Progress In Electromagnetics Research B, Vol.

12, 259295, 2009

AN ESTIMATION OF SENSOR ENERGY CONSUMP-
TION
M. N. Halgamuge, M. Zukerman, and K. Ramamohanarao
ARC Special Research Center
for Ultra-Broadband Information Networks
Department of Electrical and Electronic Engineering
The University of Melbourne
Melbourne, VIC 3010, Australia
H. L. Vu
Center for Advanced Internet Architectures
Faculty of I.C.T.
Swinburne University of Technology
PO Box 218, Hawthorn, VIC 3122, Australia
AbstractA comprehensive energy model for wireless sensor
networks is provided by considering seven key energy consumption
sources some of which are ignored by currently available models. We
demonstrate the importance of using such a comprehensive model by
comparing it to other existing energy models in terms of the lifetime
of a sensor node. We use our model to evaluate energy consumption
and node lifetime for a sensor network with xed conguration and we
validate this evaluation by simulation. We show that existing energy
models over-estimate life expectancy of a sensor node by 3058% and
also yield an optimised number of clusters which is too large. We
further make the following two observations: 1) the optimal number of
clusters increases with the increase of free space fading energy, 2) for
sensor networks with 100 sensors over an area of 10
4
10
5
[m
2
], nding
the optimal number of clusters becomes less important when free space
fading energy is very low (less than 1670 pJ/bit/m
2
), while for larger
networks, on the other hand, cluster optimization is still important
even if free space fading energy is low. Guidelines for ecient and
reliable sensor network design as well as extension to a sensor network
with rotating cluster heads are provided.
Corresponding author: M. N. Halgamuge (malka.nisha@unimelb.edu.au).
260 Halgamuge et al.
1. INTRODUCTION
Sensor networks play a major role in many aspects of society including
home automation, consumer electronics, military application [1, 2],
agriculture, environmental monitoring, health monitoring [3] and
geophysical measurement [4]. Usually sensor devices are small
and inexpensive, so they can be produced and deployed in large
numbers. Their resources of energy, memory, computational speed
and bandwidth are severely constrained [5]. Therefore, it is important
to design sensor networks aiming to maximize their life expectancy.
Dierent aspects of sensor networks such as data aggregation or
fusion [6, 7], packet size optimization [8], target localization [9],
design challenges [10], network protocols [1113] are discussed in
the literature with respect to crucial energy limitations and network
lifetime maximization [1418, 20, 21].
Accurate prediction of sensor network lifetime requires an accurate
energy consumption model. There have been various attempts to
model sensor node energy consumption. We list below several such
models and point out certain sources of energy consumption that were
not considered in those models.
(i) Heinzelman et al. [11] proposed a model that considers micro-
controller processing and radio transmission and receiving only.
This model does not consider other important sources of energy
consumption, such as transient energy, sensor sensing, sensor
logging and actuation.
(ii) The model proposed by Millie and Vaidya [22] does not
consider energy consumption of sensor sensing, sensor logging and
actuation.
(iii) The Zhu and Papavassilious model [15] does not consider energy
consumption of transient energy, sensor logging and actuation.
In this paper we provide a comprehensive energy model including
certain sources of energy consumption that are not included in previous
sensor energy models, i.e., transmit energy, sensor sensing, sensor
logging and actuation. This paper also provides guidelines for ecient
and reliable sensor network design. In Table 1, we compare these
energy models with the new model proposed in this paper.
The paper is organized as follows. In Section 2, we describe our
single node energy model and relevant energy consumption sources. In
Section 3, we extend our model to a network. We apply our results
to a sensor network with xed conguration. In Section 4 we obtain
the optimal number of clusters that maximizes the network lifetime.
Section 5 outlines numerical results where we demonstrate the benet
Progress In Electromagnetics Research B, Vol. 12, 2009 261
Table 1. Energy consumption sources considers by various energy
models.
Energy Sources Heinzelman Millie et al. Zhu et al. Our
et al. [11] [22] [15] Model
Processing

Communication

Sensing

Transient



Logging

Actuation

Cluster Formation

of our comprehensive energy model and its application to a general
sensor network with xed cluster and single-hop transmission. In
Section 6 we discuss of applying the proposed energy model to rotation
cluster heads or dynamic conguration and Section 7 concludes the
paper.
2. SENSOR ENERGY MODEL
We consider a wireless sensor network with a cluster topology, as
shown in Fig. 1, in which sensors are grouped into clusters, and
Figure 1. Cluster topology of a Sensor Network: Sensors are grouped
into clusters, and individual sensors sense data and transmit to cluster
heads (CH). Cluster heads aggregate this data and then forward it
through a unique root, depending on the tree structure, to the base
station or sink node.
262 Halgamuge et al.
individual sensors sense data and transmit to cluster heads (CH) using
single hops as in [11]. Here we assume that all sensor nodes within
a cluster use time division multiple access (TDMA) to access their
CH. Data is generated in individual sensor nodes. The CH (current
CH) processes and aggregates collected data from its own sensors,
child CHs (previous CHs) and transmit to its parent CH (next CH)
towards to the base station or sink via other CHs with multiple hops.
Every communication round base station get equal amount of data.
Clustering reduces the data to be transmitted to the base station by
processing all data locally. Data aggregation techniques can be used
to combine correlated data from sensor nodes into a small set of data
which contains only relevant information [11]. Therefore, we assume
each sensor senses b bits and transmit to CH. The CH process b
1
bits
where b
1
= b number of sensors from its own cluster and b
1
from
child CHs where 0, 1, 2, . . . , k 1 and k is the number of clusters
in the sensor network. In general the child CH, in plural: child CHs,
would have less data than the parent CH and it all depends on the
routing and topology. After processing and aggregation, CH transmits
packets with b
2
bits to BS towards parent CH, where b
2
< b
1
+ b
1
.
Figure 2. Sensor node operation.
As in [23], a sensor lifetime can be divided into rounds. In each
round a sensor node performs steps 15 as shown in Fig. 2, and any
CH performs steps 17 as shown in Fig. 3. We assume that every
sensor node generates a xed-sized packet and forwards to its CH. All
generated packets are forwarded to the base station by the CH in each
round. The base station schedules transmission time based on time
division multiple access (TDMA) to avoid collisions.
In the literature [24], network lifetime is dened as either the time
until the rst (or last) node dies or the time until a given percentage
(P
node
) of the nodes dies. We adopt the latter denition. Note that a
Progress In Electromagnetics Research B, Vol. 12, 2009 263
node/CH can die either because it runs out of battery, or because the
death of other CHs isolates it from its base station. We dene C
opt
,
optimal number of clusters, as the number of clusters that minimizes
energy dissipation. As in [25], we assume a symmetric radio channel
making the energy needed to transmit from one point to another in
both directions identical. We also assume that all sensors acquire
information at a xed rate, making data available to be sent to the
sink every round.
Figure 3. Cluster head operation.
The energy consumed by a sensor node can be attributed to seven
main basic energy consumption sources: micro controller processing,
radio transmission and receiving, transient energy, sensor sensing,
sensor logging and actuation. We assume that all sensor nodes
(except CHs) are homogeneous, therefore energy consumption for
all activities (excluding for communication energy due to dierent
transmit distances to their CHs), are the same for each sensor node.
The total number of data packets (data + control information) in one
round is sum of data packets received from the sensor nodes in its
own cluster and incoming data packets from child CHs. Therefore,
the transmission and receiving energy used by a CH is higher than
that of a normal sensor node because of the additional data processing
and aggregation tasks associated with it. Let h
i
> 1 be a weighting
factor that applies to a CH to indicate by how much it consumes more
energy than a regular sensor node for energy source i, with i = 1, 2, 3, 4
for processing, transmission and receiving, sensing and sensor logging,
respectively.
The following Sections 2.3 to 2.6 describe the energy needed in
each step (Figs. 2 and 3) in detail.
264 Halgamuge et al.
2.1. Sensor Sensing
The sensing system links the sensor node to the physical world. Sources
of sensor power consumption are: signal sampling and conversion of
physical signals to electrical signals, signal conditioning, and analog
to digital conversion (ADC). Let I
sens
be the total current required
for sensing activity and T
sense
be the time duration for sensor node
sensing. We evaluate the total energy dissipation for sensing activity
for b bit packet, E
sens
N
at the sensor node per round by
E
sens
N
(b) = bV
sup
I
sens
T
sens
,
and the total energy dissipation for sensing activity at the CH per
round by
E
sens
CH
(h
3
, b) = h
3
E
sens
N
(b).
where V
sup
is the supply voltage.
2.2. Sensor Logging
Sensor logging consumes energy used for reading b bit packet data and
writing it into memory [26]. Sensor logging energy consumption for a
sensor node per round is evaluated by
E
logg
N
(b) = E
write
+ E
read
=
bV
sup
8
(I
write
T
write
+ I
read
T
read
) ,
where E
write
is energy consumption for writing data, E
read
is energy
consumption for reading b bit packet data, I
write
and I
read
are current
for writing and reading 1 byte data. Energy consumption for logging
sensor readings at the CH per round can be evaluated by
E
logg
CH
(h
4
, b) = h
4
E
logg
N
(b).
2.3. Micro-controller Processing
The energy for processing and aggregation of the data mainly
consumed by the micro-controller, is attributed to two components:
energy loss from switching, E
switch
, and energy loss due to leakage
current, E
leak
.
This energy, dissipated by leakage current, occurs when a sub-
threshold leakage current ows between the power source and the
ground. Total energy dissipation by the sensor node used for data
Progress In Electromagnetics Research B, Vol. 12, 2009 265
processing/aggregation b bit packet, E
pro
N
, per round is given by [27]:
E
pro
N
(b
1
, N
cyc
) = b
1
N
cyc
C
avg
V
2
sup
. .
switching
+b
1
V
sup
_
I
0
e
Vsup
npV
t
__
N
cyc
f
_
. .
leakage
,
and total energy dissipation by the cluster head (CH), E
pro
CH
, per
round is given by
E
pro
CH
(h
1
, b
1
, N
cyc
) = h
1
E
pro
N
(b
1
, N
cyc
) (1)
where N
cyc
is the number of clock cycles per task, C
avg
is the average
capacitance switched per cycle, I
0
is the leakage current, n
p
is the
constant which depends on the processor, V
t
is the thermal voltage,
and f is sensor frequency. Assuming that sensor nodes only sense data
and transmit to their CH once during each round, we ignore energy
dissipation due to data processing from regular sensor nodes.
2.4. Radio Transmission and Receiving
Communication of neighboring sensor nodes is enabled by a sensor
radio. Energy dissipation by a sensor node can be attributed
to transmitting and receiving data. According to [27] the energy
dissipation due to transmit b bit packet, in a distance d
ij
from sensor
node to the CH per round is given by
E
tx
N
(b, d
ij
) = bE
elec
. .
electronics
+bd
n
ij
E
amp
. .
amplifier
, (2)
where E
elec
is the energy dissipated to transmit or receive electronics,
E
amp
is the energy dissipated by the power amplier, and n is the
distance based path loss exponent (we use n = 2 for free space fading

,
and n = 4 for multi-path fading [28]). Energy dissipation due to
receiving b bit packet from the sensor node is given by E
rx
N
(b) = bE
elec
.
Therefore energy dissipation due to transmission of a b
2
bit packet over
a distance d
j
from the CH to the parent CH per round can be estimated
by
E
tx
CH
(h
2
, b
2
, d
j
) = h
2
b
2
E
elec
. .
electronics
+b
2
d
n
j
E
amp
. .
amplifier
,
where the energy dissipation due to receiving a b bit packet from the
CH estimated by E
rx
CH
(b
2
) = h
2
b
2
E
elec
.

Free space fading refers to the attenuation of received signal strength when transmitter
and receiver have a clear unobstructed line-of-sight path between them [28].
266 Halgamuge et al.
2.5. Control Packet Overheads
These include energy dissipation from transmitting and receiving RTS,
CTS, ACK packets and retransmissions. This is only relevant to
contention based protocols like CSMA/CA, not to TDMA which
concerns us here.
2.6. Actuation
Energy dissipation for actuation, E
actu
, is hard to estimate in general
because this is highly dependent on application. Total energy
dissipation for actuation is E
actu
N
act
where N
act
is the number of
actuations per CH. For example if we use temperature sensors to drive
a fan that needs two motors, there can be a command to switch on the
two motors when temperature is beyond some value and in that case
N
act
= 2. In practice, however, actuation may not be performed by
any sensor node.
2.7. Transient Energy
Radio and micro-controller units (MCU) support dierent operating
modes including active, idle and sleep. Transitions between operating
modes involve signicant energy dissipation [22]. Changes in radio
operating mode can cause a signicant amount of power dissipation.
These are often ignored in the literature. Let T
tranON
and T
tranOFF
be the times required for sleep-to-idle and idle-to-sleep transitions,
respectively. A sensor node will listen to a busy tone of the channel,
wake up for a duration of T
A
and then sleeps for T
S
, assuming T
S
T
A
.
Similarly, CH wakes up for duration T
A
CH
, which will be discussed in
Section 3.2, and then sleeps for T
S
CH
. Let T
tr
be the time between
consecutive packet transmissions. The CH will transmit all the packets
it receives in one batch every T
tr
seconds (Fig. 4). This is given by,
T
tr
= T
A
CH
+ T
S
CH
= T
A
+ T
S
. (3)
The duty cycle for the sensor node, c
N
, can be dened as in [22]:
c
N
=
T
tranON
+ T
A
+ T
tranOFF
T
tranON
+ T
A
+ T
tranOFF
+ T
S
.
Similarly, the duty cycle for the CH, c
CH
is dened by:
c
CH
=
T
tranON
+ T
A
CH
+ T
tranOFF
T
tranON
+ T
A
CH
+ T
tranOFF
+ T
S
CH
. (4)
Progress In Electromagnetics Research B, Vol. 12, 2009 267
Figure 4. Wake-up and sleeping times of the sensor nodes and the
CHs per round.
The average current for a sensor node is given by I
N
= c
N
I
A
+(1
c
N
)I
S
. The total energy dissipation from the sensor node per round is
evaluated by
E
tran
N
= T
A
V
sup
[c
N
I
A
+ (1 c
N
)I
S
] ,
where I
A
and I
S
are current for active and sleeping mode. Similarly,
the average current for a CH is given by I
CH
= c
CH
I
A
+ (1 c
CH
)I
S
and the energy dissipation due to operating mode at the CH per round
is evaluated by
E
tran
CH
= T
A
CH
V
sup
[c
CH
I
A
+ (1 c
CH
)I
S
] .
3. NETWORK ENERGY CONSUMPTION
3.1. Applying the Proposed Energy Model to a Fixed
Cluster Head
In this section we apply the previously dened energy model to a sensor
network, assuming that N
s
sensor nodes are randomly and uniformly
distributed in a M M region.
Consider a k cluster sensor network where the clusters are laid
out in a directed tree topology whose root is a base station (sink
node). Cluster j comprises one CH denoted CH
j
and n
j
sensor nodes,
j = 1, 2, . . . , k. Thus, on average the number of sensor nodes (including
the CH) in a cluster is (N
s
/k), as some clusters will have N
s
/k| and
some other have N
s
/k| sensor nodes.
The total number of sensors is N
s
=

k
j=1
(n
j
+ 1). Sensors
transmit information to their respective CH. The CH will then forward
268 Halgamuge et al.
the packet through a unique route of CHs to the sink node (the
uniqueness results from the tree structure). All transmissions are from
the leaves through the intermediate nodes towards the root which is
the sink node. Let d
j
be the distance between CH
j
and the next
CH (or the sink node) that it transmits to, and d
ij
be the distance
between node i in cluster j and its cluster head. Here, the total energy
consumed by sensor node i in cluster j per round is
E
N
(ij) =
_

_bV
sup
I
sens
T
sens
. .
sensing
+bV
sup
(I
write
T
write
+ I
read
T
read
)
. .
datalogging
+bE
elec
+ bd
n
ij
E
amp
. .
transmit
+T
A
V
sup
[c
N
I
A
+ (1 c
N
)I
S
]
. .
transient
_

_. (5)
Similarly, the total energy consumed by cluster CH
j
per round is
E
CH
(j) =
_

_h
3
bV
sup
I
sens
T
sens
. .
sensing
+ h
4
bV
sup
(I
write
T
write
+ I
read
T
read
)
. .
datalogging
+h
1
b
1
N
cyc
C
avg
V
2
sup
. .
switching
(n
j
+ 1) + h
1
b
1
V
sup
_
I
0
e
Vsup
npV
t
__
N
cyc
f
_
. .
leakage
(n
j
+ 1)
+ h
2
b
1
E
elec
. .
receiveown
(n
j
) + h
2
b
2
(1 + )E
elec
+ b
2
(1 + )d
n
j
E
amp
. .
transmit to parent CH
+ h
2
b
2
E
elec
. .
receivechild CH
+T
CH
V
sup
[c
CH
I
A
+ (1 c
CH
)I
S
]
. .
transient
+E
actu
N
act
. .
actuation
_
_
, (6)
where n
j
= (N/k), j, for the case of equi-sized clusters (all clusters
have the same number of sensor nodes).
To compute the energy consumption of all E
N
(ij) and E
CH
(j)
values, we apply Equations (5) and (6) rst to the leaf clusters and
recursively progress down the tree until we reach the root.
Therefore the total energy consumed by the entire network per
round is given by
E
tot
=
k

j=1
_
E
CH
(j) +
n
j

i=1
E
N
(ij)
_
. (7)
Progress In Electromagnetics Research B, Vol. 12, 2009 269
3.2. Wake up Time for CH (T
A
CH
)
The above energy models require the evaluation of the sensors duty
cycle c
CH
dened in (4). In this subsection, we derive an expression
for CH wake up time, T
A
CH
, which is then used to determine c
CH
.
Figure 5. Wake up time for cluster head. This has three components:
time taken to receive data from its own sensors, receive data from its
child CHs, and transmit data to its parent CH. Data can be received
only from child CHs and can be transmitted to their parent CH.
Figure 5 shows an example of a sensor network which we will use
to illustrate the data transmission between clusters. With each CH j,
we can associate a parent p(j) and a set c(j) of child CHs. A child
c c(j) is a CH that forwards data to CH j. A parent p(j) is a CH
that will transmit data from CH i towards to the base station/sink.
For example, in Fig. 5, the parent of CH
j
is CH
j+1
and the child
CHs of CH
j
is CH
j1
, CH
j2
, therefore, c(j) = CH
j1
, CH
j2
and
p(j) = CH
j+1
.
The total wake up time for jth CH in one round is the total time
taken to:
(a) receive data packets from the sensor nodes (total of n
j
sensors) in
its own cluster, with each sensor having wake up time of T
A
,
(b) receive incoming data packets from child CHs and
(c) transmit data packets to its parent CH, p(j).
270 Halgamuge et al.
For a LEACH-type protocol, wake up time for jth CH, T
j
A
CH
, is
given by:
T
j
A
CH
= max n
j
T
A
. .
(a)
+

c c(j)
T
CH
c,j
. .
(b)
+T
CH
j,p(j)
. .
(c)
, (8)
where T
CH
c,j
is the time taken for transmission from its child CH c to
CH j, and T
CH
j,p(j)
is the time taken for transmission from CH j to its
parent CH p(j). Knowing T
A
CH
, the duty cycle of the CH in (4) can
be determined.
We nd in Sections 3.1 and 6 that the sensor nodes energy
consumption depends on the number of clusters in a network. In the
following we will seek the optimal number of clusters to maximizes the
network lifetime.
4. FINDING THE OPTIMAL NUMBER OF CLUSTERS
In this section we apply the principles discussed in the previous sections
to develop a technique for increasing network lifetime by choosing
the optimal number of clusters. Generally speaking, if we have more
clusters while maintaining the same load per CH, the transmission
distance from a sensor to its own CH is reduced. Therefore, the overall
energy consumption is also reduced. On the other hand, increasing
the number of clusters means that the transmission path between a
sensor and the BS will include more CH-to-CH hops which means
higher overall energy consumption. The aim is therefore to nd the
optimal number of clusters so that the overall energy consumption is
minimized. Note that this optimal clustering depends highly on the
energy model used [29]. Therefore, it is important to use the right
energy model, as this paper aims to do.
We will now demonstrate the use of our energy model for optimal
clustering and compare the results with other approaches. Assume
that each sensor node transmits data to its CH only once during each
round. Therefore, from (5), total energy consumed by a sensor node
during each round is
E
node
=
_
bE
sens
N
+ bE
logg
N
+ bE
elec
+ bd
2
toCH
E
fs
+E
tran
N
] , (9)
where b is the number of bits in every packet, d
toCH
is the distance
between node and CH, E
fs
is the free space fading energy.
Progress In Electromagnetics Research B, Vol. 12, 2009 271
Similarly, from (6), the total energy consumed by a CH during
each round is
E
head
=
_
bE
sens
CH
+ bE
logg
CH
+ b
1
E
pro
CH
_
N
s
k
_
+h
2
b
1
E
elec
_
N
s
k
1
_
+ h
2
b
2
E
elec
+ h
2
b
2
(1 + )E
elec
+b
2
(1 + )d
4
toBS
E
mp
+ E
tran
CH
_
, (10)
where E
mp
is the multi-path fading energy. Note that we consider a
multi path model with d
4
power loss, and assume that actuation is
not performed. Consider a square of area M M with k clusters, i.e.,
the area covered by each cluster is approximately M
2
/k. As in [11],
we assume that the CH is at the center of mass of its cluster, and we
acknowledged that the cluster area can be arbitrary shaped, but for
simplicity, we assume that it is a square. For k = 1, assuming sensors
are randomly uniformly distributed over the square area, the mean
square distance from a sensor to its CH is given by
E[d
toCH
] =
_
M
0
_
M
0
d(x, y)(x, y)dxdy
=
1
M
2
_
M
0
_
M
0
_
x
M
2
_
2
+
_
y
M
2
_
2
dxdy, (11)
where (x, y), 0 x, y M, is the joint probability density function.
(If sensors are placed uniformly then we have (x, y) =
1
M
2
.)
For k > 1, the mean square distance is given by,
E[d
toCH
2
] =
k
M
2
_ M

k
0
_ M

k
0
_
x
M
2

k
_
2
+
_
y
M
2

k
_
2
dxdy,
=
M
2
6k
. (12)
The above calculation (11) and (12) is for a square area, therefore
k = i
2
where i is an integer. As an approximation, we evaluate the
mean square distance using (11) and (12) with arbitrary value of k.
Knowing the mean square distance, we can now derive the optimal
number of clusters.
From (9) and (10) the energy dissipation in a single cluster during
each round is given by
E
cluster
= E
head
+
_
N
s
k
1
_
E
node
E
head
+
_
N
s
k
_
E
node
. (13)
272 Halgamuge et al.
The total energy for k clusters, during each round based on our
energy model is obtained using (9), (10), (12) and (13) as
E
our
= kE
cluster
= b (E
elec
N
s
+ E
pro
CH
N
s
+ d
n
toBS
E
amp
k + E
sens
CH
k
+E
tran
CH
k + E
logg
CH
k + E
elec
N
s
+ E
fs
M
2
6k
N
s
+E
tran
N
N
s
+E
sens
N
N
s
+ E
logg
N
N
s
) . (14)
We adopt the assumption of [11] that the BS is far from sensor nodes
and therefore the distance between CH to the BS for all CHs be
considered to be equal. By dierentiating (14) with respect to k and
equating to zero, the resulting optimal number of clusters, C
opt
, is
C
opt
=
_

N
s

6
M
d
2
toBS

E
fs
D

_
, (15)
where D

= (E
mp
+E
sens
CH
+E
tran
CH
+E
logg
CH
). Knowing C
opt
, for a
given network, we can evaluate the average radius of a circular cluster
as M/
_
C
opt
, the average length of square cluster as M/
_
C
opt
,
and the circum radius of the hexagon is
4
_
4
27
M/
_
C
opt
. Providing
these cluster shape alternatives allows designers to choose the one
appropriate for their work. Once we have found the optimal number
of clusters, we can calculate the network energy and compare it with
results obtained by other energy models. According to Heinzelman et
al. [11], the total energy during each round, E
Hein
, is given by
E
Hein
= b
_
E
elec
N
s
+ E
pro
CH
N
s
+ d
4
toBS
E
mp
k
+E
elec
N
s
+ E
fs
1
2
M
2
k
_
. (16)
In the following we compare the average energy dissipation of our
energy model and that of other energy models in [11, 15, 22].
In Fig. 6 and Table 2, we show that in our proposed energy model,
the range of the optimal number of clusters is 1 < C
opt
< 6, while
according to the energy models in [11, 15, 22], the range is within
2 < C
opt
< 16, when the distance between the CH and the sink
node is between 45145 m. Our simulation results agree with this
analysis and will be discussed more details in Section 5. For example,
for a distance of 55 m, the over-estimation of the optimal number of
clusters is 194.15% [11], 101.62% [22], and 74.06% [15]. This shows the
Progress In Electromagnetics Research B, Vol. 12, 2009 273
dierence between the models. Nevertheless, the dierence in optimal
number of clusters between these energy models is getting closer as the
distance between the sink and the CHs increases, as shown in Fig. 6.
This is because, as this distance increases the energy dissipation for
communication becomes more and more dominant in the cost function.
An optimal number of clusters for a given number of sensors is
found by varying the distance and comparing the energy dissipation
per round. From the simulation results, it is conrmed that the optimal
number of clusters is three for [11], two for [22] and [15], and one for
40 60 80 100 120 140 160
0
5
10
15
20
Distance from CH to Sink Node (m)
(
C
o
p
t
)

O
p
t
i
m
a
l

N
u
m
b
e
r

o
f

C
l
u
s
t
e
r
s
Heinzelman et al.
Mille and Vaidya
Zhu and Papavassiliou
Our Energy Model
Figure 6. Optimal number of clusters with distance from CHs to
sink node or base station Analytical results for N
s
= 100 nodes,
M = 100, E
fs
= 7 nJ/bit/m
2
and E
mp
= 0.0013 pJ/bit/m
4
when
45 < distance < 145 is maintained.
Table 2. Optimal number of clusters with dierent energy models,
for N
s
= 100 sensor nodes and M = 100, when the distance between
the CH and the sink node is between 45145 m.
E
fs
= 7 [nJ/bit/m
2
] E
fs
= 10 [pJ/bit/m
2
]
Energy Copt Range Copt Copt Range Copt
Model 45 < dtoBS < 145 Simu. 45 < dtoBS < 145 Simu.
Our Model 1-6 3 0-2 1
Zhu et al. 1-12 5 0-4 2
Mille et al. 2-13 6 0-5 2
Heinz. et al. 2-16 11 1-7 3
274 Halgamuge et al.
0 2 4 6 8 10 12
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
(k) Number of Clusters

A
v
g
.

E
n
e
r
g
y

D
i
s
s
i
p
a
t
i
o
n

P
e
r

R
o
u
n
d


(
m
J
)
Mille and Vaidya (k=6)
Zhu and Papavassiliou (k=5)
Our Energy Model (k = 3)
Heinzelman et.al (k=2)
Figure 7. Average energy dissipation versus number of clusters
when E
fs
= 7 nJ/bit/m
2
and N
s
= 100 nodes, M = 100 by using
(13). Optimal number of clusters, based on their energy models,
are indicated with arrows. Here the average distance from CH to
base station or sink node is 22 m. This shows the dierence between
the energy models does have signicant eect to the sensor energy
dissipation. Observe the dierence is about 8 pJ per round.
our energy model when E
fs
= 10 pJ/bit/m
2
with 100 node network.
Therefore, we can conclude that clustering will not increase eciency
when the deployment area is small.
Figure 7 shows the average energy dissipation versus number of
clusters when E
fs
= 7 nJ/bit/m
2
and N
s
= 100 nodes, M = 100
using (13). The optimal number of clusters, based on dierent energy
models, are indicated with arrows. According to Fig. 7, optimizing
the number of clusters does have signicant eect on sensor network
lifetime. We also observe that the optimal number of clusters of our
energy model is three while the energy models of [11, 15] and [22] will
lead to optimal number of clusters to be 2, 5, and 6, respectively, for the
same free space fading energy. The main reasons for this variation lie
in our use of a more comprehensive energy model with a more realistic
estimation of processing energy which turned out to be higher than
the value considered in [11, 15, 22].
From Figs. 811, we can observe that: 1) the number of clusters
varies with the energy model used, as well as the distance from the
CH to the base station, 2) energy dissipation varies with the number
Progress In Electromagnetics Research B, Vol. 12, 2009 275
0
50
100
150
0
5
10
15
0.036
0.038
0.04
0.042
0.044
0.046
0.048
(d) Distance (m)
(k) Number of Clusters
A
v
g
.

E
n
e
r
g
y

D
i
s
s
i
p
a
t
i
o
n

P
e
r

R
o
u
n
d

(
m
J
)
Figure 8. Average energy dissipation versus no. of clusters and
distance for Heinzelman et al. energy model. Distance with N
s
= 100
nodes, M = 100, E
fs
= 7 nJ/bit/m
2
and E
mp
= 0.0013 pJ/bit/m
4
.
0
50
100
150
0
5
10
15
0.144
0.146
0.148
0.15
0.152
0.154
0.156
(d) Distance (m)
(k) Number of Clusters
A
v
g
.

E
n
e
r
g
y

D
i
s
s
i
p
a
t
i
o
n

P
e
r

R
o
u
n
d

(
m
J
)
Figure 9. Average energy dissipation versus no. of clusters and
distance for our energy model. Distance with N
s
= 100 nodes,
M = 100, E
fs
= 7 nJ/bit/m
2
and E
mp
= 0.0013 pJ/bit/m
4
.
276 Halgamuge et al.
0
50
100
150
0
5
10
15
0.064
0.066
0.068
0.07
0.072
0.074
(d) Distance (m)
(k) Number of Clusters
A
v
g
.

E
n
e
r
g
y

D
i
s
s
i
p
a
t
i
o
n

P
e
r

R
o
u
n
d

(
m
J
)
Figure 10. Average energy dissipation versus no. of clusters an
distance for Mille and Vaidya energy model. Distance with N
s
= 100
nodes, M = 100, E
fs
= 7 nJ/bit/m
2
and E
mp
= 0.0013 pJ/bit/m
4
.
0
50
100
150
0
5
10
15
0.072
0.074
0.076
0.078
0.08
0.082
(d) Distance (m)
(k) Number of Clusters
A
v
g
.

E
n
e
r
g
y

D
i
s
s
i
p
a
t
i
o
n

P
e
r

R
o
u
n
d

(
m
J
)
Figure 11. Average energy dissipation versus no. of clusters and
distance for Zhu and Papavassiliou energy model. Distance with N
s
=
100 nodes, M = 100, E
fs
= 7 nJ/bit/m
2
and E
mp
= 0.0013 pJ/bit/m
4
.
Progress In Electromagnetics Research B, Vol. 12, 2009 277
of clusters, and therefore, we can nd the number of clusters that
minimizes energy dissipation.
5. SIMULATION RESULTS
Our results are based on analyses and are validated by Matlab
simulations. The hardware parameter values, such as, current required
for sensor wake up and sleeping time, used in our simulations, are those
of Mica2 Motes [32]. As in [22], we also use radio parameters, such as,
sensor wake up and sleeping time, of the radios data sheet (Chipcorn,
CC1000 datasheet) [31]. All parameter values used in our energy model
(for both simulation and the analysis) are listed in Table 3. We also
indicate in the Table the sources/references where these values are
originated. Since our results are based on real device parameters,
they are expected to reect a true energy dissipation behaviour of
the system.
In our simulations, we consider a sensor network with N
s
=
100 sensor nodes. Consider our deployment area as the square
(0, 0), (0, 100), (100, 0), (100, 100) as in [11]. The base station or
sink node is located in the coordinate (50, 175) which is outside the
deployment area and connected to an external power supply. Initially,
CHs are randomly placed within an 50 m 50 m square placed in the
middle of the 100 m 100 m deployment area. All other sensor nodes
in each cluster are randomly uniformly distributed in a circle of 25 m
radius of their respective CH. For our simulation we consider practical
sensor network with xed clusters and single-hop transmission. We
generate 1000 random setups, each with the above simulation setting.
Therefore each simulation data point is obtained by averaging over
1000 random setups. We assume that the total number of sensors in
the entire network N
s
is 100, and each node reports data once every
300 ms (T
tr
= 0.3 s). The channel bandwidth was set to 1 Mb/s as
in [11], each single packet size is b = 2 kb, as in [25], which maintains
a low average data rate requirement per node (< 12 bps). Moreover,
we assume energy dissipation for actuation, E
actu
, is 0.02 mJ as in [34]
and energy for starting up the radio, E
ini
is 1 J as in [35]. Note that
we do not account for energy dissipation in re-transmitting because of
the packets collided in the simulations. As in [22], for our simulation,
we used Mica2 Motes hardware values [32] and time values are based
on radios data sheet [31]. We assume that, being consistent with a
LEACH application, a CH and a sensor node have the same radio.
Sleeping time, T
S
= 299 ms, and wakeup time, T
A
= 1 ms, of the
sensor node are considered, as in [22]. Self-discharge of batteries is
considered, as 3% per year as in [26]. We conducted Matlab simulations
278 Halgamuge et al.
with dierent parameter settings as described later. We consider
processing and communication energy of the CH is 20% more, and
sensing and logging energy is 10% more than that of regular sensor
nodes. Therefore, we assume selection of the weighting factor, h
i
, as
h
1
, h
2
, h
3
, h
4
=1.2, 1.2, 1.1, 1.1.
Table 3. Parameter values used in energy model.
Symbol Description Value
Ncyc Number of clock cycles per task 0.97 10
6
[30]
Cavg Avg. capacitance switch per cycle 22 pF [31]
Vsup Supply voltage to sensor 2.7 V [31]
f Sensor frequency 191.42 MHz [27]
np Constant: depending on the processor 21.26 [30]
n Path loss exponent 2 or 4 [30]
I0 Leakage current 1.196 mA [30]
Vt Thermal voltage 0.2 V [27]
b Transmit packet size 2 kb [25]
E
elec
Energy dissipation: electronics 50 nJ/bit [30]
Eamp Energy dissipation: power amplier 100 pJ/bit/m
2
[30]
TtranON Time duration: sleep idle 2450 s [22]
TtranOFF Time duration: idle sleep 250 s [22]
IA Current: wakeup mode 8 mA [32]
IS Current: sleeping mode 1 A [32]
TA Active time 1 ms [22]
TS Sleeping time 299 ms [22]
Ttr Time between consecutive packets 300 ms
Tsens Time duration: sensor node sensing 0.5 mS
Isens Current: sensing activity 25 mA
Iwrite Current: ash writing 1 byte data 18.4 mA [33]
I
read
Current: ash reading 1 byte data 6.2 mA [33]
Twrite Time duration: ash writing 12.9 mS [33]
T
read
Time duration: ash reading 565 S [33]
Eactu Energy dissipation: actuation 0.02 mJ [34]
Progress In Electromagnetics Research B, Vol. 12, 2009 279
5.1. Energy Comparison
< 1%
51%
12%
10%
14%
6%
8%
Initial
Communication
Processing
Transient
Logging
Sensing
Actuation
Communication
Processing
Transient
Logging
Sensing
Acuation
Initial
Figure 12. Energy consumption pie chart for any sensor in cluster
j, when actuation is considered. Here we consider for N
s
= 100
sensor nodes, M = 100, k = 10 clusters, E
fs
= 10 pJ/bit/m
2
and
E
mp
= 0.0013 pJ/bit/m
4
.
In Fig. 12 we present a pie chart describing the energy
consumption for communication, processing, transient, sensor loggings
and sensing, 51%, 12%, 10%, 14% and 6% of the total energy
respectively. All of these sources of energy consumption are not
negligible.
The same parameters, namely N
s
= 100 sensor nodes, M = 100,
k = 10 clusters, E
fs
= 10 pJ/bit/m
2
and E
mp
= 0.0013 pJ/bit/m
4
, are
used to generate Fig. 13 where we compare the eect of the dierence
energy models on the sensor network lifetime. Each simulation data
point is obtained by averaging over 1000 random setups, but observe
that the results does not change with single simulation.
We consider actuation performed only for this pie chart (Fig. 12)
and exclude it from all other simulations, for the purpose of fair
comparison with the other energy models that also exclude it. For all
our simulation, we used an AA size alkaline battery with 1500 mAh.
However, we also repeated our simulation for AA size alkaline battery
with 700 mAh, C-cell battery with 5000 mAh and D-cell battery with
280 Halgamuge et al.
0.2 0.4 0.6 0.8 1
100
150
200
250
300
350
400
Sleeping Time (s)
N
o
d
e

L
i
f
e
t
i
m
e

(
d
a
y
s
)
Our Energy Model
Mille an Vaidya
Zhu and Papavassiliou
Heinzelman et al.
S
e
n
s
o
r
Figure 13. Sensor node lifetime verses sleeping time of the sensor
node, with dierent energy models with AA alkaline batteries by using
(5) and (6). Here we consider for N
s
= 100 sensor nodes, M = 100, k =
10 clusters, E
fs
= 10 pJ/bit/m
2
[11] and E
mp
= 0.0013 pJ/bit/m
4
[11].
9000 mAh for 1.5 V and found that all results were consistent. Node
lifetime can be computed by
node lifetime =
initial battery capacity
avg. current 365 24
[years],
where the units of the initial battery capacity is mAh and the avg.
current is mA. In Fig. 13, we show that existing energy models over-
estimate life expectancy of a sensor node by 3058%.
5.2. Eect of Free Space Fading Energy (E
fs
)
The optimal number of clusters derived in (15) is only applicable if
free space fading energy is assumed to be constant [19] , which may
not be the case in practice. For this reason, we repeated the above
simulation by varying free space fading energy, E
fs
within the interval
(1 10
4
) [pJ/bit/m
2
] and observed in Fig. 14 that when E
fs
increases,
the optimal number of clusters also increases.
We found that energy dissipation per round increases from 7.11%
to 12.81% as the optimal number of clusters changes from 1 to 3.
Moreover, we observe that when free space fading energy, E
fs
<
1670 pJ/bit/m
2
the optimal number of clusters needed is one and
Progress In Electromagnetics Research B, Vol. 12, 2009 281
0 2 4 6 8 10 12
0.138
0.140
0.142
0.144
0.146
0.148
0.150
0.152
0.154
0.156
0.158
(k) Number of Clusters

A
v
g
.

E
n
e
r
g
y

D
i
s
s
i
p
a
t
i
o
n

P
e
r

R
o
u
n
d


(
m
J
)
10000 pJ/bit/m
2
7000 pJ/bit/m
2
7000 pJ/bit/m
2
5000 pJ/bit/m
2
3000 pJ/bit/m
2
2000 pJ/bit/m
2
1 pJ/bit/m
2
Figure 14. Average energy dissipation versus number of clusters when
E
fs
varies between (1 10
4
) pJ/bit/m
2
for our energy model. This
shows when E
fs
increase, optimal number of clusters also increases.
40
60
80
100
120
140
0
0.5
1
1.5
2
x 10
9
0
2
4
6
8
10
Distance (m)
E
fs
(J)
(
C
o
p
t
)

O
p
t
i
m
a
l

N
u
m
b
e
r

o
f

C
l
u
s
t
e
r
s
-
Figure 15. Relationship between E
fs
, C
opt
, and the distance for our
energy model, when E
fs
= 10 pJ/bit/m
2
.
hence, clustering is not necessary. We repeated the simulation by
increasing E
fs
further, up to 10
5
pJ/bit/m
2
, and conrm that the
282 Halgamuge et al.
40
60
80
100
120
140
0
0.5
1
1.5
2
x 10
9
0
5
10
15
20
25
Distance (m)
E
fs
(J)
(
C
o
p
t
)

O
p
t
i
m
a
l

N
u
m
b
e
r

o
f

C
l
u
s
t
e
r
s
-
Figure 16. Relationship between E
fs
, C
opt
, and the distance for
Heinzelman et al. energy model, when E
fs
= 10 pJ/bit/m
2
.
40
60
80
100
120
140
0
0.5
1
1.5
2
x 10
9
0
5
10
15
20
Distance (m)
E
fs
(J)
(
C
o
p
t
)

O
p
t
i
m
a
l

N
u
m
b
e
r

o
f

C
l
u
s
t
e
r
s
-
Figure 17. Relationship between E
fs
, C
opt
, and the distance for Mille
and Vaidya energy model, when E
fs
= 10 pJ/bit/m
2
.
Progress In Electromagnetics Research B, Vol. 12, 2009 283
40
60
80
100
120
140
0
0.5
1
1.5
2
x 10
9
0
5
10
15
Distance (m)
E
fs
(J)
(
C
o
p
t
)

O
p
t
i
m
a
l

N
u
m
b
e
r

o
f

C
l
u
s
t
e
r
s
-
Figure 18. Relationship between E
fs
, C
opt
, and the distance for Zhu
and Papavassiliou energy model, when E
fs
= 10 pJ/bit/m
2
.
optimal number of clusters becomes more important with higher free
space fading energy dissipation. Observe that the average energy
dissipation decreases with the increasing number of clusters, this is
due to reducing communication distance.
According to Figs. 1518, the number of optimal clusters increases
with the increase of free space fading energy, E
fs
, for all the above
mentioned four energy models. Now we consider the same network
but we vary the number of clusters k, free space fading energy, E
fs
,
and the distance from the CHs to the base station or sink node. Here
we investigate the analytical results which derived in Section 3, 3.2
and 4 (see Fig. 15, 16, 17, 18, 19, 20). We observe that the optimal
number of clusters decreases dramatically with the increasing distance.
Energy dierences between the models are shown in Fig. 19. For the
particular case when the distance is equal to 100 m, the variation of
the optimal number of clusters with free space fading energy is shown
in Fig. 20.
284 Halgamuge et al.
5.3. Energy Dierence E
D
Let E
D
represent the percentage of energy dierence between our
energy model and the energy model in [11]. We found that the energy
dierence dramatically decreases when E
fs
is contained within the
interval (1 5 10
3
) [pJ/bit/m
2
]. We kept the optimal number of
clusters as three and free space fading energy E
fs
as 710
3
pJ/bit/m
2
,
and repeated the simulation. We observe that the sensor lifetime is
increased by 12.74% when the sleeping time is 0.2 s, and by 13.92%
when the sleeping time is 0.4 s, when the number of clusters used is
reduced from 10 (non optimal clusters) to 3 (optimal clusters).
5.4. Eect of Physical Area of Sensor Network
Next, we vary the number of sensor nodes N
s
and the physical area
M to their eect on the distance, free space fading energy, and the
number of optimal clusters. In addition, we consider how sensor
network lifetime can be maximized and the number of sensors required
to design a network for a given lifetime. As demonstrated in Fig. 21,
the optimal number of clusters vary linearly with M (the square root of
the physical area). Importantly, the eect of free space fading energy,
E
fs
, becomes less by increasing the distance as in Fig. 22.
40
60
80
100
120
140
0
0.5
1
1.5
2
x 10
9
0
5
10
15
20
25
Distance (m)
E
fs
(J)
(
C
o
p
t
)

O
p
t
i
m
a
l

N
u
m
b
e
r

o
f

C
l
u
s
t
e
r
s
Heinzelman et al.
Mille and Vaidya
Zhu and Papavassiliou
Our Energy Model
-
Figure 19. How optimal number of clusters, C
opt
, vary for dierent
energy models, when E
fs
= 10 pJ/bit/m
2
. Relationship between free
space fading energy, E
fs
, optimal number of clusters, C
opt
, and the
distance.
Progress In Electromagnetics Research B, Vol. 12, 2009 285
0 0.2 0.4 0.6 0.8 1
x 10
7
0
5
10
15
20
25
30
35
(
C
o
p
t
)

O
p
t
i
m
a
l

N
u
m
b
e
r

o
f

C
l
u
s
t
e
r
s
E
fs
(J)
Heinzelman et al.
Mille and Vaidya
Zhu and Papavassiliou
Our Energy Model
-
Figure 20. How optimal number of clusters, C
opt
, vary for dierent
energy models, when E
fs
= 10 pJ/bit/m
2
. Relationship between
optimal number of clusters, C
opt
, and free space fading energy, E
fs
.
50 100 150 200 250 300 350 400 450
0
5
10
15
20
25
30
35
40
(M) Square Root of the Physical Area (m)
(
C
o
p
t
)

N
u
m
b
e
r

o
f

O
p
t
i
m
a
l

C
l
u
s
t
e
r
s
Heinzelman et al.
Mille and Vaidya
Zhu and Papavassiliou
Our Energy Model
Figure 21. C
opt
versus square root of the physical area in all energy
models, when E
fs
= 10 pJ/bit/m
2
.
286 Halgamuge et al.
40
60
80
100
120
140
160
0
100
200
300
400
500
0
20
40
60
80
100
Distance (m)
(M) Square Root of Physical Area (m)
(
C
o
p
t
)

O
p
t
i
m
a
l

N
u
m
b
e
r

o
f

C
l
u
s
t
e
r
s
Heinzelman et al.
Mille and Vaidya
Zhu and Papavassiliou
Our Energy Model
Figure 22. When distance form CH to base station vary, how
C
opt
vary with square root of the physical area in all models, when
E
fs
= 10 pJ/bit/m
2
.
5.5. Eect of the Duty Cycle
Finally, we vary the number of duty cycles to investigate the aect
on energy consumption. (see Fig. 23). As shown in Fig. 23, when
the number of duty cycles increases, interestingly, the average energy
consumption of all models diverge. When the number of duty cycles
is 1, the maximum over-estimation of our energy model relative to
Heinzelman energy model is observed as 46.77%.
5.6. Percentage of Nodes Alive
The lifetime of a sensor network depends on the application where
the sensors are deployed. Therefore, we investigate how the number
of live sensor nodes varies with the number of rounds or time. An
over-estimation is shown in Fig. 24 where the number of sensor nodes
that live is plotted against the number of rounds. The maximum over-
estimation of the death of the last node of our energy model relative
to the Heinzelman energy model is 47.95%. We repeated the all above
simulations for AAA, C-cell and D-cell batteries and found the results
to be consistent.
Progress In Electromagnetics Research B, Vol. 12, 2009 287
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
Duty Cycle
A
v
g
.

E
n
e
r
g
y

D
i
s
s
i
p
a
t
i
o
n

(
m
J
)
New Energy Model
Zhu and Papavassiliou
Mille and Vaidya
Heinzelman et al.
Figure 23. Average energy dissipation per sensor node versus duty
cycle in energy models. Here we consider for N
s
= 100 sensor
nodes, M = 100, k = 10 clusters, E
fs
= 10 pJ/bit/m
2
[11] and
E
mp
= 0.0013 pJ/bit/m
4
[11].
0 1000 2000 3000 4000 5000 6000
0
20
40
60
80
100
N
u
m
b
e
r

o
f

N
o
d
e
s

A
l
i
v
e
No. of Rounds (Time)
Zhu and Papavassiliou
Mille and Vaidya
Our Energy Model
Heinzelman et al.
Figure 24. How the number of live nodes varies with the number of
rounds (time), comparing all energy models.
5.7. Eects of Number of Sensors and Distance of CHs from
Sink Node
We investigate how the number of sensors aects the optimal number
of clusters and the sensor node lifetime, in a given physical area.
288 Halgamuge et al.
0 50 100 150 200
0
1
2
3
4
5
6
7
8
9
(N
s
) Number of Sensors
(
C
o
p
t
)

N
u
m
b
e
r

o
f

O
p
t
i
m
a
l

C
l
u
s
t
e
r
s
Heinzelman et al.
Mille and Vaidya
Zhu and Papavassiliou
Our Energy Model
Figure 25. How the optimal number of clusters, C
opt
versus number
of sensors (analytical results) for a square root of the physical area
M = 100 and E
fs
= 10 pJ/bit/m
2
.
0
50
100
150
0
100
200
300
0
20
40
60
80
Distance (m)
(N
s
) Number of Sensors
(
C
o
p
t
)

O
p
t
i
m
a
l

N
u
m
b
e
r

o
f

C
l
u
s
t
e
r
s
Heinzelman et al.
Mille and Vaidya
Zhu and Papavassiliou
Our Energy Model
Figure 26. How the optimal number of clusters, C
opt
versus the
number of sensors and the distance from CH to base station for a
square root of the physical area M = 100 and E
fs
= 10 pJ/bit/m
2
.
Progress In Electromagnetics Research B, Vol. 12, 2009 289
The optimal number of clusters increases with the number of sensors
used for all energy models as shown in Fig. 25 and decreases with
the distance as in Fig. 26. It is clear that this change is far less for
the proposed energy model in comparison to the Heinzelman method.
According to our energy model, a moderate increase in the number
of sensors used may not result in the change of design parameters
concerning the optimal number of clusters. By (2), E d
2
, one may
expect that when the number of sensors is doubled the sensor lifetime
is multiplied by four. However, as we show this is not the case. Let us
consider sensor deployment with uniform distribution. According to
50 100 150 200 250 300
0
0.1
0.2
0.3
(N
s
) Number of Sensors
N
o
d
e

E
n
e
r
g
y

D
i
s
s
i
p
a
t
i
o
n

(
m
J
)
Total Energy
Communication Energy
Logging Energy
Processing Energy
Sensing Energy
Transient Energy
S
e
n
s
o
r
Figure 27. All energy components with sensor node lifetime versus
number of sensors for E
fs
= 10 pJ/bit/m
2
and a square root of the
physical area M = 100, for uniform deployment, with increasing
number of sensors.
Fig. 27, all energy dissipations converge when the number of sensors is
increased. Therefore, the change in energy dissipation with respect to
the change in the number of sensors becomes less. It should be noted
that sensor node lifetime is inversely proportional to the total energy
consumption of the sensor node:
sensor lifetime
1
total energy consumption
,
according to which, the plot of sensor node lifetime should
approximately resemble the inverse of the plot of total energy
consumption. As can be seen from Fig. 27, all energy consumption
types, except transient energy, decrease with the increasing number of
sensors.
290 Halgamuge et al.
Table 4. Sensor node lifetime when increasing number of sensors,
with uniform sensor deployment, when E
fs
= 10 pJ/bit/m
2
.
Radius = Length = Length = Avg.
No: of
M

C
opt
M

C
opt
4
_
4
27
M

C
opt
Distance
Sensors Copt [m] [m] [m] [m]
(Ns) (Circular (Square (Hexagonal Simulation
Cluster) Cluster) Cluster) NodeCH
50 1 56.4190 100 62.0403 36.9804
100 2 39.8942 70.7107 43.8691 25.5901
150 3 32.5735 57.7350 35.8190 20.1241
200 3 32.5735 57.7350 35.8190 19.1248
250 4 28.2095 50 31.0202 17.5014
300 4 28.2095 50 31.0202 17.5125
In Table 4, the same number of optimal clusters (C
opt
= 4) is
assigned when the number of sensors are 150 and 200 making cluster
radius stay the same. Interestingly, the simulation also shows that the
average number of cluster radius increases from 17.5014 to 17.5125 m,
also increasing the sensor lifetime. This is mainly due to the decrease
in the number of bits to be transmitted. Note that when the number
of sensors increases, the average number of bits sent by each sensor
decreases so that the total amount of information in a network is kept
constant at 10
5
bits. Knowing C
opt
and M for a given network, we
evaluate and present the average radius of a circular cluster in the
3rd column of Table 4, the average length of a square cluster in the
4th column, and the average circum radius of a hexagon in the 5th
column. Network designers can then use these values to optimize
network lifetime.
We investigate how the optimal number of clusters, C
opt
, varies as
a function of the number of sensors and area, when one million sensors
are deployed, as shown in Table 5. We observe that the optimal number
of clusters increases with the increased number of sensors and area.
For example, it is observed that 58 clusters are needed for a million
sensors and no clustering for 100 sensors, and M = 100 m, when the
distance from the CH to the sink node or base station is 145 m when
E
fs
= 10 pJ/bit/m
2
is maintained. We also observe that network
lifetime can be increased up to 7.3337 years for a million sensors by
Progress In Electromagnetics Research B, Vol. 12, 2009 291
assuming that the minimum required number of bits to be transmitted
is 200 per round. We cannot decrease the number of bits of the data
to be transmitted, as the number of sensors grow, or sensors will fail
to transmit the information. Depending on the application we may
increase the sensor sleep time instead of reducing the number of bits.
Table 5. Optimal clusters (C
opt
) with square root of physical area
(M) and number of sensors (N
s
), when E
fs
= 10 pJ/bit/m
2
.
M (Ns) Number of Sensors
[m] 100 400 800 10 10
3
10 10
4
10 10
5
100 1 2 2 6 19 58
200 1 3 4 12 37 116
300 1 4 5 18 55 174
400 2 5 7 24 74 232
500 2 6 9 29 92 290
6. DISCUSSION
LEACH is a hierachical routing protocol which used cluster based
approach in wireless sensor network. It is an example that directly
extends the cellular TDMA model to sensor networks [36]. It uses
rotation CH method in each communication round. Therefore, LEACH
forms new clusters and CHs in each round. These include energy
dissipation due new cluster formation including rotating CH, E
form
.
This is only relevant to rotating CH based protocols, such as LEACH.
By rotating cluster heads, LEACH distributes the energy load among
all the nodes, so that the networks lifetime is increased. Unfortunately
rotating cluster heads in every communication round dissipates battery
energy unnecessarily.
To dene a network scenario in which a particular choice of sensor
nodes are CHs one in each cluster. Let S be the set of network
scenarios. As there are n
j
+1 sensors in a cluster j, the total number of
network scenarios (the cardinality of S) is given by [S[ =

k
j=1
(n
j
+1).
Let d
j
(s) be the distance from the CH
j
to the next CH (or sink) in
a network scenario s (s S). For each s S and each cluster j, let
E
CH
(s
j
) be the energy consumed by CH
j
and E
N
(s
ij
) be the energy
consumed by sensor i in cluster j, both in scenario s. Replacing d
j
with d
j
(s) in (5) and (6) enables us to obtain E
N
(ij) and E
CH
(j)
values respectively. Let T
s
be the proportion of time the system spends
292 Halgamuge et al.
in network scenario s, s S. The average energy consumption of
any sensor in cluster j in such a LEACH-type protocol per round is
estimated by
E
L
(j) =
1
(n
j
+ 1)
|S|

s=1
T
s
_
E
CH
(s
j
) +
n
j

i=1
E
N
(s
ij
) + E
form
_
, (17)
and the total energy consumed by the entire network per round is given
by
E
tot
L
=
k

j=1
E
L
(j).
This gives rise to many interesting questions of how to optimize
the T
s
values and the sleep and active times to maximize network
lifetime.
7. CONCLUSION
In this paper, we have proposed a new, realistic and comprehensive
energy model for wireless sensor networks. The energy consumption
between dierent sources in the considered set up of a sensor node
were analyzed. The results indicate that, simple energy models over-
estimate the real sensor node lifetime. We also have applied our
model to a LEACH-type protocol to obtain an accurate evaluation
of the energy consumption and node lifetime. The paper inspires new
interesting and useful avenues to maximize sensor network lifetime.
Energy consumption by Heinzelman et al. [11] is over-estimated life of
a sensor node by 5158%, Zhu and Papavassiliou [15] by 3241%, and
Mille and Vaidya [22] by 3035%.
We have concluded that the number of clusters does not play
signicant role for moderate size sensor networks if the free space
fading energy is low. For large networks, on the other hand, cluster
optimization is still important even if free space fading energy is low.
Moreover, we have shown that the optimal number of clusters is very
sensitive to the energy model used. We observe that over-estimation
of the last node death is 30.1% when the number of the sensor nodes is
plotted against the number of rounds (time). This paper also provides
an estimation of the number of sensor nodes needed to design a network
for a given lifetime, with all the important factors that inuence to the
life expectancy of sensor networks.
Progress In Electromagnetics Research B, Vol. 12, 2009 293
REFERENCES
1. Moustafa, K. and K. F. A. Hussein, Performance evaluation of
separated aperture sensor GPR system for land mine detection,
Progress In Electromagnetics Research, PIER 72, 2137, 2007.
2. Chan, Y. K. K., B.-K. Chung, and H.-T. Chuah Transmitter
and receiver design of an experimental airborne synthetic aperture
radar sensor, Progress In Electromagnetics Research, PIER 49,
203218, 2004.
3. Callaway, E. H., Wireless Sensor Networks, Architectures and
Protocols, Auerbach Publications, 2004.
4. De Swiet, T. M., An R.F. sensor for logging-while-drilling geo-
physical measurements, Progress In Electromagnetics Research,
PIER 17, 124, 1997.
5. Chen, F., Simulation of wireless sensor nodes using S-
MAC, Masters thesis, Department of Computer Science,
University of Erlangen-Neuremberg, Sept. 2005, [online]
http://dcg.ethz.ch/theses/ss05/mics-embedding-report.pdf.
6. Boulis, A., S. Ganeriwal, and M. B. Srivastava, Aggregation in
sensor networks: An energy-accuracy trade-o, Proc. Int. Sensor
Network Protocols and Applications, 128138, 2003.
7. Cayirci, E., Data aggregation and dilution by modulus addressing
in wireless sensor networks, IEEE Commun. Lett., Vol. 7, No. 8,
355357, Aug. 2003.
8. Sankarasubramaniam, Y., I. F. Akyildiz, and S. W. McLaughlin,
Energy eciency based packet size optimization in wireless
sensor networks, Proc. IEEE Int. Sensor Network Protocols and
Applications Conf., 18, 2003.
9. Zou, Y. and K. Chakrabarty, Target localization based on energy
considerations in distributed sensor networks, Proc. IEEE Int.
Sensor Network Protocols and Applications Conf., 5158, May
2003.
10. Goldsmith, A. J. and S. B. Wicker, Design challenges for energy-
constrained ad hoc wireless networks, IEEE Wireless Commun.,
Vol. 9, No. 4, 827, Aug. 2002.
11. Heinzelman, W. R., A. Chandrakasan, and H. Balakrishnan, An
application-specic protocol architecture for wireless microsensor
networks, IEEE Tran. on Wireless Comm., Vol. 1, No. 4, 660
670, Oct. 2002.
12. Intanagonwiwat, C., R. Govindan, and D. Estrin, Directed
diusion: A scalable and robust communication paradigm for
sensor networks, Tech. Rep. 00-732, University of Southern
294 Halgamuge et al.
California, Los Angeles, 2000.
13. Ye, W., J. Heidemann, and D. Estrin, An energy-ecient MAC
protocol for wireless sensor networks, Proc. IEEE INFOCOM
Conf., Vol. 3, 15671576, June 2002.
14. Lim, Y., H. Seo, and B. G. Lee, Lifetime maximization under
data convergence latency constraint in wireless sensor networks,
Proc. IEEE GLOBECOM Conf., Vol. 5, 31873192, Nov.Dec.
2004.
15. Zhu, J. and S. Papavassiliou, On the energy-ecient organization
and the lifetime of multi-hop sensor networks, IEEE Commun.
Lett., Vol. 7, No. 11, 537539, Nov. 2003.
16. Schurgers, C., V. Tsiatsis, S. Ganeriwa, and M. Srivastava,
Optimizing sensor networks in the energy-latency-density design
space, IEEE Trans. Mobile Computing, Vol. 1, No. 1, 7080, Mar.
2002.
17. Min, R. and A. Chandrakasan, Energy-ecient communication
for ad-hoc wireless sensor networks, Proc. 35th Signals, Systems
and Computers Asilomar Conf., Vol. 1, 139143, Nov. 2001.
18. Shah, R. C. and J. M. Rabaey, Energy aware routing for low
energy ad hoc sensor networks, Proc. IEEE WCNC Conf., Vol. 1,
350355, Mar. 2002.
19. Ghosh, S. and A. Chakrabarty, Performance analysis of emi
sensor in dierent test sites with dierent wave impedances,
Progress In Electromagnetics Research, PIER 62, 127142, 2006.
20. Halgamuge, M. N., R. Kotagiri, and M. Zukerman, High powered
cluster heads for extending sensor network lifetime, Proc. IEEE
ISSPIT Conf., 6469, Vancouver, Canada, Aug. 2006.
21. Zhang, H. and J. C. Hou, On the upper bound of & alpha
-lifetime for large sensor networks, ACM Trans. on Sensor
Networks, Vol. 1, No. 2, 272300, 2005.
22. Mille, M. J. and N. H. Vaidya, A mac protocol to reduce sensor
network energy consumption using a wakeup radio, IEEE Trans.
Mobile Computing, Vol. 4, No. 3, 228242, May 2005.
23. Soro, S. and W. B. Heinzelman, Prolonging the lifetime of
wireless sensor networks via unequal clustering, Proc. IEEE Int.
Parallel and Distributed Processing Symposium Conf., 236b, Apr.
2005.
24. Dong, Q., Maximizing system lifetime in wireless sensor
networks, Proc. IEEE Int. Symp. Info. Processing in Sensor
Networks Conf., 1319, Apr. 2005.
25. Heinzelman, W. R., A. Chandrakasan, and H. Balakrishnan,
Progress In Electromagnetics Research B, Vol. 12, 2009 295
Energy-ecient communication protocol for wireless microsensor
networks, Proc. 33rd Annu. Int. Hawaii System Sciences Conf.,
Vol. 2, 110, Jan. 2000.
26. Skadas, S., K. Saleem, M. Halpern, W. Qiu, and H. Gan,
Wireless sensor networks, Lecture Notes, University of
Melbourne, Australia, 2005.
27. Wang, A. and A. Chandrakasan, Energy-ecient DSPs for
wireless sensor networks, IEEE Signal Processing Mag., Vol. 19,
No. 4, 6878, July 2002.
28. Rappaport, T. S., Wireless Communications: Principles and
Practice, Prentice Hall, New Jersey, 1996.
29. Smaragdakis, G., I. Matta, and A. Bestavros, SEP: A stable
election protocol for clustered heterogeneous wireless sensor
networks, Proc. 2nd Int. Workshop on Sensor and Actuator
Network Protocols and Applications, SANPA 04, Boston, Aug.
2004.
30. Heinzelman, W. R., A. Sinha, A. Wang, and A. P. Chandrakasan,
Energy-scalable algorithms and protocols for wireless microsen-
sor networks, Proc. IEEE Acoustics, Speech and Signal Process-
ing Conf., Vol. 6, 37223725, June 2000.
31. Chipcon, CC1000 datasheet, Aug. 2005, [on-
line]http://www.chipcon. com/lies/CC1000-Data-Sheet-2-1.pdf
32. MICA2 mote datasheet, 2004, [online] http://www.xbow.com/
Products/Product-pdf-les/Wireless-pdf/MICA2-Datasheet.pdf
33. Shnayder, V., M. Hempstead, B. Chen, G. W. Allen, and
M. Welsh, Simulating the power consumption of large-scale
sensor network applications, Proc. 2nd Int. Embedded Networked
Sensor Systems Conf. SenSys 04, 188200, NY, USA, 2004.
34. Rahimi, M., H. Shah, G. S. Sukhatme, J. Heideman, and
D. Estrin, Studying the feasibility of energy harvesting in a
mobile sensor network, Proc. IEEE Int. Robotics and Automation
Conf. ICRA 04, Vol. 1, 1924, Sep. 2003.
35. Raghunathan, V., C. Schurgers, S. Park, and M. B. Srivastava,
Energy-aware wireless microsensor networks, IEEE Signal
Processing Mag., Vol. 19, No. 2, 4050, Mar. 2002.
36. Ye, W. and J. Heidemann, Medium access control in wireless
sensor networks, USC/ISI TECHNICAL REPORT, Tech. Rep.
ISI-TR-580, Oct. 2003.