Professional Documents
Culture Documents
Research Journal
Volume 9, No. 25, March 2015, pages 7–11 Original Article
DOI: 10.12913/22998624/1918
1
Department of Computer Engineering, Marvdasht Branch, Islamic Azad University, Marvdasht, Iran, e-mail:
Afsharzade@gmail.com
2
Department of Computer Engineering, Bushehr Branch, Islamic Azad University, Bushehr, Iran, e-mail:
JelodarH@gmail.com
3
Department of Computer Engineering, Sari Branch, Islamic Azad University, Sari, Iran, e-mail: Javad_
Aram_66@yahoo.com
7
Advances in Science and Technology Research Journal Vol. 9 (25) 2015
a result, necessary and processed data is transmit- placed, sensor failure is highly probable. Due to
ted; this decreases the amount of transmitted data. the Large number of sensors in these networks,
It will decrease network traffic as well as result- design and operation of such networks requires a
ing in better data fusion. Mentioned characteris- scalable architecture and applying specific man-
tics have caused these networks to be applied to agement. Furthermore, sensors in such environ-
various fields, such as military, hygienic, industrial ments have considerable energy consumption.
and smart home. Table 1 and Figure 1 demonstrate Hence, designing energy saving algorithms sig-
components and applications of wireless sensor nificantly prolongs sensor lifetimes and, in turn,
networks, respectively. network lifetime increases. Clustering is a meth-
od for decreasing energy consumption and in-
creasing network lifetime. Besides, it offers other
CLUSTERING CONCEPT IN WIRELESS advantages such as higher security, decreasing
SENSOR NETWORKS redundant data and increasing scalability. Clus-
tering divides nodes into groups which are called
In wireless sensor networks, sensors are ran- clusters. Each cluster is managed by a cluster
domly distributed in a wide area. Since battery head and all members are coordinated with the
lifetime is short and they cannot be charged or re- cluster head. Cluster heads are responsible for
communicating with their own members and
Table 1. Components of wireless network other cluster heads. There are diverse methods
Wireless network
Definition
for selecting cluster head. In some methods clus-
components ter head is selected by cluster members; while,
Sensor A device which detects an object
or event or measures a physical
in other methods cluster head is selected by net-
quantity and converts it to electric work designer. Cluster head may either remain
signal. There are different types constant during network lifetime or change based
of sensors such as temperature
sensors, pressure, humidity, photo on the algorithm. The same is true for members
and acceleration sensors of clusters. In clustering methods each sensor is
Sensor node A node consisting of one or more either a cluster head which introduces itself in a
sensors
specific region or is a member which must intro-
Network A network composed of sensor
nodes with a specific objective duce itself to cluster head and become its mem-
Base station A node which is responsible for ber. The members are only able to communicate
receiving data and establishes with their own cluster head and transmit their
communication between nodes and
task manager node. data to it. Each cluster head receives data from
Task manager A node by which a person as its members, processes obtained data (if neces-
operator or network manager sary), aggregates processed data and transmits it
communicates with the network
to the base station [2].
LEACH algorithm
It is a self-organization clustering algorithm
which distributes energy consumption between
all sensors of the network. In this algorithm sen-
sors are organized in local networks. In each lo-
cal network one sensor is selected as cluster head.
According to this algorithm all sensors might be
selected as cluster head with a specific probability;
Fig. 1. Applications of wireless sensor networks [1] however, due to importance of energy as a crucial
8
Advances in Science and Technology Research Journal Vol. 9 (25) 2015
Algorithm operation
LEACH could be divided into several rounds.
Each round starts with a setup phase and finishes
with steady state phase. In the setup phase clus-
ters are organized and in steady state information
is transmitted to base station.
Setup phase
First off, each sensor determines whether it
wants to be cluster head or not considering its re-
sidual energy and number of times it is selected as
cluster head. The decision is made based on a ran-
dom number between zero and one. Afterwards,
if selected number is less than threshold that node
will be cluster head in a new round. This trust
threshold is derived from following equation:
Fig. 2. Classification of clustering algorithms
9
Advances in Science and Technology Research Journal Vol. 9 (25) 2015
transmission the node turns off its radio hardware tentative cluster head, final cluster head or cov-
to avoid unnecessary energy consumption. It is ered. At the beginning of grouping phase a node
noteworthy that radio hardware of cluster head with higher energy level is more probable to be
must be on as long as it is playing cluster head selected as tentative cluster head. It transmits a
role in the cluster. When all nodes transmit their message to all sensor nodes existing in its group
data to cluster head, it processes received data range and informs them about its new condition.
to compress it as a unit. This is the operation of For this purpose, AMRP (average power) is intro-
network in steady state conditions. Then, cluster duced which is a cost criterion and is exploited
head transmits information to base station. When for breaking constraints in procedure of cluster
this phase finishes, based on determined time head selection.
span the setup phase starts again to determine This algorithm has four main goals:
next cluster head. If one node is constantly select- •• Prolonging network lifetime by distributing
ed as cluster head it loses its energy very soon and energy consumption
dies. When a cluster head is dead all dependent •• Terminating clustering process by constant
nodes will be out of work. Hence, the principle number of iterations
of this algorithm is involving all nodes as cluster •• Minimizing control overload
heads in the network. Using this algorithm cluster •• Providing appropriate distributed cluster
heads change alternatively to address the above- heads and compressed clusters.
mentioned problem.
HEED does not consider any assumptions re-
garding distribution, density or ability of nodes
(such as nodes ability relative to their location).
Since the algorithm aims to prolong network life-
time, it decides based on residual energy of each
node. To improve energy efficiency and network
lifetime, inter cluster communication cost is also
In the next round all nodes which were clus- taken into account. Usually each node has a lim-
ter heads will set their probability to zero and the ited number of transmission power levels. As the
other nodes will increase their probability. If the power increases the covered area is increased as
probability of being cluster head is p, its prob- well. Thus, cluster range is determined by trans-
ability in the next round is pr. In the first round p mitted power level used in inter cluster communi-
is zero which means that the node was not clus- cation and during clustering.
ter head. p/1 is the total number of rounds. This
procedure continues till all nodes are selected as Algorithm operation
cluster head once. When several nodes are candidate for being
cluster head, the node which has the lowest inter
Advantages of this algorithm cluster cost will be selected as cluster head.
•• This algorithm does not need information of This cost is a function of:
whole system and is completely distributed. •• Cluster characteristics, such as size of a cluster.
•• Network lifetime is prolonged in comparison •• Variability or non-variability of power levels
with other existing algorithms. used in intra cluster communications.
•• With scheduling the nodes, they can stay in
sleep mode and save energy. Additionally, If power levels for cluster nodes are the same,
data packet contention is avoided [1, 3]. communication cost cloud be dependent on the
following parameters:
HEED algorithm Node degree, if it is necessary for load
to be distributed on cluster heads.
HEED clustering algorithm is a distributive Inverse of node degree, if denser clusters
clustering method which considers energy and are needed
communication cost simultaneously. HEED is an
iterative grouping algorithm which utilizes resid- Now consider a case where variable power
ual energy of nodes and their communication cost levels could be utilized for intra cluster commu-
to choose the best set as cluster head nodes. Dur- nications. In this condition MinPwri is the mini-
ing clustering procedure a sensor node might be mum power level required for node i (1<i<M)
10
Advances in Science and Technology Research Journal Vol. 9 (25) 2015
so that VI is able to communicate with a cluster CH. If at the end of HEED algorithm one node
head, such as U. M denotes the number of nodes does not select a cluster head that node would be
in the cluster range. Average Minimum Reach- final cluster head. In the next step if a cluster head
ability Power (AMRP) is considered as average with lower cost appears, a tentative cluster head
needed power level for all M nodes in the cluster might be converted to a normal node. Notice that
range for reaching u. if a node has large residual energy and low cost it
could select itself as cluster head successively [5].
REFERENCES
1. Abbasi A.A., Younis M. A survey on clustering algo-
Assuming that an optimum percent cannot be
rithms for wireless sensor networks. Elsevier 2007.
determined before executing the algorithm, at the
2. Malik M., Singh Y. Analysis of LEACH protocol
beginning an arbitrary percent is considered for
in wireless sensor networks. International Journal
probability of being cluster head (Cprob = 0.5%).
of Advanced Research in Computer Science and
In this equation Eresidual is the energy remained Software Engineering, Feb. 2013.
in the node and Emax is the maximum energy of
3. Xuxun Liu 2012. A survey on clustering routing
node. The value of Chprob for each node cannot protocols in wireless sensor networks.
be less than a threshold value pmin (4–10) which
4. Heinzelman W.B., Chandrakasan A.P. An applica-
is proportionate to inverse of Emax. This limita- tion specific protocol architecture for wireless mi-
tion is considered to terminate the algorithm after crosensor networks. IEEE, 2002.
Niter = o(1) iterations. In each iteration, i uncov-
5. Kour H., Sharma A.K. Hybrid energy efficientdis-
ered nodes select themselves as cluster head with tributed protocol for heterogeneous wireless sensor
Emax probability. After step i SCH cluster heads network. International Journal of Computer Appli-
are equal to union of cluster head sets after i–1 cations, 2010.
and new cluster heads in i iteration. Each node vi 6. Yun Li, Nan Yu, Weiyi Zhang, Weiliang Zhao,
selects a node with the lowest cost in the SCH as Xiaohu You, Daneshmand M. Enhancing the per-
its cluster head which can even be the node itself. formance of LEACH protocol in wireless sensor
Afterwards, each node doubles its Chprob and networks. IEEE, 2011.
goes to next step. Pseudo algorithm is illustrated 7. Demirbas M., Arora A., Mittal V., Kulathumani V.,
in figure 7. As the neighbors are detected once, it A fault-local self-stabilizing clustering service for
is not necessary to detect neighbors in each itera- wireless ad hoc networks. IEEE, 2006.
tion of algorithms. Nodes detect their neighbors 8. Demirbas M., Arora A., Mittal V. FLOC: A fast lo-
by transmitting and receiving messages. Each cal clustering service for wireless sensor networks.
node which selects itself as cluster head transmits 9. Demirbas M., Arora A., Mittal V., Kulathumani V.
a message in the form of clusterhead msg (Node Design and analysis of a fast local clustering ser-
ID, selection status, cost). vice forwireless sensor networks. IEEE.
If Chprob is less than 1, the node is tentative 10. Mamalis B., Gavalas D., Konstantopoulos C.,
CH, and if it is equal to one, it would be final Pantziou G. Clustering in wireless sensor networks.
11