You are on page 1of 4

31

INVESTIGATION OF ENERGY EFFICIENCY FOR


SINGLEHOP AND MULTIHOP ROUTING SHCEMES IN A
WIRELESS SENSOR NETWORK
D. A. Khan and X.-H. Peng
Electronic Eng., School of Eng. & App. Eng., Aston University, UK
dkukl9@hotmnai1.corand x-h.peng@aston.ac.uk

Keywords: Wireless sensor networks, energy efficiency. nodes and the sink node. Furthermore, energy efficiency of
the network is determined by the total amount of energy
Abstract dissipated at all the sensor nodes involved versus a certain
amount of data successfully received at the sink node during
Wireless sensor nodes run on small battery power and it is the same period of time.
sometimes impossible to replace the batteries due to the
environment and cost factors associated with them. Therefore, The motivation behind this work is to investigate data
energy efficiency is a critical issue in the design and dissemination via various routing schemes in a WSN and to
management of a wireless sensor network. In this paper, find how the routing schemes and other factors affect the
following the configuration and protocol implementation network performance in terms of energy efficiency. The
within a wireless sensor network, energy efficiencies for investigation mainly concentrates on energy consumption and
different routing schemes have been analysed based on the network throughput for both singlehop and multihop routing
testing results on both energy consumption and data yield techniques. Based on different network configurations and
(throughput) for the whole network. The real world tests show testing results, the energy efficiency for both cases will be
that for a dense deployment multihop transmission is more analysed and compared.
energy efficient compared to singlehop transmission, while
the opposite may be true when the sensors are deployed There has been extensive research in this area but most of the
sparsely. work is simulation oriented. Our approach is based on real
world tests giving realistic accounts on all practical issues
such as electronic and RF circuitry, noise, attenuation, multi-
1 Introduction path, etc.
The wireless sensor network technology is emerging as one of
the very attractive research areas, and advanced electronic 2 System setup
design and cost reduction have made the future of this
technology very bright. A wireless sensor network (WSN) is A WSN is set up in a laboratory environment for the
composed of numerous small independent sensor nodes and investigation. The system operated by TinyOS [7] is
has many applications such as in military, environment and comprised of three remote sensor nodes (or source nodes) and
health monitoring. There are, however, significant constraints one sink node, which are configured to a certain network
on resources available for WSNs, compared to the wired topology and in line of sight to each other. In our
networks. In particular, wireless sensor nodes can only have a experiments, all sensors are sensing the environment
limited power source, and their working duration largely constantly at a fixed rate and therefore the sensor nodes
relies on battery lifetime. Therefore the key challenge here is always have data to send to the sink node.
how to efficiently disseminate the data from remote sensor
nodes to the sink node through a network that is limited in For measuring purposes each sensor node in the network is
power and prone to failures. connected to two mulitmeters, one measuring the current
flowing through the battery and the other measuring the
There are a variety of routing algorithms [1-6] suggested for voltage across it. The throughput readings are taken at the
this environment, and data can be disseminated through sink node using the surge view graphical interface [8]. All the
singlehop (or called direct communication) or multihop sensor nodes are kept in the same environmental conditions,
depending on the scenarios given. Each of these routing although they may need to be configured in different routing
protocols has its energy consumption characteristics, which in schemes, to minimise experimental errors.
general relate to the operations in radio electronics and
amplifier units, and the distance between the remote sensor Two main configurations used for testing:
32

(1) Singlehop: All the sensor nodes send data directly to signal spreading; whereas camp is the amplifier energy
the sink node, e.g. node 1, node 2, and node 3 e sink dissipation expended by the transmitter in achieving an
node (node 0), respectively, as shown in Fig. 1(a). acceptable SNR ratio. d is the distance between the sensor
(2) Multihop: Sensor nodes follow a multihop path to send node and the sink node. Typical values for Eec and camps
their data to the sink node, e.g. node 3 node 2-node given in the first order radio model [9] are Eelec =5OnJ/bit and
1 - node 0, as showninFig.l(b). Camp = 10 pJ/bit/m2. However, the energy consumption
measured in our experiments refers to the total consumption
by the sensor nodes including radio transmission and
1 o reception,
2 sensing, signal processing, and component &
s*i ash W2 **4a
@ *-- t I1 -4BB} Wm v1Ar.4-1 Wm
~~~~~~~~~~~~~circuit
maintenance.
For a WSN having n sensor nodes, the energy consumption of
the system for the singlehop scheme is given by

2v ', 3 Esingle =n[l (Eelec +amp d +Eother] (2)

.............5... M i W
i
where Eother is the other energy consumption per node in
addition to the energy used for data transmission and
0 reception. For the multihop scheme, there are n-I
/S > intermediate sensor nodes which operate as both the
1V|7 transmitter and the receiver. Assuming they are aligned in one
dimension and separated with an equal distance r, we then
2 have
3 -EmuIti = n[l * (Eelec + camp * r2) + Eother] + (n-1) * I Eelec (3)
2.2 Testing results
Fig. 1(a): Singlehop configuration
Based on the networks set up with different routing schemes,
field tests are carried out by monitoring the energy
o consumption (covering all types of operation) at each of the
- tE ~ a>;t sensor nodes for both singlehop and multihop configurations.
The overall energy consumption for the whole network is
2 EvEg iilEthen obtained y adding the consumption of all individual
nodes at each measuring point. At the same time, system
throughput is measured at the sink node by counting the total
amount of data successfully received per unit time by the
sink, which are contributed individually (in the singlehop
configuration) or collaboratively (in the multihop
configuration) by all the sensor nodes involved.
Fig. 1(b): Multihop configuration

2.1 Energy consumption model 60

In a WSN, the general energy consumption models for a


0 inglehop
sensor node to transmit and receive an /-bit message are , UMultihop
given, respectively, as below:

ET eE e amp
_l*d2
+E 818 1
()
40

30 60 90 120
elec
~~~~~~~~~~~~~~~~~~~~~~~~~~~Time (sec)

Here EeIeC is the electronics energy dissipation for


transmitting or receiving a bit, depending on some functionFi.2a:Eegcosmtncmpro:
features such as digital coding, modulation, flutering, anduigeo s uthp( m
33

Testing results show that in a dense deployment (r is


relatively small) the total energy consumption of the network
in multihop configuration is slightly more than that in 155
singlehop configuration, as shown in Fig.2(a). But, for a 1.4
larger r multihop configuration consumes considerably less
energy than singlehop configuration does, as shown in 1.25
Fig.2(b). This is because when distance increases nodes in X
signlehop will raise their transmitting power in order to
maintain the signal strength. This increase will offset the E 095 - Singlehop
advantages of the singlehop configuration in which sensor M ultihop
nodes do not receive data and have fewer operations in data 0.8
processing. 10 20 30 40 50 60 70 80 90 100 110 120
Time (sec)

Fig.3(b): Throughput comparison:


70 singlehop vs. multihop (r 4m).
2465
0 inglehop
60 mEMutihop 3 Analysis and discussion
55
_________________Energy efficiency, also called energy cost per bit, is a key
30
50
60 90 120 factor in evaluating the performance of a WSN. In our work,
Time (sec) this measurement takes accounts for both energy consumption
and throughput of the network, which is calculated as the total
energy consumed across the network for a period of time
Figure 2(b): Energy consumption comparison: observed versus the total amount of data accepted at the sink
singlehop vs. multihop (r= 4m) node during the same period of time. The sample values of
energy cost per bit for both singlehop and multihop
For the throughput, as shown in Fig.3(a)&(b), it is on more or configurations and for different r are given in Fig.4(a)&4(b),
less the same scale in the singlehop configuration for both c tively
respectively.
cases: r=2m and r=4m. However, it varies quite substantially
in the multihop configuration, which is significantly lower
than that in the singlehop configuration when r=4m although 164
slightly higher than the singlehop configuration when r=2m. 162
This mainly due to the sensor nodes in the multihop 160
configuration that need to wait for their parent nodes to relay e 158
their messages. 156
154

Each intermediate node will spend certain time to help other ^ 152
nodes by receiving and forwarding their messages while 150 Slfglehop
processing and sending its own data. This leads to the 148
146
Mutihop
reduction of the number of packets received at the sink node 144
per unit time. 30 60 90 120
Time (sec)

1.65 Fig.4(a): Energy cost per bit comparison:


singlehop vs. multihop (r =2m).

>
1.5 ;As we can see,
singlehop when
multihop is more energy efficient than
r=2m, since multihop has the throughput
advantage here although its energy consumption is slightly
° 1.35 higher, compared to singlehop. However, singlehop is better
3~ ~ ~~- SinglehOp on the same account than multihop when r=4m. Even though
12 1.2 Multihlop singlehop consumes
~~~~~~~~~~~for r,
a larger significantly
multihop has muchmorelower
energy than multihop
throughput than
10 20 30 40 50 60 70 80 90 100 110 120 ..
'
Time (sec)
_________________________________________ ~~~singlehop in this case. Because of lower energy consumption,
nodes employed in the multihop configuration can operate for
Fig.3(a): Throughput comparison: a longer period than those in the singlehop configuration
singlehop vs. multihop (r= 2m). under the loose deployment of sensor nodes (large r). This is
34

particularly useful in some applications where lifetime is a [2] Jamil Ibriq, Imad Mahgoub, "Cluster-Based Routing in
more dominant factor than throughput. Wireless Sensor Networks: Issues and Challenges," in
Proc. of the 2004 Symp. on Performance Evaluation of
Computer Telecomm. Systems, pp.759-766, Sept. 2004.
265 [3] M. Ettus, "System capacity, latency, and power
consumption in multihop-routed SS-CDMA wireless
245 X Siglehop
S'Mglehop

networks", in Proc. Radio and Wireless Conf. Colorado


Multihop Springs, USA, pp.55-58, Aug. 1998.
= 225 [4] S. Paark and M. Srivastava, "Power aware routing in
Et205 sensor networks using dynamic source routing", ACM
MONET Special Issue on Energy Conserving Protocols
18
__________ =in Wireless Networks, 1999.
W15 , [5] J.N. Al-Karaki and A.E. Karmal, "Routing techniques in
165 F wireless sensor networks: a survey", Wireless
30 60 90 120 Communications, vol. 1 1, pp.6-28, Dec. 2004.
Time (sec) [6] W.B. Heinzelman, P. Chandrakasan, H. Balakrishnan,
"An application-specific protocol architecture for
wireless microsensor networks", IEEE Trans. on Wireles
Fig.4(b): Energy cost per bit comparison:
singlehop vs. multihop (r 4m).= Communication, Vol. 1, pp.660-669, No.4, Oct. 2002.
[7] P. Levis, S. Madden, D. Gay, J. Polastre, R. Szewczyk,
A. Woo ' E. Brewer ' and D. Culler "The emergence of
In general, energy efficiency of a WSN depends on the
networking 'g in TinyOS", in
and techniques
network
networ configuration including
configuration including topology, the
the relative
relative Proc. of 1stabstractions
USENIX/ACM Symp. on Networked
distances of remote sensor nodes to the sink node, routing or St Dsi ImplM (NSDI 2004).
diffusion protocols, and node synchronisation techniques. The [8]TiyO Tuoia l, mhttp wnyoSD ne/tiyos
choice as to whether to use singlehop or the multihop T
./o /Tutorindex. htmt
configuration will..depend upon the requirements of certain
. resources and, ~~~~~[9]
W.B. . Heinzelman,' A. Chandrakasan, and H.
applications. Due to. the . constraints. in . '
experimental conditions, this work is caffied out using a Balakrishnan, "Energy-efficient communication
limited number of sensor nodes. It would be desirable that a protocaols for wireless microsensor networks", in Proc.
sensor network with considerably more nodes can be tested
Hawaii
of IEEEHawaii
MauI 2000.on System Science, vol 8,
U Int.n.Conf.
Mau Haai S,Jn00
practically in the future investigations.practically intefturinvs[10] A. Manjeeshwar and D. Agrawal, "TEEN: A protocol
for enhanced efficiency in WSNs", in Proc. of Pt Int.
Energy efficiency remains to be the central issue in WSNs. WorkshonP ld Dinstrib Compuin Issue
Data transmission and reception are most energy intensive Wires Networks and Mobile Computing san
operations, which need to be minimised or properly regulated Francisco, USA, April 2001.
through system and protocol designs. At the same time, radio [11]s5. Lns 'An CS. R v , a P
range and routing scheme considerations in connection to the efficie..n ang in. sensorenfrat systems", in
relative distances between the sensor nodes and between Proc. o aterospacerConf.,U a,iol 3, p i
sensor nodes and the sink node are also vital. All the 1130 M arch2002.
techniques identified in this research field should be 1130, March 2002.
developed to achieve the maximum system throughput at the
minimum energy cost.
There are many examples of the WSN design which is driven
by energy efficiency. For instance, the WSNs adopting
cluster-based routing achieve excellent performance results in
this account. LEACH [9], TEEN [10], and Chain-based 3
level PEGASIS [ 1] are three well known hierarchical
cluster-based routing protocols are. All of them take the
factors discussed above into account in their design strategies,
and show significant reduction in energy consumption by
each node over other non-clustering protocols.

References
[1] I.F. Akyildiz, W. Su, Y. Sankarasubramaniam, and E.
Cayirci, "A survey on sensor networks", IEEE
Communications Magazine, August 2002.

You might also like