Energy Efficient Search Protocol

Utsav Kakkad, Dr. Asis Kumar Tripathy,

School of Information Technology and Engineering,
VIT University,Vellore
Abstract—A semi-centralized stochastic election cluster heads only, it helps in conserving a good deal of
approach named SEARCH is studied in this paper and energy.
some changes to it are proposed. The main advantage of However, the situation of no cluster heads emerging during
this protocol is that it boosts the cluster head threshold of a round in WSN is one of the most severe since it means that
sensor nodes in a favorable position while deteriorating it no packets will be transmitted to the base station and hence,
otherwise. Further, the SEARCH protocol assures an no data would reach the destination it is intended for.
optimal number of cluster heads and a low time cost. It SEARCH avoids this by using an optimal number of cluster
further prolongs the stable period by prolonging the heads which are then compared to nodes eligible to be
round of half alive nodes surviving. We aim to extend this elected as a cluster head. If not, it elects the given optimal
to include energy efficiency. Using SEARCH, an number of cluster heads otherwise it elects the nodes eligible
improvement in the performance of the wireless sensor to be a cluster head.
network is observed. This is achieved by eliminating the Energy efficiency will serve a twofold purpose:
scenario in which no cluster head emerges. This is ⚫ network lifetime will be lengthened
particularly advantageous since it ensures that data will ⚫ reduce the number of dead nodes
reach the sink (base station). By making this stochastic which means there is a higher probability of a greater
election protocol energy efficient the gains are namely, the number of cluster heads will be chosen which would serve to
number of dead nodes would drastically reduce which deliver a greater amount of data to the end destination.
increases the possibility of a node to be chosen as a cluster
head. Thus, the presence of a cluster head in every round II. PROBLEM DEFINITION
is assured. Therefore, it is ensured that the data collected One of the major problems faced with the traditional cluster
by the sensor nodes would reach the base station. One of based WSN routing protocols like LEACH is the number of
the parameters being judged is the number of packets to dead nodes keeps increasing drastically till all the sensor
the base station. This helps us in determining effectiveness nodes in the network are dead nodes. The situation of no
of the WSN topology thus created. The base station is the cluster head emerging during a round of transmission is a
end destination of the data collected by the network and a possibility in LEACH and LEACH like protocols since they
greater number of packets being sent to the base station have a high number of dead nodes. Further, so far only simple
indicates that the WSN is operational. Another parameter clustering methodology in which the clusters are at the four
we judge is the count of cluster heads being elected per corners of the area under consideration is tried for in the
round which is also reflected in the preceding graph. A simulations, the effect of other clustering methods needs to be
round in which a lower number of cluster heads are explored. The difference in results with random clustering is
elected would send fewer packets of data to the base also observed.
station. Further, we try to find the number of dead nodes
as the rounds progress which would help in the III. LITERATURE SURVEY
determination of the network lifetime. [1] proposes SEARCH which is stochastic election of
Keywords: optimal number of cluster heads, dead nodes, appropriate range cluster heads. SEARCH enhances the
energy efficiency, alive nodes, base station cluster head threshold for a node in a favorable position else
decrements it. Further, it eliminates the scenario of no cluster
I. INTRODUCTION head emerging during a round of operation of WSN. [3]
proposes a modified version of the SEARCH protocol. Here,
T HE POPULARITY OF WIRELESS SENSOR NETWORKS HAS
the SEARCH protocol is adapted with data thresholding
which enhances the lifetime of the network and removes the
BEEN INCREASING DAY BY DAY. THE DISCOVERY OF THE redundancy in the data sent. [6] proposes a new algorithm
CLUSTERING PROTOCOLS REVOLUTIONIZED THE FIELD. based on the residual energy of isolated sensor nodes. Cluster
Clustering in WSNs gives us a scalable topology that is Head selection is based on both the average residual energy
much easier to manage than the flat topology and can of the sensors and the regional average energy of all the sensor
respond much better to changes in the environment. Further, nodes contained in each cluster. Preliminary simulation
it conserves energy by means of data aggregation from results for this protocol hold much promise. A fuzzy cluster
various sensor nodes to provide the final aggregated data to head selection scheme is proposed in the [2]. The cluster head
the base station. This also helps in eliminating redundant selection is based on three parameters: residual energy of the
data. Further, the data aggregation helps in reducing the node, centrality of the cluster and the distance between the
amount of data transmitted and thus, helps in saving energy. sensor node and the mobile sink. [4] proposes the LEACH
Moreover, clustering with intra-cluster and inter-cluster protocol in which cluster heads are rotated randomly in order
communications can reduce the number of sensor nodes to uniformly distribute the energy load in the network. It also
performing the task of long-distance communications, thus includes data fusion which reduces the amount of data to be
allowing less energy consumption for the entire network. transmitted to the base station. [7] gives us a firm idea on the
Further, since the transmission of data to the sink(base idea about the optimal number of cluster heads that must be
station) in the cluster-based routing scheme is done by the taken into consideration. This idea will be utilized in the
cluster head election algorithm that is proposed in [1]. [11] heads. Therefore, if the number of nodes having a non-zero
lists out partitioning as a method to attain energy efficiency. G value(t) is less than the ideal number of cluster heads(k),
IV. DETAILED DESIGN then t candidates are chosen as cluster heads with the G
The following figure lists the steps followed in the design of values of the remaining non-dead nodes will be brought up
the SEARCH protocol for WSN routing: to date. It is from these nodes whose G value has been
updated, the remaining k-t cluster heads will be chosen. If
the number of nodes with a non-zero G value is greater than
or equal to the ideal (optimum) number of cluster heads then
the sink chooses k cluster heads at once.
V. Changes Proposed to the original SEARCH
protocol
One of the changes proposed is the partitioning the network
into 4 discrete clusters with an aim to increase the energy
efficiency. With the greater amount of initial energy
available to advanced nodes and the greater energy loss in
the transmission of data to the sink it stands to reason that
only an advanced node is eligible to be a cluster head which
would help in the reduction of the number of dead nodes and
thereby increase the energy efficiency of the network.

Figure1. search protocol flowchart

Fig3. Proposed WSN topology, the yellow triangles indicate the dead
nodes

Rather than the randomized clustering implemented in figure

4, ordered clustering(partitioning) has been followed in figure
3. This has an advantage of energy efficiency, fault tolerance
and maximizes the network lifetime and provides secure data
communication. Similar topology has been followed in [10]
and [11].
Figure2. refined G value updating rule followed in search
Figure 1 lists out the steps involved in the functioning of the In case of a topology which involves random arrangement of
SEARCH protocol. Figure 2 explains the cluster head the sensor nodes around the sink is shown in the figure4 below.
selection approach followed in SEARCH protocol via a
flowchart. Here t is the number of nodes that have a non-
zero G value while k is the optimum number of cluster
 Average Residual Energy per round

Fig4. Random clustering of the sensor nodes

Fig5. Average Residual Energy per Round
Figure 5 highlights the differences in residual energy
VI. Results observed for both the protocols in terms of the average
The simulation of the SEARCH protocol with the residual energy per round. In case of random clustering the
proposed topology has been done using Matlab R2018a. average residual energy quickly falls to zero which means
The simulation of the SEARCH protocol with random that all the sensor nodes in the WSN are dead within those
clustering has also been done in Matlab R2018a. many rounds. It is noted that when the sensor nodes are
Number of sensor nodes: 300 partitioned into separate groups the residual energy remains
initial energy of the node:0.1 Joule above zero, implying that the sensor nodes in the WSN are
message size: 2000bits not dead and still capable of transmitting the data collected
150 X 150 field is considered for the WSNs operation to the sink or the cluster head.
Sink is placed at the center of the field Number of Dead Nodes per Round
Etx=Erx:50 nano joules
Eelec:50 nano joules
Free space amplification energy Efs: 10picoJoules per bit-
meters^2
Multipath amplification energy Emp: 0.0013picoJoules
per bit-meters^4
optimal probability of node becoming a cluster head:
p=0.2
Data aggregation energy: 5X0.000000001 joules
fraction of the advanced nodes: 0.5
Additional energy factors: alpha=2: beta=3
Maximum number of rounds: 50

We compare the performance of the topology in which the

sensor nodes are grouped together into separate clusters at
the four corners of the field under observation (partitioned
case) with random clustering (non-partitioned case) in terms
of the number of dead nodes, the average residual energy per
round, the packets sent to the base station and the cluster
head and the number of cluster head candidates per round.

Fig6.Number of Dead Nodes per round

In Figure 6, we observe that for 50 rounds there are almost In figure 8 it is observed that in case of partitioning the
no dead nodes for the partitioned case while the number of number of cluster head candidates per round is greater than
dead nodes is equal to the size of the sensor network within random clustering case. In case of no cluster head being
10 rounds in the case of random clustering. This is expected elected during a round of operation the data collected by the
since in the case of random clustering the average residual sensor nodes in that cluster would not reach the end
energy falls to zero within twenty to twenty-five rounds. destination. We, thus, note that the scenario of no cluster
Packets to Base Station per round head being elected during a round of operation is avoided in
case of the SEARCH protocol whereas it was one of the
major shortcomings of the LEACH protocol.
VII. Network Lifetime
For the SEARCH protocol implemented with the random
arrangement of the sensor nodes around the base station, it is
observed that the first dead node occurs in the sixth round
and all the nodes in the sensor network are dead by the tenth
round. For SEARCH protocol with partitioning, the first
dead node occurs in the 24th round. Network lifetime is
defined as the time at which the first sensor node runs out of
energy to send a packet since losing a sensor node amounts
to a wireless sensor network losing some of its functionality.
Therefore, an increase in the network lifetime is observed in
the case of a partitioned SEARCH.
VIII. CONCLUSION
We see that clustering the sensor nodes into fixed clusters
gives us a superior performance in terms of the number of
dead0 nodes and the packets sent to the base station. We
have only distributed the sensor nodes into fixed clusters in
four quadrants of the area (100 X 100) under consideration
Further work in trying to find the optimal number of clusters
Fig7. Packets to Base Station per round is now being attempted.
In Figure 7 it is observed that the packets to the base REFERENCES
station (end destination of the data collected) are zero in 1. SEARCH: A Stochastic Election Approach for
the case of random clustering which is expected since all Heterogeneous Wireless Sensor Networks : in-Yi Wang,
the nodes in the sensor network have exhausted their Jie Ding, Wan-Pei Chen, and Wen-Qiang Guan
energy within twenty-five rounds while they have some 2. An Energy Efficient Fuzzy Cluster Head Selection
significant value in the case of partitioning. Algorithm for WSNs: Jin Wang, Junming Niu, Kai Wang,
Count of Cluster Head candidates per round Wei Liu
3. Extending the Lifetime of Wireless Sensor Networks
with Threshold Sensitive Stochastic Election Approach:
Indrani Chakraborty
4. Energy-Efficient Communication Protocol forWireless
Microsensor Networks:Wendi Rabiner Heinzelman,
Anantha Chandrakasan, and Hari Balakrishnan
5.”An Energy Efficient Fuzzy Cluster Head Selection
Algorithm for WSNs”: Jin Wang ,Junming Niu , Kai Wang
and Wei Liu
6.”Energy Efficient Clustering Scheme for Prolonging
the Lifetime of Wireless Sensor Network With Isolated
Nodes”: Jenq-Shiou Leu ; Tung-Hung Chiang ; Min-Chieh
Yu ; Kuan-Wu Su
7."Research on the optimal number of cluster heads of
wireless sensor networks based on Multi hop –
LEACH":Zhichao Qin, Zheng Zhou ,
Xiaochuan Zhao
8.”Clustering Analysis in Wireless Sensor Networks: The
Ambit of Performance Metrics and Schemes
Taxonomy”: Asim Zeb,A. K. M.Muzahidul Islam,Mahdi
Zareei,Ishtiak AlMamoon,Nafees Mansoor, Sabariah
Fig8.Number of Cluster Head candidates per round
Baharun, Yoshiaki Katayama,and Shozo Komaki
9.”Particle Swarm Optimization-Based Unequal and
Fault Tolerant Clustering Protocol for Wireless Sensor
Networks” :Tarunpreet Kaur and Dilip Kumar
10.”LEACH Partition Topology for
Wireless Sensor Network”: M. Udin Harun Al Rasyid1,
Bih-Hwang Lee, Iwan Syarif, Mokhammad Muqoffi
Arkham
11.”Energy-Efficient Partitioning Clustering Algorithm
for Wireless Sensor Network”: Koffi V. C. Kevin de
Souza, Catherine Almhana1, Philippe Fournier-Viger,Jalal
Almhana