A Novel Weighted Clustering Based Approach for Improving the Wireless Sensor Network lifetime

Vivek kumar et al. Int. Journal of Engineering Research and Applications www.ijera.com
ISSN: 2248-9622, Vol. 5, Issue 11, (Part - 3) November 2015, pp.23-31
www.ijera.com 23 | P a g e
A Novel Weighted Clustering Based Approach for Improving the
Wireless Sensor Network lifetime
Vivek kumar *, Ramratan Ahirwal**, Yogendra Kumar Jain***
* Research Scholar (CSE), Samrat Ashok Technological Institute, Vidisha (M.P.), India
** Ramratan Ahirwar, Asst Professor (CSE), Samrat Ashok Technological Institute, Vidisha (M.P.), India
*** Y. K. Jain, Head of the Department (CSE), Samrat Ashok Technological Institute, Vidisha (M.P.), India
ABSTRACT
Great lifetime and reliability is the key aim of Wireless Sensor Network (WSN) design. As for prolonging
lifetime of this type of network, energy is the most important resource; all recent researches are focused on more
and more energy efficient techniques. Proposed work is Weighted Clustering Approach based on Weighted
Cluster Head Selection, which is highly energy efficient and reliable in mobile network scenario. Weight
calculation using different attributes of the nodes like SNR (Signal to Noise Ratio), Remaining Energy, Node
Degree, Mobility, and Buffer Length gives efficient Cluster Head (CH) on regular interval of time. CH rotation
helps in optimum utilization of energy available with all nodes; results in prolonged network lifetime.
Implementation is done using the NS2 network simulator and performance evaluation is carried out in terms of
PDR (Packet Delivery Ratio), End to End Delay, Throughput, and Energy Consumption. Demonstration of the
obtained results shows that proposed work is adaptable for improving the performance. In order to justify the
solution, the performance of proposed technique is compared with the performance of traditional approach. The
performance of proposed technique is found optimum as compared to the traditional techniques.
Keywords - Wireless Sensor Network, Energy preserving, Weighted clustering, Network lifetime, Cluster head
selection.
I. INTRODUCTION
Among different technologies, Communication
technology really plays a big role in exploring human
life and making world a small place. As a part of
communication technology, quick growing networks
are taking considerable attention in recent years.
Modern techniques are serving revolution in this
domain. According to the structure of network, it is
categorized in two main domains wired and wireless.
While, according to their utility and applications the
wireless communication is also illustrated in two
parts, short range or indoor communication and broad
range or outdoor communication. Among them the
wireless technologies are growing more rapidly as
compared to the wired network.
The wireless sensor network is one of the most
popular wireless networks. It is a distributed real-
time system which contains all the basic
computational functions inbuilt. Unfortunately yet
very little work is applied in this new system and
always new improvements are essential in all areas of
the system [1]. Most of the earlier distributed systems
are developed with limited power and ﬁxed
resources. In contrast, the designing of a wireless
sensor network should be formulated, keeping in
view terms and conditions, such as the systems is
completely ad-hoc and works with wireless channel,
have scarce power, is real-time, utilizes the sensors
and actuators as interfaces with dynamically
changing sets of resources; aggregate behaviour is
also important there and location is very critical.
Various wireless sensor networks also exploit
negligible capability devices which place a further
strain on the ability to use precedent solutions [2].
In this work the wireless sensor network has
been investigated for finding the solution for the
energy efficiency and exploring the issue to preserve
energy during communication. Additionally for
overcoming the gap a new solution using weighted
clustering technique is proposed.
II. BACKGROUND
A more general issue in sensor network design
for static network is coverage problem. In order to
make it clear, let it given a set of sensors deployed in
an area, required to determine if the area is
sufficiently k-covered, it means every point in the
target area is covered by at least k sensors, where k is
a given parameter. For finding the result, the
aforementioned work [3] can be taken as reference in
this special case with k = 1. Applications requiring k
> 1 may be needed in situations, where monitoring
capability is desired for more sensitive and
unpredictable environments, such as military
applications. It also happens when multiple sensors
are required to detect an event. For example, the
triangulation-based positioning protocols [4] require
at least three sensors (i.e., k ≥ 3) at any moment to
RESEARCH ARTICLE OPEN ACCESS

monitor a moving object. Enforcing k ≥ 2 is also
desirable for fault-tolerant purpose. The article also
considers the same coverage problem combined with
the communication connectivity issue [5].
Unlike the static network, mobile network
doesn’t having the positioning problem but they have
the connectivity issues [6].Therefore the key aim is to
find the solution against the mobility and
connectivity issues in wireless sensor networks [7].
In order to find the solution of connectivity issues in
mobile network one of the essential contribution is
cluster based network functioning [8]. Thus the next
section provides the recent contributions which are
based on clustering approaches to enhance the
performance of network and reduce the losses due to
connectivity issues.
III. RELATED WORK
This section provides the study of various
approaches and algorithms, which have been recently
developed.
From the power preservation viewpoint in
WSNs, grouping of antenna nodes is a challenging
assignment. Grouping method in routing procedures
occupies a key position to extend the constancy phase
and the lifetime of the network. Knowing the
applications of WSNs in wide range and looking
forward for energy efficient ways of communication
in the network, Heinzelman et al [9] proposed a
clustering protocol called LEACH (Low-Energy
Adaptive Clustering Hierarchy), which is a clustering
protocol uses random rotations of cluster heads in
order to achieve equal distribution of energy among
the sensors of the network. Furthermore, it cuts down
the overall energy utilization by performing a load
distribution to various points of the network. The
authors presented evidences of its working by
depicting the simulation results. They further
concluded that it offers 8 times more energy
efficiency in the network.
WSNs comprise a number of network sensors
alternatively actuators. These are implemented at
large geographical areas and often unattended. Also,
requirement of energy efficiency in these networks
remains as it is. Therefore, Yan Yu, Ramesh
Govindan and Deborah Estrin [10] presented an
energy efficient routing protocol, designed for
wireless sensor networks implemented in
geographical areas. They named it as Geographic and
Energy Aware Routing (GEAR) algorithm that
incorporates neighbour selection procedure, attentive
towards energy for routing the packet towards the
destination. For dissemination of the packet inside
the destination region, flooding procedure is used. To
demonstrate positive effects of the algorithm, they
also presented simulation results and confirmed the
longer energy lifetime of the network.
WSNs are different from those typical networks,
in terms of energy constraints, redundancy and low
rate of data and many-to-one flows. In this scenario,
end-to-end routing cannot be applied. Therefore,
data-centric techniques are needed here in order to
have energy efficient packet dissemination by data
aggregation. Into this context Bhaskar
Krishnamachari, Deborah Estrin and Stephen Wicker
[11] proposed a data-centric routing mechanism and
demonstrated its performance in comparison with
traditional networks. Basic idea of this method is to
combine data coming from different sources and then
remove redundancy as well as minimizing the
transmissions. To this, the conclusion drawn clearly
states that this routing scheme puts positive effects on
source-destination placement and communication
network density, on the energy costs, delay, and
robustness of data aggregation.
Above mentioned technique of data-centric
routing falls short in some terms. Since network these
days contain large amount of data and data
redundancy, it is likely that data may be dissociated
from the specific node. Therefore, it becomes natural
to deliver queries to the events being occurred in the
network. David Braginsky and Deborah Estrin [12]
proposed a novel scheme named as “Rumor Routing”
for routing queries to nodes that have observed a
particular network without having concern to
geography or addressing method of the underlying
network. It is designed to be adaptable in a variety of
applications and their requirements. The authors
demonstrated its working and claimed that this
algorithm is capable to deal with node failure as well
as supports promising delivery rate and route repair.
Imaging and video data transmission demands
higher amount of energy QoS. In such cases special
mechanisms of routing must be implemented with
significant awareness of energy and Quality of
Service to achieve efficient sensor usage. For this
Kemal Akkaya and Mohamed Younis [13] projected
a routing mechanism based on energy and QoS
effective constraint. It considered only end-to-end
delay, and using a path function least cost path is
calculated for each link. A path is chosen which has
least delay and maximum throughput, in order to
achieve a best effort routing service. Simulation
results show that our protocol consistently performs
well with respect to QoS metrics, e.g. throughput and
average delay as well as energy-based metric such as
average lifetime of a node.
Along with the advances of internet and
communication technology, mobile ad hoc networks
(MANETs) and wireless sensor networks have
attracted extensive research efforts in recent years. In
the past, Chatterjee et al. proposed an efficient
approach, called the weighted clustering algorithm, to
determine the cluster heads dynamically in mobile ad
hoc networks. Wireless sensor networks are,

however, a little different from traditional networks
due to some more constraints. Besides, in wireless
sensor networks, prolonging network lifetime is
usually an important issue. Tzung-Pei Hong et al
[14], an improved algorithm based on the weighted
clustering algorithm is proposed with additional
constraints for selection of cluster heads in mobile
wireless sensor networks. The cluster heads chosen
will act as the application nodes in a two-tired
wireless sensor network and may change in different
time intervals. After a fixed interval of time, the
proposed algorithm re-run again to find new
applications nodes such that the system lifetime can
be expected to last longer. Experimental results also
show the proposed algorithm behaves better than
Chatterjee’s on wireless sensor networks for long
system lifetime.
This section provides the details about the sensor
network routing protocols that help to improve the
connectivity issues and for enhancing the
performance in terms of energy consumption.
IV. PROPOSED WORK
WSNs (wireless sensor Networks) is a
technology with huge potential which is mainly used
in difficult sensitive environments like battlefield
surveillance (in military), disaster management,
combat field reconnaissance, border protection etc.
where human involvement is not possible for solving
any type of network related problem or improvement.
In Wireless Sensor Networks, optimum utilization of
energy is the most important concern, because in this
type of network it is not possible to change or
recharge the batteries of nodes at all. Due to this
reason to prolong network lifetime is a challenging
task.
In either static or dynamic WSNs, active nodes
lose its energy uninterruptedly and readily because of
being in continuous working state. Hence the limited
energy of these nodes has been exhausted earlier
which results in network failure.
In static WSNs since nodes are static, they can’t
change their positions and the energy of all nodes is
not being used optimally. The nodes which are nearer
to the target will be more active as compare to the
other nodes which are far away from the target.
Active node consumes more energy because they
have more loads. Due to this reason limited energy of
active node has been exhausted and it results into a
network failure despite of remaining unused energy
present in those nodes which are away from the
target. So, for making the nodes again useable which
was remained unused in static network, we are
proposing a new dynamic cluster head based
Clustering Approach for improving the lifetime of
whole network.
In our proposed technique, Network has been
divided in different clusters. Each cluster posses
some nodes. For each cluster one random node is
being selected as CH (Cluster Head). Due to the
mobility of nodes, single node is not bound to work
as cluster head all the time, as the process of cluster
head selection has been repeated after particular time
interval, because of this all nodes have the
opportunity to be selected as cluster head. That node
is being selected as cluster head which have higher
percentage of available resources and due to this
rotation energy utilisation is optimum here and it
helps to protect network failure by using energy
available with all nodes.
In proposed dynamic WSNs approach, to select
the cluster head, first we create the clusters according
to the clustering method [15] and generally select
those nodes which are directly connected to the
randomly selected node.
After creating the cluster, our proposed
technique works according to following steps for
selection of better nodes on regular interval of time.
1. Node quality estimation
After completing the clustering process, five
important attributes or parameter we identify, as
Signal to Noise Ratio (SNR), Connectivity (c),
Remaining Energy (e), Mobility (M) and Buffer
Length (B) of each node. With the help of these
parameters or attributes, first we estimate quality of
each node and then calculated comparative weight of
nodes for selection of cluster head. SNR gives the
comparison between the levels of the desired signal
to the levels of background noise. It is defined as the
ratio of signal power to the noise power, often
expressed in decibels. The node sends a data packet
to neighbour node and between both the nodes during
the data exchange signal strength values are
estimated [16]. In order to obtain the SNR for the
network nodes the sender node sends the packet to
receiver nodeA_signal and receiver node send
acknowledge to sender nodeA_noise. During
receiving the acknowledgement the sender node
compute the SNR of receiver node for receiving the
signal. This calculation finds that node, which
performs data delivery in critical scenarios. Therefore
optimal SNR ratio is less for good quality of node
selection. Second attribute Connectivity facilitates to
find such node which is connected with maximum
number of nodes. In this type network the nodes are
said to connect, which are in radio range of a node.
This is sometimes also called the Degree of node.
Therefore the selected node that has higher degree
can communicate with greater number of nodes [17].
Further Remaining Energy is essential attribute for
node service quality estimation. As nodes in WSNs,
are with limited amount of energy and loses energy
for sending and receiving data and for its liveliness.
Therefore with the help of remaining energy, we
become able to find such nodes which possess

sufficient energy for better performance [18].
Another attribute of nodes in WSNs is mobility.
Nodes frequently move from one place to other, in
the dynamic networks. The computation of node
mobility demonstrates the rate of change in their
mobility pattern. If a node is less frequently changes
its place, then, that node can create such group of
nodes which will be connected for long time with
each other. In other words the low mobile nodes are
able to form more stable clusters. [19]. As far as
Buffer Length is concerned. It makes us able to find
better performing nodes for receiving more data. The
buffer is a temporary storage that holds the
information during the data exchange. If the buffer is
almost full then the node will not be able to accept
new coming data and that can drop the essential
information. In this way free node will serve better
for the network. Thus the available buffer for
accepting upcoming stream is given as buffer length.
During calculations the buffer length is denoted using
B [20].
Therefore with the help of these five attributes
we estimate quality of nodes. But, only by estimation
on the basis of separate attribute we can’t be able to
say that which node will be most suitable to become
cluster head node. So, for selection of such node
there is need of combined analysis of all attributes for
all nodes. And for this our next step is weight
calculation.
2. Weight calculation
In weight calculation we use all five attributes,
Signal to Noise Ratio (SNR), Connectivity (c),
Remaining Energy (e), Mobility (M) and Buffer
Length (B) as discussed in first step. In a cluster, all
nodes broadcast estimated value of its attributes to
neighbour nodes. Then all nodes store values of
attribute broadcasted by neighbour nodes, with its
own attributes value, in a table available with each
node known as evaluated parameters table. For this
we take an example as shown in table 1 for evaluated
parameters.
Table 1: Evaluated parameters
Node id C E M SNR B
1 3 87 2 23.54 46
2 2 62 6 12.53 67
3 3 65 3 17.43 82
4 5 59 5 23.53 41
As evaluated values of all the attributes are
shown in table1 for each node, but the problem here
is that all attributes have not similar measurement
scale because measurement unit for each attribute is
entirely different from other. So, to calculate the
weight it is necessary to find values of attributes at
similar scale. For this, first of all we normalize the
values of evaluated parameters with the help of
proposed formula 1 as show below.
𝑁𝑣𝑎𝑙𝑢𝑒 =
𝑐𝑢𝑟𝑟𝑒𝑛𝑡 𝑣𝑎𝑙𝑢𝑒 − 𝑚𝑖𝑛𝑖𝑚𝑢𝑚 𝑣𝑎𝑙𝑢𝑒
𝑚𝑎𝑥𝑖𝑚𝑢𝑚 𝑣𝑎𝑙𝑢𝑒 − 𝑚𝑖𝑛𝑖𝑚𝑢𝑚 𝑣𝑎𝑙𝑢𝑒
Where,
𝑁𝑉𝑎𝑙𝑢𝑒 = 𝑁𝑜𝑟𝑚𝑎𝑙𝑖𝑧𝑒𝑑 𝑣𝑎𝑙𝑢𝑒
Current value = value of particular attribute of given
node
Minimum value=minimum value of particular
attribute in the table.
Maximum value=maximum value of particular
attribute in the table.
After normalizing the values of attributes given
in table 1, following normalized values are obtained
which are shown in table 2.
Table 2: Normalized parameters
Node id C E M SNR B
1 0.33 1.00 0.00 1.00 0.12
2 0.00 0.11 1.00 0.00 0.63
3 0.33 0.21 0.25 0.44 1.00
4 1.00 0.00 0.75 0.99 0.00
Now for more nearer approximated values, these
normalized values of attributes are factorised using
the factor 𝑤1, 𝑤2, 𝑤3, 𝑤4 and 𝑤5, for Connectivity
(c), Remaining Energy (e), Mobility (M), SNR, and
Buffer Length comparatively. For constructing these
factors, the random values can be used between 0-1
and it is required to have sum of these factors exactly
to be 1. In other words 𝑤1 + 𝑤2 + 𝑤3 + 𝑤4 + 𝑤5 =
1.For example, in a specified scenario, where the
nodes having good energy, and almost they are
having 100% energy, here the factor energy is less
important to utilize for energy constrain thus the
weight distribution can be 0.3
+0.05+0.25+0.20+0.20=1. In other case where the
mobility is rare in all the nodes, the factor mobility is
less important for weight computation thus that can
be 0.30+0.25+0.05+0.20+0.20=1.
After that by using formula 2, we calculate final
weight of all attributes of each node as given below.
𝑊 = 𝑤1 ∗ 𝑐 + 𝑤2 ∗ 𝑒 + 𝑤3 ∗ 𝑀 + 𝑤4 ∗ 𝑆𝑁𝑅 − 𝑤5
∗ 𝐵
In order to compute final weight of normalized
parameters of nodes given in table 2.
Let,
𝑤1 = 0.30
𝑤2 = 0.25
𝑤3 = 0.05
𝑤4 = 0.20
𝑤5 = 0.20
By using above assumed value
of 𝑤1, 𝑤2, 𝑤3, 𝑤4 and 𝑤5, we calculate final weight as
shown in table 3.

Table 3: Weight estimation
Node
Id
C E M SNR B W
1 0.33*
0.30
1.00*
0.25
0.00*
0.20
1.00*
0.20
0.12*
0.20
0.525
2 0.00*
0.30
0.11*
0.25
1.00*
0.05
0.00*
0.20
0.63*
0.20
-
0.048
3 0.33*
0.30
0.21*
0.25
0.25*
0.05
0.44*
0.20
1.00*
0.20
0.025
4 1.00*
0..30
0.00*
0.25
0.75*
0.05
0.99*
0.20
0.00*
0.20
0.565
3. Cluster head selection
Further after weight calculation, now in order to
select a head node for each cluster the next most
important step is cluster head selection.
As with the help of weight calculation method, all
nodes compute the weight of their neighbour nodes
using a neighbour table. The example of neighbour
table is shown as table (4) for four nodes defined in
the figure1.
Figure 1: Sample network nodes
Table 4: Neighbour table
nodes 1 2 3 4
1 0 T T T
2 T 0 F T
3 T F 0 T
4 T T T 0
Figure 1 shows sample network node 1,2,3,4.
Above nodes with the use of neighbour table (4), find
their neighbour nodes as shown in table. In given
table T represents direct connection between two
nodes.
The nodes use this neighbour table to identify its
neighbours and become able to broadcast its
attributes to neighbour nodes and receive information
about attributes broadcasted by neighbour nodes.
Further nodes use this information about attributes in
calculation of combined weight for all nodes with its
own weight as described in detail in weight
calculation method.
Thus as shown in table 3, now each node has
information about weights of its neighbour nodes
with its own weight, Finally cluster head selection
depends on these weight information available with
all nodes in a cluster. All nodes compare own weight
with weight of its neighbour. If a node finds that self
weight is greater than weight of its neighbour nodes
then it broadcast self as cluster head. Otherwise if it
finds self weight is smaller than weight of any
neighbour node then it ignore self weight and update
itself with cluster head information and declare self
as Clint node. In rare case, there may be a situation in
which weights of more than one node are optimum
and equal. In such case there will be confusion in
selection of cluster head. So, in this condition to deal
with this confusion, further energy of equal weighted
nodes is being compared. Among these nodes, that
node will be selected as cluster head, which possess
optimum energy.
The entire process of cluster head based
clustering approach is shown below in the form of
proposed algorithm.
Algorithm:
1. Start
2. Initialize set of Node in network
3. Anonymous node A initiating cluster request by
broadcasting a hello packet
4. If entire node’s acknowledgement is received
5. Initiate step 9.
6. Else
7. Wait for acknowledgement
8. End if
9. Broadcast SNR, C, E, M and B to first hop
neighbours.
10. For each node in network
I. Signal to noise ratio SNR
II. Remaining Energy E
III. Connectivity C
IV. Mobility M
V. Buffer Length B
End for
11. Weight Calculation
𝑊 = 𝑤1 ∗ 𝑐 + 𝑤2 ∗ 𝑒 + 𝑤3 ∗ 𝑀 + 𝑤4 ∗ 𝑆𝑁𝑅
− 𝑤5 ∗ 𝐵
12. Weight Comparison
If (self-weight > neighbours weight)
I. broadcast self as Cluster Head
13. Else if(self-weight < neighbours weight)
I. ignore self-weight and update self with cluster
head information
II. Declare self as client node
14. Else if (self-weight=neighbour weight)
If (𝐸𝑛𝑒𝑟𝑔𝑦𝑠𝑒𝑙𝑓 > 𝐸𝑛𝑒𝑟𝑔𝑦 𝑛𝑒𝑖𝑔 𝑕𝑏𝑜𝑢𝑟 )
a. Broadcast self as cluster head
b. Else
c. Broadcast neighbour as cluster head
d. End if
15. End if
16. Wait for new clustering request
The summarized steps of the algorithm are
demonstrated using the flow chart of the algorithm
using the figure 2. In this diagram the processes

involved in the cluster head selection is demonstrated
in a flow.
Figure 2: Algorithm flow
V. SIMULATION
The proposed cluster head selection algorithm is
simulated using the NS2 network simulator. That
simulation is configured on the basis of the following
simulation setup and scenarios.
A. Simulation setup
This section provides the required network
parameters which are used for network simulation.
The key parameters and their corresponding values
are reported using table 5.
Table 5: Simulation setup
Simulation properties Values
Antenna model Omni Antenna
Dimension 750 X 550
Radio-propagation Two Ray Ground
Channel Type Wireless Channel
No of Mobile Nodes 15
Routing protocol AODV
Time of simulation 10.0 Sec.
B. Simulation scenario
This section provides the simulation scenarios on
which the proposed and traditional algorithm [4] for
network performance evaluation is based.
1. Simulation using the traditional approach: In
this simulation a static network is prepared using the
given network configuration. Additionally in this
static network the issues of covering set and the
performance improvement in terms of network life
time is simulated. The simulation scenario is given
using figure 3.
Figure 3: Traditional network
2. Simulation of proposed clustering scheme: In
this scenario the proposed cluster head selection
technique is implemented in the mobile network
scenario and their performance improvement is
provided and compared with the previous network
technology. The simulation screen of the proposed
simulation scenario is given using figure 4.
Figure 4: Proposed network
In this section the network simulation strategy is
provided and further in next section the results
evaluation of proposed methodology is provided.
VI. RESULTS ANALYSIS
This section provides the study of the
performance parameters obtained during simulation
of the proposed cluster head selection approach.
Additionally the detailed description of each
parameter is also reported in this section.
A. End to end delay
End to end delay is the amount of additional time
consumed during communication scenario in the
network, which refers to the time taken by a packet to
be transmitted across a network from source to
destination device, this delay is calculated using the
below given formula.
𝐸2𝐸 𝑑𝑒𝑙𝑎𝑦 = 𝑟𝑒𝑐𝑒𝑖𝑣𝑖𝑛𝑔 𝑡𝑖𝑚𝑒 – 𝑠𝑒𝑛𝑑𝑖𝑛𝑔 𝑡𝑖𝑚𝑒

Figure 5: End to end delay
The comparative end to end delays for both the
network techniques are given using figure 5. In this
graph the amount of time overhead is given using Y
axis, and the X axis shows the number of nodes in
network. According to the obtained performance of
the proposed technique, it has less end to end delay as
compared to the traditional (old) technique [4]. Thus
the proposed technique is much promising as
compared to the traditional one.
B. Packet drop ratio
The amount of packet which is transmitted to
destination node is not successfully delivered is
known as the packet drop ratio. The figure 6 shows
the packet drop ratio of both the techniques. In order
to show the performance of network protocols the X
axis contains the number of nodes in network and the
Y axis shows the amount of dropped packets in terms
of percentage. According to the obtained result the
proposed technique drop less amount of packets as
compared to the traditional technique.
Figure 6: Packet drop ratio
C. Energy consumption
The amount of energy reduced from initial
energy of nodes during the active communication
sessions of network is known as energy consumption.
The energy consumption is responsible to provide the
efficient network life time
Figure 7: Energy consumption
The comparative performance in terms of energy
consumption is given in the figure 7. In this graph the
X axis shows the amount of nodes in network and the
Y axis shows the consumed energy in terms of Jules.
According to the obtained results the proposed
technique consumes less amount of energy as
compared to the traditional technique.
D. Packet delivery ratio
The performance parameter Packet delivery ratio
sometimes termed as the PDR ratio provides
information about the performance of any routing
protocols by the successfully delivered packets to the
destination, where PDR can be estimated using the
formula given
𝑝𝑎𝑐𝑘𝑒𝑡 𝑑𝑒𝑙𝑖𝑣𝑒𝑟𝑦 𝑟𝑎𝑡𝑖𝑜 =
𝑡𝑜𝑡𝑎𝑙 𝑑𝑒𝑙𝑖𝑣𝑒𝑟𝑒𝑑 𝑝𝑎𝑐𝑘𝑒𝑡𝑠
𝑡𝑜𝑡𝑎𝑙 𝑠𝑒𝑛𝑡 𝑝𝑎𝑐𝑘𝑒𝑡𝑠
The packet delivery ratio of the proposed and
traditional network performance improvement
technique is given using figure 8. In this graph the X
axis shows the amount of nodes in network and the Y
axis shows the packet delivery ratio in terms of
percentage. According to the obtained performance
the proposed technique provides higher PDR as
compared to traditional technique.
Figure 8: Packet delivery ratio
E. Throughput
Network throughput is the average rate of
successful message delivery over a communication
channel. This data may be delivered over a physical
or logical link, or pass through a certain network
node. The throughput is usually measured in bits per

second (bit/s or bps), and sometimes in data packets
per second or data packets per time slot.The obtained
performance of both the algorithms for network life
time improvement is given using figure 9. In this
graph the green line shows the performance of
proposed technique and red line shows the
performance of the traditional technique. In order to
shows the performance the X axis contains the
number of nodes in network and the Y axis shows the
performance in terms of consumed bandwidth in
terms of kbps.
Figure 9: Throughput
This section reports the performance analysis of
the proposed and traditional algorithms. Additionally
their comparative performances with increasing
number of nodes are presented. The next section
provides the conclusion of the entire project
development and the results analysis.
VII. CONCLUSIONS
In this paper, techniques are explored and
studied for improving the performance of wireless
sensor networks, focused on connectivity issues and
energy efficiency. The cover set problem in static
network is studied and their different solutions are
also investigated. Further it is found that the cover set
problem of static network can be converted into
connectivity issue in mobile network. Therefore for
optimizing the connectivity issues the clustering
algorithms are studied. Among different clustering
algorithms, weighted clustering algorithm is found
more suitable for improving the performance of
network.
Thus a weighted clustering scheme is proposed
and implemented using the NS2 simulator. The given
approach utilizes the network parameters to find the
optimum cluster head (CH) for each cluster on
regular time interval. It provides uniform load
distribution to the nodes during network functioning
in mobile network scenarios, results in optimum
energy utilization and avoidance of network failure
like situation. Furthermore the proposed clustering
approach is compared with the exiting covering set
solution. The performance evaluation of the proposed
cluster head selection approach is provided in terms
of end to end delay, packet delivery ratio, packet drop
ratio, throughput, and energy consumption.
According to the obtained performance the proposed
cluster head algorithm consumes less energy and
increases the packet delivery ratio even for the
mobile networks. Thus the proposed clustering
algorithm is more efficient and effective.
VIII. Acknowledgements
We would like to thank the anonymous referees
for their helpful guidance that have improved the
quality of this paper.
REFERENCES
[1] John A. Stankovic, “Wireless Sensor
Networks”, Department of Computer
Science University of Virginia
Charlottesville, Virginia 22904,June 19,
2006.
[2] Jennifer Yick, Biswanath Mukherjee,
DipakGhosal, “Wireless sensor network
survey”, Computer Networks, 2008 Elsevier
B.V. All rights reserved.
[3] D. Tian and N.D. Georganas, A coverage-
preserving node scheduling scheme for large
wireless sensor networks, in: ACM Int’l
Workshop on Wireless Sensor Networks and
Applications (WSNA) (2002).
[4] Hosein Mohamadi, Shaharuddin Salleh,
Mohd Norsyarizad Razali, Sara Marouf, “A
new learning automata-based approach for
maximizing network lifetime in wireless
sensor networks with adjustable sensing
ranges”, Neurocomputing,& 2014 Elsevier
B.V. All rights reserved.
[5] CHI-FU HUANG and YU-CHEE TSENG,
“The Coverage Problem in a Wireless
Sensor Network”, Mobile Networks and
Applications 10, 519–528, 2005 C 2005
Springer Science + Business Media, Inc.
Manufactured in The Netherlands.
[6] You-Chiun Wang, “A Two-Phase Dispatch
Heuristic to Schedule the Movement of
Multi-Attribute Mobile Sensors in a Hybrid
Wireless Sensor Network”, IEEE
TRANSACTIONS ON MOBILE
COMPUTING, Volume:13 , Issue: 4, 11
March 2014
[7] Chengchao Liang and F. Richard Yu,
“Wireless Network Virtualization: A
Survey, Some Research Issues and
Challenges”, IEEE COMMUNICATIONS
SURVEYS & TUTORIALS • JULY 2014
[8] Xuxun Liu, “A Survey on Clustering
Routing Protocols in Wireless Sensor
Networks”, Sensors 2012, 12,

[9] Heinzelman, Wendi Rabiner,
AnanthaChandrakasan, and
HariBalakrishnan. "Energy-efficient
communication protocol for wireless
microsensor networks." System sciences,
2000. Proceedings of the 33rd annual
Hawaii international conference on. IEEE,
2000.
[10] Yu, Yan, Ramesh Govindan, and Deborah
Estrin. “Geographical and energy aware
routing: A recursive data dissemination
protocol for wireless sensor networks.”
Technical report ucla/csd-tr-01-0023, UCLA
Computer Science Department, 2001.
[11] Krishnamachari, Bhaskar, Deborah Estrin,
and Stephen Wicker. "Modelling data-
centric routing in wireless sensor networks."
IEEE infocom. Vol. 2. 2002.
[12] Braginsky, David, and Deborah Estrin.
"Rumor routing algorthim for sensor
networks." Proceedings of the 1st ACM
international workshop on Wireless sensor
networks and applications. ACM, 2002.
[13] Akkaya, Kemal, and Mohamed Younis. "An
energy-aware QoS routing protocol for
wireless sensor networks." Distributed
Computing Systems Workshops, 2003.
Proceedings. 23rd International Conference
on. IEEE, 2003.
[14] Tzung-Pei Hong and Cheng-Hsi Wu, “An
Improved Weighted Clustering Algorithm
for Determination of Application Nodes in
Heterogeneous Sensor Networks”.
[15] RatishAgarwal, Roopam Gupta, Mahesh
Motwani, “REVIEW OF WEIGHTED
CLUSTERING ALGORITHMS FOR
MOBILE ADHOC NETWORKS”,GESJ:
Computer Science and Telecommunications
2012|No.1(33).
[16] Huse ¨ yin Arslan, Sharath Reddy, “Noise
Power and SNR Estimation for OFDM
Based Wireless Communication Systems”.
[17] Mohamed Younis, Izzet F. Senturk, Kemal
Akkaya, Sookyoung Lee, FatihSenel,
“Topology management techniques for
tolerating node failures in wireless sensor
networks: A survey”, Computer Networks,
2013 Elsevier B.V. All rights reserved.
[18] María Gabriela Calle Torres, “ENERGY
CONSUMPTION IN WIRELESS SENSOR
NETWORKS USING GSP”, Electronics
Engineer, Universidad
PontificiaBolivariana, Medellín, Colombia,
1995
[19] Rajeev Patel, VikashPorwal, Dr. R. K.
Kapoor, “A review of Mobility-Based
Wireless Sensor Networks”, International
Journal of Advanced Research in Computer
Engineering & Technology (IJARCET)
Volume 3 Issue 8, August 2014.
[20] Tie Qiu, Lin Feng, Feng Xia, Guowei Wu,
and Yu Zhou, “A Packet Buffer Evaluation
Method Exploiting Queueing Theory for
Wireless Sensor Networks”, DOI: 10.2298/
CSIS110227057Q.
.

A Novel Weighted Clustering Based Approach for Improving the Wireless Sensor Network lifetime

Recommended

More Related Content

What's hot (17)

Viewers also liked (20)

Similar to A Novel Weighted Clustering Based Approach for Improving the Wireless Sensor Network lifetime (20)

Recently uploaded (20)

A Novel Weighted Clustering Based Approach for Improving the Wireless Sensor Network lifetime