Energy Efficient Cluster Head Election using Fuzzy Logic in Wireless Sensor Networks by ijcsiseditor


									                                                               (IJCSIS) International Journal of Computer Science and Information Security,
                                                                                                                   Vol. 9, No. 5, May 2011

  Energy Efficient Cluster Head Election Using Fuzzy
         Logic In Wireless Sensor Networks
                  Mostafa Basirnezhad                                                  Dr. Masood Niazi Torshiz
           Department of Computer Engineering                                     Department of Computer Engineering
           Islamic Azad University of Mashhad                                     Islamic Azad University of Mashhad
                     Mashhad, Iran                                                          Mashhad, Iran

Abstract— Routing problem is one of important issues in                   we can provide them in dangerous or inaccessible places we
wireless sensor networks and many methods for this purpose is             leave. Each sensor node contains a processor on your board
presented. One of these methods is the method of clustering. In           and instead of sending all the raw data to center or node
this method suitable cluster head can be effective for efficiency         responsible for information processing and concluded, his
and network lifetime. Nodes in the clustering approach can be             first in a series of basic and simple processing on the
divided into several groups and each group has a node with the            information gained, performs and will send the data semi-
name of cluster head which collect data from the rest nodes of            processed [2].
the cluster and leads them to the sink. Although clustering
reduces energy consumption, there might be some problem.                      Work on wireless sensor networks in the first began
The main problem is that the most energy consumption is                   defense and military purposes, but became also many other
gathered on the cluster head. To overcome this problem in                 applications quickly that some of the applications of this
clustering methods, energy consumption should be distributed              technology is in military and security applications, monitor
that could be done by choosing appropriate Cluster head. In               internal and external environments (used in buildings
this paper a clustering trend based on the clusters                       Intelligent, traffic control, detection of natural disasters,
reorganization and selecting fuzzy Cluster head is introduced.            agriculture and environmental monitoring), industrial
Fuzzy variables remaining energy of node distance of node to              applications and medical applications (health care and
sink and average distance of node to other nodes to select
                                                                          surgery) [3].
Cluster heads considered. We compare the proposed algorithm
terms of energy consumption and network lifetime with                         Other articles seek to be as follows. In second section
LEACH, Gupta and CHEF methods. Simulation results show                    LEACH algorithm and two fuzzy logic methods for selecting
that the proposed method in comparison with other methods                 Cluster head use is discussed. The third section main model
have been improved considerably.                                          in wireless sensor networks, we define. In Fourth section, the
                                                                          proposed model is introduced. In fifth Section simulation
    Keywords- increased network lifetime; clustering; fuzzy               results and charts shown and in sixth section we express
logic; wireless sensor networks

                                                                                               II.    RELATED WORK
                      I.    INTRODUCTION
                                                                               One of the introduced and old clustering protocols is
    In recent years are technological advances and the                    LEACH protocol [4] in wireless sensor networks that has
telecommunications industry, small electrical and electronic              been established two Setup phase and steady state phase.
components, leading to construction of small and relatively               Steady state phase in a single data transmission step occurs.
inexpensive sensors which through a wireless network relate               Each node in the cluster such as Cluster head is elected. Data
to each other. The networks that wireless sensor networks to              collected from member node, before a base station or sink,
be known, to a suitable tool for extracting data from the                 locally processed data in Cluster head added if any of it is
environment and monitoring environmental events have                      removed and then form a new package will be sent to the
become and their Application in household, industrial and                 base station, the Because energy consumption more than
military day to day is increasing. In wireless sensor network             node Cluster head is normal after a while their energy is all.
design major issue is limited sensor energy source. On the                Therefore, the use of clustering in LEACH has been dynamic
other hand there because too many sensor in network or lack               [5].
of access to them, or replace the sensor battery is not
practical. Therefore provide ways to optimize energy                          This means that after each course and receive operations,
consumption, which ultimately will increase the network                   in other words after each launch the implementation phase,
lifetime, is strongly felt [1].                                           Cluster head is changed and another node randomly chosen
                                                                          to be as Cluster head. But random selection Cluster head
    A sensor network consists of many sensor nodes in an                  improper distribution may lead them in the network. This
environment which is widely distributed and collected data                means that part of the area covered by two or more are
from the environment. Not necessarily located where the                   gathered Cluster head while in another area there are no
sensor nodes are clear. Such properties to the possibility that           Cluster head.

                                                                                                     ISSN 1947-5500
                                                             (IJCSIS) International Journal of Computer Science and Information Security,
                                                                                                                 Vol. 9, No. 5, May 2011
    LEACH clustering algorithm in the model locally and is                     In Gupta method is consumed energy for all nodes send
based on probability models. In this algorithm for each node               location information to the BS. Considering wireless sensor
n Cluster head being a random number between zero and one                  networks for development on a geographical area of the main
will choose. If the number is smaller than a threshold T (n),              objectives and receive collected information is meaningful.
the node for the current round is Cluster head. Threshold is               The assumption is that nodes at least are moving, so send
defined by equation (1).                                                   location information during setup phase is sufficient.
                                                                           Operation of fuzzy cluster head selected is divided into two
                                                                           stages, each is including a setup phase and steady state phase
                                                                           similar to LEACH algorithm. During the setup phase, Cluster
                                                                          head are determined by fuzzy knowledge process and then
                                                                           clusters are formed. In steady state phase Cluster head given
   In equation (1) we have:                                                by aggregation are collected and signal processing functions
                                                                           for data compression are performed in a signal. Then the
   P is probability of being Cluster head.                                 combined signal is sent to the BS. Radio model used in this
   r is current round number.                                              method is the same model of LEACH [9].
   G is a set of nodes which in 1 / p before round have not                   In this method, all nodes based on their chance are
been Cluster head.                                                         compared and node with the highest chances to be selected
                                                                           as Cluster head. Each node in the cluster joins itself to
    When Cluster head were selected, a message to other                    Cluster head and will start sending data. Data transmission
network nodes as their introduction will send. The nodes,                  phase is similar to steady state phase of LEACH algorithm.
their clusters are selected. This choice is based on the
received signal strength. After the time clusters formed,                      This method increased somewhat network lifetime, but
Cluster head a TDMA schedule to create and to each node of                 the problem is central selection mechanism, i.e., sink should
a period when the node can send information to offers. This                be all about energy and the distance of the sensor node can
schedule is broadcast in the entire cluster.                               collect this information and in accordance with fuzzy logic,
                                                                           to select Cluster head and other problems that is that each
    During steady state phase, sensors located in the cluster              round is only one choice Cluster head and are causing many
can receive and operate the data to do Cluster head. This                  overhead in sink [9].
information is Cluster head gathering, compression, and then
they are submitted to the sink. After a specified time phase                   CHEF method [10] was introduced a mechanism elected
network is launching new and Cluster head are elected. In                  Cluster head locally using fuzzy logic. This method on
clusters to communicate and to reduce interference nodes                   factors that affects the network lifetime defines and
belonging to different clusters reject CDMA is used [6].                   according to these factors and IF-Then fuzzy rules, Cluster
                                                                           head is select. In this method, two fuzzy sets of local energy
    Although LEACH increases the network lifetime, but in                  and distance (total distance between node and node which
this protocol there are several assumes. LEACH assumes that                are at r distance.) Is defined and using two sets Cluster head
all nodes have enough energy to have communication with                    chances for getting a node is calculated. Fig. 1 and equation
the sink and each node has computational power to support                  (2) shows to calculate the local distance.
the MAC protocol is different. Therefore applicable to large
networks is not. Also assume that nodes always have data to
send. Consequently nodes near the data they send a similar.
Also unknown is how Cluster head are in the distribution                                                                              
network means may all together in a corner of the network
and therefore are there will be sensors that are no Cluster                    r mean radius of clusters is desired. Equation (3) shows
head.                                                                      the calculated r. In this Equation n is number of sensors in
    So the idea of dynamic clusters overhead brings a lot.                 wireless sensor network.
Cluster head such changes and confirmation messages that
may be reduced energy. Finally, the protocol assumes that all
nodes with a certain amount of energy in each round are
elected and assumes that energy consumption Cluster head
almost equal with the other nodes [7].
    In Gupta method Cluster head at each round by                                                                          
calculating the chance of each node for to be Cluster head
described with three fuzzy term node density, energy level
and the centrality of each node is selected. In This algorithm                 CHEF method avoids problems of LEACH method and
central control is in BS that has the network global                       the sensor node that has more energy is being better chance
information and select best Cluster head. BS is many times                 for Cluster head. In addition, fuzzy variable local distance,
more powerful than sensor nodes, and has enough memory                     locally to be selected optimal node that has the most chances
and power [8].                                                             as Cluster head because energy consumption in wireless
                                                                           communication depends on distance between two nodes and

                                                                                                    ISSN 1947-5500
                                                                               (IJCSIS) International Journal of Computer Science and Information Security,
                                                                                                                                   Vol. 9, No. 5, May 2011
node density. According to the local distance and energy,
CHEF can choose optimal Cluster head at any time and to                                                              Erx  l * Eelec
increase the network lifetime [11].                                                                                                              

                                                                                                           IV.    PROPOSED ALGORITHM
                                                                                               In this method sensor nodes randomly in a circular area
                                                                                           of radius R are placed. Each node will send position and its
                                                                                           remaining energy to the sink located in the center area. Sink
                                                                                           based on the percentage of expected Cluster head (p) can
                                                                                           share the network a few sectors. The total network is divided
                                                                                           into N * P sector that N is number of wireless sensor network
                                                                                           node and the angle between the two sectors (θ) are obtained
                                                                                           by equation (7).

               Figure 1. Calculate the local distance
    CHEF also overcome the problems of method Gupta and
BS will not need to collect fuzzy information of nodes and                                     Then each of these sectors according to Fig. 2 lays a
will cause additional overhead. Furthermore, the method of                                 cluster with the angle θ and in any cluster selects a Cluster
Gupta choice only one Cluster head at each round.                                          head using fuzzy sets.

                     III.     SYSTEM MODEL
    In this case the system model is given to each sensor
node sends its Cluster head and Cluster head, the collected
data aggregation and that they will send to the sink. Some of
the assumptions of this method follow [3]:
   • Sensor nodes are homogeneous.
   • Distance can be by wireless radio signals can be
   • Nodes are fixed and no motion.
   • All sensor nodes have the same primary energy.                                                        Figure 2. Proposed system model

   Sink in the center of the sensor network is located.                                        Select Cluster head in proposed method is doing with
    Equation (4) the amount of energy consumed in a L bit                                  defining three fuzzy sets the residual energy of each node,
depending on the distance d shows. Eelec are amount of                                     the distance between node to sink and average distance of
energy consumed per bit to run radio transmitter and receiver                              node to other node. Output fuzzy variable is the chance of
circuits. D0 values of the equation (5) are obtained and that                              node for select Cluster head. In a next stage by adding a
the Efs and Emp are amount of energy wasted by the amplifier                               fixed amount to θ that we show it with φ, the clusters are re-
radio bits [2], [13].                                                                      organized and new Cluster head such prior steps are select to
                                                                                           the fuzzy methods. With this work nodes that at previous
                                                                                           stages have not been selected Cluster head have better
            l * (
                   E elec    E        *d
                                                     d  d0                                chance to be Cluster head. Angle φ can be according to the
                                 fs
       E tx l * (                                                                         number of sectors such as Fig. 3 was determined. This
                   E elec                           d  d0
                             E        *d                                                  continues until over first energy sensor node.

                                            E   fs
                              d   0
                                            E   mp

    Amount of energy consumed in receiving a L-bit package
as equation (6) is calculated.

                                                                                                              Figure 3. Changing clusters

                                                                                                                     ISSN 1947-5500
                                                                       (IJCSIS) International Journal of Computer Science and Information Security,
                                                                                                                           Vol. 9, No. 5, May 2011
    Input fuzzy variables (residual energy of each node, the                               TABLE I.     FUZZY RULES OF PROPOSED ALGORITHM

distance between node to sink and average distance of node
                                                                                      ID      Energy   Distance   AverageDistance        Chance
to other node) and output fuzzy variable (the node chance to
select Cluster head) are defined as follows:                                          1        Low       Far             Far            VeryWeak

    Fuzzy variable residual energy of each node are divided                           2        Low       Far            Med             VeryWeak
into three levels Low, Med and High and fuzzy variables                               3        Low       Far            Near              Weak
distance to sink and average distance node to other node are
divided into three levels Near, Med and Far. Output means                             4        Low       Med             Far            VeryWeak
chance to choose Cluster head are divided into nine levels                            5        Low       Med            Med               Weak
VeryWeak, Weak, LittleWeak, LowMed, Med, HighMed,
                                                                                      6        Low       Med            Near            LittleWeak
LittleStrong, Strong, and VeryStrong. Membership functions
of residual energy, distance to sink, the average distance to                         7        Low       Near            Far              Weak
other nodes and the chances can see in Fig. 4, 5, 6 and 7                             8        Low       Near           Med             LittleWeak
                                                                                      9        Low       Near           Near             LowMed
                                                                                      10       Med       Far             Far             LowMed
                                                                                      11       Med       Far            Med              LowMed
                                                                                      12       Med       Far            Near               Med
                                                                                      13       Med       Med             Far             LowMed
                                                                                      14       Med       Med            Med                Med
                                                                                      15       Med       Med            Near               Med
       Figure 4. Membership functions of residual energy of node                      16       Med       Near            Far               Med
                                                                                      17       Med       Near           Med             HighMed
                                                                                      18       Med       Near           Near            HighMed
                                                                                      19       High      Far             Far           LittleStrong
                                                                                      20       High      Far            Med            LittleStrong
                                                                                      21       High      Far            Near              Strong
                                                                                      22       High      Med             Far           LittleStrong
                                                                                      23       High      Med            Med               Strong
        Figure 5. Membership functions of distance node to sink                       24       High      Med            Near              Strong
                                                                                      25       High      Near            Far              Strong
                                                                                      26       High      Near           Med            VeryStrong
                                                                                      27       High      Near           Near           VeryStrong

                                                                                     According to the fuzzy input and output variables, fuzzy
                                                                                 rules base is including 27 law that is shown in Table (1). Our
                                                                                 fuzzy logic control model includes a fuzzifier, fuzzy rules,
Figure 6. Membership functions of average distance of node to other node         fuzzy inference engine and is a de-fuzzifier. In this paper, the
                                                                                 most famous model named fuzzy inference technique is used
                                                                                 Mamdani. For the output membership function is expressed
                                                                                 center of gravity method is used as equation (8).


                                                                                 In equation (8) μA(x) is Membership function of fuzzy
           Figure 7. Membership functions of chance of node                      set A.

                                                                                                            ISSN 1947-5500
                                                             (IJCSIS) International Journal of Computer Science and Information Security,
                                                                                                                 Vol. 9, No. 5, May 2011
    The proposed algorithm was assessed in MATLAB. For
simulation a wireless sensor network with 100 nodes in a
circular environment with radius 100m is considered. Nodes
distribute randomly in the environment with same energy of
one joule and the sink were located in the center region is
assumed that no sink in terms of energy restriction and all
processing is done in the sink. Position of each node to
determine the distance to the sink and angular node that has
the X axis in polar coordinates is considered. We compare
the proposed algorithm with the LEACH, Gupta and CHEF
algorithm of network lifetime. Here the network lifetime
when the first network node is lost we have considered. The
Fig. 8 shows simulation result after 200 stages. As in Fig. 8
                                                                                   Figure 8. Cmpare network lifetime for 100 nodes
can be seen, the network lifetime compared to other methods
has increased. This increase is due to in optimal selection
cluster head in each cluster and the appropriate distribution
of clusters over the simulation environment.
    Network residual energy in each round is good scale to
measure performance of algorithm. Lower slope in the
diagram of energy balance and better distribution of
appropriate energy nodes will specify. The Fig. 9 shows
compared the rate of energy consumption in the four
methods mentioned above. The Fig. 9 is determined by the
network residual energy in the proposed method more than
the three other methods.
    Distance between Cluster head is one of the important
parameters in clustering. In this algorithm order to the region
is equal and in each region one Cluster head are selected
                                                                                          Figure 9. Network residual energy
from rally Cluster head in the near and from distance falling
between cluster heads prevents and increase network lifetime
and energy consumption will be better.
   Fig. 10 shows energy consumption in the proposed
algorithm compared with the three other methods mentioned.
As received from the chart increase rate of energy
consumption in the proposed method than the other three
methods is reduced. This reduction of energy consumption
due to right distribution of cluster Heads in the environment
and the use of fuzzy logic is to select Cluster head.

                      VI.   CONCLUSION
    Optimize energy consumption in wireless sensor
networks is very important, so that the optimal energy
consumption lead to increase network lifetime. In this paper,                             Figure 10. Energy consumption rate
using the fuzzy as sink run in Central, we have determined
place of Cluster heads and we have the minimum energy                   References
consumption in the network. Also, with reorganization of                [1] W. Heinzelman, A. Chandrakasan, and H. Balakrishnan, “An
clusters we could increase wireless sensor network lifetime.                 Application-Specific Protocol Architecture for Wireless Microsensor
                                                                             Networks,” IEEE Transactions on Wireless Communications, vol. 1,
                                                                             no. 4, pp. 660-670, October 2002.
                                                                        [2] S.H. Park, J.S. Cho, S.M. Jung, Y.J. Han, and T.M. Chung,
                                                                             “Architecture of Context-Aware Integrated Security Management
                                                                             Systems for Smart Home Environment” , APNOMS2007, LNCS
                                                                             4773, pp. 543-546, October 2007.
                                                                        [3] K. Akkaya, M. Younis, “A Survey of Routing Protocols in Wireless
                                                                             Sensor Networks,” in the Elsevier Ad Hoc Network Journal,
                                                                             Vol.3/3,pp. 325-349, 2005.
                                                                        [4] W. Heinzelman, A. Chandrakasan and H. Balakrishnan, ”Energy-
                                                                             efficient communication protocols for wireless microsensor

                                                                                                     ISSN 1947-5500
                                                                      (IJCSIS) International Journal of Computer Science and Information Security,
                                                                                                                          Vol. 9, No. 5, May 2011
      networks” Proceedings of the Hawaii International Conference on            [10] J. Myoung Kim, S. Park, Y. Han and T. Chung, “CHEF: Cluster Head
      Systems Sciences, Jan. 2000.                                                    Election mechanism using Fuzzy logic in Wireless Sensor Networks,”
[5]   A.Khadem Zadeh, A.Hosseinalipour, S. Babaie "New Clustering                     10th International Conference on Advanced Communication
      protocol for increasing Wireless Sensor Networks Lifetime"                      Technology(ICACT), pp.654-659, February 2008.
      International Journal of Computer and Network Security,Vol. 2, No.         [11] Kumar 0, Aseri TC and Patel RB, "EEHC : energy efficient
      1, January 2010.                                                                heterogeneous clustered scheme for wireless sensor networks,"
[6]   Y.J. Han, S.H. Park, J.H. Eom, and T.M. Chung, “Energy-Efficient                Computer Communication, 32(4), 2009, p p. 662-667.
      Distance Based Clustering Routing Scheme for Wireless Sensor               [12] Israr N and A wan I, "Coverage based inter cluster communication for
      Networks”, ICCSA2007, LNCS 4706, Part II, pp. 195-206, 2007.                    load balancing in heterogeneous wireless sensor networks,"
[7]   Namhoon Kim, Jongman Heo, Hyung Seok Kim, Wook Hyun Kwon                        Telecommunication. Sits, 38(3-4), 2008, pp. 12 1-132.
      "Reconfiguration of cluster heads for load balancing in wireless           [13] Haosong G., Gang L.,Younghwan Y.,” A Partition-Based Centralized
      sensor networks" Elsevier, Computer Communications 31 (2008)                    LEACH Algorithm for Wireless Sensor Networks using Solar
      153–159.                                                                        Energy”. in Proceedings of ICHIT2009.
[8]   I.Gupta, D. Riordan and S. Sampalli, “Cluster-head Election using          [14] Giuseppe Anastasi, Marco Conti, Mario Di Francesco, Andrea
      Fuzzy Logic for Wireless Sensor Networks,” In Proceedings of the                Passarella, Energy Conservation in Wireless Sensor Networks: a
      3rd Annual Communication Networks and Services Research                         Survey, Ad Hoc Networks, 2009.
      Conference, pp.255-260, May 2005.
[9]   M.R. Minhas, S. Gopalakrishnan and V. Leung, “Fuzzy Algorithms
      for Maximum Lifetime Routing in Wireless Sensor Networks”,
      Global Telecommunications Conference, 2008, pp. 1-6.

                                                                                                              ISSN 1947-5500

To top