Application of Fuzzy Inference Systems in the Transmission of
Wireless Sensor Networks
Pedro Henrique Gouvea Coelho, J. F. M. do Amaral, N. N. de Almeida, A. C. X. S. Fonseca
and K. P. Guimarães
State Univ. of Rio de Janeiro, FEN/DETEL, R. S. Francisco Xavier, 524/Sala 5001E, Maracanã, RJ, 20550-900,Brazil
Keywords: Fuzzy Logic Systems, Artificial Intelligence Applications, Node Routing, Wireless Sensor Networks.
Abstract: The purpose of this paper is to apply fuzzy logic techniques for determining the data transmission path in
wireless sensor networks. Wireless sensor networks are used in many applications and require efficient
operation with speedy transmission of information and long lifespan. For suitable sensor transmission, a
fuzzy system is defined considering as the goal to minimize the information transmission distance and
maximize the battery lifespan of the network routers. Case studies simulations were carried out and the
results indicate satisfactory performance of the method.
1 INTRODUCTION
Industrial wireless sensor networks are an emerging
class of wireless sensor networks that faces specic
constraints linked to the peculiarities of the
industrial production. (Akyildiz et al., 2002).
The main advantages of wireless networks are
reduction of devices installation time, no need for
cabling structure, savings in the cost of projects and
infrastructure, flexibility of devices configuration,
savings in assembly costs, flexibility in changing
architectures and the possibility of installing sensors
in places that are difficult to reach.
However, industrial wireless sensor networks face
several challenges such as the reliability and
robustness in harsh environments, as well as the
ability to properly execute and achieve the goal in
parallel with all the other industrial processes.
Besides, industrial wireless sensor networks
solutions should be versatile, simple to use and
install, long lifespan and low-cost devices – as a
matter of fact, the combination of requirements hard
to meet (Matin, 2012).
The deployment and the setup of wireless sensor are
overwhelmingly challenging tasks that become even
more challenging in industrial applications. The
environment where industrial wireless sensor
networks are deployed is extremely dynamic, it can
depend on the specic product, the phase of life of
the product and the kind of service provision
considered. Indeed, each kind of product or phase of
life has different requirements and imposes on the
monitoring system different constraints. One of the
challenges to face is the impact of the propagation
environment. When the industrial wireless sensor
network is deployed inside a factory to assess the
production process quality, one has to tackle the
interference and the radio environment produced by
the production machines. For that matter, such
network has to be deployed and calibrated not only
to guarantee the correct assessment of the production
process, but above all not to interfere with the
production process. The same logic holds for
networks used to monitor electricity, water and gas
consumption (Kumari and Prachi, 2015).
The environment in which sensors will be deployed
needs to be considered in order to find the optimal
locations for sensors. Operation lifetime, due the
power management policy, is one of the key issues
in all the wireless sensor network applications,
including industrial applications. Many industrial
wireless sensor networks applications, particularly in
the eld of environmental monitoring, require the
autonomous power supply from alternative power
sources, such as wind or solar power. Even though it
is possible to have a constant power supply in some
industrial environments, sensors tend to be battery
powered in order to keep the monitoring non-
intrusive. But, in most cases, batteries are not
expected to be reloaded or changed. So, energy
618
Coelho, P., Amaral, J., Almeida, N., Fonseca, A. and Guimarães, K.
Application of Fuzzy Inference Systems in the Transmission of Wireless Sensor Networks.
DOI: 10.5220/0006338406180624
In Proceedings of the 19th International Conference on Enterprise Information Systems (ICEIS 2017) - Volume 1, pages 618-624
ISBN: 978-989-758-247-9
Copyright © 2017 by SCITEPRESS Science and Technology Publications, Lda. All rights reserved
should be saved. There are many ways to obtain that,
both in software and hardware.
The hardware solution requires to carefully choosing
the components. These latter should be low energy
consuming while providing the needed capacity. In
some particular applications, energy harvesting
modules can be envisioned, like solar cells or
kinematic sensors, etc., but their usage is still
marginal.
The software solution focuses energy preservation
by controlling the number of messages to be sent
and the range. As a matter of fact, radio activities,
such as sending and receiving data are the activities
that consume more energy in wireless sensor
networks compared with processing and sensing
activities. Therefore, it is of paramount importance
to monitor carefully the amount of data to send and
the frequency at which it is sent, i.e. the number and
size of messages, while preserving the quality of
service expected by the application. Similarly, the
further the messages are sent the more the energy
needed and the more the interferences generated.
Thus, it is important to monitor the transmission
range based on the target to reach (Lonare and
Wahane, 2013).
Usually, mesh or tree topologies are used to
minimize these interference effects on the
transmission system. Mesh or tree topologies
transmit the message from one node to another using
other nodes (sensor and repeaters), which act as
intermediate routers, directing messages to the
others until it reaches its destination (gateway or
manager node), as shown in figure 1.
Figure 1: Mesh network characteristics.
The importance of data transmission in wireless
networks motivates research works for improvement
in routing algorithms. Some studies have already
been carried out, such as in (Alshawi et al., 2012) in
which it is proposed to include fuzzy logic
techniques to the gossip algorithm with the purpose
of reducing energy waste in data transmission. Other
researchers (Shah et al., 2015) propose a fuzzy logic
algorithm in which the main goal is to achieve fast
real-time communication. An algorithm that
performs routing through clusters and search for the
best cluster-head to maximize the number of
transmitted information is presented in (Amri et al.,
2014). The present work uses a fuzzy system to find
an efficient route for transmission between a sensor
and the gateway in order to obtain the shortest
possible time, avoid transmission failures in case
there is an inactive router, and save the battery
consumption of the routers and thus increase their
lifespan. The article is organized in six sections.
Section 2 presents a brief description of wireless
sensor networks. Section 3 does a brief discussion to
routers energy management. Section 4 describes the
proposed method. In section 5 a case study is
described and section 6 presents conclusions and
future developments.
2 WIRELESS SENSOR NETS
A Wireless Sensor Network (WSN) is an
autonomous network of smart sensors with a high
degree of cooperation among them. It is responsible
for monitoring a process or an environment,
processing the collected information, classifying the
degree of importance of the information, and
spreading it to the other sensors or routers closest to
the gateway. The sensor network consists of the
elements: sensor, observer, phenomenon, router and
gateway, as shown in figure 2.
Figure 2: Elements of a wireless sensor networks.
The sensors measure the physical quantities and
generate measurement reports by means of signal
diffusion. For each type of phenomenon, there is a
different type of sensor, which has a specific
physical characteristic of the process. The observer
is a user, or multiple users, who receive and request,
when necessary, the information of the phenomena
collected by the sensors, and broadcast by the
wireless sensor net. The phenomenon is the process
that will be monitored by the sensor element, and
Application of Fuzzy Inference Systems in the Transmission of Wireless Sensor Networks
619
diffused by the wireless sensor network for the final
evaluation of the observer.
The sensors measure the physical quantities and
generate measurement reports by means of signal
diffusion. For each type of phenomenon, there is a
different type of sensor, which presents a specific
physical characteristic of the process. The observer
is a user, or multiple users, who receive and request,
when necessary, the information of the phenomena
collected by the sensors and broadcast by the
wireless sensor network. The phenomenon is the
process that will be monitored by the sensor
element, and diffused by the wireless sensor network
for the final evaluation of the observer element. The
router directs the information generated by the
sensors to another element of the network within its
transmission radius, so that the information can
reach the gateway. The gateway is responsible for
receiving all the electromagnetic information sent by
the sensor nodes, decoding them into physical
quantities of the monitored phenomenon so that the
observer is able to understand them and trigger, for
example, an alarm if the physical magnitude is
outside of the acceptable value. WSNs are classic
examples of the so-called Low-Power and Lossy
Networks (LLNs). LLNs are made up of many
embedded devices with limited power, memory, and
processing resources (Akyildiz et al, 2002). Thus,
WSNs are classified as a type of a LLN network
with a specific purpose for remote monitoring of
physical quantities (e.g. temperature, pollution level,
pressure, interference, etc.) and can also work as
actuators on certain control elements. There is also
another important question regarding the nature of
the data used in the routing decision-making
process. Many of them are dynamic, that is, they
change their value over time. There are two
traditionally used routing algorithms: flooding and
gossiping algorithms (Chandna and Singla, 2015).
The flooding algorithm has the strategy of sending
each packet of data from one node to all nodes that
are within its reach, and so it proceeds until it
reaches the gateway. This algorithm has the
disadvantage of transmitting unnecessary data,
increasing the power consumption. The gossiping
algorithm is based on a random choice between
nodes that are in range, thus causing a delay in data
propagation.
3 ENERGY MANAGEMENT
In recent years, the number of WSNs deployments
for real-life applications has rapidly increased.
However, the energy problem remains one of the
main barriers limiting the full exploitation of this
technology. Sensor nodes are usually powered by
battery with limited capacity and even when there is
possibility of obtaining additional energy from
external environment, for example solar or piezo-
electric, it remains a limited resource to be
consumed judiciously. Among the characteristics of
the WSNs, regardless of their application, the energy
issue is the one that imposes the most restriction on
the lifespan of the network. This causes several
solutions to be presented as a way to maximize
network durability without sacrificing the system
reliability. Most of the sensor nodes have the
characteristic of being disposable, because in certain
applications of the network its maintenance is
impracticable. The physical resources of the sensor
nodes can be represented by an energy model, in the
consumption and in the interaction level with a
model of functions. Such energy model can be seen
as an energy provider for consumer elements,
through a battery with finite stored energy capacity.
Each sensor informs the level of energy available to
its provider, called an observer or sink. The energy
model consists of the following elements: battery,
radio (communication interface), processing and
sensing unit. From the power model it is possible to
obtain individual information from each sensor node
of the network and perform a survey of the power
map of the network, which in turn can be used to
make decisions about what can and cannot be done
to improve performance the network. It is important
to highlight that in this energy model, the element
that consumes the most energy is communication,
mainly in the function of transmission and routing of
the data collected by the sensing units. A sensor
node is internally divided into four main units:
i) processing unit, consisting of a
microprocessor or microcontroller;
ii) transmission unit (communication)
consisting of a short-range radio for -
wireless communication;
iii) sensing unit that connects the node to
the physical world, and consists of a
group of sensors and actuators;
iv) power unit.
It should be stressed that the path loss of the
transmitted signal between two sensor nodes may be
as high as the fourth order exponent of the distance
between them, because the antennas of the sensor
nodes are near to the ground (Akyildiz et al, 2002).
Figure 3 shows the main components of a sensor
structure.
ICEIS 2017 - 19th International Conference on Enterprise Information Systems
620
Figure 3: Sensor node structure of a WSN (Matin, 2012).
Due to the difficulty in having access to a dense
WSN in full operation to carry out simulations
regarding the energy consumption among the
sensors, some proposals of models of energy
dissipation for the sensor nodes are found in the
literature. The main objective of these models is to
characterize the energy consumption in each sensor
node in a WSN. The common models used are:
uniform dissipation model, hotspot based dissipation
model and model based on four modes of operation.
Among other authors, (Kashani and Ziafat, 2011), in
addition to being concerned with Dynamic Power
Management (DPM), present a new approach to
energy management based on a centralized adaptive
clustering protocol for self-organization of the
sensor map through neural networks, which can
group the sensor nodes based on several parameters:
available energy level and coordinates of sensor
nodes, in order to distribute the energy consumption
between the sensor nodes and extend the lifetime of
the network.
4 METHODOLOGY
For the sensors collected information be transmitted
to the gateway, it is often necessary for it to pass
through other sensors or routers until it reaches the
sink or observer, which will take an action. If a
sensor or router is not functioning properly, the
information must find another path to go to the
gateway. In addition, these sensors and routers are
usually battery operated so they have a limited
lifespan and it becomes impractical to frequently
change their power source. To extend the period of
use of the sensors it is interesting to optimize the
data transfer in order to save the unnecessary use of
routers and avoid transmission failures due to lack of
power. One of the main goals in the transmission of
information is to ensure that the data arrives at the
gateway in the shortest possible time, since each
additional router uses extra time. One way to
minimize the delay time is to use a shorter, more
efficient path in which a minimum number of
routers are used to transmit information. In order to
achieve that, a fuzzy inference system is
implemented that will determine for each sensor or
router what the next element to receive the
information should be. For this, the system uses
fuzzy inference rules that take into account the
distance that the information will travel until it
reaches the gateway and the state of conservation of
the router, in other words, if it is in operation and
has a good battery charge. The details of that
modeling are found in the next section where a case
study is elaborated to illustrate better the method.
5 CASE STUDY
In this section the simulation of a wireless routing
system is elaborated, in which the sensors measure
the physical quantities and transmit the collected
data to the gateway through a network of routers and
sensors. We used the MATLAB software and the
Fuzzy Logic toolbox as a tool to illustrate the
method for a case study. Initially an algorithm was
developed capable of generating a configuration
simulating the location of sensors / routers in an
industrial area. The sensors monitor various physical
quantities such as temperature, pressure and flow.
We consider that all sensors are also routers, so all
points can be used for routing. One hundred nodes
were distributed in a square area with 1km² and the
gateway was placed in the center of the environment
in order to avoid great distances between it and some
routers. The transmission range of each router was
defined as 200m. At first we calculate which nodes
can be reached by each of the mesh sensors / routers.
This information will be used to define which of
them will be evaluated by the fuzzy system. The
choice of nodes to be used in the transmission of the
information collected by the sensors is performed by
the fuzzy system and is done based on the battery
level and the distance between the nodes that are
within its range and the gateway. We set the node
closest to the gateway and with the highest battery
level as the priority. Thus, we will be preventing
transmission failure due to lack of energy, while at
the same time, a minimum number of components in
the process will be used in the transmission of
information. The fuzzy system has as inputs the
distance between the sensors / routers and the
gateway, only those that can receive the information,
i. e. the routers that are within reach of the
Application of Fuzzy Inference Systems in the Transmission of Wireless Sensor Networks
621
transmitter sensor, and the battery level of these. The
output of the system is a classification of the quality
of the node to receive the information. After
executing the fuzzy system, the node with the
highest grade is chosen to carry out the transmission.
Membership functions with triangular and
trapezoidal format were used, and the linguistic
variables are:
i) Distance: near / reasonable / far;
ii) Battery: low / medium / high;
iii) Output: poor / fair / good.
The membership function for an input variable is
shown in figure 4 and the membership function for
an output variable is shown in figure 5.
Figure 4: Membership function for input variable.
The rule matrix of the implemented fuzzy system is
presented in Table 1.
Table 1: Fuzzy matrix.
Battery
Distance
Low Medium High
Near Fair Good Good
Reasonable Poor Fair Fair
Far Poor Poor Fair
In order to verify the effectiveness of the proposed
fuzzy system, some simulations were performed.
Initially the sensor node where the information
would be sent from is defined. Then the fuzzy
system was executed to choose the node to which
the information would be directed. The node with
the best classification was chosen and the
information transmitted to it. After this, the fuzzy
system is executed again, using now the information
regarding this new router. This process is repeated
successively until the gateway is reached. In a first
Figure 5: Membership function for output variable.
simulation, a sensor was chosen to the left of the
gateway. The path traveled and the amount of
battery existing in the used routers can be seen in
figure 6.
Figure 6: Path followed by information collected from a
sensor on the left of the gateway.
In a second simulation a sensor was chosen in the
lower right corner. The path obtained for
transmission is shown in figure 7. The amount of
battery in the nearby routers is also indicated.
In this simulation it was observed that the
information passed through four nodes of the
environment until reaching the gateway. It is also
observed that the third chosen node was the one with
the highest level of battery among the nearby nodes.
A new simulation was performed in which the same
source sensor was maintained, but the battery of the
third node traveled in the previous path was
modified. Initially it had the value of 53% and in this
simulation a very low value of 5% was used. The
ICEIS 2017 - 19th International Conference on Enterprise Information Systems
622
new transmission path obtained can be seen in figure
8.
Figure 7: Path covered by information collected from a
sensor in the lower right corner of the gateway.
Figure 8: New path covered by information in the
simulation shown in figure 7 after changes carried out.
Because the previously used battery node is now
at a low level, it is preferable to save this router for
cases where it needs to be used. It is observed that
the new path used went through the node that had
the highest level of battery still acceptable, that is,
30%, which was also close to the gateway. A new
simulation was performed by changing the battery of
this node, which was decreased to 10%. The result
obtained is shown in figure 9.
Figure 9: New path covered by information in the
simulation shown in figure 8 after additional changes.
In this case, the nearest node with a higher battery
level, 24%, was on the left and was chosen instead
of the routes previously found. In a real system, this
process would run simultaneously on several nodes
according to the need of the plant. Therefore, the
system was run again simulating five sensors
transmitting information at the same time. The found
trajectories can be seen in figure 10.
Figure 10: Path covered by information from five sensors
transmitting concurrently.
The simulation was not performed with more
sensors to allow a better visualization of the
behavior at each node. Through the implemented
simulations it is possible to observe the proper
behavior of the proposed fuzzy system, which
focused the choice of a shorter path to the gateway,
but considered the amount of residual energy in each
router in order to extend the useful life of the
network.
Application of Fuzzy Inference Systems in the Transmission of Wireless Sensor Networks
623
6 CONCLUSIONS
This paper proposed the use of fuzzy system
techniques for use in wireless sensor networks
focusing on shortest path and energy saving for the
sensors in routing applications. Case studies of a
wireless routing system was carried out, in which
sensors measuring physical quantities were
simulated, transmitting the data collected by them to
the gateway through a network of routers and
sensors. Five simulations were presented in the
present paper in which favorable results were
achieved. In the simulations, we performed the fuzzy
inference system that determined the best route for
routing taking into account the node's reach, the
battery level of the possible nodes that could
propagate the information and the distances among
them and the gateway. The decision on which way
to go was established based on the matrix of rules
and functions of pertinence of the fuzzy inference
system presented in this work. Through these
simulations it was possible to illustrate the trajectory
of a chosen path in the transmission of the
information collected by a sensor until its reception,
in the gateway. The results show that the fuzzy
systems represent a suitable method for this
application, presenting satisfactory results, since
they chose a shorter path to the gateway considering
the amount of energy in each router, extending the
lifespan of the network. More complex networks
involving new case studies are under way as well as
a comprehensive comparison taking into account
traditional techniques and algorithms. The huge
amount of information and time needed to perform
that task did not allow the authors to finish that task.
There is sure a long way to go.
REFERENCES
Lonare, S., and Wahane, G., 2013. A Survey on Energy
Efficient Routing Protocols in Wireless Sensor
Network. In Proc. Fourth International Conference on
Computing, Communications and Networking
Technologies (ICCCNT). Tiruchengode, India.
Akyildiz, F., Su W., Sankarasubramaniam, Y. and Cayirci,
E., 2002. Wireless Sensor Networks: A Survey,
Computer Networks, 38, pp. 393-422.
Coelho, P. H. G., Amaral, J. L. M., Amaral, J. F. M.,
Barreira, L.F.A. and, Barros, A. V., 2013. Deploying
Nodes for Industrial Wireless Networks by Artificial
Immune Systems Techniques. In 15
th
International
Conference on Enterprise Information Systems,
Angers, France.
Coelho, P. H. G., Amaral, J. L. M., Amaral, J. F. M.,
Barreira, L.F.A. and, Barros, A. V., 2014. Router
Nodes Positioning for Wireless Networks Using
Artificial Immune Systems. In 16
th
International
Conference on Enterprise Information Systems,
Lisbon, Portugal.
Kumari, J., and, Prachi, A., 2015. A Comprehensive
Survey of Routing Protocols in Wireless Sensor
Networks. In 2
nd
International Conference on
Computing for Sustainable Global Development
(INDIACom), India.
Alshawi, I. S., Yan, L., Pan, W., and Luo, B., 2012. A
Fuzzy-Gossip Routing Protocol for an Energy
Efficient Wireless Sensor Networks. 2012 IEEE
Sensors.
Shah, B., Khattak, A. M., and Kim, K., 2015. A Fuzzy
Logic Scheme for Real-time Routing in Wireless
Sensor Networks. In 6
th
International Conference on
Computing, Communication and Networking
Technologies (ICCCNT). Denton, USA.
Amri, S., Kaddachi, M. L., Trad, A., 2014. Energy-
Efficient Multi-hop Hierarchical Routing Protocol
using Fuzzy Logic (EMHR-FL) for Wireless Sensor
Networks. World Congress on Computer Applications
and Information Systems (WCCAIS).
Prathap, U., Deepa, P. S., Venugopal, K. R., and Patnaik,
L. M., 2013. Wireless Sensor Networks Applications
and Routing Protocols: Survey and Research
Challenges. In Proc. 2012 International Symposium on
Cloud and Services Computing. Post Mangalore,
India, pp. 49-56.
Chandna, M. and Singla, B., 2015. Comparative Analysis
of Flooding and Gossiping in Wireless Sensor
Networks Using SIR. International Journal of
Computer Science and Information Technologies,
(IJCSIT),Vol. 6 (4), pp. 4020-4023.
Matin, M. A., editor, 2012. Wireless Sensor Networks –
Technology and Protocols. Applications. InTech,
Croatia.
Kashani, M. A. A., and Ziafat, H., 2011. A Method for
Reduction of Energy Consumption in Wireless Sensor
Network with Using Neural Networks. 6th
International Conference on Computer Sciences and
Convergence Information Technology (ICCIT),
Seogwipo, Korea (South), pages 476–481.
ICEIS 2017 - 19th International Conference on Enterprise Information Systems
624