newly elected
Leach K Means: In Low Energy Adaptive Clustering CH informs other nodes about the change of the CH. All other
Hierarchy (LEACH) (Lin-Feng, 2014), a hierarchical protocol remaining node joins the nearest cluster channel head and the
in which most nodes transmit to cluster heads. This section centroid of each cluster is calculated.
describes the original k-means clustering algorithm.
10178 International Journal of Current Research, Vol. 12, Issue, 04, pp.10176-10181, April, 2020

ID number is given to each node in which smaller number is the number of cluster members with number of neighbors
given to nearest one. The selected cluster head is compared to nodes close to the threshold will have the higher chance to be
residual energy is to be greater than threshold energy and node selected as CH.SP4 is defined by Eq. (4).
parameters as follows:
( )
=1− (4)
Calculation of Residual Energy Between Nodes: The CH
node consumes more energy than its member nodes (8). They
SP4 will be equal to 1 if the number of the node’s neighbors
combine data together and send them to the BS. Therefore, the
counting is equal to the optimal number of neighbors
clustering process assigns CH nodes with the highest residual
threshold), and less than 1 if counting is either higher or lower
energy. So, score parameter (SP) is defined as a criterion for a
than the threshold.The threshold is obtained as follows:
node with higher residual energy to obtain more score than
other nodes to be qualified as the cluster head. SP1 is defined
as follows:
ℎ ℎ = (5)
= (1)
where, countnig denotes the number of neighboring nodes of
Where, Ecurrent denotes the volume of residual energy and each node and threshold stands for the optimal number of
Emax is the maximum volume of the energy of a node when it neighbors for each node. The combination of SP1, SP2, SP3
is fully charged. Since residual energy appears in the and SP4, score function (SF) for CH selection is designedas
numerator, nodes with higher residual energy b\is selected as per the midpoint algorithm sensor nodes are collected into
CH. nearest channel heads and the process is carried out by
following condition
Distance Between Node And Bs: After the selection of CH it
should carried out in collecting data from member nodes, if (Residual energy of cluster head ≥ E threshold) and (SP1) and
deleting repeated data, conducting a limited processing (SP2) and (SP3) and (SP4)
operation on the data, and finally sending them to the BS (20). then
The CH selects its nearby node to the BS so the energy
consumption will be lower. It results in increase of CH lifetime The node will remain as cluster head
and the duration of the steady-state phase. In each round of
clustering that allows to lower energy consumption for Else
reselecting CHs. Thus, nodes that are closer to the BS more
likely become a CH. SP2 is defined by below equation as a Check ID number of all sensor nodes in that cluster
criterion for the closer node to BS, in order to obtain more The node I the next order of ID number is selected as a new
score than other nodes as a prerequisite to become a cluster CH. Then the newly elected channel heads inform other nodes
head. about the change until the cluster heads are not changed any
more. After then sensor nodes sends data packet to their
channel heads and calculates the distance between elected CH
= 1− (2) by satisfying the following condition

where, dbs the distance between each node and BS. The if (dbs<dthreshold)
dimension of the network in this paper is 200 _ 200 and the
location of BS is 100 _ 250. Therefore, distance of the farthest then
node to BS is 269.25 m.
Cluster head directly connected to the BS
The Distance Between Node and Its Neighboring Nodes:
The direct relationship is carried out between distance and the Else
energy consumed for transmitting data in which each member
node sends the data nearby is sensed by its CH. SP3 is defined It selects the nearest neighbor cluster head whose dbsis less
by Eq. (3) as a criterion for the node that is closer to its than thedthreshold to communicate to BS
neighboring nodes to obtain more score than others as a pre-
requisite to become qualifies as a cluster head. In our approach, an estimation of threshold residual energy is
given. Here the threshold energy is the amount of energy
∑ required to receive, aggregate and transmit the average number
=1− (3)
of sensor nodes in the cluster.Sensor nodes send data packet.
From the proposed algorithm noticed if it is a higher energy
where, disi denotes the distance between the node and its ith level of the node indicates that it has a higher chance to
neighbor. dismax denotes the distance between the node and its become a CH.
remotest neighboring node and N is the number of neighboring
nodes. Since the total distance is the numerator and N _ dismax
is the denominator in Eq. (3), multiplied by a negative number,
SIMULATION RESULTS
the smaller the total distance between a node and its neighbors
is, the lower the SP3 is obtained Energy Model: Eelec is the power consumption of sending
and receiving, determined by the circuit. If the transmission
Number of Neighboring Nodes: The number and size of distance is less than threshold d0, the power amplifier Efs of
clusters should be balanced (7). It is must to set a threshold for free space model is adopted.
10179 Arunkumar et al. Energy efficient algorithm for wireless sensor networks based on cluster head selection

Fig. 3. Residual Energy Parameter in Reducing Energy

Consumption

Fig.1. Unbalanced clustering

Fig. 4. Comparison of Energy Consumption by Nodes

Fig. 2. Channel heads and centroids are selected using ID number

Table1. Parameters for clustering

Parameter Value
Number of sensor nodes (N) 100 Fig. 5CH Centroid in Clusters
Network size 100x100 m2
Base station (0,0)
Number of clusters (Kopt) 4,5
Initial energy of node 1J
Data packet 3200 bits
Eelec 50 nJ/bit
0.0013 pJ/bit/m’
10 pJ/bit/m2
Energy for data aggregation (EDA) 5 nJ/bit/signal
dBS 85_100 m
dCH dBS/2
dthreshold 88 m

Conversely, when the transmission distance is equal or greater

than the value, the power amplifier Ԑamp of multi path
attenuation model is adopted.

Ԑfs
d0 = (6) Fig. 6 Distribution of Channel Heads
Ԑamp
10180 International Journal of Current Research, Vol. 12, Issue, 04, pp.10176-10181, April, 2020

Parameter Selection for Simulation: Following table IEEE, 1st International Conference on Futuristic trend in
presents the list of parameter required for the simulation. The Computational Analysis and Knowledge Management.
proposed method is simulated using, MATLAB 7.7 and C Barati, H., Movaghar, A., Rahmani, A.M. and Sarmast, A.,
language. Table 1 is summarized with Simulation parameters. 2012. A Distributed Energy Aware Clustering Approach
As per given in base paper considering in d BS = 100, gets the for Large Scale Wireless Sensor Network. International
number of desired CH = 4 having 100 sensor nodes in 100 × Journal on Technical and Physical Problems of
100 m2 sensing region. As well as, our proposed approach has Engineering (IJTPE), 4(13), pp.125-132.
been compared with existing approaches (Chen, 2006; Barati Celebi, M.E., Kingravi, H.A. and Vela, P.A., 2013. A
et al., 2012; Jenq-Shiou Leu, 2015) with respect to different comparative study of efficient initialization methods for the
network parameters. By considering BS = 85, the number of k-means clustering algorithm. Expert Systems with
desired CH = 5 having 100 sensor nodes in 100 × 100 m2 Applications, 40(1), pp.200-210.
sensing region. It forms unbalanced cluster formation like 4- Chamam, A. and Pierre, S., 2010. A distributed energy-
clustered network. Here amongst the seven observations, it is efficient clustering protocol for wireless sensor
found that a particular cluster contains 33 nodes in cluster 4, networks. Computers & electrical engineering,36(2),
which is much higher than the average number of sensor nodes pp.303-312.
(20) to be present in a cluster. At the same time for same Chamam, A. and Pierre, S., 2010. A distributed energy-
observation 1, cluster 5 contains only 11 sensor nodes, which efficient clustering protocol for wireless sensor
is much lower than the average number of sensor nodes. Using networks. Computers & electrical engineering,36(2),
our proposed protocol, the seven observations a particular pp.303-312.
cluster contains maximum 24 nodes and minimum 16 nodes, Chen, H. and Megerian, S., 2006, April. Cluster sizing and
which is much closer to the average number of nodes to be head selection for efficient data aggregation and routing in
present (20) in a particular cluster is found. Therefore our sensor networks. In Wireless Communications and
proposed approach produces balanced cluster compared to Networking Conference, 2006. WCNC 2006. IEEE (Vol. 4,
existing methods. Simulations results are shown in Fig. 3 and pp. 2318-2323). IEEE.
4. Heinzelman, W.B., Chandrakasan, A.P. and Balakrishnan, H.,
2002. An application-specific protocol architecture for
Fig. 3 and 4 shows the energy consumption of the protocols in wireless microsensornetworks.IEEE Transactions on
different rounds. It is clear from above graph that proposed wireless communications, 1(4), pp.660-670.
protocol MPLK (Mid Point Leach K-means) outperforms Heinzelman, W.R., Chandrakasan, A. and Balakrishnan, H.,
LEACH (Low Energy Adaptive Clustering Hierarchy) and 2000, January. Energy-efficient communication protocol
EECF (Energy Efficient Cluster Formation) in energy for wireless microsensor networks. In System sciences,
consumption and network lifetime. Fig. 5 and 6 shows the 2000. Proceedings of the 33rd annual Hawaii international
selected CHs of proposed protocol MPLK (Mid Point Leach conference on (pp. 10-pp). IEEE.
K-means) are well distributed across the network due to cluster Jenq-Shiou Leu, Tung-Hung Chiang, 2015. ” Energy Efficient
formation proposed protocol outperforms LEACH (Low Clustering Scheme for Prolonging the Lifetime of Wireless
Energy Adaptive Clustering Hierarchy) and EECF (Energy Sensor Network With Isolated Nodes”, IEEE
Efficient Cluster Formation). Including the parameters: (1) communications letters, vol. 19, no. 2, february.
distance between node and BS and (2) distance between a node Krishnasamy, G., Kulkarni, A.J. and Paramesran, R., 2014. A
and its neighboring nodes in the CHs selection.The proper hybrid approach for data clustering based on modified
distribution of CHs across the network provides connected cohort intelligence and K-means. Expert systems with
CHs for multi-hop communications. applications, 41(13), pp.6009-6016.
Kuila, P. and Jana, P.K. 2014. Energy efficient clustering and
Conclusion
routing algorithms for wireless sensor networks: Particle
swarm optimization approach. Engineering Applications of
In this paper, by considering the advantages and disadvantages
Artificial Intelligence, 33, pp.127-140.
of previous works, proposed a MPLK (Mid-Point Leach K-
Kumar, D., 2014. Performance analysis of energy efficient
means) new Energy-Efficient Clustering algorithm for
clustering protocols for maximizing lifetime of wireless
Wireless Sensor Networks. Evaluated the performance of
sensor networks. IET Wireless Sensor Systems, 4(1), pp.9-
proposed scheme and compared with standard protocols (Low
16.
Energy Adaptive Clustering Hierarchy) and EECF (Energy
L. Periyasamy, S., Khara, S. and Thangavelu, S., 2016.
Efficient Cluster Formation) in different network scenarios,
Balanced cluster head selection based on modified k-means
simulation result shows that the proposed scheme provides
in a distributed wireless sensor network. International
better results than the other protocols with respect to low
Journal of Distributed Sensor Networks, 12(3), p.5040475.
energy consumption and increase in lifetime of nodes. It also
Lin-Feng, L., Ping, G., Juan, Z. and Ning, L., 2014. An
provides to select cluster head from the selected cluster heads
Improved LEACH Protocol in Wireless Sensor
of cluster and try to consume more energy with less delay.
Networks. Applied Mechanics & Materials, 743.
Park, G.Y., Kim, H., Jeong, H.W. and Youn, H.Y. 2013.
REFERENCES March. A novel cluster head selection method based on K-
means algorithm for energy efficient wireless sensor
Aggarwal, N., Aggarwal, K.A. A mid–point based k-mean network. In Advanced Information Networking and
clustering algorithm for data mining’, Int. J. Comput. Sci. Applications Workshops (WAINA), 2013 27th
Eng., 2012, 4, (6), pp. 1174–1180 International Conference on (pp. 910-915). IEEE.
Ashish Upadhyay, Raghvendra Kumar, Sujeet, K. Tiwari, Park, G.Y., Kim, H., Jeong, H.W. and Youn, H.Y., 2013,
2015. ”Modified LEACH Protocol for Sensor Network”, March. A novel cluster head selection method based on K-
means algorithm for energy efficient wireless sensor
10181 Arunkumar et al. Energy efficient algorithm for wireless sensor networks based on cluster head selection

network. In Advanced Information Networking and Sha, K., Gehlot, J. and Greve, R., 2013. Multipath routing
Applications Workshops (WAINA), 2013 27th techniques in wireless sensor networks: A survey. Wireless
International Conference on(pp. 910-915). IEEE. personal communications,70(2), pp.807-829.
Prerna, Sanjay Kumar, 2015. ”Energy Efficient Clustering Singh, S., Chand, S., Kumar, R., Malik, A. and Kumar, B.,
Algorithm for WSN”, IEEE, 2nd International Conference 2016. NEECP: Novel energy-efficient clustering protocol
on Signal Processing and Integrated Networks. for prolonging lifetime of WSNs. IET Wireless Sensor
Systems, 6(5), pp.151-157.

*******