al-yad: a wearable sensor network over dds middleware for … · 2017-01-18 · al., developed a...
Post on 08-Jul-2020
2 Views
Preview:
TRANSCRIPT
Procedia Computer Science 56 ( 2015 ) 333 – 340
Available online at www.sciencedirect.com
1877-0509 © 2015 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).Peer-review under responsibility of the Conference Program Chairsdoi: 10.1016/j.procs.2015.07.216
ScienceDirect
10th International Conference on Future Networks and Communications, FNC-2015
Al-Yad: A Wearable Sensor Network over DDS Middleware for
Industrial Application
Basem Almadania, Farouq Muhammad Aliyua,∗, Elhadi M. Shakshukib
aComputer Engineering Department, King Fahd University of Petroleum and Minerals (KFUPM), Dhahran, Saudi ArabiabJodrey School of Computer Science, Acadia University, Wolfville, NS Canada
Abstract
Recently, it is estimated that in the coming five years, 700 million wearable devices for all kinds of applications will be sold
worldwide. They owe their fame to their ease of accessibility, which in turn ensures productivity. Meanwhile, the need for increase
in productivity in industries is at its highest. Many companies are looking for tools to increase the productivity of their employees.
One of the main tools welcomed by the industries is handheld devices. This includes tablets, PDAs and mobile phones. However,
these tools are not designed for applications that require marathon use — for applications such as machinery maintenance and
inventory, require more subtle tools. This paper presents a proposed intelligent glove named Al-Yad. Al-yad is designed to control
actuators in an industrial environment.
© 2015 The Authors. Published by Elsevier B.V.
Peer-review under responsibility of the Conference Program Chairs.
Keywords: AlYad; RTI; DDS; RIT Connext DDS; ubiquitous computing; Internet of Things; Embedded systems.
1. Introduction
Wearable Sensor Network (WeSN) is a field of Wireless Sensor Networks (WSN), where sensor nodes are designed
to gather physiological and kinetic data from wearers’ body1. Embedded devices are becoming more intelligent and
soon become the third layer of the Internet, i.e. Internet of Things2.
Despite the advantages of WeSN, little research is available in this field with regards to industrial applications.
This is due to the common belief that wireless communication is not reliable and therefore is not usable in such
applications. However, recent advances in wireless technology enable it to compete with its wired counterpart.
This paper presents and discusses the development and analysis of a proposed, reliable industrial based Networked
Control System (NCS), using WeSN. NCS is a Distributed Control System (DCS) where sensors, actuators and con-
trollers communicate over a network3. In order to ensure real-time performance, Real Time Innovations (RTI) Con-
next middleware is utilized to allow the system to connect with other devices over wireless network4. The main aim
of this work is to design intelligent wearable device with the ability to replace the current hand-held devices used in
industrial NCS applications. This provides a great help in the industrial environments that require mobile users with
∗ Corresponding author. Tel.: +9-665-415-08971 ; fax: +9-663-860-3059.
E-mail address: g201303650@kfupm.edu.sa
© 2015 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).Peer-review under responsibility of the Conference Program Chairs
334 Basem Almadani et al. / Procedia Computer Science 56 ( 2015 ) 333 – 340
tasks such as maintenance, mobile monitoring and control5. We do believe that users are in favor and keen to use
wearable devices than handheld devices. This is because they are lighter, cheaper, consume less energy and more
convenient to carry around.
The remaining sections of this paper are as follows: Section 2 presents recent research carried out on wire-
less/wearable sensor network over DDS middle-ware. Section 3 provides detail description of the proposed Al-Yad
wearable sensor. Section 4 explains the performed experiments to evaluate the performance of the proposed system.
Section 5 discusses the achieved results. Finally, section 6 concludes the achievements of the proposed research and
presents the proposed future work.
2. Previous Work
Wired control systems have been engineers’ favorite, because they are fast and reliable. However, this method
of control is expensive to install and maintain6. Current wireless communication systems are as reliable and fast as
most wired systems, thereby making handheld devices applicable to NCS. Gorlich et al.7 developed a NCS where a
Profibus-to-Bluetooth connection is used to control and parameterize a Flow Unit via Nokia 6280 mobile phone. The
system suffers from incompatibility with other phones, limited communication range and low bandwidth. Huang etal.5, developed an oil and gas storage and transportation simulation system using wireless local area network and an
Android tablet PC, with the main objective to address the shortcomings of range and bandwidth problems encountered
in7. The tablet PC communicates with programmable logic controllers (PLC) via PC station. However, the PC station
that is between tablet PC and the PLC delays communication.
Recent developments in embedded systems allow researchers to replace tablets with Single Board Computers
(SBC) like Beagle Bone, Intel Galileo and Raspberry Pi8 9 10. SBCs have smaller footprint yet they are powerful
processing devices capable of replacing phones, tablets and computers in controls systems and WeSN11. Wang et
al., developed a wearable sensor network system using Raspberry Pi12. The developed system is used for Crowd-
sensing. It uses transcoding algorithm to reduce resolution of video files, as its memory is getting full. Khelil etal.13, developed a Body Sensor Network using Raspberry Pi. The system uses Cooking Hacks e-Health shield, which
contains an airflow sensor for measuring wearer’s respiratory rate, pulse oximeter for measuring oxygen saturation
of hemoglobin and position sensor for motion tracking. Jutila et al.14, use Intel Galileo as a sink node (or gateway)
for a wearable system that helps to track children at school. All the aforementioned SBC systems showed promising
results. However, they have one common problem that is incompatibility between the different types of SBCs. This
forces scientists to stick to one type of device throughout their system.
In order to reduce the incompatibility and mask the heterogeneity of sensor nodes in mission-critical applications,
Park et al. in15 proposes a Middleware. This Middleware is capable of hiding complexity and heterogeneity be-
tween system components. It eases the management of the system’s resources and improves the predictability of the
system16. Zhai and colleagues developed a publish/subscribe middleware called Wireless Message-Oriented System
on top of TinyOS17. They managed to reduce overall the energy consumption of the network through the use of
self-adaptive QoS and content/topic model. Zug et al.18 successfully applied a publish/subscribe middleware called
Family of Adaptive Middleware for autonomOUs Sentient Objects to an industrial automation system using Katanta
robot. The robot publishes its coordinates and angles and it subscribes to movement–, speed– and emergency-stop
commands. In all cases, authors reported increase in the performance using the proposed middleware.
The proposed wearable sensor system architecture is developed in the form of intelligent hand glove that accepts
hand gesture as input, process it using ARM microprocessor (Raspberry Pi) and publish it using RTI connext profes-
sional (publish/subscribe middleware) to all subscribing equipment over wireless network. With the proposed system
architecture we are able to address the data processing constraints, the large hardware footprint problem and the
communication delays issue.
3. The Proposed System
The proposed system architecture is shown in Figure 1. It consists of the following four main components; 1) A
wearable sensor, Al-Yad. The main function of this component, is to convert the physical movements into electric sig-
nal. The electric signal is then conditioned so that it can be processed by the 3.3v ARM-6 microprocessor, Raspberry
335 Basem Almadani et al. / Procedia Computer Science 56 ( 2015 ) 333 – 340
ARM-6 Microprocessor
Transducer
SignalConditioning
RTI ConnextPublisher
Data-writer
ARM-6 Microprocessor
RTI ConnextSubscriber
Data-reader Actuator
Middleware
Al-YadProcessing Device
Fig. 1. System Architecture
Pi. 2) A Real-time Innovations Connext Professional is selected as a middleware to mediate between the wearable
sensor and other computing devices of the system. 3) A computing device is connected to the targeted actuator. The
computing device can be any device capable of running Connext Professional (e.g. PC, Beagle Bone, Intel Galileo or
Raspberry Pi). 4) Actuator that represents any industrial machinery and that is required to be controlled remotely.
3.1. Al-Yad
Figure 2 shows an intelligent hand glove named Al-Yad. It comprises of a combination of light sources, e.g. Light
Emitting Diode (LED) and light sinks, e.g. Light Dependent Resistor (LDR). The LDR and the LED are in line-
of-sight when fingers are straight, thereby sending a high signal to the Raspberry Pi through a signal conditioning
circuit, as shown in Figure 3. However, if the user bends his finger the light stops reaching the LDR and this raises
Fig. 2. Proposed system
336 Basem Almadani et al. / Procedia Computer Science 56 ( 2015 ) 333 – 340
Fig. 3. Proposed systems complete circuit diagram
Fig. 4. Transducer and signal conditioning circuit diagram
the resistance of the LDR. Hence, sending a low output to the Raspberry Pi. In order to ensure no ambient light
contaminates the signals, the LDR is placed in a hollow cylinder of 1.45cm depth. Furthermore, the light source is
pulsed at a frequency of 5KHz, which is the highest frequency of PWM allowed by ”wiringpi.h” implementation19.
The highest frequency available is chosen, because the higher the frequency the smaller the size of the capacitors
required. This approach ensures a lighter hand glove.
The signal conditioning circuit consists of a network of capacitors, diodes and inverting buffers, as shown in Figure
4. The circuit is responsible for smoothing the pulsing signal to Direct Current (DC) signal, which can be understood
by the Raspberry Pi. The Raspberry Pi is programmed to scan the glove five times in one second (5Hz), which is
little higher than value used in20. Consequently, the Raspberry Pi converts the gloves signals to an integer value and
publishes it over the middleware.
3.2. Middleware
Research has shown that DDS Middleware can achieve up to 75% efficiency in mobile nodes21. This efficiency
value has encouraged us to utilize a middleware approach in our proposed solution. The middleware chosen for
337 Basem Almadani et al. / Procedia Computer Science 56 ( 2015 ) 333 – 340
Table 1. QoS used in the proposed systems middleware.
Quality of Service Value
Durability kind = VOLAT ILEReliability kind = RELIABLE, max blocking time = 100msLiveliness kind = AUTOMAT IC, lease duration = 1sHistory kind = KEEP LAS T , depth = 5
Presentation access scope = TOPIC, ordered access = True, coherent access = FalseOwner kind = EXCLUS IVEOwner Strength 5
Time Based Filter minimum seperation = 0.2s
this task is RTI Connext Middleware. This middleware is chosen because of its strict compliance to the Object
Management Group (OMG) specification for Data Distribution Service middleware22. Moreover, certain Quality of
Services (QoS) is set in order to ensure maximum efficiency, as recommended in22 23.
Table 1 shows the quality of services used in the proposed system. In order to ensure compatibility, the QoS is
configured in both the sensor and the actuator.
4. Experiments
The LDRs casings are designed using Blender2.6 3D modeling software24 and printed with Replicator2 3D printer.
The different electronic components are assembled and described in an environment, as shown in Figure 5. The RTI
subscriber is installed on the PC labeled Testbed and the publisher on the Raspberry Pi. The PC receives the gesture
commands sent by the Raspberry Pi. All experiment is carried out base on this setup.
During these experiments, three key parameters are measured in order to assess the performance of the proposed
system. This includes throughput, latency and power consumption of the system.
In order to test latency and throughput, RTI’s performance toolkit ’Perftest’ is used25. To test latency of the system,
packets are sequenced and time stamped before they are transmitted to the Data-reader (receiver). The packets are
immediately echoed back to the transmitter where the Round Trip Time (RTT) is calculated. In other to calculate
the throughput the receiver counts the number of packets it receives per second. At the end of the experiments,
the transmitter reports the latency of the system while the receiver reports its throughput. The smallest packets size
allowed by RTI connext is chosen (i.e. 28 byte). This is due to the fact that, our proposed system is sending integer
values only. This experiment is carried out using QoS settings (see also: Table 1) and without QoS so as to investigate
the cost of applying QoS to the system. Figures 6 and 7 show the results obtained for both throughput and latency of
the system respectively.
Signal Conditioning
Circuit
SingleBoard
ComputerTestbedNetwork
LEDLDR
LEGEND
Fig. 5. Experiment setup for measuring Latency and Throughput
338 Basem Almadani et al. / Procedia Computer Science 56 ( 2015 ) 333 – 340
The second experiment is carried out to ascertain the time response of the transducer itself. To carry out this test,
the hand glove is connected to an oscilloscope and the output voltage across the GPIO port is fed to an oscilloscope.
A delay of 20ms is observed.
Finally, an experiment is setup to measure power consumption of the whole system. A 1Ω resistor is connected in
series to the system so that the same amount of current passes through the system and the resistor, as such the voltage
drop across the 1Ω resistor equals to the current passing through the system. An Analogue to Digital Converter is
connected across the 1Ω resistor. It samples the voltage drop across the resistor every 200ms. The acquired data is
then transmitted to a computer via USB, where it is stored. The current flowing through the system is recorded for
two transmissions after which the Wi-Fi dongle is immediately removed so as to record the energy overhead of the
dongle itself.
0 1 2 3 4 5 6 7 8 9 10x 104
0
200
400
600
800
1000
No. of Packets sent
Thro
ughp
ut (P
acke
ts/s
)
Throughput without QoS Vs No.of Packets
0 1 2 3 4 5 6 7 8 9 10x 104
0
200
400
600
800
1000
No. of Packets sent
Thro
ughp
ut (P
acke
ts/s
)
Throughput with QoS Vs No.of Packets
Fig. 6. Throughput of the proposed system
1 2 3 4 5 6 7 8 9 101000
2000
3000
4000
5000
6000
7000
No. of Iterations
Lath
ency
(us)
Latency Vs No.of Iterations
LatencyLatency with QoS
Fig. 7. Latency of the proposed system
339 Basem Almadani et al. / Procedia Computer Science 56 ( 2015 ) 333 – 340
Fig. 8. Current consumed by system while running
5. Results Discussion
During our experiments, we observed that there is a little difference in the throughput of the system with or without
using QoS as shown in Figure 6. It is found that the average throughput of the system is 769 Mbps with QoS and 778
Mbps without QoS. In both cases no packets were lost during the 100,000 packets transmission. Since we are only
sending integer values, this result is acceptable. Moreover, the QoS setting has little effect on the performance of the
system.
With respect to our experiments towards latency, the results achieved are demonstrated in Figure 7 for the system
with and without QoS. From Figure 7, it can be seen that there is also a little difference (1ms). The total latency of the
system is the sum of latency due to transmission and latency due to the hand glove. In this paper, the latency of the
smart glove is define as the duration from a fixed hand gesture to when a steady signal is available on the output bus
of the glove. The latency of the hand glove is found to be 20 ms. Therefore, the total delay of the system is 23 ms.
Finally, Figure 8 shows the current (top) and the moving average of the current (bottom) that is flowing through the
system during and after data transmission. Immediately after the second transmission, the WiFi dongle is removed;
hence the spike at 375th second. However, no change of flow in current is encountered after removing the dongle.
This shows that the energy overhead comes from data transmission itself. The average power dessipated is 310mW
during transmission and 220mW while the system is not transmitting.
6. Conclusions and Future Work
In this paper, the possibility of using WeSN in wireless industrial control systems to complement mobile phones,
tablets and computers is investigated. It is found that the system has a delay of approximately 23ms, which considered
is fast enough for controlling motors, pneumatic actuators and other mechanical actuators. Furthermore, the maximum
power dissipated by the system is 310mW. Therefore, the system can be powered by a mobile phone’s Lithium-ion
battery (3100mAh) for up to 10hrs of continuous data transmission. As such, the system can be battery powered, since
it can go into sleep mode when it is not transmitting.
Our plan for future work is to focus on developing a smart actuator with the ability to publish its state. This makes
the smart glove able to receive feedback from the actuator. Moreover, the actuator can be programmed such that it can
publish system failure and its possible cause.
340 Basem Almadani et al. / Procedia Computer Science 56 ( 2015 ) 333 – 340
Acknowledgment
The authors would like to acknowledge King Fahd University of Petroleum and Minerals and Real-Time Innova-
tions Inc. for their support.
References
1. Elsarnagawy, T., Farrag, M., Haueisen, J., Abulaal, M., Mahmoud, K., Fouad, H., et al. A wearable wireless respiration rate monitoring
system based on fiber optic sensors. Sensor Letters 2014;12(9):1331–1336.
2. Konieczny, M.. Enriching wsn environment with context information. Computer Science 2012;13(4).
3. Wang, F.Y., Liu, D.. Networked control systems. Springer; 2008.
4. Al-Madani, B., Al-Roubaiey, A., Al-Hammouri, M.F.. Performance enhancement of limited-bandwidth industrial control systems. Ad-vanced Materials Research 2013;739:608–615.
5. Huang, Z., Lu, Y.. Wireless monitoring and control system via android tablet pc. In: International Symposium on Computer, Communication,Control and Automation (3CA 2013). Atlantis Press; 2013, p. 449–452.
6. Hespanha, J., Naghshtabrizi, P., Xu, Y.. A survey of recent results in networked control systems. Proceedings of the IEEE 2007;95(1):138–
162. doi:10.1109/JPROC.2006.887288.
7. Gorlich, D., Stephan, P., Quadflieg, J.. Demonstrating remote operation of industrial devices using mobile phones. In: Proceedings ofthe 4th international conference on mobile technology, applications, and systems and the 1st international symposium on Computer humaninteraction in mobile technology. ACM; 2007, p. 474–477.
8. Coley, G.. Beaglebone black system reference manual. 2013.
9. Ramon, M.C.. Intel galileo and intel galileo gen 2. In: Intel® Galileo and Intel® Galileo Gen 2. Springer; 2014, p. 1–33.
10. Pi, R.. An arm gnu/linux box for $25. Take a byte 2012;.
11. Shah, J., Patel, V.P.. Real time interfacing & control techniques using an open source. International Journal of Science, Engineering andTechnology Research (IJSETR) 2014;3(3).
12. Wang, S.T., Fan, C.L., Huang, Y.Y., Hsu, C.H.. Transcoding on wearable devices: Technical report. 2015.
13. Khelil, A., Shaikh, F., Sheikh, A., Felemban, E., Bojan, H.. Digiaid: A wearable health platform for automated self-tagging in emergency
cases. In: Wireless Mobile Communication and Healthcare (Mobihealth), 2014 EAI 4th International Conference on. 2014, p. 296–299.
doi:10.1109/MOBIHEALTH.2014.7015969.
14. Jutila, M., Strmmer, E., Ervasti, M., Hillukkala, M., Karhula, P., Laitakari, J.. Safety services for children: a wearable sensor vest
with wireless charging. Personal and Ubiquitous Computing 2015;:1–13URL: http://dx.doi.org/10.1007/s00779-015-0838-z.
doi:10.1007/s00779-015-0838-z.
15. Park, S., Hosen, A.S., Cho, G.. An dds based architecture in supporting of data centric wireless sensor network environments. InternationalJournal of Control and Automation 2014;7(1):251–258.
16. Mascolo, C., Hailes, S., Lymberopoulos, L., Picco, G.P., Costa, P., Blair, G., et al. Survey of middleware for networked embedded
systems. Project Report: http://www ist-runes org/docs/deliverables/D5 01 pdf 2005;.
17. Zhai, L., Li, C., Sun, L.. Research on the message-oriented middleware for wireless sensor networks. Journal of computers 2011;
6(5):1040–1046.
18. Zug, S., Schulze, M., Dietrich, A., Kaiser, J.. Programming abstractions and middleware for building control systems as networks
of smart sensors and actuators. In: Emerging Technologies and Factory Automation (ETFA), 2010 IEEE Conference on. 2010, p. 1–8.
doi:10.1109/ETFA.2010.5641341.
19. Wiringpi, . Software tone library. 2014. URL: http://wiringpi.com/reference/software-tone-library/.
20. Yuan, H., Perdoni, C., He, B.. Relationship between speed and eeg activity during imagined and executed hand movements. Journal ofneural engineering 2010;7(2):026001.
21. Shen, H., Li, T.. Data management of wireless sensor networks. In: Consumer Communications and Networking Conference, 2009. CCNC2009. 6th IEEE. 2009, p. 1–2. doi:10.1109/CCNC.2009.4784851.
22. (OMG), O.M.G.. Data distribution service for real-time systems version 1.2. 2007. URL: http://www.omg.org/spec/DDS/1.2/PDF/.
23. Hunt, G.. Is your data secured? 2014. URL: http://www.slideshare.net/RealTimeInnovations/is-your-datasecure.
24. Brito, A.. Blender 3D: Architecture, Buildings, and Scenery: Create photorealistic 3D architectural visualizations of buildings, interiors,and environmental scenery. Packt Publishing Ltd; 2008.
25. RTI, . Combined latency and throughput performance test getting started guide version 5.1.0. 2013. URL: http://community.rti.com/
rti-doc/510/RTI_Performance_Test_5.1.0/doc/RTI_ConnextDDS_PerformanceTest_GettingStarted_5.1.0.pdf.
top related