Energy Efficient Routing Protocol Using

Local Mobile Agent for Large Scale WSNs

Ali Seirafi, Mohammad Hadi Moazzam, Farshad Kabiri Samani
Abstract The energy consumption is always an important topic for research on routing protocols of wireless sensor networks.
Many algorithms are introduced in this field and each of them can propose solutions in order to improve the energy consumption
issue. Energy Efficient Cluster Head Selection (EECHS) is one of the most well known protocols in WSN, but its performance
decreases with increase of the network scale. To solve this problem, we introduce an improved routing protocol uses a new
multi hop transmission method to balance energy consumption among the whole network. This approach reduces the time
distance between failing of the first and the last nodes at minimum and keeps network coverage until the last moments. The
simulation results performed in MATLAB, show that the proposed algorithm is more efficient and increases the network lifetime.
Index Terms Local Mobile Agent, Clustering, Routing Protocol, Network Lifetime.

1 INTRODUCTION

n the WSN, the network composed of a huge number of
small self-contained devices with computational, sens-
ing and wireless communication capabilities covering
the monitoring area.
Energy consumption is one of the major challenges and
many studies have attempted to tackle this problem.
Nodes are powered by a small battery, which may run
out in a short time [1]. Hence, an efficient routing protocol
can increase the network lifetime.
According to the final topology, the routing protocols of
WSN are categorized into two types; flat routing and hi-
erarchical routing protocols. As flat routing protocols
need to maintain large routing table information, they are
unsuitable for large-scale sensor networks. Hierarchical
can solve this problem to a certain extent.
Low Energy Adaptive Clustering Hierarchy (LEACH)
[2] is one of the first major algorithms based on hierar-
chical method. LEACH is a completely distributed rout-
ing algorithm which does not require any global infor-
mation for cluster formation and routing. The clustering
strategy is mainly depends on the randomly selected clus-
ter-heads and does not consider the residual energy of
nodes. [3]
Many clustering algorithms are based on LEACH ar-
chitecture and improve the network lifetime. Energy Effi-
cient Cluster Head Selection (EECHS) [4] is one of them.
It considers the factor of residual energy, distance from
the Base Station and the number of consecutive rounds in
which a node has not been a cluster head.
In this study, EECHS algorithm is improved by the
addition of two factors; distance between cluster heads
and minimum energy that each node require to be a clus-
ter head.
Simulation result shows that proposed protocol has
better performance than LEACH and EECHS.
2 RELATED WORK
2.1 LEACH
LEACH is a time division multiple access based on
distributed clustering approach in homogeneous WSNs.
[5] The algorithm is self-organized, employs a single hop
approach and the data combination technique can de-
crease the rate of data transfer. This algorithm is divided
into two stages; setup stage and steady stage.
During the first stage, all nodes are organized into
some clusters through communicating with short mes-
sages and one node is selected as a cluster head according
to the cluster head selection algorithm. At the beginning
of this phase every node in the network must decide
whether it will become a cluster head or not. This deci-
sion is made by the node n choosing a random number
between 0 and 1. If the number is less than a threshold
T(n), the node becomes a cluster head for the current
round. [6] The threshold value T(n) is calculated by (1),
where p is the predetermined percentage of cluster heads,
r is the current round number, and G is the set of nodes
that have not been cluster head in the last 1/p rounds.


G n if n T
G n if
p
r p
p
n T
e =
e
|
|
.
|

\
| 1

=
0 ) (
mod 1
) (
(1)



- Ali Seirafi is with the department of Electrical and Computer Engineering,
Lenjan Branch, Islamic Azad University, Isfahan, Iran.
- Mohammad Hadi Moazzam is with the Engineering Faculty, Payam Noor
University, Tehran, Iran
- Farshad Kabiri Samani is with the department of Electrical and Computer
Engineering, Lenjan Branch, Islamic Azad University, Isfahan, Iran.





I
JOURNAL OF COMPUTING, VOLUME 5, ISSUE 2, FEBRUARY 2013, ISSN (Online) 2151-9617
https://sites.google.com/site/journalofcomputing
WWW.JOURNALOFCOMPUTING.ORG 1


After the setup stage, cluster heads simultaneously
broadcast TDMA scheduling messages using spreading
codes to start the steady stage, which involves data signal
sending from non-CH nodes to the CH nodes by TDMA
scheduling, aggregation and routing to the Base. LEACH
uses a single hop routing algorithm and each CHs send
the aggregated data to the Base Station in a single hop
manner. [7]
LEACH is based on rounds and the system repeats the
clustering and transmission for each round. LEACH has
its own defects, but improved algorithms will definitely
solve these problems. Some of these problems are as fol-
lows:


1. Cluster heads are predefined in LEACH protocol
typically taken as 5% or 10% of total deployed
nodes. If sensor nodes in the network are not uni-
formly distributed, specified number of clusters
might not be sufficient to cover the entire area.
2. Cluster heads send aggregated data to the Base
Station in a single hop manner, so consume a lot
of energy.
3. Probability function T(n) does not take residual
energy of nodes into account while selecting
cluster heads, so if a less energy node becomes
CH, it will fail and subsequently the cluster net-
work becomes dysfunctional.
4. It consist of rounds while in each round, all sen-
sor nodes take part in reconstructing new clus-
ters and this action consumes a lot of energy. [4]

2.2 EECHS
EECHS [2] improves the LEACH protocol by improving
the election strategy of the cluster heads, considering re-
sidual energy, distance from the Base Station, number of
consecutive rounds in which a node has not been a cluster
head.
In EECHS, stochastic cluster head selection algorithm
is extended by adjusting the threshold T(n) denoted in (1),
considering the residual energy of the nodes, distance
between the nodes and the Base Station and the number
of consecutive rounds in which a node has not been a
cluster head as follows:


)] ( )) ( 1 ( ) ( [
mod 1
) ( r div s i D i E
p
r p
p
n T + +
|
|
.
|

\
| 1

=

(2)


Here the residual energy factor is:
E(i) = (E
residual
(i) /E
initial
) Where E
residual
is the remaining
amount of energy and E
initial
is the initial energy of a node
before transmission and Variable i indicates the serial
number of nodes.
The Distance factor D(i)=d
iB
/d
farthest
Where debt is the dis-
tance from node i to Base as shown in (3):

2 2
) ( ) (
BS i BS i iB
Y Y X X d + =
(3)


And d
farthest
is the distance of the farthest node from the
Base Station. Here X
BS
and Y
BS
are the location of the Base
Station.
In the third parameter (s div r), s is the number of consec-
utive rounds in which a node has not been a cluster head
and r is the number of the current round.
Instead of considering these three important parame-
ters, EECHS does not consider the factor that whether the
nodes remaining energy are sufficient enough to send the
aggregated data to the Base Station or not. Besides, in
EECHS, there is no limitation for location of cluster heads.
3 PROPOSED APPROACH
In 2.1 and 2.2 sections, LEACH and EECHS Algorithms
are investigated along with problems existed in each ap-
proach. A method will be presented in this part doesnt
have any drawbacks of the two previous methods. To
improve the insufficiency of LEACH and EECHS, we in-
troduce an improved protocol in this paper. This protocol
like LEACH and EECHS divided into two stages include
setup stage and steady stage.
Before describing the algorithm, network model
should be investigated.


3.1 Network Model
For our proposed approach, we adopt a few reasonable
assumptions as follows: [2, 3]

1. Sensor nodes are immobile.
2. Nodes are not able to access Base directly.
3. The Base Station is located at a far distance
from the sensor nodes and is immobile.
4. All nodes are able to control their transmit
power to change the communication range.
5. Access of nodes to the Base is performed by a
special node called Local Mobile Agent (LMA)
located in the center of the network, in a circle
with L radius.
6. All sensor nodes are homogeneous and have
the same hardware components except LMA
which includes specific components.
7. The LMA is the only mobile node in the net-
work.
8. Regarding to the access of Local Mobile Agent
to the wireless charging equipment, the ener-
gy of this node will be considered unlimited.
9. Radio and energy model is the same as radio
model in [2], which is called first radio model.

3.2 Setup Stage
1. CH selection
In addition to the factors that EECHS algorithm is uti-
lized in the cluster head selection process, two other fac-
tors that are ignored in this algorithm will be added to the
JOURNAL OF COMPUTING, VOLUME 5, ISSUE 2, FEBRUARY 2013, ISSN (Online) 2151-9617
https://sites.google.com/site/journalofcomputing
WWW.JOURNALOFCOMPUTING.ORG 2


previous set, although they have significant impact on
energy consumption.
The first factor is the minimum energy required. If the
nodes remaining energy is not sufficient enough to send
the aggregated data to the Base Station, then a flag varia-
ble called MER (Minimum Energy Required) will be tak-
en as 0. And if the remaining energy is sufficient, then it
will be taken as 1.
Considering the additional factor MER with all three fac-
tors as in [2] the modified threshold T(n) becomes:
)] ( ) ( ) ( 1 ( ) ( [
mod 1
) ( i MER r div s i D i E
p
r p
p
n T + +
|
|
.
|

\
| 1

=
(4)

This factor guarantees that if it doesnt have enough en-
ergy for operation as a cluster head, but it can be operated
on the network as an usual node for a while doesnt be-
come a cluster head.
Another factor is the distance between cluster heads in
the network. In order to prevent the congestion of cluster
heads in a particular area, the distance between the CHs
cannot be less than d (d is an ideal radius, and its calcula-
tion is given by (5) ), so it can ensure that the CHs will not
be assigned to a centralized location. [8]

=
k
y x
d

(5)
Where x is the length of the sensor network , y is the
width of the sensor network, the value of is 3.14 , k is
the number of CHs in the sensor network, the value of
is between (1, 2), here we select 1.1.













Fig. 1. The distribution of nodes.

In the process of CH selection, CHs will broadcast the
advertisement messages, when other nodes receive the
messages, they will calculate the distance from the signal
strength of the messages, and record the minimum value.
If the value is greater than d, this node can turn into CH,
otherwise, this node will not. As shown in Fig.1, two
nodes 21 and 47 have been acting as CH in the network.
All nodes in the two circular areas have lost their eligibil-
ity to turn into CH, as nodes 7, 15, 41, 68, 72, 94.
The process of CHs selection is shown in Fig.2

































Fig. 2. CH selection process.


2. Cluster Formation
In the process of CH selection, CHs will inform all nodes
that act as non-CH in this round by sending CH broadcast
message. In order to avoid the collision of individual
messages, the CSMA mechanism is used to send broad-
cast messages. The broadcast message is a very small
message including the nodes ID and a header that distin-
guishes this message. Each non-CH node determines the
distance to CH node based on the received signal strength
of the broadcast message. In order to reduce the energy
consumption that communicates with CH, non-CH will
choose the nearest CH to join. After each node has decid-
ed to which cluster it belongs to, they must inform the CH
node that it will be a member of the cluster by sending a
join request message to the chosen CH using CSMA
mechanism. CH node confirms the members of this clus-
ter according to the join request messages that received.
When all non-CH nodes are finished to join the cluster,
CH will produce a slot table based on the members of the
cluster using TDMA mechanism. At the same time, CH
will get a cluster channel based on the sequence of be-
coming CH. In addition to these cluster channels, the im-
proved protocol also assigned a common channel for
transmitting the control information in cluster setup stage
Y
Y
Y
Y
N
N
N

Node generates a random
number between zero and one.


This node serve as
CH and send
broadcast message


Measurement of the distance of node
from LMA and determination
of the farthest node to LMA

LMA placement in
the specified location


Has
The node been a CH
during this loop?


Number
of received broadcast
message < k


The beginning
of CH selection


Is this
number less than threshold
of T(n)?


Is the
distance between cluster heads
more than d?

This node can
not be a CH
CH selection
is completed
N
e
x
t

n
o
d
e

N
e
x
t

n
o
d
e

Cluster head
21
47
Member node

d
d
21
47
15
7
68
72
94
41
JOURNAL OF COMPUTING, VOLUME 5, ISSUE 2, FEBRUARY 2013, ISSN (Online) 2151-9617
https://sites.google.com/site/journalofcomputing
WWW.JOURNALOFCOMPUTING.ORG 3


and the data between the CH and Local Mobile Agent.
CH will send the message containing the slot table and
cluster channel to each member of this cluster using
CSMA mechanism. The member node that received the
message containing slot table and cluster channel will
determine to its transmitting data slot and transmitting
data channel based on the content of the message. Be-
cause of using TDMA mechanism, each cluster member
sends data in different time slot to avoid the data conflict
within the cluster. The data transfer between the clusters
using different channels, so they will not influence each
other. At this point, the setup stage is completed and the
sensor network enters the steady stage.


3.3 The steady stage
After setup stage, the sensor network will enter the
steady stage. In the steady stage, each non-CH sends data
in the assigned time slot, using cluster channel. On the
other time, non-CH node will go to sleep in order to re-
duce the energy consumption. The CH must be awake to
receive all the data from the member nodes in the cluster.
Before CH receives the data that cluster members send,
the CH must judge whether the working channel is clus-
ter channel or not. If the working channel is cluster chan-
nel, CH will directly receive the data, otherwise, the CH
will receive the data after switching the working channel
to cluster channel. Once the CH receives all the data, it
performs data aggregation. After data aggregation and
fusion which are performed by cluster heads, it is time for
the most important step i.e. sending data outside the clus-
ter.
In LEACH data transfer phase, cluster heads send the
packages to the sink node directly as shown in Fig. 3.
This aim is easy to be realized, but it is not suitable for
large networks. In this case, we proposed an improved
protocol and introduce how it works.










Fig. 3. Transmit style of LEACH.
When there are too many sensors in a large area, the
number of cluster heads is also tremendous. So, we can
make these cluster-heads cluster like previous. Then the
new cluster-heads will be chosen, we can call them super
cluster-heads. Now the super cluster-heads are used to
send data packages to the sink node. This can be de-
scribed as follows in Fig. 4.
c2 send packages to sink node with the help of c1.














Fig. 4. Transmit style of the proposed method.

But how do we choose the super cluster-heads? As
shown in the formula, we can draw a conclusion that
when is an obtuse angle (as shown in Fig. 5), it shows
that a
2
+
b
2
<
c
2
. In this case, multi layer clustering is nec-
essary and effective but when it is not obtuse angle direct
transmission is Affordable. Pythagorean Theorem proves
this statement.








Fig. 5. Distribution model.

In the process of clustering, each cluster-head stores
other cluster heads locations and they also know the lo-
cation of the Local Mobile Agent. By computing, they can
decide which one can be their super cluster-head. In real
network, the cost of transmitting signal between two
nodes is smaller than that of sending directly. According
to the conclusion we draw before, the multi-layer cluster-
ing saves the energy of the cluster-heads. After sending
data via cluster head and their collection by LMA, this
element sends data to Base directly according to unlim-
ited energy and high power radio.

The Fig.6 indicates the operation of suggested algo-
rithm. Regarding to the proposed topology, the source
node transmits its data to its own cluster head and clus-
ter head transmit data to the Local Mobile Agent using
above routing protocol. Agent with unlimited energy and
high transmission power, sends data to the Base Station
directly.

In this figure, the Local Mobile Agent is located in a
circle with L radius in the center of the network. Place-
ment of Local Mobile Agent in fixed location causes of
neighbors nodes energy becomes discharged quickly.
Hence, this agent is able to move and change its position
once every a few rounds, but this movement depends on
special conditions. Firstly, the range of movement of this
agent is limited and is confined to the inside of the ex-
plained circle. (That is the reason for nomination of Local
Mobile Agent). The primary location of the agent is in the
center of the network and the other locations are assigned
C1
C2
Sink node
a
b
c
A
S
C

B
C1
C2
Sink node

JOURNAL OF COMPUTING, VOLUME 5, ISSUE 2, FEBRUARY 2013, ISSN (Online) 2151-9617
https://sites.google.com/site/journalofcomputing
WWW.JOURNALOFCOMPUTING.ORG 4


by the average of neighbor nodes energy. In this way, the
agent should be placed where the average of neighbor
nodes energy of that place is more than the previous lo-
cation. If there are several locations with mentioned fea-
tures, one of them will be chosen randomly.




























Fig. 6. Distribution and transmit model.

Routing from a specified node to the Local Mobile
Agent includes two stages;
In the first stage, a node called forwarding node is se-
lected that is in sender domain and has the least distance
from the Local Mobile Agent. In the second stage of rout-
ing, if there is a neighbor between the sender node and
the current forwarding node that its direction make ob-
tuse angle with the two previous nodes, this node is nom-
inated forwarding node. If there is no neighbor node with
these characteristics, transmission process occurs directly.
This way goes on until the reaching of data to the Local
Mobile Agent.

4 SIMULATION RESULTS AND ANALYSIS

The simulation environment is composed of 200 sensor
nodes and sensor nodes are distributed in a region of
100m100m randomly. Table 1 lists the simulation
parameters.


















Fig. 7 indicates the number of alive nodes in each three
algorithms. As you can see in the following figure, the
proposed algorithm grantees the network coverage over
time more than the two previous methods. One of the
considerable results in this paper is reduction of time dis-
tance between failing of the first and the last nodes at
minimum. This way keeps network coverage until the last
moments completely.


CONCLUSION
Due to the scarce energy resources of sensors, energy
efficiency is one of the main challenges in the design of
protocols for WSNs. Our aim is to specify shortcomings
related to classical routing protocols and discuss some
updated algorithms which define new methods for pro-
longing lifetime. In this paper we examined LEACH and
EECHS protocols, then a new CH selection and multi hop
routing mechanism is proposed. We compared our algo-
rithm with two other protocols. The simulation results
show that the proposed algorithm can effectively reduce
network energy consumption and prolong the network
lifetime. In this study, the minimum distance between
cluster heads is fixed. This factor can be changed effec-
tively by changing the network conditions. The findings
of this study will be utilized in future research.

Fig. 7. The network lifetime.
TABLE 1
SIMULATION PARAMETERS
Parameter Value
No. of nodes 200
Network size 100 m100 m
Cluster Head probability 0.1
Initial energy of node 0.2 Joule
Node Distribution Randomly distributed
Data Packet 4000 bits
E elec 50 nJ/bit
mp 0.0013 pJ/bit/m4
fs 10pJ/bit/m2
d Farthest 84 m

The corrected path by proposed method
L
CH
Shortest path
LMA
JOURNAL OF COMPUTING, VOLUME 5, ISSUE 2, FEBRUARY 2013, ISSN (Online) 2151-9617
https://sites.google.com/site/journalofcomputing
WWW.JOURNALOFCOMPUTING.ORG 5


REFERENCES
[1] QU.Wei-Qing, Cluster head Selection Approach Based on
Energy and Distance, IEEE International Conference on
Computer Science and Network Technology, 2011.
[2] W.B. Heinzelman A.P. Chandrakasan, and H.Balakrish,
An Application-Specific Protocol Architecture for Wire-
less Micro Sensor Networks , IEEE Trans.Wireless Com-
mun, vol. 1,no.4 , 2002.
[3] M.Tong and M.Tang, LEACH-B:An Improved LEACH
Protocol for Wireless Sensor Network, 6th IEEE Interna-
tional Conference on Wireless Communications Network-
ing and Mobile Computing, 2010.
[4] Anindita Ray and Debashis De, Energy Efficient Cluster-
ing Hierarchy Protocol for Wireless Sensor Network,
IEEE International Conference on Communication and In-
dustrial Application, 2011.
[5] A.Azim and MM.Islam, Deterministic Cluster Head Se-
lection for Wireless Sensor Networks, 25th IEEE Canadi-
an Conference on Electrical and Computer Engineering
(CCECE), 2012.
[6] M.Arshad and M.Alsalem and F.Siddqui and N.Kamel,
Efficient Cluster Selection Scheme in Mobile Data Collec-
tor Based Routing Protocol, 4
th
IEEE International Con-
ference on Intelligent and Advanced Systems, 2012.
[7] S.H. Kang and T.Nguyen, Distance Based Thresholds for
Cluster Head Selection in Wireless Sensor Networks,
IEEE COMMUNICATIONS LETTERS, VOL. 16, NO. 9,
2012.
[8] Wang Yao, Liu Quanli
*
, Gao Guangen, An Improved
LEACH Protocol with Determined Number and Fair Dis-
tribution of Cluster Heads , IEEE International Confer-
ence on System Science and Engineering, 2012.




Ali Seirafi received his B.S. degree in computer engineering
(Hardware) from University of Najaf Abad, Iran in 2007 and M.S. in
computer engineering (Software) from PNU, Iran in 2013. He is cur-
rently working as a lecturer and researcher in Electrical and Com-
puter Engineering department of IAU, Lenjan Branch, Iran. His re-
search interests mainly focus on wireless communication and net-
work systems.

Mohammad Hadi Moazzam has a PhD in Systems and Con-
trol Engineering. At present, he is an Assistant Professor at the Fac-
ulty of Engineering of Payam Noor University, Tehran, Iran.

Farshad Kabiri Samani received his B.S. and M.S. degree in
computer engineering (Hardware) from university of Najaf Abad, Iran
and university of Arak, Iran in 2007 and 2010 respectively. He is
currently working as a lecturer and researcher in Electrical and
Computer Engineering department of IAU, Lenjan Branch, Iran. His
research interests mainly focus on computer arithmetic algorithms
and circuits, microprocessor architecture and VLSI hardware design-
ing.

JOURNAL OF COMPUTING, VOLUME 5, ISSUE 2, FEBRUARY 2013, ISSN (Online) 2151-9617
https://sites.google.com/site/journalofcomputing
WWW.JOURNALOFCOMPUTING.ORG 6