You are on page 1of 6

JOURNAL OF COMPUTING, VOLUME 2, ISSUE 11, NOVEMBER 2010, ISSN 2151-9617

HTTPS://SITES.GOOGLE.COM/SITE/JOURNALOFCOMPUTING/
WWW.JOURNALOFCOMPUTING.ORG 34

An Oriented Cluster Formation for


Heterogeneous Wireless Sensor Networks
Moulay Lahcen HASNAOUI, Abderrahim BENI HSSANE, Abadellah EZZATI and Said BENALLA

Abstract— In wireless sensor networks clustering is one of the crucial methods for increasing the network’s lifetime. In this
paper, we propose an Oriented Cluster Formation to enhance the performance of LEACH-E protocol, we call this clustering
protocol (OCFLE). In OCFLE, we extend LEACH-E by introducing an oriented cluster formation based on the position of the
cluster-head and the best way to transmit data with the minimum distance between the normal nodes and the base station using
an adequate intermediate cluster-head. Also, the cluster-head are elected by a probability based on the ratio between residual
energy of each node and the average energy of the network. Moreover, we use a 2-level hierarchy concept by selecting an
intermediate befitting cluster-head for data transmission and we utilize an adapted formula to estimate the network lifetime, thus
avoiding the need of assistance by routing protocol in LEACH-E. Simulation results show that OCFLE increases the lifetime of
the whole network and performs better than LEACH by about 109%, than SEP by about 62% and than LEACH-E by about 35%
in term of first node dies.

Index Terms— Clustering algorithm, Energy-efficient, Heterogeneous environment, Lifetime, Wireless sensor networks

——————————  ——————————

1 INTRODUCTION

W IRELESS sensor networks (WSNs) pose new re-


search challenges related to the design of algo-
rithms, network protocols, and software that will
such sensor nodes. These tiny sensor nodes consist of
sensing, data processing and communicating compo-
nents. Data are collected at sensor nodes and transmitted
enable the development of applications based on sensor to the Base Station (BS) directly or by means of other in-
devices. WSNs are envisioned to be applied in different termediate nodes to give the end-user the ability to in-
applications, including, among others, habitat, environ- strument, observe, and react to events and phenomena in
mental, and industrial monitoring, which have great po- that environment. Since the cost of transmitting informa-
tential benefits for the society as a whole. Also various tion over a long distance is higher than computation,
and useful applications, such as detection of chemical clustering sensors into groups so that sensors communi-
activity in military field, healthcare monitoring, and wild cate information only to Cluster-Heads (CHs) and then
life sensing, exploit the strengths of WSNs. Wireless sen- the CHs communicate information to the BS. This tech-
sor nodes are typically less mobile and more densely nique is used to reduce energy consumption and then
deployed than mobile ad hoc networks (MANETs), and increase the lifetime of the network [2], [3]. The CHs,
they must be left in hostile environments, which makes it which are elected periodically by certain clustering algo-
difficult or impossible to recharge or replace their batte- rithms, aggregate the data of their cluster’s members and
ries. This necessitates devising novel energy-efficient send it to the BS, from where the end-users can access the
solutions to some of the conventional wireless network sensor data. Thus, only some nodes are required to
problems, such as medium access control (MAC), routing, transmit data over a long distance and the rest of the
self-organization and bandwidth sharing. Thus, exploit- nodes will need to complete short distance transmission
ing the tradeoffs among energy, scalability, and latency is only. Therefore, more energy is saved and the overall
crucial for prolong WSN’s lifetime which is a critical as- network lifetime can be extended.
pect of their applications. In [3], it is proposed to elect the CHs according to the
The clustering algorithm is a kind of key technique used energy left in each node. In [4], this clustering protocol is
to reduce energy consumption and to increase the scala- called LEACH-E. Based on LEACH-E, we propose an
bility and lifetime of the network. Many efforts have been Oriented Cluster Formation to enhance the performance
made to improve the energy efficiency and extend the of LEACH-E protocol; we call this clustering protocol
lifetime by designing energy efficient networking proto- (OCFLE). OCFLE is an oriented cluster formation based
cols [1],[2],[3]. A WSN contains hundreds or thousands of on the postion of the CHs in the sensing area and the
minimum distance between Non-Cluster-Head (NCH)
———————————————— and the BS using a suitable intermediate CH. Also, the
 Moulay Lahcen HASNAOUI and Abderrahim BENI HSSANE with the CHs are elected by a probability based on the ratio be-
Department of Mathematics and Computer Science, MATIC Laboratory
Chouaïb Doukkali University, El Jadida Morcco.
tween residual energy of each node and the average
 Abadellah EZZATI and Said BENALLA with the Department of Mathe- energy of the network. Moreover, we use a 2-level hie-
matics and Computer Sciences, Systems Analysis and Processing Informa- rarchy concept by selecting an adequate intermediate CH
tion Laboratory, FST Settat Morocco. for data transmission and and utilize an adapted formula
to estimate the network lifetime, thus avoiding the need
JOURNAL OF COMPUTING, VOLUME 2, ISSUE 11, NOVEMBER 2010, ISSN 2151-9617
HTTPS://SITES.GOOGLE.COM/SITE/JOURNALOFCOMPUTING/
WWW.JOURNALOFCOMPUTING.ORG 35

of assistance by routing protocol in LEACH-E. energy resources of the network consequently the overall
The use of an oriented clusters formation and 2-level network lifetime can be reduced. Fortehrmore this proto-
hierarchy concept for transmitting data to the BS, leve- col requires the assistance of routing protocol, which
rages the advantages of small transmit distances and should allow each node to know the total energy of net-
reduces the number of transmission data to the BS. As a work.
consequence, fewer cluster heads are required to transmit In order to avoid these problems and enhance the per-
far distances to the BS. This permits a better distribution formance of LEACH-E protocol we introduce the OCFLE
of the energy load through the sensors in the network and protocol, which is an oriented cluster formation for hete-
then increases the whole network lifetime by reducing the rogeneous network. In OCFLE, we use a 2-level hierarchy
total energy dissipation on the network. concept by selecting an intermediate suitable cluster head
The remainder of this paper is organized as follows. for data transmission to the BS and we utilize an adapted
Section 2 presents the related work. Section 3 describes formula to estimate the network lifetime, thus avoiding
the problem outline. Section 4 exhibits the details and the need of assistance by routing protocol in LEACH-E.
analyzes the proposal OCFLE protocol. Section 5 eva-
luates the performance of OCFLE protocol by simulations 3 PROBLEM OUTLINE
and compares it with LEACH, SEP and LEACH-E. Final- 3.1 Hetrogeneous WSN Network
ly, Section 6 gives a conclusion.
We assume that there are N sensor nodes, which are
uniformly dispersed within a M x M square region. The
2 RELATED WORK
nodes always have data to transmit to a BS, which is often
The WSN design often employs some approaches as far from the sensing area. The network is organized into a
energy-aware techniques, in-network processing, multi- clustering hierarchy, and the CHs execute fusion function
hop communication, and density control techniques to to reduce correlated data produced by the sensor nodes
extend the network lifetime. In most WSN applications within the clusters.
the power supply is limited, so preserving the consumed We assume that the nodes are stationary as supposed in
energy of the network is a challenge that must be consi- [3]. In the 2-level heterogeneous networks, there are two
dered when developing a routing protocol for WSNs. types of sensor nodes the advanced nodes and normal
A comprehensive survey of the routing protocols for nodes. We denote by the initial energy of the normal
WSNs can be found in [5]. In general, these protocols can nodes, and the fraction of the advanced nodes, which
be categorized into two classes according to the node’s
own a times more energy than the normal ones. Thus,
participating style: flat protocols and clustering protocols.
there are Nm advanced nodes equipped with initial ener-
Those in [6], [7], [8], [9] belong to the first class. The
gy of 1 , and 1 normal nodes equipped
second class can be also categorized into two subclasses:
with initial energy of . The total initial energy of the 2-
the clustering algorithms applied in homogeneous net-
level heterogeneous networks is given by:
works are called homogeneous schemes, where all nodes
have the same initial energy and the clustering algorithms
applied in heterogeneous networks are referred to as 1 1 1 (1)
heterogeneous clustering schemes, where all the nodes of
the sensor network are equipped with different amount 3.2 Radio Energy Dissipation Model
of energy. We use in this study a similar energy model as pro-
Many homogeneous clustering algorithms exist in lite- posed in [3]. According to the radio energy dissipation
rature such as LEACH [2], PEGASIS [10] and HEED [11]. model illustrated in (Fig. 1), and in order to achieve an
Low-Energy Adaptive Clustering Hierarchy (LEACH), acceptable Signal-to-Noise Ratio (SNR) in transmitting an
which is one of the most fundamental protocol frame- L-bit message over a distance d.
works in the literature, utilizes randomized rotation of
the (CHs) to uniformly distribute the energy budget
across the network. The sensor nodes are grouped into
several clusters and in each cluster, one of the sensor
nodes is selected to be CH. Each NCH will transmit its
data to its own CH which forwards the sensed data to the
BS. Both the communication between sensor nodes and
CH and that between CHs and the BS are direct, single-
hop transmission.
Many heterogeneous clustering algorithms exist in lite- Fig. 1. Radio Energy Dissipation Model.
rature such as LEACH-E[3], DEEC[4], SEP[12], M-LEACH
[13] , EECS[14], and LEACH-B[15]. The energy dissipated by the radio is given by:
The drawbacks of LEACH-E protocol are that it utilizes ,
direct transmission from CHs to the BS and some of , (2)
,
NCHs send their data to the further CH and afterwards
their data travels back over a long distance distance to Where is the energy dissipated per bit to run the
reach the BS. These kinds of transmissions waste the transmitter or the receiver circuit, and and
JOURNAL OF COMPUTING, VOLUME 2, ISSUE 11, NOVEMBER 2010, ISSN 2151-9617
HTTPS://SITES.GOOGLE.COM/SITE/JOURNALOFCOMPUTING/
WWW.JOURNALOFCOMPUTING.ORG 36

depend on the transmitter amplifier model used and ing an intermediate befitting CH for data transmission.
d is the distance between the sender and the receiver. We The CH with more energy is likelihood to be the interme-
have fixed the value of at 87.7 meters. diate CH used to transmit aggregate data to the BS, we
will call this suitable CH a MaxCH.
4 THE OCFLE PROTOCOL We consider a network with N nodes, uniformly distri-
buted within M×M square region and that the network
The OCFLE protocol is a hierarchical heterogeneous
topology remains unchanged over time and the BS loca-
WSN routing protocol which utilizes uses the same strat-
egy in Cluster Heads selection, and Schedule Creation tion is (x = 50, y = 175). In OCFLE, We get the probability
(TDMA) as LEACH-E but differs in Clusters formation threshold, which each node s uses to determine whether
and Data transmission. In OCFLE, we enhance the per- itself to become a cluster-head in each round, as follow:
formance of LEACH-E by introducing an oriented cluster
formation Fig. 2 based on the postion of the CH and the
best way to transmit data with the minimum distance
between the NCH and the BS using an adequate interme- (3)
diate CH. The NCHs form a cluster with the nearest CH 0
in front of him in the same direction to the BS in order to
minimize the target for transmitting data to the BS as a
Where G is the set of nodes that are eligible to be cluster
result the lost of the energy is minimized so the overall
heads at round . In each round r, when node finds it is
network lifetime can be extended.
eligible to be a CH, it will choose a random number be-
tween 0 and 1. If the number is less than threshold ,
the node becomes a cluster head during the current
round. Also, is defined in [3] as follow:

,1 (4)

Where is the current energy of node , k is the


desired number of cluster, and is an estimation
of the remaining energy of the network per round :

1 (5)

Where is the average initial energy of the 2-level


heterogeneous networks and R denotes the total rounds of
the network lifetime. It means that every node consumes
the same amount of energy in each round, which is also
the target that energy-efficient algorithms should try to
achieve. The value of R is:

(6)

Where denote the total energy dissipated in the


network during a round r is given by:

2 1 _ _
_ _ (7)

Where k is the number of clusters, EDA is the data ag-


Fig. 2. Dynamic oriented cluster formation by OCFLE
gregation cost expended in the CHs, _ is the average
Also, the CHs are elected by a probability based on the distance between the CH and the base station, _ is
ratio between residual energy of each node and the aver- the average distance between the cluster members and
age energy of the network and a particular algorithm is the CH, finally _ _ is the average distance between
used to estimate the network lifetime, thus avoiding the the CHs and the CH with more energy (MaxCH).
need of assistance by routing protocol in LEACH-E. The OCFLE algorithm can be summarized as follow in
Moreover, we use a 2-level hierarchy concept by select- Fig. 3:
JOURNAL OF COMPUTING, VOLUME 2, ISSUE 11, NOVEMBER 2010, ISSN 2151-9617
HTTPS://SITES.GOOGLE.COM/SITE/JOURNALOFCOMPUTING/
WWW.JOURNALOFCOMPUTING.ORG 37

Each NCH sends its data during their allocated trans-


mission time (TDMA) to the respective CH. The CH must
For each node s keep its receiver on in order to receive all the data from
If (node s is NCH) then the nodes in the cluster. When all the data is received, the
Form a cluster with a suitable CH CH node performs signal processing functions to com-
(see OCFLE detail in section 4) press the data into a single signal. When this phase is
Send data to CH completed, each CH can send the aggregated data to the
Else MaxCH. After that, each non cluster head can turn off the
If(node s is not MaxCH) then sleep mode.
Data aggregation(nodes) Furthermore, to improve the performance of communica-
If(d_to_BS>d_to_MaxCH) then tion of LEACH-E and save more energy, we are used in
Send to MaxCH OCFLE a 2-level hierarchy concept by selecting an interme-
Else diate maximum residual energy cluster head for data trans-
Send to BS mission. Based on the residual energy and the information
End if coordinates included on the message broadcasted, the
Else CHs elected can select the Maximum energy CH which is
First data aggregation (nodes) called MaxCH. Consequently, the CH with more energy
Second data aggregation (CHs) will be the MaxCH in this round. This last node collects
Send to BS all data coming from all CHs, compress it into a single
End if signal and send it directly to the base station. We have
End if chosen the MaxCH as intermediate hierarchical level,
End for because the MaxCH granted the transmission for long
time. In fact, they have not waste energy in long trans-
mission to the BS.
Fig. 3. Summary of OCFLE algorithm

5 SIMULATION RESULTS
Assuming that the nodes are uniformly distributed, by In this section, we evaluate the performance of OCFLE
using the result in [3, 16] we can get the equations as fol- protocol using MATLAB. We consider a wireless sensor
low: network with N = 100 nodes randomly distributed in a
100m 100m field. We assume that the BS is far away
_ (8) from the sensing region. It was placed at location (x=50,

y=175). To compare the performance of OCFLE with oth-
er protocols. The radio parameters used in our simula-
_ (9) tions are shown in TABLE 1. We assume that all nodes
know their location coordinates and. The protocols com-
_ √2 (10) pared with OCFLE include LEACH, SEP and LEACH-E.
We define stable time as time until the first node dies, and
√ unstable time the time from the fist node dies until the
(11) last node dies. We will consider following scenarios and

examine several performance measures.
Substituting equations (11, 10, 9, 8, 7, 6, 5, and 4) into
equation (3), we obtain the probability threshold , TABLE 1
for our OCFLE, which is used to decide if node should Radio characteristics used in our simulations
be a cluster-head in current round and broadcasts an Parameter Value
advertisement message to the rest of the nodes. For this E 5 nJ/bit
cluster-head’s advertisement phase, the CHs use a CSMA
10 pJ/bit/m
MAC protocol, and all CHs transmit their advertisement
using the same transmit energy. The NCHs nodes must 0.0013 pJ/bit/m
keep their receivers on during this phase of set-up to hear E 0.5 J
the advertisements of all the CHs. The sent messages
EDA 5 nJ/bit/message
content in addition the Id nodes, the information coordi-
nates. d 87.7 m
After this phase is complete, In OCFLE each NCH node Message size 4000 bits
decides the cluster to which it will belong for this round. p 0.1
This decision is based on the oriented cluster formation
based on the postion of the cluster-head and the best way to
transmit data with the minimum distance between the NCH
and the BS using an adequate intermediate cluster-head.
JOURNAL OF COMPUTING, VOLUME 2, ISSUE 11, NOVEMBER 2010, ISSN 2151-9617
HTTPS://SITES.GOOGLE.COM/SITE/JOURNALOFCOMPUTING/
WWW.JOURNALOFCOMPUTING.ORG 38

a=5 and m=0.1 4


x 10 a=5 and m=0.1
100 3.5

LEACH
90
LEACH-E 3 LEACH
SEP

Number of messages received at the BS


80 LEACH-E
OCFLE
SEP
70 2.5 OCFLE
Number of nodes alive

60
2

50
1.5
40

30 1

20
0.5
10

0 0
0 500 1000 1500 2000 2500 3000 0 500 1000 1500 2000 2500 3000
Time(Round) Time(Round)

Fig. 4. Performance of LEACH, SEP, LEACH-E and OCFLE under 2- Fig.6. Performance of LEACH, SEP, LEACH-E, and OCFLE under 2-
level heterogeneous networks a=5, m =0.1: Number of nodes alive level heterogeneous networks a =5, m =0.1: Number of message
over time. received in base station over time.

First, we run simulation to study the performance of Fig. 6 shows the results of the case with m = 0.2 and a =
LEACH, SEP, LEACH-E and OCFLE under two kinds of 3. We conclude that the number of messages delivered by
2-level heterogeneous networks Fig. 4 and Fig. 5. OCFLE to the BS over time are greater than the others
Fig. 4 shows the results of the case with m = 0.1 and a = ones, this means that OCFLE is a more efficient protocol.
5, and Fig. 5 shows the results of the case with m = 0.2 Third, we run simulation for our proposed protocol
and a = 3. We observe clearly that the stable time of OC- OCFLE to compute the number of received messages at
FLE is prolonged compared to that LEACH, SEP and the BS over energy dissipation, in the case with a =5, m
LEACH-E. The introduction of the orientation in cluster =0.1 and compare the results of OCFLE prototcol with
formation and 2-level concept extend the lifetime of the others LEACH, SEP and LEACH-E protocols.
whole network and performs better than LEACH by
Fig. 7 shows the results of the case with m = 0.1 and a =
about 109%; than SEP by about 62% and than LEACH-E
5. It’s obvious that the number of messages delivered by
by about 35% in term of first node dies.
OCFLE over energy dissipation to the BS are greater than
Second, we run simulation for our proposed protocol
the others ones, this means that OCFLE is a more efficient
OCFLE to compute the number of received messages by
of energy consumption protocol. In other words, it is an
the BS over time and compare the results of OCLFE pro-
energy-aware adaptive clustering protocol.
tocol with others, LEACH, LEACH-E and SEP protocols.
a=3 and m=0.2 x 10
4
a=5 and m=0.1
100 3
LEACH
90 LEACH
LEACH-E
Number of messages received at the BS

2.5 LEACH-E
80 SEP
OCFLE SEP
70 OCFLE
Number of nodes alive

60
1.5
50

40
1
30

20 0.5

10
0
0 0 10 20 30 40 50 60 70 80
0 500 1000 1500 2000 2500 3000 Energy dissipation
Time(Round)
Fig.5. Performance of LEACH, LEACH-E, SEP and OCFLE under 2- Fig.7. Performance of LEACH, SEP, LEACH-E and OCFLE under
level heterogeneous networks a=3, m =0.2: Number of nodes alive 2-level heterogeneous networks a =5, m =0.1: Number of mes-
over time sages received at BS over Energy dissipation (Joules).
JOURNAL OF COMPUTING, VOLUME 2, ISSUE 11, NOVEMBER 2010, ISSN 2151-9617
HTTPS://SITES.GOOGLE.COM/SITE/JOURNALOFCOMPUTING/
WWW.JOURNALOFCOMPUTING.ORG 39

6 CONCLUSION Sensor Networks (IEEE IWSEEASN2005), Phoenix, Arizona,


April 7-9, 2005.
In this paper, we have presented the OCFLE protocol, [15] A. Depedri, A. Zanella, R. Verdone, An energy efficient proto-
which si an extension of LEACH-E Protocol. The main col for wireless sensor networks, in: Autonomous Intelligent
Networks and Systems (AINS 2003), Menlo Park, CA, June 30-
features in our proposal extension is the use of an
July 1, 2003.
oriented cluster formation based on the postion of the [16] S. Bandyopadhyay, E.J. Coyle, An energy efficient hierarchical
CHs in the sensing area and the minimum distance be- clustering algorithm for wireless sensor networks, in: Proceed-
tween non-cluster-head and the base station using a ing of INFOCOM 2003, April 2003.
suitable intermediate cluster-head and the use of a 2-
level hierarchy concept for data transmission. Accord-
ing to the simulation results, we can obviously con-
clude that OCFLE reduce the energy dissipation, ex-
tend the lifetime of the whole network and performs
better than LEACH by about 109%, than SEP by about
62% and than LEACH-E by about 35% in term of first
node dies. Finally, we can conclude that OCFLE is an
energy-aware adaptive clustering protocol for hetero-
geneous WSNs.

REFERENCES
[1] F. Akyildiz, W. Su, Y. Sankarasubramaniam, E. Cayirici, A
survey on sensor networks, IEEE communications magazine
40 (8) (2002)102–114.
[2] W.R. Heinzelman, A.P. Chandrakasan, H. Balakrishnan, Ener-
gyefficient communication protocol for wireless microsensor
networks, in: Proceedings of the 33rd Hawaii International
Conference on System Sciences (HICSS-33), January 2000.
[3] W.R. Heinzelman, A.P. Chandrakasan, H. Balakrishnan, An
application-specific protocol architecture for wireless micro-
sensor networks, IEEE Transactions on Wireless Communica-
tions1 (4) (2002) 660-670.
[4] L. Qing, Q. Zhu, M. Wang, ”Design of a distributed energy-
efficient clustering algorithm for heterogeneous wireless sen-
sor networks”. ELSEVIER, Computer Communications 29, pp
22302237, 2006.
[5] Jiang, Q. and Manivannan, D. (2004) ’Routing protocols for
sensor networks’, IEEE Consumer Communications and Net-
working Conference, January, pp.93-98.
[6] Hedetniemi, S. and Liestman, A. (1988) ’A survey of gossiping
and broadcasting in communication networks’, Networks, Vol.
18, pp.319-349.
[7] Heinzelman, W.R., Kulik, J. and Balakrishnan, H. (1999)
’Adaptive protocols for information dissemination in wireless
sensor networks’, ACM MobiCom, Seattle, pp.174-185.
[8] Sohrabi, K., Gao, J., Ailawadhi, V. and Pottie, D.J. (2002) ’Pro-
tocols for self-organization of a wireless sensor network’, IEEE
Personal Communications, October, pp.16-27.
[9] Intanagonwiwat, C., Govindan, R. and Estrin, D. (2000) ’Di-
rected diffusion: a scalable and robust communication para-
digm for sensor networks’, ACM MobiCom, pp.56-67.
[10] S. Lindsey, C.S. Raghavenda, PEGASIS: power efficient gather-
ing in sensor information systems, in: Proceeding of the IEEE
Aerospace Conference, Big Sky, Montana, March 2002.
[11] O. Younis, S. Fahmy, HEED: A hybrid, energy-efficient, distri-
buted clustering approach for ad hoc sensor networks, IEEE
Transactions on Mobile Computing 3 (4) (2004) 660-669.
[12] G. Smaragdakis, I. Matta, A. Bestavros, SEP: A Stable Election
Protocol for clustered heterogeneous wireless sensor net-
works, in: Second International Workshop on Sensor and Ac-
tor Network Protocols and Applications (SANPA 2004), 2004.
[13] V. Mhatre, C. Rosenberg, Design guidelines for wireless sensor
networks: communication, clustering and aggregation, Ad
Hoc Network Journal 2 (1) (2004) 45-63.
[14] M. Ye, C. Li, G. Chen, J. Wu, EECS: an energy efficient cluster
scheme in wireless sensor networks, in: IEEE International
Workshop on Strategies for Energy Efficiency in Ad Hoc and