A New Method for Reducing Energy Consumption in Wireless Sensor Networks using Fuzzy Clustering and Fault Tolerance

International Journal of Computer Applications Technology and Research
Volume 3– Issue 2, 109 - 113, 2014
www.ijcat.com 109
A New Method for Reducing Energy Consumption in
Wireless Sensor Networks using Fuzzy Clustering and
Fault Tolerance
Vahab ashrafian
Department of Computer
Science and Research Branch
Islamic Azad University
Ardabil, Iran
Ali Harounabadi
Member of Science Board of
Computer Group in Azad
Islamic university of Tehran
Center, Iran
Mehdi Sadeghzadeh
Member of Science Board of
Computer group in Azad
Islamic university of
Mahshahr, Iran
Abstract: Nowadays, wireless sensor networks, clustering protocol based on the neighboring nodes into separate clusters and fault
tolerance for each cluster exists for sensors to send information to the base station, to gain the best performance in terms of increased
longevity and maintain tolerance than with other routing methods. However, most clustering protocols proposed so far, only
geographical proximity (neighboring) cluster formation is considered as a parameter. In this study, a new clustering protocol and fault
tolerance based on the fuzzy algorithms are able to clustering nodes in sensor networks based on fuzzy logic and fault tolerance. This
protocol uses clustering sensor nodes and fault tolerance exist in the network to reduce energy consumption, so that faulty sensors
from neighboring nodes are used to cover the errors, work based on the most criteria overlay neighbor sensors with defective sensors,
distance neighbor sensors from fault sensor and distance neighbor sensors from central station is done. Superior performance of the
protocol can be seen in terms of increasing the network lifetime and maintain the best network tolerance in comparison with previous
protocols such as LEACH in the simulation results.
Keywords: Wireless sensor networks; fault tolerance; fuzzy algorithms; clustering; faulty sensor.
1. INTRODUCTION
Nowadays, remote control and monitoring systems are one of
the challenging issues in the field of electronics and computer
science. This investigation whenever looking for a solution to
take into accounts the specific conditions and expectations for
the answer. In terms of quality and the same thing, whatever
the cost /effectiveness ratio is lower, the same way as its
popularity is higher. Aware of their surroundings or changes
the state of each set of equipment that can be used as sensors
are known. Each sensor can take into account the changing
environment in terms of certain parameters such as
temperature, humidity, pressure ... Senses and offer. Based on
information from a set of sensors embedded in the
environment, can controls environment and its changes.
Recent advances in electronics and wireless sensors make it
possible multiple-purpose sensors provided with low energy
consumption and cost. These sensors are able to communicate
with each other over short distances. A node is a very small
sensor with sensing equipment, data processing and wireless
communication. In fact, a sensor network is a collection of
numerous sensor nodes are distributed in the environment and
each autonomously and in collaboration with other nodes to
follow a particular purpose. Nodes are close together, and
each node can communicate with another node and other
nodes located in their data available to the environment,
finally the status of desired environment to be reported to
central point.
Techniques and methods used in sensor networks are highly
dependent on the nature of the application, network topology
structure, and environmental conditions, limits and efficiency
and cost parameters. So today, throughout the reputable
universities and computer research centers, electronics and
telecommunications, wireless sensor networks are considered
as a very attractive and popular research field. Many
suggestions and research on various topics sensor network has
been presented and volume of research in this field is
increasing. The main goal of all these efforts and provide
solutions, having a system with a simple control methods,
which is easy and low cost to meet the desired needs
(bandwidth, energy, and environmental interventions . . .) that
could stood up against constraints and provide general
conditions in accordance with the wishes and aspirations
(transfer bulk data content, continuity, long life, low cost,
etc.).
In order to neither do tasks in a wireless sensor network must
consider the time and consumption energy to loss duly works
nor drastic decline in lifetime of the network. In other words,
the constraints facing these networks, and how much and
manner of consume energy is of particular importance
because complete loss of battery in the sensors means they
are destroy and due to the environment and use networks
changing of thousands of sensor node batteries is virtually
impossible.
The reduction of energy consumption in wireless sensor
networks has a direct relationship with increased longevity.
Ideal mood in the sensor networks is that the energy of all
nodes to finish concurrent and it is a mood in which highest
lifetime is possible for the network. Thus, to increase in
network lifetime trying to distribute the load on the network is
a uniform distribution to mineralize the time between death of
the first node and the last node.

Volume 3– Issue 2, 109 - 113, 2014
www.ijcat.com 110
To achieve this goal, several communication protocols have
been proposed so far, in which the protocols are based on
clustering, significantly lowers energy consumption.
The Protocols of the entire network is partitioned into several
cluster-ware and each cluster has a node is selected as cluster
heads. In this case, instead of each node sends its data directly
to the base station, send it to the cluster heads, finally, the
cluster heads then collecting and combining data from all of
the nodes in the cluster, and data are sent to the central station.
In this protocol, choose a sensor as a replace faulty cluster and
appropriate clustering significantly increase the network
lifetime, scalability and efficiency.
2. Previous Work
Ahmadinia and his colleagues are provided a clustering
method of nodes in sensor networks based on ICLA. In the
clustering method, various parameters such as balancing the
cluster size, clusters energy and ... taken into consideration
and compared with other clustering methods, create clusters
with more balanced and increases network lifetime [1].
Akbarzadeh and his colleagues declared that was originally
selected cluster heads by fuzzy logic and considering the
energy parameters of the sensor, numbers of neighbor sensor
and spacing parameters for proper distribution of cluster heads
done in the network then optimize point to move the base
station using a genetic algorithm are determined to energy
efficiency consumption of cluster heads [2].
Attention to the problem of clustering based routing in
wireless sensor networks aimed to reduce energy consumption
and maintain network coverage has been paid. For this
purpose, neural network self-organizing map (SOM) is used
to provide energy-based clustering protocols. The new
protocol based on the energy of the self-organizing clustering
protocol (EBCS) is called clustering according to three
criteria: level of energy and spatial coordinates of each node
are performed and its superiority in terms of longevity and
maintains network covering (simulation) are proved.
(Supervisor: Dr. Reza Askari Moghadam), (Supervisor: Dr. A.
Taraghi Haghighat) [3].
Roshan Zadeh et al are provided an optimize algorithm for
energy consumption and send the matched packets. In the
proposed algorithm is called (PT-Multipath) routing decisions
make is based on the residual energy of nodules. Also, select
nodes for racing with information node one to one and double
jumping node were sender. Simulation results show that the
proposed algorithm efficiently distributed terrific load of
network between nodes overall lifetime wireless sensor
networks has increased. Furthermore, the proposed algorithm
reduced the number of packets sent to the destination during
the release [4].
Toloe Honari using genetic algorithm is implemented as a
central base station, where the cluster heads are determined so
that the network will have minimum energy consumption. In
fact, during the course of operations data, selection criteria for
the new generation of "minimum energy difference grid" have
taken place. Balance and uniformity of the energy
consumption of nodes and long life of network is outcome of
use genetic algorithms in this study [5].
Dechene and his colleagues have been concerned on how
clustering, necessity, benefits, and various combinations of
these patterns in wireless sensor networks and effect of
clustering in wireless sensor networks, both energy efficiency
and optimal use of resources are examined.
In addition to the mentioned issues quality of service in
wireless sensor networks, a high level of importance. Cases
include delay rates, packet loss and network fault tolerance
can be achieved by clustering topics. Protocols introduced in
this paper meet the above mechanism by mentioned
clustering. It should be noted that the optimal number of
clusters is a factor that reduces network soldiers, increase
efficiency and improve the routing and load distribution in the
network. After clustering, in this article some cases need to re-
clustering [6].
Handy and his colleagues presented a method for selecting
cluster heads in a tree structure with the lowest cost. The
protocol based on adaptive clustering or selecting definite
cluster heads cause of death postponement of the first node,
intermediate node and the last node in the network. That is a
smart way to choose cluster heads that select the cluster heads
without the knowledge of their location in the network. Some
of the ideas in this thesis will be based on the same reference
[7].
Mao and his colleagues are considered data collect as a
common but critical operation in various applications in
wireless sensor networks, using the innovative techniques to
improvement the electronic energy consumption and as a
result are necessary to long life of the network. Clustering is a
useful method for topology control in wireless sensor
networks which can increase network scalability and lifetime.
In this paper a clustering approach called EECS offered select
better cluster heads on accordance with energy consumption
loads in the network. Decision-making methods selection
cluster heads taking into account the residual energy through
local radio communication is effective load balancing among
cluster heads. Simulation results show that EECS significantly
increasing the network lifetime over 35 % effective [8].
Ray has been applied his original idea on LEACH protocol
with the aim of improving the energy consumption and
increase the lifetime of the network. In this paper, based on
the residual energy of cluster heads nodes, the distance to sink
and the nodes in their cluster heads will be periodically
decision on future processes. So that cluster heads node with
low residual energy, the distance from the sink is high and
exist in the network as cluster heads will have any chance to
cluster head. Simulation results indicate that the death of the
first node, the intermediate node and the last node in the
proposed protocol, respectively, 41%, 36% and 25% occur
later than LEACH protocol [9].

Volume 3– Issue 2, 109 - 113, 2014
www.ijcat.com 111
3. Proposed Protocol
In the proposed protocol, it is assumed that the base station is
located in a square in LEACH protocol for data transmission
nodes to the base station will consume more energy. This
could be one of the causes of the failure are not considered
faults in the network, which is not covered the sensor nodes
running out their energy and incorrect clustering nodes. This
protocol uses fuzzy logical error coverage reduce network
lifetime, we will work to increase the network lifetime. At
each step of the algorithm based on fuzzy system sensors,
each sensor prioritized and the highest priority will be
selected to move.
The fuzzy fault tolerant algorithm three criteria for selecting
neighbors considered: The criterion measures the amount of
overlay neighbor's node with broken sensor, distance neighbor
from broken sensor and distance neighbor from the base
station. For each neighbor, the values are given as input and
the fuzzy system considered priority for each neighbors in
comparison to other neighbors for each group, the overall
relationship of fuzzy system to choose neighbors would be
equation (1).
(1)
The relation (1), m is the number of fuzzy rules and n is the
number of inputs of the fuzzy system, the system will be 3
times. f (x) x is neighbor priority Xth defines each groups. Iy
is the answer bet Lth and μ (xi) is equal to amount of input
fuzzy sets. μ (xi)l given amount of input overlay and distance
of each sensor and fuzzy triangular used for each of them.
In the algorithm priority of each neighbor in a group
according to its distance from the failed sensor and the
overlapping range of the observations will be determined
using a fuzzy system. Linguistic variables are used to describe
the distance between neighbor and failed sensors, in the short,
medium and long are expressed.
The amount of observations overlapping area of each sensor
neighbor with his neighbors than the sum of all overlapping
neighboring failed sensors sensor is calculated. To obtain this
ratio, the total observations overlapping area of all
neighboring failed sensor with its neighbors is calculated.
The overlapping ratio each neighbor is equal to the ratio of the
observations overlapping area with its neighbors and total
observations overlapping area of all the neighboring sensors.
So much overlapping neighbors of each sensor are a relative
value. Those values said, are normalized value and expressed
in the range 0-100, the proportion is amount overlapping
neighbor to other neighbors. To express the observations
overlapping area of the sensor with their neighbors the
linguistic variables very low, low, moderate, high and very
high are used. To determine the priority of linguistic variables
very bad neighbors, bad, acceptable, good, and very well
used. Fuzzy rule system is designed to select neighbors
summarized in Table 1.
Table 1. Fuzzy rules for selecting neighbors
4. Simulation Results
The simulation of the proposed protocol used ns-2 version
2.29 software package to better display the results of the
second and third series of simulations have been done. In the
first series of nodes in the network 25 nodes, in the second
series 50 nodes and in the third series 100 nodes have been
compared. To determine the performance the proposed
algorithm with the well-known LEACH algorithm has been
evaluated. As the graphs will show the proposed algorithm
could performs better than other proactive and reactive
algorithms.
4.1 Distance of Neighbor to Central Station
It is some nodes based on the location of nodes compare to the
central station, using fuzzy variables center is characterized.
Central chart is displayed in the fuzzy set. To find the central
node, selects the base station of each sensor nodes and
average distance nodes from the base station calculates.
4.2 Distance of Neighbor to failed sensor
It is some nodes based on location of node are compare to
faulty nodes by fuzzy variables interval are specified.
Distance chart is displayed in the fuzzy set. To find the
Cases
Distance
neighboring
nodes of
faulty sensor
Distance of
central
station
from sensor
Overlapping
Probability of
selection
1 High High Low Very low
2 High High Middle Low
3 High High High Relatively low
4 High Middle Low Low
5 High Middle Middle Relatively low
6 High Middle High Middle
7 High Low Low Relatively low
8 High Low Middle Middle
9 High Low High Relatively high
10 Middle High Low Low
11 Middle High Middle Relatively low
12 Middle High High Middle
13 Middle Middle Low Relatively low
14 Middle Middle Middle Middle
15 Middle Middle High Relatively high
16 Middle Low Low Middle
17 Middle Low Middle Relatively high
18 Middle Low High High
19 Low High Low Relatively low
20 Low High Middle Middle
21 Low High High Relatively high
22 Low Middle Low Middle
23 Low Middle Middle Relatively high
24 Low Middle High High
25 Low Low Low Relatively high
26 Low Low Middle High
27 Low Low High Very high

Volume 3– Issue 2, 109 - 113, 2014
www.ijcat.com 112
distance select any sensor and the average distance from the
failed sensor calculates.
4.3 The amount of overlap
It is some nodes based on location of node than failed sensor
is classified, by fuzzy variable the amount of overlapping
neighboring sensors than failed sensor in the fuzzy set is
shown.
Linguistic variables used to represent overlapping, divided to
three level respectively low , medium and high, and three
level to represent the central node and the distance to the base
station , there are near , intermediate and far away,
respectively.
Probability to selects to neighboring sensors are divided
motion direction into seven categories: very low, low,
relatively low, medium, relatively high, high and very high.
Fuzzy base low includes the following rules: If the faulty
sensor distance from the nearest neighbor, distance between
neighboring sensors near the central station overlapping the
neighboring sensors with failed sensor is too high so the
probability selection motion direction is very high.
So we'll use 3
3
=27 rules for fuzzy rule base. In order to
provide fuzzy sets near, far and medium we use the average of
the triangular membership functions and to represent fuzzy
sets low, medium, and large, trapezoidal membership
functions are used.
Development of Membership functions and related linguistic
states is shown in fig.1 and all nodes are compared based on
the probability that the sensor is selected motion direction
with most likely.
Figure 1. fuzzy sets and probability fuzzy variables
Figure 2. comparing the amount of sensor motion in the
network environment
Figure 3. overlapping sensor ratio in the networks
Figures 1 and 2 shown the results of 100 times the
performance of the proposed algorithm on network with 100
sensor nodes in a 50 x 50 environment. The results obtained in
the above figures is the result of the proposed algorithm both
in terms of overlapping amount of sensors in the environment
to cover the lost areas and amount of observation overlapping
area of sensors. As is clear from the figure, failed sensors how
many are reduced down, lost space due to downtime to easily
compensate as well as improving network coverage.
Simulation results show that the proposed fuzzy algorithm
provided better coverage than the conventional fuzzy
algorithm and the network's energy consumption is reduced.
Figure 4. first node dead
As figure 4 is shown; the proposed protocol has better
efficiency than LEACH protocol in the parameter of first node
death.
As can be seen in 25 nodes that is 5.069293 increasing the
number of nodes randomly reduces or increases the duration
of the first node death. In total, 100 nodes have shown an
increase than in the 50 nodes. But the number has dropped to
50 nodes compared to 25 nodes. The results show a decrease
in energy consumption in the network.

Volume 3– Issue 2, 109 - 113, 2014
www.ijcat.com 113
5. REFERENCES
[1] Ahmadinia, M., Meybodi, M.R. 2008. "Clustering in
Wireless Sensor Networks Using Cellular Learning
Automata", Computer Society of Iran.
[2] Akbarzadeh Totonchi, M., Yaghma'ee Moghaddam, M.H
2008. "Clustering self-organizing sensor with movable
base stations in wireless sensor networks using fuzzy
logic and genetic algorithms", Computer Society of Iran.
[3] Anami, N. 2009. "Reducing energy consumption in
wireless sensor networks using neural networks SOM",
Dissertation for the degree of Master of Engineering,
Computer-oriented software, Information and
Communication Technology Department, Payam Noor
University Central Tehran Branch.
[4] Roshanzadeh, M. 2009. "Providing an optimal algorithm
in terms of energy consumption and reliable sending in
the wireless data network", National Conference on Soft
Computing and Information Technology, Islamic Azad
University, mahshahr.
[5] Toloe Honary, M., Tashtarian. 2007. "Node clustering in
the wireless sensor network using genetic algorithm",
First Joint Congress on Fuzzy and Intelligent Systems.
[6] Dechene. D. J, Jardali. A. El, Luccini. M and Sauer. A.
2008. ”A Survey of Clustering Algorithms for Wireless
Sensor Networks”, Department of Electrical and
Computer Engineering the University Of Western
Ontario London MS.Thesis, Ontario, Canada.
[7] Handy. M. J, Haase. M, Timmermann. D. 2002. “Low
Energy Adaptive Clustering Hierarchy with
Deterministic Cluster-Head Selection”, Fourth IEEE
Conference on Mobile and Wireless Communications
Networks, Stockholm, Erschienen in Proceedings, pp.5.
[8] Mao. Ye, Chengfa. Li, Guihai. C and Jie.Wu. 2005.
”EECS: An Energy Efficient Clustering Scheme in
Wireless Sensor Networks”, in Performance, Computing,
and Communications Conference, IPCCC 2005., 24th
IEEE International, pp. 535-540.
[9] Ray. A. 2012. “Energy efficient cluster head selection in
wireless sensor network”, 1st International IEEE
Conference in Recent Advances in Information
Technology (RAIT), Kolkata, India, pp. 306-311.

A New Method for Reducing Energy Consumption in Wireless Sensor Networks using Fuzzy Clustering and Fault Tolerance

Related slideshows

More Related Content

A New Method for Reducing Energy Consumption in Wireless Sensor Networks using Fuzzy Clustering and Fault Tolerance