Performance Evaluation of Routing Schemes in Wireless Sensor
Networks for Train Monitoring
Oussama Drissi
1
, Adel Omar Dahmane
1
and Tayeb Medjeldi
2
1
Microsystems and Telecommunications Laboratory, Universite du Quebec a Trois-Rivieres, Trois-Rivieres, QC, Canada
2
College Center for the transfer of telecommunication technology, Trois-Rivieres, QC, Canada
Keywords: Routing Scheme, Wireless Sensor Network (WSN), Mesh Network, WI-FI, Power Consumption, Multi-tier
Architecture, Multi Hop, Shadowing.
Abstract: Wireless sensor technologies offer new opportunities in different applications thanks to the great
technological progress in the development of smart sensors, powerful processors and wireless
communication protocols. In this paper, performance evaluation of two network topologies based on routing
strategies for train monitoring has been conducted in a realistic mesh sensing system. Results conducted in
NS2 using Mannasim extension show that Multi-tier multi-hop topology outperforms the classic multi-hop
topology in terms of end-to-end delay, throughput and residual energy level.
1 INTRODUCTION
Today, wireless sensor networks attract interest in
both industrial and research community network.
The technical developments in micro-electro-
mechanical systems (MEMS) and wireless
communications allow the realization of wireless
sensor networks with a large number of sensor nodes
at low cost (Kiziroglou et al., 2011). The wireless
sensor networks are composed of nodes with limited
power and processing. They can be deployed
quickly in sensitive and inaccessible areas. Their
mission is most often to monitor an area, take
regular measurements and to trace alarms to certain
nodes of the network called sink, capable of relaying
information on a large scale ( Akyildiz et al., 2007).
Many WSN applications are emerging in areas as
diverse as defense, security, health, agriculture,
smart homes, and transportations. For example, the
rail network needs to improve services to its trains to
satisfy the customer expectations and to deal with
the increased demand for railway services.
Therefore, WSN is a good choice to ensure a
reliable, secure and comfortable service by sensing a
set of parameters in each wagon such as
temperature, acceleration and humidity (Viani et al.,
2010).
WSN can be deployed with a star topology
where PAN coordinators are needed; or with a mesh
network where the network is self-formed and self-
healed. Although many applications involve WSN,
they have to overcome several constraints (Wang et
al., 2011) including end-to-end delay, throughput,
power consumption and number of hops. The latter
becomes a serious threat to the deployment of WSN
in trains. In (
Mahasukhon et al., 2010), authors have
proposed a new protocol scheme based on muti-tier.
They have reduced the number of hops by dividing
the train wagons into several small multi-hop
segments based on ZigBee. These segments were
connected through Wi-Fi. However, in their study,
they have considered only one sensor per wagon.
Moreover, the WSN chosen is based on ZigBee with
the star topology where a PAN coordinator is
mandatory. The use of PAN coordinator is not
recommended for energy constrained applications.
In this case, mesh networks are preferred. In this
paper, we consider data transmission from sensor
nodes placed in wagons to the sink located in the
train headboard via mesh network. We propose to
use chain-topology multi-hop wireless sensor
networks. This work implements this approach using
AODV routing protocol in a realistic shadowed
environment. The performance study will focus on
reducing the number of hops to the sink and
compare results between the routing schemes aboard
a train using NS2 and Mannasim extension.
The remainder of this paper is structured as
follows: In Section 2, we introduce the main WSN
82
Drissi O., Omar Dahmane A. and Medjeldi T..
Performance Evaluation of Routing Schemes in Wireless Sensor Networks for Train Monitoring.
DOI: 10.5220/0004697200820086
In Proceedings of the 3rd International Conference on Sensor Networks (SENSORNETS-2014), pages 82-86
ISBN: 978-989-758-001-7
Copyright
c
2014 SCITEPRESS (Science and Technology Publications, Lda.)
routing schemes that are considered in this paper.
The simulation environment and setup are presented
in section 3. Then, in Section 4, performance metrics
and comparison results are presented. Finally,
conclusions are drawn in section 5.
2 SYSTEM MODEL
2.1 Network Topology
The design of protocols for routing information of a
WSN is closely related to the network topology
considered. In this paper, two network topologies
are presented in order to study the impact of
reducing the number of hops in the overall
performance of the network for train monitoring.
2.1.1 Classic Multi-Hop Network Topology
Our model consists on 37 nodes distributed as
illustrated in Figure 1 on a square area of 100 x 100
m². We consider a train that contains 9 wagons and a
headboard that contains the sink node. Each wagon
contains 4 sensor nodes. The nodes sense,
continuously, a chosen parameter. Then, it sends
data to the sink via multi-hop. Consequently, the
network has 36 similar nodes and we consider that
the sink node has ten times more energy level than
simple nodes. The mesh network formed by the 37
nodes is based on the IEEE 802.15.4 standard (IEEE
SA Standards Board, 2003) for the physical and
MAC (Medium Access Control) layers to ensure low
cost and low energy consumption.
Figure 1: Classic multi hop network topology.
2.1.2 Multi-Tier Multi-Hop Network
Topology
In this section, we adopt the Wi-Fi to form an upper
layer multi hop network. Thus, we add 3 access
points to the previous topology as shown below. We
consider that these access points are powered by the
train on the same square area of 100 x 100 m². In
real deployment of such scheme, existing Wi-Fi
access points in modern trains can be used. The total
number of nodes is then 40 nodes.
Figure 2: Classic multi hop network topology.
2.2 Routing Protocol
Routing protocols in WSN are influenced by energy
consumption constraint. The sensors use their energy
for the purpose of data processing and transmission.
The lifetime of a sensor depends mainly on its
battery. Sensor node failure can change significantly
the network topology and can impose a costly
reorganization of the latter. In this work, AODV (Ad
hoc On Demand Distance Vector) routing protocol is
considered. AODV is a distance vector protocol
(Shuangyin et al., 2012). It uses sequence numbers
to avoid routing loops and to indicate the new paths
to the destination node.
An entry in the routing table contains essentially
the address of the destination, the address of the next
node, the distance in number of hops and the
destination sequence number. AODV path discovery
process begins when a node send a RREQ packet
(Route request message), which is relayed by
intermediate nodes until the destination responds
with a RREP (Route reply). The routing tables of
nodes are updated after each retransmission of the
RREQ and RREP messages.
Although AODV is considered in this paper,
other routing protocols can be evaluated using the
same network model.
3 SIMULATION SETUP
To evaluate the performance of the two network
topologies, we have used NS2 along with
Mannassim extension.
Mannassim allows extending the functions of
NS2 by adding new modules to the design,
development and analysis of various WSN
applications such as detection of temperature
PerformanceEvaluationofRoutingSchemesinWirelessSensorNetworksforTrainMonitoring
83
change, battery model and radio propagation model
(The Manna Research Group, 2006). Table 1
summarizes the simulation parameters.
Table 1: Global simulation Parameters.
Parameter Value
Channel type
Channel/Wireless
channel
Radio-propagation
model
Shadowing
visibility
Network interface type Phy/WirelessPhy
MAC type
Mac/802.11,
802.15.4
Antenna
Antenna/Omni
Antenna
Energy model EnergyModel/Battery
Interface queue Type Queue/Drop Tail
Initial energy level (J) 10
Area (m*m) 100*100
3.1 Radio Propagation Model
A new radio propagation model called shadowing
visibility is provided by Mannassin that changes the
shadowing parameters depending on the visibility
between sensor nodes. If both nodes are in line of
sight, this model uses a good propagation paradigm.
Otherwise, it changes the shadowing parameters
to ensure bad propagation paradigm (Rhattoy et al.,
2012). Thus, shadowing visibility switches between
two paradigms depending on the visibility between
sensor nodes which allows a more accurate
simulation model to represent a realistic
environment as illustrated in table 2.
3.2 Node Configuration
A sensor node ensures temperature change sensing,
processing task and data generation. The creation of
an application layer using Mannasim, that detects
the change in temperature and generates data to the
sink, allows accurate results at the expense of a more
complex node configuration. Table 3 shows all
node’s parameters that are possible to configure
using Mica 2 mote setup.
Table 2: Radio propagation model parameters.
Shadowing paradigm Parameter Value
Good propagation
conditions
Path loss exponent 2.0
Deviation (dB) 3.0
Close-in reference
distance
30.0
Bad propagation
conditions
Path loss exponent 3.0
Deviation (dB) 5.0
Close-in reference
distance
1.0
Table 3: Full node configuration.
Parameter Value
Sensing type Continuous/On demand
Disseminating interval (s) 2.0
Reception power (J) 0.024
Transmission power (J) 0.036
Sensing power (J) 0.015
Processing power (J) 0.024
4 RESULTS
To evaluate the performance of network topologies,
we have used AODV routing protocol in a realistic
environment by exploiting shadowing visibility
radio propagation model.
We assume that all nodes have a fixed position
during the period of simulation using the parameters
mentioned in the previous tables. The performance
comparison between the two topologies is based on
the following metrics:
End-to-end delay: is the time spent by a
packet to travel across the network from source to
destination
SENSORNETS2014-InternationalConferenceonSensorNetworks
84
Throughput: is the average rate of successful
message delivery over a communication channel.
These data may be delivered over a physical or
logical link, or pass through a certain network node.
The Residual Energy: is the level of the
residual energy of nodes relative to the initial energy
during the simulation.
4.1 End-to-end Delay
Figure 3 reports the end-to-end delay of traffic
generated by sensing temperature changes at nodes
obtained by both topologies. The sensing type is
continuous and the aggregated data is then sent to
the sink at a regular interval of 2 s.
Figure 3: End-to-end Delay.
The end-to-end delay of the classic multi hop
approach reached 1.9 s during the period of path
research in a shadowed environment compared to
about 2.4 s for the multi-tier multi-hop topology.
This result was expected since the latter comes
with the expenses of an extra setup time due to the
upper layer. However, as shown in figure 3, this
approach outperforms the classic multi hop scheme
on the end-to-end average delay once the network
reaches the state of convergence.
4.2 Throughput
Figure 4 shows the network throughput for each
topology. We notice that the instant network
throughput of multi-tier multi-hop routing scheme in
shadowed environment is about 5 Kbits while the
classic multi-hop routing scheme gives a lower
instant throughput.
Figure 4: Network throughput.
4.3 Residual Energy Level
All nodes in the network simulation start by initial
energy level equal to 10 (J). Each node in the
network will consume energy to ensure sensing task,
data processing and communication with other
sensor nodes. To compute the energy expended for
each sent or received data, we used the model of
radio power dissipation proposed in table 3. Figure 5
shows the residual energy of sensor node for each
simulation scenario.
Figure 5: Residual energy of sensor node.
In a shadowed environment, the use of Wi-Fi
network to reduce the number of hops along the train
from the source node to the sink was efficient in
term of power conservation. Consequently, a sensor
node will be alive for a longer time compared to the
classic multi hop routing scheme.
PerformanceEvaluationofRoutingSchemesinWirelessSensorNetworksforTrainMonitoring
85
4.4 Summary of the Comparison
between Routing Schemes
The proposed multi-tier multi-hop routing scheme
provides better performances in terms of average
throughput and end-to-end average delay in realistic
environment as summarized in table
4. Moreover,
this approach extends the network lifetime to ensure
sensing tasks for a longer duration as shown in
Figure 6. Indeed, simulation results show that the
multi-tier multi-hop routing scheme keeps 50 % of
sensor nodes alive for duration three times more
than the classic multi hop scheme.
Table 4: Comparison between routing schemes.
Average
Throughput
[kbps]
End-to-end
average delay
(ms)
Multi-tier multi-
hop routing
scheme
5.06 29.03
Classic multi-hop
routing scheme
3.11 312.80
Figure 6: Network lifetime.
5 CONCLUSIONS
In this paper, performance evaluation of routing
schemes using AODV routing protocol in a realistic
environment for train monitoring has been
conducted. Multi-tier multi-hop routing scheme, by
reducing the number of hops, has been shown better
performance in terms of end-to-end delay,
throughput and the residual energy of sensor node.
The use of Wi-Fi to divide the train into three
parts has allowed reaching a network lifetime three
times longer than the classic multi-hop scheme. In
future work, AODV and other routing protocol
techniques will be investigated when mobile nodes
are present in the mesh network.
REFERENCES
Kiziroglou, H. C., Michail, E., Yates, D. C., Yeatman, E.
M, 2011. A MEMS Self-Powered Sensor and RF
Transmission Platform for WSN Nodes, Sensors
Journal. IEEE.
Akyildiz, I., Su, W., Sankarasubramaniam, Y., and
Chowdury, K. R., 2007. Wireless multimedia sensor
networks: A survey. Wireless Communications. IEEE.
Viani, F., Oliveri, G., Donelli, M., Lizzi, L., 2010. WSN-
based solutions for security and surveillance, Wireless
Technology Conference (EuWIT). IEEE.
Wang, Y., Li, X.Y., Song, W.Z, Huang, M., and Dahlberg,
T. A., 2011. Energyefficient localized routing in
random multihop wireless networks, Parallel and
Distributed Systems, IEEE Transactions on. IEEE.
Mahasukhon, P., Sharif, H., Hempel, M., Zhou, T., 2010.
Multi-tier Multi-hop Routing in Large-scale Wireless
Sensor Networks for Real-time Monitoring, Sensors,
IEEE.
IEEE SA Standards Board, 2003. IEEE 802.15.4, Part
15.4: Wireless Medium Access Control (MAC) and
Physical Layer (PHY) Specifications for Low Rate
WirelessPersonal Area Networks (LR WPANs), IEEE.
Shuangyin, R., Huanmin, H., Baoliang, L., Jia, L., Cgao
P., Wenhua D., 2012. An Improved Wireless Sensor
Networks Routing Protocol Based on AODV,
Computer and Information Technology (CIT), 2012
IEEE 12th International Conference on. IEEE.
The Manna Research Group, 2006. Mannasim Framework
Classes Manual, Sensornet project. Mannasim.
Rhattoy, A., Zatni, A., 2012. The Impact of Radio
Propagation Models on Ad Hoc Networks
Performances, Journal of Computer Science. Science
Publications.
SENSORNETS2014-InternationalConferenceonSensorNetworks
86
