236 915 MHz Wireless Sensor Network Reliability Evaluation when

236 915 MHz Wireless Sensor Network Reliability Evaluation when
915 MHz Wireless Sensor Network Reliability Evaluation when Working in an Industrial Electrical
Power Substation
Eduardo H. Ferroni¹, Hugo Rodrigues Vieira¹, Omar Carvalho Branquinho², Valceres V. R. Silva³
Centro Universitário do Sul de Minas - UNIS-MG¹, Pontifícia Universidade Católica de Campinas PUC-Campinas²,
Universidade Federal de São João del-Rei – UFSJ³
ABSTRACT
The electrical energy spent in manufacturing process is responsible for a large amount of the
operational costs. Thus, the consumed energy control and monitoring is important to improve the
energy quality use and minimize losses. This task is more convenient for wireless sensors
networks, that needs to operate in an industrial environment, exposed to spurious radiations, and
in industrial substations of medium voltage lines. This work aims to evaluate the effect of medium
voltage substation radiation in a sensor network, operating at 915 MHz, compared with an
environment without this type of infrastructure. The work characterizes and compares the two
environments, considering the path loss, standard deviation and signal statistic. The results show
that the interference effect from substations does not have significant effect in the system
performance.
Keywords: Sensors’ network, industrial substations monitoring, 915 MHz wireless sensors.
RESUMO
A energia eléctrica utilizada no processo industrial é responsável por uma grande parcela dos
custos operacionais. Assim, o controle da energia consumida e monitoramento é importante para
melhorar a qualidade de energia e minimizar as perdas. Esta tarefa é mais conveniente para redes
de sensores sem fio, que precisa operar em um ambiente industrial, as quais ficam expostas a
radiações espúrias, e em subestações industriais de linhas de média tensão. Este trabalho tem
como objetivo avaliar o efeito da radiação subestação de média tensão em uma rede de sensores,
operando em 915 MHz, em comparação com um ambiente sem este tipo de infraestrutura. O
trabalho caracteriza e compara os dois ambientes, considerando a perda de percurso, desvio
padrão e dados estatísticos do sinal. Os resultados mostram que o efeito de interferência de
subestações não tem relevância significativa no desempenho do sistema.
236
Palavras-chave: Rede de sensores, monitorização subestações industriais, sensores sem fios de
915 MHz.
INTRODUCTION
The industrial sector is responsible for operating more than 42% of the total energy, and for more
than half consumed electricity (including cogeneration). The connected loads management
combined with the production system is essential to achieve better efficiency; in other words,
higher productivity with lower energy consumption, (ALLEN, 2005). Thus, a fast and automatic
power monitoring and control system is essential.
Many questions regarding efficiency and reliability come up when controllers and sensors are
connected via a radio system. Some professionals believe that a radio system is not reliable when
it is immersed in an atmosphere contaminated by electrical field and/or electromagnetic
interferences caused by large motors or 13.8 KV transmission lines, (NORDMAN, 2004).
This work evaluates a wireless sensor network (WSN) operating in the 915 MHz band. The
objective is the evaluation of radio link characteristics in two situations: in an electrical power
substation environment and outside of this environment.
An experiment set up was utilized to evaluate the effect of the substation compared with a
condition without the effect of the substation. The characteristics analyzed com compare the two
situations are: path loss, standard deviation and signal statistics. The results show that the level of
interference caused by the substation does not change the WSN reliability.
This work is organized as follow. Section II presents the fundamental of signal propagation utilized
to analyze the experiment results. Section III presents the Radiuino platform utilized to evaluate
the propagation conditions. Section IV presents the wireless communication system parameters.
Section V presents the results and analyzes. Section VI presents the conclusions.
SIGNALS PROPAGATION
To characterize the wireless communications it is essential to understand the propagation
processes. While the environment that uses cables is almost deterministic, the wireless
communication is probabilistic, and thus, an uncertainty level must be considered. The radio
237
waves in the atmosphere scatter and mix up, depending on the obstacles, (NASCIMENTO, 1992).
The follow items consider the models utilized in the work.
Reference free space loss
The free space model developed by Friis, (RAPPAPORT, 2009) is the reference for all the signal
intensity prediction methods currently used. It begins with the free space attenuation to
determine the received signal intensity in a given distance. Equation 1 represents the Friis formula
with the system loss.
2
 4d 
L RL  10 log
  LS
  
(1)
Where d is the distance, λ is the wave length and LS is the system loss, (KARL, 2005). This last
factor considers the non-ideal feature of the radio frequency circuits, like connectors and antenna
mismatch, caused by the RF circuit, cable loss, etc.
The power received in the reference distance could be calculated by the Equation 2.
PRX d0   PTX  Gbase  Gsensor  L RL
(2)
Log-Distance Model
The Log-Distance model is widely used to characterize the environment type. This model assumes
that the received power of a signal, in a distance d, can be calculated considering a factor called
path loss (β) and a received power in a reference distance d0 closer to the transmitter.
From the transmitter, a reference distance is defined, and in this space it is considered the free
space attenuation. Beyond this distance, the signal will undergo a higher attenuation than in free
space, and the path loss dependent. Usually, the distance d0 must be at least 10 times lower than
the distance d. In this work is used a reference distance of 1 meter, as recommended by
NASCIMENTO, 2004). This is a practical consideration that can be changed depending on the
environment. Then, the power in a distance d can be determined, taking account of the power
measured in d0, as indicated in Equation 3.
 d
PRX d  PRX d0   10  log
 d0



(3)
238
Where  is the path loss coefficient that characterizes the propagation environment. Measuring
the powers in d0 and d it is possible to calculate β. Thus, total loss is the log-distance attenuation
plus the system loss can be calculated by Equation 4.
 d
 4 d0 
 10 log
  L S  10  log
  
 d0
2
L Tot



(4)
The path loss β could be calculated with Equation 5.

PRX d0   PRX d
 d
10 l og 
 d0 
(5)
When is not possible to measure the received power in the reference distance, and the system
loss is known, the PRX(d0) could be calculated with the Expression 2.
Shadowing Model
Using the model with obstacles, the signal attenuation calculations depend on the environment
and use information such as constant power, antenna gains and attenuation factor. These are
deterministic parameters that depend on the environment. However, there is an uncertainty
degree in the signal level. The Equation 6 gives this uncertainty degree as a random variable XdB.
 d
Pr (d)[dBm]  Pr (d o )[dB]  10log
 do

  X dB

(6)
Where XdB is a log-normal random variable and the standard deviation σ is another factor that
characterizes the environment.
Radiuino Platform
The experiment was conducted with the Radiuino open source platform specially designed to
implement WSN. This platform was developed with a structured stack that simplifies the
implementation of WSN. This approach is easy to use compared with others platforms like
SimpliciTI (Texas Instruments), SMAC (Freescale) and MiWi (Microchip).
239
The differentiation is initially based on an easy firmware development environment. Unlike the
conventional platforms, which use complex and extremely detail-oriented IDE’s (Integrated
Development Environment), Radiuino is based on Arduino, (MCROBERTS, 2010). This platform
uses a simplified IDE and is easily adaptable, even for non-experts programmers.
Another interesting feature of this platform is its conceptual similarity with the TCP/IP stack. As in
the Internet protocol stack, the Radiuino stack has five layers: Physical, Data Link, Network,
Transport and Application. This distinction is very clear when designing and implementing the
application, while maintaining the firmware modularity and maintainability. Figure 1 shows the
structure of the protocol stack of Radiuino in its three main elements.
Figure 1: Protocol stacks in a Radiuino platform.
It can be seen that the sensor’s protocol stack have all layers. In the computer and base the
structure is:

in the sink: physical, data link and network

in the host: transport and application.
Thus, the communication and addressing characteristics are handled in a hardware base, leaving
the host to control the transport layer and the application itself.
The Radiuino communication platform sketch is shown in Figure 2.
Figure 2: Radiuino communication platform.
240
The host is set with software to establish the communication with the sensors. The computer
communicates through serial with the base, which sends a wireless requisition to the sensor.
Then, the sensor executes the commands sent by the host and reads the data from the
transducers. The sensor sends back to the base an answer to the requisition. Other information
received by the base is the RSSI (Radio Signal Strength Indicator) with the signal received powers
of the up and down link. The computer processes these data and let them available for
manipulation.
Both the requisition and the answer packet have the same size (52 bytes) and the same format.
The packet composition is:
 4 bytes for the layers: physical, mac, net and transportation;
 36 bytes for the payload.
The module with the radio and microcontroller used for the experiment is made in Brazil, called
BE900, which contains an Atmega328 microcontroller from Atmel, and the transceiver CC1101
from Texas Instruments (Figure 3).
Figure 3: The Radio BE 900.
The connection with the antenna uses a SMA connector that provides a robust connection
between the radio and the antenna.
To collect the results it was used a free Brazilian SCADA (Supervisory Control Acquisition Data
Automation) called ScadaBR (Figure 4).
241
Situation 2
Situation 1
Figure 4: ScadaBR screen.
The graphical presents the RSSI of up and down link used to characterize the two situations. The
ScadaBR has a data base that stores the data for off line processing.
Wireless Communication System
The wireless network project needs to consider the environment type and the system
characteristics. Figure 5 illustrates the radio communication elements.
Figure 5: Wireless communication system.
System parameters:

PTX – transmission Power = 10 dBm;

Gbase – sectorial base antenna gain = 12 dBi;

Gsensor – omnidirectional sensor antenna gain = 8 dBi;

Sensibility – the lower receiving power needed to obtain an specific BER = -97 dBm;

Modulation format = 2-FSK;

Deviation = 177.734375 kHz;
242

Base frequency = 915.999725 MHz;

Carrier frequency = 915.999725 MHz;

Sync word qualifier mode = 30/32 sync word bits detected;

Preamble count = 4;

Channel spacing = 405.456543 kHz;

Data rate = 4.79794 bps;

Rx filter BW = 541.666667 kHz.
Figures 6, 7 and 8 show the test bed photographs.
Figure 6: Base sectorial antenna.
Figure 7: View of substation and sectorial antenna.
243
Situation 1
Situation 2
Figure 8: The two situations analyzed.
The sensor antenna was Omni-directional and the base antenna Panel Standalone
900-928 GHz, 12 dBi, which radiation diagram can be seen in Figure 9.
Figure 9: Base antenna with 60 degree of aperture angle.
RESULTS AND ANALYSIS
A - Test Characterization
Two data sets were collected considering the two situations showed in Figure 8:
1. Without the substation between the base and the sensor node;
2. the substation between the base and the node sensor.
The experiment aims to evaluate these two situations calculating the path loss β and the standard
deviation σ. The signal is collected in dBm in a 2 seconds period. On Situation 1, 424
measurements were made and on situation 2, 824 measurements were made. The number of
samples was sufficient to cognize a regularity of the data, as showed in Figure 4. The RSSI of up
and down link as measured to verify a possible asymmetry in the two paths.
244
Firstly the data must be converted from dBm to Watt for the signal average calculation. Then, the
results was converted to dBm again to determine the path loss β.
To determine the standard deviation σ the set of data was manipulated directly in dB
(RAPPAPORT, 2009). First is calculated the average, which is subtract of each data point. The result
is a random variable with zero average, as mentioned in Section II, where is determined the
probability distribution and the standard deviation σ.
B - Path Loss β
The system loss (LS) is 10dB measured in laboratory and the reference distance is considerate 1
meter to compare the two situations.
Considering the values of the parameters of system, and using Equation 2, is calculated the P RX (d0)
= -11.68 dBm. Considering the average power of each situation is determined the values of β.
Table 1 shows the path loss.
Table 1: The β value for the two situations.
Down Link
Up Link
β
PRX (d ) [dBm]
β
PRX (d ) [dBm]
Situation 1
3.143
-73,1
3.189
-74,0
Situation 2
3.065
-75,4
3.094
-76,0
C - Statistics Analysis.
The Probability Mass Function (PMF) curve can be plotted. Figures 10 and 11 show the distribution
curves for each situation, (BUSSAB, 2003), (DANTAS, 2000).
Figure 10: Probability function for the 1st situation.
245
Figure 11: Probability function for the 2nd situation.
The two curves present a symmetrical behavior, which demonstrate that the Gaussian function
shape can be considerate. But, the standard deviation and average have differences as showed in
Table 2.
Table 2: The σ value for two situations.
Down
Link
Up Link
σ [dB]
σ [dB]
Situation 1
2.62
2.39
Situation 2
1.10
1.15
Figures 12 and 13 show the Cumulative Distribution Function for the two situations.
Figure 12: CDF for the 1st situation.
246
Figure 13: CDF for the 2nd situation
Comparing the two analyzed cases, it is possible to see the difference of 1.7 dB. This is a small
difference that indicates that the two situations are similar.
The data analyzed show that the use of 915 MHz sensors networks did not commit the quality and
reliability, when operating in a substation environment. For the two cases studied, the standard
deviation, lower for the 2nd situation, occurred due to the fact that the measurements made inside
the substation did not suffer the effects of the vehicles transit nearby the antennas as showed in
Figure 8; for the 1st situation, the cars and buses traffic caused higher standard deviation.
Therefore, it can be noticed through this experiment that, the power transformer operation in 60
Hz did not cause any radio link reliability loss or damage.
CONCLUSION
The work shows that an electrical equipment’s monitoring and/or controlling system, using 915
MHz sensors networks, are not influenced anyhow by high power electrical devices operating
nearby. Its reliability depends only of the design accuracy, considering the environment where the
sensors network is going to operate.
ACKNOWLEDGMENT
247
The authors gratefully acknowledge the financial support of this research by Fundação de Amparo
a Pesquisa do Estado de Minas Gerais (FAPEMIG) and Universidade Federal de São João del Rei
(UFSJ), Brazil.
REFERENCES
ALLEN, W.; Lee, T., “Esquema de Rejeição de Cargas Flexível e de Alta Velocidade Utilizando Crosspoint Switchs,” Artigo Técnico da Schweitzer Enginee-ring Laboratories Comercial Ltda, 2005.
BUSSAB, W. O. and Morettin, P. A., Estatística Básica, 5th ed., Ed. Saraiva, 2003, 103-140
DANTAS, C. A. B., Probabilidade: Um Curso Introdutório. 2sd ed., Ed. USP, 2000, 183-220
KARL, Holger; Willig, Andreas. Protocols and Architectures for Wireless Sensor Networks. 1 st ed.,
Wiley, 2005
NASCIMENTO, J., Telecomunicações, 2sd ed., Makron Books, 1992. pp, 07-19
NORDMAN, M., Na Arquitecture for Wireless Sensors in Distributed Management of Electrical
Distribution Systems. Hilsinki, Finland: Hilsinki University of Tecnology, 2004.
MCROBERTS, M., Beginning Arduino. Publisher Apress. 2010
RAPPAPORT, T. S., Wirelles Comunications – Principles and Practice – 2sd ed. Dorling Kindersley,
2009.
248
Was this manual useful for you? yes no
Thank you for your participation!

* Your assessment is very important for improving the work of artificial intelligence, which forms the content of this project

Download PDF

advertising