Sunteți pe pagina 1din 3

2018 IEEE International Conference on Big Data (Big Data)

A Study on Big Data Cluster in Smart Factory


using Raspberry-Pi
Chae-Soo Kim Seung-Beom Son
Dept. Of Industrial Engineering R&D Center
Dong-A university SDS Ltd
Busan, Korea Busan, Korea
cskim@dau.ac.kr seungbeom.son@gmail.com

Abstract—In smart factory, the IoT (Internet of things) is artifacts can communicate and negotiate with each other
applied to capture the data generated by the production through the IWN to implement self-organization, and the
facility, store and analyze data generated in real time using Big massive data could be uploaded. This enables the system-
Data technology. A parallel distributed processing system for wide feedback and coordination based on big data analytics
big data analysis and processing requires a large-capacity
to optimize system performance. Wang proposed a brief
server and a high cost to build infrastructure. In acquisition of
shop floor data, Raspberry-Pi which is a low cost single board framework that incorporates industrial wireless networks,
computer is very efficient device for big data clustering. In this cloud, and fixed or mobile terminals with smart artifacts
paper, we show an case study on the feasibility of Raspberry-Pi such as machines, products, and conveyors [5]. Almada
as a big data cluster in smart factory. insisted that Industry 4.0 dictates the end of traditional
centralized applications for production control. Its vision of
Keywords—big data cluster, smart factory, Raspberry-Pi ecosystems of smart factories with intelligent and
autonomous shop floor entities is inherently decentralized
I. INTRODUCTION [6].
Smart factory refers to an intelligent production plant that By the way, it is not easy to collect the status data from the
improves productivity, quality and customer satisfaction by individual facilities in shop floor stably and effectively.
applying information and communication technology (ICT) Because there are so many old facilities which do not equip
combined with digital automation solution to design, any communication port in themselves. To overcome this
development, manufacturing, distribution and logistics. It is difficulty, we may need to utilize an additional data
also characterized by the coordination and control of collection device which should be robust in function and
machinery and personnel including engineer and efficient in operation.
operators. Thanks to smart factory, an idea to create a Big data is so large and diverse that it needs to build a
remotely controllable work center is not new, but the cluster that collects, filters, and transmits large amounts of
challenges with interaction scenarios between machines and structured or unstructured data in a timely manner, and a
humans became more and more complex as shown in Figure large parallel distributed processing system. In spite of
1. Usually industrial automation is mainly organized by smart factory benefit, small manufacturing companies are
microcontroller systems and the main disadvantages of such hesitating to the elemental technologies on account of the
systems is focusing on specific tasks, needs of external amount of financial investment for those systems.
specialist and advanced level of user's programming skills. Consequently this study introduces the use of raspberry pie
With the advent of various mobile devices, the development to enable small and medium enterprises or organizations to
of SNS(Social Network Service), the demand and supply of handle big data processing at low cost, and introduces the
various multimedia contents, interest in big data processing developed software and hardware architecture.
is continuously increasing and its importance is also And the rest of this article is organized as follows. In
emphasized [1]. In smart factory, it collects the data chapter 2, we describe the big data cluster based on
Raspberry-Pi, the characteristics of Raspberry-Pi, system
generated from shop floor using DCS(Data Collection
configuration, and the casing unit for the cluster in the view
System), and communicate through IoT technology, and
of hardware. Finally concluding remarks are summarized in
stores the collected data in repository. Eventually we chapter 4, and references are followed in a row.
analyze the data gathered in repository using big data
technology, and generate appropriate alternatives to
determine the optimal solution based on AI(Artificial II. BIG DATA CLUSTER USING RASPBERRY-PI
Intelligence) technology [2]. In order to take advantage of The big data processing technology to utilize large
those technologies and feedback system, we need to amount of data has a wide range of applications such as
configure a big data cluster, and the information gathered in research, user analysis and trend analysis, and it has great
it will be stored on the server computer for big data analysis potential value. In order to solve the difficulties and
via low power wireless network standards which have disadvantages of high cost and large scale server systems in
enabled supervisors to manage and control wide ranges of domestic and foreign countries, researches have been
sensor networks and system [3]. In order to facilitate the conducted on a small data cluster with a small size using
interaction between sensor network and server computer for SBC(Single Board Computer) [7] [8] and [9].
analysis of big data in smart factory, IWN(Industrial In reality, most of the shop floor control system in small
Wireless Network) could be suitable in reality by business deals with much more simple information in
implementing IoT and services [4]. By this means, the smart comparison with large scaled business, even though they

XXX-X-XXXX-XXXX-X/XX/$XX.00 ©20XX IEEE


978-1-5386-5035-6/18/$31.00 ©2018 IEEE 5360
make the same products in the same process. And the basic function of the big data cluster could be
MES(manufacturing execution system) have been pivotal in described as follows.
the performance, quality and agility needed for the 1) Time and place of data generation
challenges created by globalized manufacturing business and 2) Saving and editing collected information
will most likely continue to be. However, a completely new 3) Automatic conversion of communication channel
generation is required to cope with the new challenges 4) Confirm communication, and resend
created by Industry 4.0. Though a very simple model of MES, 5) Real-time inquiry and storage of work instruction
we need to test operation for feasibility check and check the 6) Manual input such as work accomplishment report
effectiveness of big data cluster in advance. Since the
number of data collection terminal could be increased
2.2 Hardware
according to the manufacturing capability and process, we
need to configure a big data cluster. In this study, we try to In the view of robustness of Raspberry-Pi devices as a big
take advantage of Raspberry-Pies as a big data collection data cluster, we developed hardware case as an individual
cluster. protection gear. As an example for the Raspberry-Pi case,
we can show the image and specification as shown in Figure
The Raspberry-Pi is a series of credit card sized single- 2. The case includes some spaces for multiple Raspberry-
board computers developed in the United Kingdom by the Pies with GPIO. Ethernet, power supply, USB, and so on.
Raspberry-Pi Foundation to promote the teaching of basic
computer science in schools and in developing countries.
Accessories including keyboards, mice and cases are not
included with the Raspberry-Pi. There are a number of
developers and applications that are leveraging the
Raspberry-Pi for home automation. These programmers are
making an effort to modify the Raspberry-Pi into a cost
affordable solution in energy monitoring and power
consumption. Because of the relatively low cost of the
Raspberry-Pi, this has become a popular and economical Fig. 2. A big data cluster included multiple Raspberry-Pies
solution to the more expensive commercial alternatives. This
trend has been extended to the manufacturing environment in
the virtue of cost effectiveness in comparison with other
devices. Raspberry-Pi has been evolved in 2018, in the name
III. EXPERIMENT PLANNING
of 3B+ which equips the additional function of networking.
The cluster will be composed of more than three
Raspberry-Pi (nodes) and connected to each other by using
2.1 Software network cable. One node is a master node and the remaining
In order to develop big data cluster using Raspberry-Pi, nodes are slave (data) nodes. Since most of SBC based big
we utilize Raspbian as an operating system, which has been data clusters have been implemented using Hadoop, which
authenticated by Raspberry-Pi foundation. Raspbian makes supports HDFS(Hadoop Distributed File System), we will
analysis and control system more effective in the terms of install Hadoop and use it to share data between nodes in a
economic value. For instance, the system measures cluster. Though an experiment to evaluate the performance
dimensions of a WIP(work in process), and it will keep the of big data cluster using Raspberry-Pi has not been
measurement data in the memory of Raspberry-Pi. The data performed yet, we plan to proceed in the order shown in
will be transferred to the server computer through the Figure 3 in the near future.
transmission module inside of Raspberry-Pi. Raspberry-Pi
serves wired and unwired communication module, and
could be switched by itself. We developed software for the
big data cluster including successful Wi-Fi communication
in shop floor. Figure 1 shows the system configuration for
the Raspberry-Pi based big data cluster.

Fig. 3. Experiment planning for performance evaluation

IV. CONCLUSION
In recent years, smart factories have been actively
Fig. 1. System configuration for big data cluster introduced in advanced countries including Korea, and the
development of the elemental technologies for smart factory
is also progressing very rapidly. The facility status data and

5361
Real-time working situation in shop floor should be
collected and analyzed for diagnosis, prediction, and
decision making for the smart factory. To do this, we
configured a big data cluster using Raspberry-Pi on account
of its usability and economical feature. In order to evaluate
the performance of the Raspberry-Pi based cluster, we need
to conduct an experiment for speed and accuracy in the near
future.
As another further study, we will compare the
performance of using Arduino as a big data cluster instead of
using Raspberry-Pi..

ACKNOWLEDGMENT
This research was financially supported by the
DAURP(Dong-A University Research Park) in Korea.

REFERENCES
[1] Ahn, C. W., & Hwang, S. K., “Big Data technologies and main issues,”
Journal of Korean Institute of Information Scientist and Engineers, vol.
30, no. 6, 2012, pp.10-17.
[2] Choi, S. W., Kim, H. Y., & Kim, Y., “Research Trend of Big data
Technology and Analysis Technique,” The Journal of the Korean
Institute of Information Scientists and Engineers, vol. 39, no. 2, 2012,
pp.194-196.
[3] Aditya Gaur, Bryan Scotney, Gerard Parr, Sally McClean, “Smart City
Architecture and its Applications based on IoT,” Procedia Computer
Science, vol. 52, 2015, pp.1089-1094.
[4] J. Liu, Q. Wang, J. Wan, J. Xiong, and B. Zeng, “Towards key issues of
disaster aid based on wireless body area networks,” KSII Transactions
on Internet and Information Systems, vol. 7, no. 5, 2013, pp.1014–1035.
[5] Shiyong Wang, JiafuWan, Di Li, and Chunhua Zhang, “Implementing
Smart Factory of Industry 4.0,” International Journal of Distributed
Sensor Networks, 2016, pp.1-10.
[6] Almada-Lobo, “The Industry 4.0 revolution and the future of
Manufacturing Execution Systems,” Journal of Innovation
Management, vol. 3, no. 4, 2016, pp.16-21.
[7] Cha, B. R., Kim, N. H., Lee, S. H., & Kim, J. W., “Integrated
Verification of Hadoop Cluster Prototypes and Analysis Software for
SMB,” 2014.
[8] d’Amore, M., Baggio, R., & Valdani, E., “A Practical Approach to Big
Data in Tourism: A Low Cost Raspberry Pi Cluster,” Information and
Communication Technologies in Tourism, 2015, pp.169-181.
[9] Dye, B., “Distributed computing with the Raspberry Pi,” Doctoral
dissertation, Kansas State University, 2014.

5362

S-ar putea să vă placă și