IEEE International Conference On Recent Trends In Electronics Information Communication Technology, May 20-21, 2016, India

Energy Efficient Cluster Head Selection to

Enhance Network Connectivity for Wireless
Sensor Network
MadhuPatil, Chirag Sharma
AbstractWireless sensor based communication system is an enables them to connect through each other to form a wireless
ever growing sector in the industry of communication. Wireless infrastructure. WSNs is embedded with sensor node that has
infrastructure is a network that enables correspondence between capabilities to commutate and transmit information.The
various devices associated through a infrastructure protocol. In critical part of sensor node is to provide security due to device
enhancing the life span of network depends on the energy remote deployment nature and improve the lifetime of sensor
dissipation of the sensor devices. Reducing the energy dissipation
devices as they are battery powered.
of sensor devices will improve the lifetime and device failure
which help in better connectivity and coverage of sensor network. The low energy adaptive clustering hierarchy ()
There has been various approach that has been developed to strategy is a revolutionary protocol [1], [2]. The
improve energy efficiency of sensor network among that strategy forms many clusters of sensor devices and assign a
clustering is a significant technique that help in improving the cluster head (CH) node for each cluster, with the aim of
network lifetime by reducing the energy consumption but the reducing the energy depletion of . A centralized approach
issues with existing methodology is the energy inefficiency in of namely , where the base station takes
selecting the cluster head which result in loss of connectivity the cluster formation decision using all composed information
which reducing the life time of network. To overcome this here that provided better outcomes over the distributed approach in
the author proposes a design for cluster head selection based on terms of the bandwidth utilization and lifetime of network [1].
connectivity of node. The experimental result shows that the Recently may approach have been developed by enhancing
proposed PS-LEAH perform better than existing LEACH in term the . In [3], each cluster head sensor devices avoid
of network lifetime, number of active device. the energy exhaustion in its roles by assigning vice-cluster head
sensor devices in its cluster. [4] Considers the
KeywordsWsn, Clustering, data aggression, Hierarchy residual energy information of the candidate cluster head
clustering technique. sensor devices more actively. In [5] adopted a
refined clustering technique based on K-mean clustering
technique rather than simulated annealing [6] which is
I. INTRODUCTION considered to be an effective clustering methodologies. Below
In wireless sensor network, communication has enabled are some of the challenges faced in sensor networks
low cost, power, dynamic sensor devices for the development A. Issues and challenges faced in wireless sensor network
that are tiny and short communication range. The WSN used
tiny (small) sensor nodes, which consist of sensing data Communication among sensor devices: Huge amount
processing, and collaborating (communication) component. A packet transmission in wireless sensor network is merely not
sensor device is a device in a WSN which perform the task like practical contrasting to other wireless infrastructure devices
sensing activity for data transmission and interaction among such as router, network interface card, etc., since these sensor
sensor devices in the wireless sensor infrastructure. devices are designed to cater low power requirement.

We are living in an environment loaded with sensors Networking: Sensor networks which loaded with numerous
devices. Sensors are used for checking temperature, sensor devices often lead to volatile outcome due to topology
inhabitance, pollution and safety. Our automobiles contain dynamic as result it leads to an unbalance consumption of
handfuls if not many sensors, checking temperature status of energy, higher collision rates, and channel contention among
engines, braking, and traveler wellbeing hardware and so on. nodes.
Fabricating industries need sensors in light of the fact that you Multihop and clustering method has been considered by
can't control what you can't quantify. Manufacturing a product, various researchers that help in reducing one to one association
while meeting wellbeing, quality and proficiency targets, i.e. between the base stations and the devices. Though these
requires a great deal of sensors. Sensors have turned out to be methodology could not give assurance for integrity of data and
much tinier, less costly and low powered force in the most security or safety against data loss.
recent couple of decades.
Management: Sensor network require proficient protocol
A WSNs has key/important application/apps that are used strategy to extract their potential of following thing such as
for remote tracking ecological monitoring, detecting wildfire or power management, radio device interference management,
sign of any catastrophic phenomenon, intruder detection in battery management of sensor devices, hardware and software
military application. This has been enabled by the availability, bug management, node failure, data loss due node failure etc...
especially in recent years, of sensors that are small in size.
These nodes are furnished with wireless interface (WI) that Data Aggregation: In wireless sensor network routing of
information or packet needs a new secure and energy effective
framework/technique that aiding fusion of huge dimension of
Madhu Patil, Assoc Prof, Dept ECE, NMIT, Bangalore, India
([email protected] )
data. Data aggregation has to be done for high dimension data
is accurate that has constraint battery issue, assurance of data
Chirag Sharma, Nitte Research Education Academy, Research Guide, obtained etc
University of Mysore, Mysore

978-1-5090-0774-5/16/$31.00 2016 IEEE

175
 IEEE International Conference On Recent Trends In Electronics Information Communication Technology, May 20-21, 2016, India

There are various approach that has been developed energy consumption model as in figure 1. From figure 1 the
recently to enhance lifetime of sensor network which is amp represent the energy requirement of transmitter power
explained in section two and to overcome these shortcoming amplification to transmit a bit to the receiver at distance = 1
here the author propose an energy efficient design for cluster and denotes the energy required to handle trans-receiver
head selection based on connectivity of sensor devices. Our circuitry which is measured in (nano joule). Therefore the
method provide a cost effective solution by considering radio energy requirement of transmitting message bit to a distant
signal strength of sensor device for cluster selection and the receiving device can be evaluated considering equation (2) is as
message are broadcasted by using a based channel. follows
The paper organization is as follows: The proposed model
considering are presented in Section two. The results and the
experimental study are presented in the section three. The
concluding remark is discussed in the last section.
II. PROPOSED HIERARCHIAL CLUSTERING MODEL
Here the author adopts energy efficient design to for better
cluster selection that aid in improving network node
connectivity and improve lifetime of network. The proposed
model is compared with existing based clustering
protocol.
A. LEACH Working Model
In protocol each and every sensor device serve as
cluster head in round based on a random and unbiased manner. Fig.1: First order energy model
The protocol is further divided in to multiple rounds of
fixed time duration and each round is divided into two stage
and they are as follows the setup stage and the steady stage. In (, ) = ( + ). (2)
setup stage certain devices are chosen to behave as cluster
heads by adopting a distributed strategy performed in every
device. Consequently, the clusters equivalent to those chosen The energy required to obtain a bit of data by the receiver is
as cluster head are organized. In steady stage every device as follows
performs the sensing activity and generates the data and it is
forwarded to the corresponding cluster heads and then the () = . (3)
cluster head transmit this data to the sink or base station.
Selection of cluster head in is as follows. In every
round, for a particular sensor device a random uniform value C. Cluster Head Selection
between 0 and 1 is obtained and the obtained value is Generally the sensor devices are deployed in a random
compared with the threshold () of corresponded to this manner over the entire area for various applications , there exist
sensor devices. If the obtained value is less than (), then an overlapping region of the sensing areas of different sensor
this sensor devices elect himself as cluster-head in that devices. If the sensor device density of a local area is
particular round and the value of the threshold is updated in significantly lower than mean, than a target location is covered
each and every round. by only one sensor device and similarly when a local area is
significantly higher than mean sensor device density, a target
(1)
, ; location may be covered by several sensor devices. Therefore
() = { 1 [(1 )] in order to obtain overlapping area OF of two sensor device
0, . with distance , 0 2, apart is estimated using a
geometric theory of intersection of two circles which is as
follows.
Where represent mean ratio of cluster head in every round
to the total sensor devices, , 0 < , is the current round
number, and is the collection of sensor devices that has not
elected as cluster head of period 1/ rounds, that is, rounds
0~ 1 1, rounds 1 ~ 2 1 , and so on. Based on the
equation (1) every sensor devices behaves as for a
particular period in a round. In the next round this sensor
devices is removed for cluster head selection candidate.
B. System Channel Modelling
The author considers that there are devices of sensor
placed in a sensing field of area . The author consider a
random deployment of sensor nodes with in its area limit which
employ a routing strategy to forward the sensed data to the base
station which is placed outside the region of sensing area. The
sensor device are considered to be homogeneous in nature
which has same energy and sensing range () therefore the Fig2: intersection of two circles
sensing area is 2 . Here the author considers a first order

176
 IEEE International Conference On Recent Trends In Electronics Information Communication Technology, May 20-21, 2016, India

(4) Where depict the mean amount of cluster head. Now the
2 proposed threshold () considering sensor device can be
2
= 2 [ 1 ( ) ],
2 2 rewritten as follows
() (9)
, ;
() = { 1 () [(1())]
Where = cos 1 (2 ). To obtain a normalized
0, .
overlapping of sensor device in a network is as follows

Where is a collection sensor devices that has not yet been

= 2 = 2 [cos 1 () 1 2 ], (5) as a cluster head in the present round of period, is the sensor
node that behave as cluster head in round1(). Based on this
every sensor devices are selected to be cluster head with varied
Where 0 1 and = 2 and the value of is in likelihoods.
range of 0 and 1. The sensor device with a large quantum of ()is been
To enhance the cluster head selection to improve network given smaller likelihood of being cluster head where as a
lifetime by better network coverage different sensor device are sensor device with small () is given higher value of ()
assigned different likelihood of being cluster head and to that is given as input in equation 9 and the sensor device will
calculate this likelihood it depend on the normalized active behave as cluster head for a short period of time. Subsequently
sensing coverage area of sensor devices. The active sensing the sensor devices the higher () will aid in improving
region is represented as the ratio of active sensing coverage energy efficiency due to being cluster head, however the
region to the maximum sensing region of a sensor device. smaller value () sensor device will carry the burden of being
cluster head. From this it seen that the proposed cluster
Let us consider a sensor device that 0 % of this sensor selection strategy based on normalized active sensing range
devices area of sensing is enclosed by this sensor device only improved the network lifetime and its coverage since the
and out of that % of this sensor devices area of sensing is sensor device with smaller normalized active sensing range
enclosed by this sensor device and apart other adjacent dies first. The simulation and experimental proof of proposed
sensor devices. Now let us consider an instance that 3 % of an methodology is proved in next section below.
sensor devices area of sensing enclosed by 4 sensor devices
that include this sensor device and out of other 3 adjacent
devices. Therefore the normalized active sensing region of this
III. SIMULATION RESULT AND ANNALYSIS
sensor device is represented by following equation
The system environment used is windows 10 enterprises
(6) 64-bit operating system with 12GB of RAM. We have used
= 0 + . sensoria simulator which is a dot net based simulator that uses
+1
=1 C# as a programming language. We have conducted simulation
study on following parameter for network lifetime and
communication overhead and compared our proposed
The assessment of is in between (0,1). If the area of with existing based protocol and we have
sensing of a sensor device is overlapped by many sensor varied node size by 500 and 1000 and conducted simulation
devices the assessment of is less than 1 were as for the non- sturdy and the simulation parameter is shown in below table 1
overlapped with any other device than the assessment of is below.
equal to 1.
Consequently by using equation (5) and when = 2 , TABLE I. SIMULATION PARAMETER CONSIDERED
we obtain the normalized overlapping region () for sensor Network Parameter Value
device . Then based on equation (6) the normalized active
Network Size 30m * 30m
area of sensing is as follows
Number of sensor nodes 500, 750, 1000
1 () () (7) Number of Base station 1
() = 0 + = [1 ()] + = 1 .
2 2 2 Initial energy of sensor nodes 0.1 J
Radio energy dissipation 50 nj/bit
From the above equation it can be assured that a sensor Idle energy consumption (Eelec) 50 nj/bit
device with smaller value of () is been given higher
likelihood of being cluster head and similarly for larger () is Data packet processing delay 0.1 ms
been given the smaller likelihood of being cluster head. Amplification energy (Emp) 100 pJ/bit/m2
Therefore the threshold selection in equation (1) and Data packets length 2000 bits
the parameter is changed to be parameter proportional to a
sensor devices normalized overlapping region, i.e. a particular Transmission speed 100 bit/s
sensor device , we obtain the following value as follows Bandwidth 5000 bit/s

() = (), (8)
In In Fig. 3 and 4, we can see that the proposed
clustering technique performs better than the existing
algorithm in term of network lifetime efficiency. The
experimental result shows that the energy efficiency of the

177
 IEEE International Conference On Recent Trends In Electronics Information Communication Technology, May 20-21, 2016, India

proposed algorithm over the existing

.The proposed improves the lifetime of Network Lifetime Analysis
sensor network by over 88.73% and 90.72% when sensor node
equal to 500 and 1000 respectively over . From the 2500
experimental result we can see that when we increase the
sensor node the performance of proposed get
better but the performance of protocol decreases with 2000
increasing number of nodes which is shown in figure 3.

Life time(S)
1500 Leach
Lifetime of sensor network for 500 sensor
devices
1000 PS-LEACH
520
500 500
480
Number of sensor node

0
460 500 1000
440 Number of sensor nodes
420
LEACH
400 Fig. 5: Network lifetime analysis
380 PS-LEACH
In Fig. 5we can see that the proposed
360 clustering performs better than in term of active node
340 percentage. The experimental result shows that the active node
percentage of decreases when we increase the number
320 of node to 500and 1000 and the active node percentage
130
259
388
517
646
775
904
1033
1162
1291
1420
1549
1678
1807
1

performance our proposed algorithm is

comparatively high when compared to when node is
Number of round increase to 500 and1000.The proposed improves
the node active rate of network by over 32.7% and 34.3% when
sensor node equal to 500 and 1000 respectively. From this we
Fig.3: Network lifetime analysis for 500 nodes can conclude that the proposed clustering
performs better than in term of node active rate.

Lifetime of sensor network for 1000

sensor devices
% Of Active Nodes

1032 Leach

982 PS-LEACH
Number of sensor node

100
932
80
Active Nodes

882
LEACH 60
832
PS-LEACH 40
782
20
732
0
682 500 1000
131
261
391
521
651
781
911
1041
1171
1301
1431
1561
1691
1821
1

No of sensor nodes

Number of round

Fig. 4:Network lifetime analysis for 1000 nodes Fig. 6: Number of active nodes

IV. CONCLUSION
The existing clustering protocol designed based on
is not efficient in term of life time of network so there was a
necessity for a better clustering protocol to increase network
lifetime. Here the author proposed an energy efficient
clustering protocol namely to improve energy
efficiency of sensor network. Our proposed

178
 IEEE International Conference On Recent Trends In Electronics Information Communication Technology, May 20-21, 2016, India

algorithm improved lifetime of network by over

88.73% and 90.72% over when sensor node equal to
500 and 1000 respectively and the network node active rate
performance is improved by 32.7% and 34.3%over
when sensor node equal to 500and 1000 respectively.
Experimental results show that the proposed
performs better than in term lifetime efficiency and
node active rate. In future work we would conduct simulation
study to check the performance of other network parameter and
develop a cross layer design considering radio signal strength
for neighbor discovery to reduce the overhead of the network.
REFERENCES
[1] W. B. Heinzelman et al.,An application-specific protocol architecture
for wireless microsensor networks, IEEE Trans. Wirel. Commun., vol.
1, no. 4, pp. 660-670, Oct. 2002.
[2] X. Liu, A survey on clustering routing protocols in wireless sensor
networks, Sensors, vol. 12, no. 12, pp. 11113-11153, Aug. 2012.
[3] F. Zhao, Y. Xu, and R. Li, Improved leach routing communication
protocol for a wireless sensor network, Int. J. Distr. Sensor Netw., vol.
2012, Article ID 649609, 6 pages, 2012.
[4] M. Tripathi et al.,Energy Efficient Clustered Routing for Wireless
Sensor Network, in Proc. IEEE MSN13, pp. 330-335, 11-13 Dec.
2013.
[5] D. Mechta, S. Harous, I. Alem, and D. Khebbab, LEACH-CKM: Low
energy adaptive clustering hierarchy protocol with K-means and
MTE,in IIT14, Al Ain, pp. 99-103, 9-11 Nov. 2014.
[6] S. Z. Selim and K. Alsultan, A simulated annealing algorithm for the
clustering problem, Patt. Recog., vol. 24, no. 10, pp. 1003-1008, 1991.

179