SPECIAL SECTION ON KEY TECHNOLOGIES FOR SMART FACTORY OF INDUSTRY 4.

Received October 24, 2017, accepted December 9, 2017, date of publication December 14, 2017, date of current version March 9, 2018.
Digital Object Identifier 10.1109/ACCESS.2017.2783682

Smart Factory of Industry 4.0: Key Technologies,

Application Case, and Challenges
BAOTONG CHEN1 , JIAFU WAN 1 , (Member, IEEE), LEI SHU2,3 , (Senior Member, IEEE),
PENG LI4 , MITHUN MUKHERJEE5 , (Member, IEEE), AND BOXING YIN6
1 School of Mechanical and Automotive Engineering, South China University of Technology, Guangzhou 510641, China
2 School of Engineering, Nanjing Agricultural University, Nanjing 210014, China
3 School of Engineering, University of Lincoln, Lincoln, U.K.
4 Institute Industrial IT, Ostwestfalen-Lippe University of Applied Science, 32657 Lemgo, Germany
5 Guangdong Provincial Key Laboratory of Petrochemical Equipment Fault Diagnosis, Guangdong University of Petrochemical Technology,

Maoming 525000, China

6 School of Electrical Engineering and Automation, Jiangxi University of Science and Technology, Ganzhou 314000, China

Corresponding author: Jiafu Wan ([email protected])

This work was supported in part by the National Key Research and Development Project under Grant 2017YFE0101000, in part by the Key
Program of Natural Science Foundation of Guangdong Province, China, under Grant 2017B030311008, in part by the Maoming
Engineering Research Center of Industrial Internet of Things under Grant 517018, and in part by the National Natural Science Foundation
of China under Grant 61401107.

ABSTRACT Due to the current structure of digital factory, it is necessary to build the smart factory to
upgrade the manufacturing industry. Smart factory adopts the combination of physical technology and cyber
technology and deeply integrates previously independent discrete systems making the involved technologies
more complex and precise than they are now. In this paper, a hierarchical architecture of the smart factory was
proposed first, and then the key technologies were analyzed from the aspects of the physical resource layer,
the network layer, and the data application layer. In addition, we discussed the major issues and potential
solutions to key emerging technologies, such as Internet of Things (IoT), big data, and cloud computing,
which are embedded in the manufacturing process. Finally, a candy packing line was used to verify the key
technologies of smart factory, which showed that the overall equipment effectiveness of the equipment is
significantly improved.

INDEX TERMS Smart factory, big data, cloud computing, cyber-physical systems, industrial Internet of
Things.

I. INTRODUCTION In the context of intelligent manufacturing, it is important

With the rapid development of electric and electronic tech- to establish the smart factory to achieve advanced manu-
nology, information technology and advanced manufacturing facturing based on network technologies and manufacturing
technology, the production mode of manufacturing enter- data. In addition, the implementation of smart factory should
prises is being transferred from digital to intelligent. The new take into account the status quo and manufacturing require-
era that combines virtual reality technology based on the ments. Due to the different characteristics of manufacturing
Cyber-Physical System (CPS) is coming [1]–[4]. Due to field and information field, there are still many technical
the new challenges, the advantages of traditional manufac- problems to be solved in order to accelerate the path of smart
turing industries have been gradually diminished. Conse- factory. The main goals of this paper are summarized as
quently, the intelligent manufacturing technology is one of follows:
high technology areas where industrialized countries highly • In the physical resources layer, physical equipment
pay more attention to. Europe 2020 strategy [5], Industry needs to have support for real-time information acqui-
4.0 strategy [6] and China manufacturing 2025 [7] have sition, and communication devices should provide a
been proposed. United States has gradually accelerated the high-speed transmission of heterogeneous information.
speed of reindustrialization and manufacturing reflow [8]. The workshop should ensure fast reconfiguration and
The transformation of intelligent manufacturing intrigued adaptability. In addition, the intelligence of underlying
the profound and lasting effect on the future manufacturing equipment should be enhanced in order to meet the
worldwide. requirements of Internet of Things (IoT).

2169-3536
2017 IEEE. Translations and content mining are permitted for academic research only.
VOLUME 6, 2018 Personal use is also permitted, but republication/redistribution requires IEEE permission. 6505
See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
 B. Chen et al.: Smart Factory of Industry 4.0: Key Technologies, Application Case, and Challenges

• In the network layer, Industrial Internet of Things (IIoT) smart factory, which is based on digital and automated
should support new protocols and new data format factory, uses information technology (e.g., cloud platform
with high flexibility and scalability, whereas the Indus- and IIoT) to improve the management of manufacturing
trial Wireless Sensor Networks (IWSNs) bring new resources and QoS [13], [14]. In order to build the smart fac-
opportunities for development of industrial network. tory, manufacturing enterprises should improve production
Additionally, the other relevant technologies (e.g., OLE and marketing, enhance controllability of production process,
for Process Control Unified Architecture (OPC UA), and reduce manual intervention in workshop. Through the
Software defined Networks (SDNs), and Device-to- analysis of manufacturing data, the smart factory can realize
Device (D2D) communication) should be introduced flexible manufacturing, dynamic reconfiguration, and pro-
to guarantee Quality-of-Service (QoS) of the net- duction optimization, which are aimed to adapt the system
work, reliable communication, and cooperation among to the changes of business model and consumer shopping
equipment. behavior [15].
• In the data application layer, the cloud platform should In the implementation of smart factory, the IIoT is
be able to analyze the semantics of various data. There- employed to integrate the underlying equipment resources.
fore, ontology is being employed in modeling of the Accordingly, the manufacturing system has abilities of per-
smart factory, which can provide the abilities of self- ception, interconnection and data integration. The data anal-
organization, self-learning and self-adaption. Moreover, ysis and scientific decision are used to achieve production
data analysis could provide the scientific basis for scheduling, equipment service and quality control of products
decision-making, while data mining could be used to in smart factory. Further, the Internet of services is intro-
ensure design optimization and active maintenance. duced to virtualize the manufacturing resources from a local
In general, we developed and analyzed the hierarchical archi- database to the cloud server. Through the human-machine
tecture of smart factory with the focus on key technologies interaction, the global collaborative process of intelligent
of every layer. The remainder of the paper is organized manufacturing oriented to the order-driven market is built.
as follows. In Section II, the architecture of smart factory Therefore, the smart factory represents an engineering system
is presented, and approaches for implementation of smart that mainly consists of three aspects: interconnection, collab-
factory based on hierarchical architecture are discussed. oration and execution. As shown in Fig. 1, the architecture of
In Section III, the key technologies of physical resource layer smart factory [16], [17] includes four layers, namely physical
are studied. In Section IV, technologies related to the network resource layer, network layer, data application layer, and
layer are given. The methods used for data processing in terminal layer. With the aim to convert modern factory into
data application layer are presented in Section V. Its research smart factory, key technologies involved in all layers have to
results and development trends of these technologies are also be studied in detail.
presented. In Section VI, the implementation of smart factory,
namely the prototype platform of candy packing line, was III. PHYSICAL RESOURCES LAYER
introduced, wherein the key technologies of the smart fac- The physical resources include all manufacturing resources
tory were employed and verified in practice. In Section VII, involved in the entire life cycle of manufacturing, which rep-
main issues and challenges of key technologies are presented. resent the basis for achievement of intelligent manufacturing.
Finally, in Section VIII a brief conclusion and envisions of The efficient manufacturing of customized products puts for-
further development of smart factory are given. ward new demands on manufacturing equipment, production
line and data acquisition. Therefore, to meet the requirements
II. SMART FACTORY ARCHITECTURE of smart factory, present problems of key technologies should
In the context of Industry 4.0, the intelligent manufacturing be solved.
attracts enormous interest from government, enterprises and
academic researchers [9]. Therefore, the construction pat- A. RECONFIGURABLE MANUFACTURING UNIT
terns of smart factory are widely discussed. However, the Due to the lack of flexible and configurable construction,
standards for smart factory implementation have not been current manufacturing equipment in workshop has strong
established yet. Benkamoun et al. [10] proposed a class specificity and relatively narrow application range, which
diagram which can be used to represent the manufacturing results in weak adaptation to manufacturing environment
system from different perspectives of entities and functions. changes. The manufacturing unit, which is modularized by
Radziwon et al. [11] expounded former research from the manufacturing equipment (e.g., industrial robot, mechanical
concept of smart factory, and they pointed out that smart arm, and machining center), improves dynamic scheduling.
factory is actually an exploring of adaptive and flexible Moreover, the controller is reconfigurable and provides an
manufacturing. Lin et al. [12] proposed an architecture for extension of manufacturing equipment functions.
cloud manufacturing systems oriented to aerospace conglom-
erate, which facilitates optimal configuration of manufac- 1) MODULAR MANUFACTURING UNITS
turing resources. The above-mentioned authors provided a From the aspect of modular manufacturing units,
guidance architecture for smart factory. In summary, the Fiasché et al. [18] proposed a modular-adaptive and

6506 VOLUME 6, 2018

B. Chen et al.: Smart Factory of Industry 4.0: Key Technologies, Application Case, and Challenges

FIGURE 1. Hierarchical architecture of smart factory.

self-contained reconfigurable robotic island to improve In addition, heterogeneity of interaction should be took
assembly capacity of the workshop, where flexible manu- account.
facturing ability was also enhanced by integrated manage- • The functions of different modular manufacturing units
ment framework which controls and organizes the modular may be redundant for a certain product, therefore it is
manufacturing unit. Piranda and Bourgeois [19] proposed crucial to make an optimal or suboptimal combinatorial
a distributed algorithm for reconfiguration of lattice-based scheme.
modular self-reconfigurable robots, which drastically sim- • Each manufacturing unit can not only meet the manu-
plifies complexity of robots configuration through iterative facturing requirements of products, but also improve the
approach. Further, Valente [20] proposed a vertical cyber smart factory efficiency in a self-organized way.
physical integration of cognitive robots in manufacturing. In the intelligent manufacturing, small quantities of vari-
The cognitive robots, which were vertically integrated into ous products enter into manufacturing system disorderly,
the manufacturing industry, and coordinated with the man- which certainly leads to the deadlock phenomenon. Currently,
ufacturing execution system. In the context of intelligent approach for deadlock prevention in flexible manufacturing
manufacturing, the cognitive robots can perceive information systems is a hot research topic [21], [22].
uncertainty, change scheduling management and adjust man-
ufacturing behavior to independently cope with a complex 2) CONFIGURABLE CONTROLLER
manufacturing problem. The intelligent level of smart factory The configurability of the control system refers to the ability
is closely related to the modular manufacturing unit, and to integrate, extend, replace, and reuse hardware or software
the above-mentioned literature describes the intelligent robot components of the system. Proper configuration of controller
units from different perspectives. The modular manufacturing can improve configurability of manufacturing unit, which
units can work independently and deal with a changeable expands unit’s function in multi-application scenarios. In this
scheduling of smart factory. way, manufacturing unit would be able to adapt swiftly to
Therefore, it is very important to enhance the intelligence operation condition changes [23], [24].
of robots units. Several suggestions for construction of mod- Short and Burn [25] upgraded robot function using
ular manufacturing units are proposed as follows: available software components, which were used to adapt
• Modular manufacturing units should cooperate with the robot to changeable operating conditions. Moreover,
each other to accomplish the common tasks, where they introduced an advanced design of robot controller
the emphasis is on the mutual perception and the which can be used in the new generation of robots.
collaboration mechanism between intelligent modules. Morales-Velazquez et al. [26] developed a new multi-agent

VOLUME 6, 2018 6507

 B. Chen et al.: Smart Factory of Industry 4.0: Key Technologies, Application Case, and Challenges

distributed controller system to meet the requirements technology convergence to support the re-configurability of
of intelligent reconfigurable Computer Numerical Control door trim assembly line, where the emphases is on mini-
(CNC). Specifically, they proposed an open architecture for mization of reconstruction complexity and maximization of
reconfigurable applications based on a Field Programmable workbench recyclability. Järvenpää et al. [30] proposed an
Gate Array (FPGA) and utilized an eXtensible Markup Lan- ontology model and a description concept for manufacturing
guage (XML) structure to describe system files. Since the resources based on XML, and developed a set of integrated
component models of a control system can improve the tools to reconfigure production system rapidly and auto-
reconfiguration processes by standardization and universal- matically. Gyulai et al. [31] introduced a hierarchical plan-
ity, the embedded design tools such as Ptolemy II have ning decision workflow to assign products to dedicated and
been matured for component model. The service oriented reconfigurable production lines, which denotes an integrated
technology with the semantic web theory is employed to way to optimize system configuration and production plan.
adapt external demand on industry control domain. However, Goyal and Jain [32] divided the optimization design of
this static-related reconfigurable technology cannot meet reconfigurable production line into two stages. In the
dynamic requirements. Lepuschitz et al. [27] proposed an first stage, a multiple-objective particle swarm algorithm
automation agent-based architecture which subdivides the was proposed to optimize cost, machine utilization, oper-
control layer into two layers, where the high-level layer is ational capability, and configuration convertibility. At the
combined with the ontology while the low-level layer is based next stage, a maximum deviation method was used to
on the IEC 61499 standard. The low-level control is modified choose an optimal scheme from the alternatives in order to
by the reasoning and initiating reconfiguration processes of avoid both subjectivity and uncertainty in decision-making
high-level layer. Hence, the control system provides the self- process.
reconfiguration and self-adaption services. However, the cur- The reconfigurable production line can generate a large
rent research still lacks the validation techniques for the strict range of different products due to its variability, scalability
background. Based to normalized structure of reconfigurable and schedulability, which is the basis of flexible manufac-
logic controller, Adamski and Tkacz [28] proposed a hierar- turing in smart factory. Presently, the problem of production
chical modular control logic described by a petri net where line is strong specialty, which can be enhanced by advance
the formal reasoning method is used to predict results of a planning and control methods. Here, we build a reconfig-
logic design. With the development of equipment computing urable system to simulate the production line and propose
ability, it has become a trend to apply the multi-agent technol- a scheme for reconfiguration. The feasibility of proposed
ogy, knowledge modeling, and reasoning technology in the scheme is verified by the manufacturing scenario. With the
control system. aim to respond timely to the market requirements, it is nec-
The research of configurable controller mainly focuses on essary to implement a reconfigurable production line in the
its structure and function. General, an information model smart factory.
is built in the control system to perceive operation con-
dition, while reasoning decision is made according to the C. INTELLIGENT DATA ACQUISITION
information of evaluation module. Moreover, the controller The manufacturing resource data denote basic information
system completes reconfiguration planning and implements for workshop scheduling and intelligent service in smart
processes automatically for specific needs. However, lack factory. The wireless sensor networks (WSNs) are employed
of support for interoperable technology, such as interactive in smart factory for data monitoring, acquisition and logging.
interface, results in poor flexibility. The basic theory from Based on data analysis and using an intelligent equipment in
reconfigurable aspect (e.g., ontology modeling and verifica- manufacturing scenarios, the manufacturing execution sys-
tion) is still weak. Moreover, dynamic re-configuration mech- tem can properly implement production scheduling. The most
anism is not perfect and research results are only applicable common types of wireless sensor networks are Radio Fre-
to the specific scenarios. Thus, more attention should be quency Identification (RFID) [33], ZigBee, and Bluetooth.
paid to the embedded component model in real-time run- Zhong et al. [34] proposed a RFID-enabled real-time man-
ning framework. Therefore, the knowledge-driven adaptive ufacturing execution system for mass-customization produc-
re-configuration is crucial for the implementation of recon- tion, where RFID identifies anomalous information and feeds
figurable service. it back to the decision-making production system in real-
time, thus the smart factory has more efficient execution plans
B. RECONFIGURABLE PRODUCTION LINE and makes accurate scheduling decisions. Zhang et al. [35]
Nowadays the manufacturing products in the market are proposed a new data cleaning algorithm based on WSNs and
characterized by small batches and multiple varieties. There- RFID, where WSNs and RFID system provide an excellent
fore, the production-line should reconfigure its process paths infrastructure for data acquisition, distribution, and process-
and recombine manufacturing units dynamically. Accord- ing. With the development of RFID, the Near-Field Commu-
ingly, the smart factory should adjust product type and pro- nication (NFC) has been studied as an access to manufactur-
duction capacity in real-time. Kim et al. [29] proposed an ing resources [36]. Further, both Bluetooth and ZigBee meet
application service based on information and communication the requirements for cost of industrial automation of wireless

6508 VOLUME 6, 2018

B. Chen et al.: Smart Factory of Industry 4.0: Key Technologies, Application Case, and Challenges

standardized [39], [40]. However, there are still many issues

such as routing, congestion control, errors handling and seg-
mentation technology in network layer. The IWSNs and the
other related technologies involved in the smart factory are
discussed in the following.

A. INDUSTRIAL WIRELESS SENSOR NETWORKS

The Industrial Wireless Sensor Networks (IWSNs) represent
the expansion and promotion of the existing wireless com-
munication technology intended for industrial application
and they lead to the revolution of measurement and control
mode in the traditional industrial field. In the meantime,
the deployment of industrial networks has become flexible,
reliable, and low-cost. Currently, the universal and mature
FIGURE 2. Dynamic electrocardiogram system. wireless network communication standards mainly include
wireless HART, WIA-AP, and ISA100.11a. The application
of industrial networks is complex, and it is difficulty to intro-
communication technologies (e.g., low price and low energy duce a generic wireless network communication standard.
consumption) [37]. The industrial wireless network technology should have low
Various kinds of special sensors are used to collect data latency, high reliability, and high synchronous accuracy when
in manufacturing area, wherein devices are independent of dealing with a control service, and should have high access
each other. The collected data are heterogeneous with uneven density and low power consumption in data acquisition. The
quality. Therefore, the communication interface of intelli- networks should also have a high transmission rate of inter-
gent equipment should be compatible with a series of com- active service.
munication protocols. For instance, it should be compatible Reliability and latency are the core requirements of indus-
with OPC, Open Database Connectivity (ODBC), RS232, trial wireless network communication [41], [42]. The wireless
Dynamic data exchange (DDE), etc., which are used to con- network technology has gradually permeated into the indus-
nect control systems such as Supervisory Control and Data trial field including data acquisition and production control.
Acquisition (SCADA), Distributed Control System (DCS), The workshop is characterized by a complicated electro-
and Process Control System (PCS). In addition, data acquisi- magnetic environment. Meanwhile, data transmission may
tion devices should be easily set up and their interfaces should be unsynchronized with the clock causing packet loss and
be flexible and scalable. The manufacturing resources should delay or some other interference factors, which have negative
support a fine-grained data acquisition in process visibility effect on QoS of IWSNs. Niu et al. [43] presented a reli-
system. As show in Fig. 2, the equipment operating state is able reactive routing enhancement to increase the resilience
similar to the human heart beating. The equipment health of WSNs/IWSNs to link dynamics, which provides reliable
management system, which is based on the monitoring equip- and energy-efficient packet delivery regardless the unreliable
ment status obtained by the ‘‘electrocardiograph (EGC)’’, can wireless links. Therefore, both high reliability and low deliv-
predict system failure and arrange necessary maintenance ery latency are guaranteed. Dobslaw et al. [44] proposed a
activity in advance. However, current pattern of data acqui- generic heuristic scheduling algorithm extension to improve
sition is still a bottleneck of intelligent manufacturing. the end-to-end reliability of WSNs. In summary, IWSNs
are complex dynamical networks which couple real-time
IV. NETWORK LAYER control with high-speed communication using a multi-node
Industrial networks represent the integration of various kinds collaboration.
of network technologies such as field bus and sensor net- Energy efficiency is an important factor that affects adapt-
works. The network layer, which is characterized by per- ability of wireless sensor networks [45]. Due to the large
ception and control, plays an important role in the smart number of nodes in IWSNs, a dynamic scheduling is needed
factory. Due to the improvements of cloud computing tech- in intelligent manufacturing. Gao et al. [46] improved energy
nology [38], real-time and reliable network techniques are efficiency with the coded preambles in low-power sensor
required for data transmission, information sharing between networks. Lee et al. [47] proposed FlexiCast which uses an
intelligent equipment, and manufacturing cloud platform. energy-efficient method to check the integrity of software
The advanced information technologies (e.g., IWSNs and objects, and it significantly reduces energy consumption for
field bus) and their related technologies provide an impor- both software object updating and modification checking.
tant way to meet above mentioned requirements. Field buses Luo et al. [48] achieved energy saving using an opportunistic
(e.g., Foundation Fieldbus (FF), Profibus, and Hart) grad- routing algorithm that ensures a minimal-power low-residual
ually meet enterprise requirements of open, universal and energy, and which is focused on minimization of energy con-
compatible networks, and most of them have already been sumption and maximization of network lifetime of data relay

VOLUME 6, 2018 6509

 B. Chen et al.: Smart Factory of Industry 4.0: Key Technologies, Application Case, and Challenges

in one-dimensional queue networks. The current researches agent is characterized by autonomy, heterogeneity and decen-
on energy efficiency are mainly based on following aspects: tralization [22]. Leitão et al. [52] pointed out that OPC UA
improvement of data forwarding mechanism of sensor itself, is transferred from the original communication architecture
optimization of deployment of network nodes, and devel- (e.g., the Component Object Model (COM) standard) to the
opment of new energy-saving algorithms. Along with the service-oriented architecture with a cross platform, in which
expanded deployment of IWSNs to enhance energy efficiency way the communication among equipment and intelligent
more attention should be paid to continuity. manufacturing system is provided. With the standardization
Since the wireless communication has a character of radio of OPC UA, a new opportunity for multi-agent technology
broadcasting, wireless networks are easily accessed by both is created. Girbea et al. [53] designed a service-oriented
authorized users and unauthorized visitors. For that reason, architecture for optimization of industrial application, where
IWSNs are more vulnerable than industrial wire networks, OPC UA is used to connect everysub-manufacturing system
and invasive intrusion may lead to serious consequences for and ensure real-time communication between devices in stan-
smart factory. Sun et al. [49] proposed a fountain-coding (FC) dardized and unified manner. Fernbach et al. [54] proposed an
aided relaying scheme, where a secure cooperative transmis- OPC UA information modelling framework which provides a
sion is guaranteed by accumulating the required number of holistic information base. Hence, the multi-agent architecture
FC packets. Zou and Wang [50] proposed an optimal sensor can integrate agent-based system into the existing manufac-
scheduling scheme to improve the physical-layer security of turing site using a semantic method.
industrial WSNs against the eavesdropping attack, and they The multi-agent system is a system of coordinated solving
also conducted an asymptotic intercept probability analy- of large-scale complex problems with the cooperation of
sis to characterize diversity gains of round-robin scheduling every agent. Besides data transmission, the OPC UA can be
and optimal sensor scheduling. Dong and Liu [51] proposed used to integrate production data into manufacturing environ-
robust and secure time-synchronization protocol to defend ment and provide semantic annotation for every agent. Due
from Sybil attacks to the sensor networks. Due to limited to the construction of data source and receiver, the OPC UA
computing power and networks resources, it is impossible to can allocate all manufacturing resources including embedded
apply generic security protocols and mechanisms directly to system to the specific areas and extensible computing nodes
the IWSNs. Therefore, IWSNs are faced with greater security through the address space and pre-defined model. Further,
challenges than public networks. the OPC UA solves the problem related to unified access
The equipment monitoring and information interaction are to the information of different systems. The OPC UA-based
convenient for smart factory with IWSNs because they accel- interaction in multi-agent systems makes all agent coordinate
erate the progress of intelligent manufacturing. The IWSNs with each other to solve problems in a parallel manner, which
should meet the requirements for reliability and real-time effectively improves the problem solving efficiency. Thus,
data transmission. Because of uncertainty of position, energy the development of OPC UA is of high importance because
efficiency of IWSNs constrains its expansion. When a large- of redundancy, which is featured with adjustable overdue
scale equipment accesses to the industrial networks, network settings, error detection, and automatic correction. Therefore,
security is very important. In the IWSNs scenarios such as the communication tools which are in accordance with the
mass connection, low-power consumption, and wide cover- OPC UA standard can deal with the communication errors
age, new wireless communication technologies (e.g., NB-IoT, and failures easily.
5G, LTE, and 3GPP) provide low latency and high reliabil-
ity. However, certain measures should be taken to improve 2) SOFTWARE DEFINED INDUSTRIAL NETWORKS
robustness of signal propagation in complex electromagnetic Network technology is used to integrate internal and exter-
environment and extend working hours of networks nodes. nal resources in smart factory in order to form compre-
Accordingly, networks standardization and security defense hensive, unified, and high covering manufacturing resource.
need further improvement. Therefore, it is crucial to achieve network scalability and
adaptive transmission [55], [56]. Making network config-
B. RELATED TECHNOLOGIES uration flexible is a key issue for researchers. The SDNs
The application of network technology in smart factory pro- simplify the required hardware with centralized control
vides reliable and supportable information service. However, by a software, which facilitates network management and
the implementation of industrial internet still faces with many satisfies requirements of dynamic networks for intelligent
problems such as information interaction between equipment, manufacturing [57], [58].
flexible configuration of networks, and transmission delay. As the IIoT advances, the network becomes overloaded
Some of key related technologies are described in detail in due to a large number of connections between heterogeneous
the following. equipment. Bizanis and Kuipers [59] supposed that SDNs
and networks virtualization technology provide scale and
1) OPC UA-BASED INTERACTION IN MULTI-AGENT SYSTEMS versatility necessary for IoT service, and they pointed out that
The intelligent manufacturing system is a multi-agent system SDNs could be used together with machine learning tech-
which consists of task-driven intelligent equipment, where nology to make network more intelligent and self-adaptive.

6510 VOLUME 6, 2018

B. Chen et al.: Smart Factory of Industry 4.0: Key Technologies, Application Case, and Challenges

Wan et al. [60] proposed a software-defined IIoT architec- control of communication system. In common cellular net-
ture which can determine network resource allocation and works, an intelligent equipment can use D2D communication
accelerate information exchange by an easily customizable technology for direct communication via an isotropic antenna
networking protocol. Using this architecture, some innovative at every base station. This technology provides a new way
industrial applications can be realized through well-defined for low-latency communication, large data transmission, and
APIs. The OpenFlow [61], [62], which is a novel switched massive access to the mobile terminal. Holfeld et al. [68]
internetwork model, promotes implementation of SDNs [63]. stated that LTE-based wireless technology can provide new
The application of OpenFlow drastically reduces complexity services for factory automation. On the other hand, 5G com-
of control plane, increases programmability and extensibility, munication technology brings new opportunities to the auto-
and provides a gradual adoption path [64]. Though the Open- mated wireless communication technology. Liu et al. [69]
Flow is not a sole standard that supports SDN technology, its described the challenges and potential solutions to the imple-
rules have been generally accepted. Namely, the OpenFlow- mentation of machine-type communication system and pre-
based SDN deployment has been already used in industrial sented a roadmap of current cellular technologies toward
scenarios, such as intranets and data center. 5G mobile systems capable of fully-machine type commu-
Currently, the basic network system of manufacturing nication(MTC). Pratas and Popovski [70] proposed network-
enterprises is consisted of different types of communica- assisted D2D schemes that enable the cooperation between
tion devices (e.g., routers, switches, and services). There- machine-type devices and standard cellular devices, and meet
fore, present network management problems are as follows: the MTC outage requirement by maximizing the rate of
(1) work-flow of network security is complex [65], and it is broadband services. Bagaa et al. [71] proposed an opti-
difficult to locate network failure; (2) traditional networks mal solution that uses D2D communication to reduce the
cannot effectively support a real-time cloud manufacturing; overhead of MTC devices in 5G networks, and which aims
in addition, it is crucial to establish the coordination in multi- to decrease energy consumption and data transfer delay at
layer network; and lastly, (3) it is difficult to adjust net- eNodeBs. Yu et al. [72] proposed a D2D-communication-
work bandwidth in real time according to the flow demand based system underlying cellular networks to improve local
of data acquisition, which leads to low utilization of net- services, where optimal resource allocation and power con-
work resources. The solutions to the above problems require trol between cellular and D2D connections were analyzed for
an intelligent network management. The SDNs framework, different resource sharing modes.
which makes network programmable by separating the con-
trol layer from the data transmission layer is presented in
Fig. 3. This separation meets demand on centralized control
of network and increases flexibility of configuration and
operation of data center. The OpenFlow-based SDNs pro-
vide a good platform for innovative network applications and
development of IIoT.

FIGURE 4. The D2D communication in smart factory.

Once the D2D communication link is established in com-

munication network, data transmission is free of core device
and interference from the middleware, which can reduce
pressure on core network in communication system, improve
FIGURE 3. The SDN framework in smart factory.
spectrum utilization and networks throughput, and expand
network capacity. The D2D-based communication in smart
factory is presented in Fig. 4. The D2D communication tech-
3) DEVICE TO DEVICE COMMUNICATION nology brings high-quality network services to the intelli-
The contradictions between rapid growth of intelligent equip- gent manufacturing. Although, a small-scale distribution of
ment and limited bandwidth of industrial networks become data packets is more convenient, there are still many chal-
increasingly prominent in smart factory [66], [67]. The D2D lenges related to the introduction of D2D communication
communication refers to the communication wherein devices technology into cellular networks. For instance, it is difficult
directly exchange information with neighbors under the to switch network interaction from D2D communication to

VOLUME 6, 2018 6511

 B. Chen et al.: Smart Factory of Industry 4.0: Key Technologies, Application Case, and Challenges

cellular communication, and spectrum reuse in D2D commu- system interacts with physical system with high stability.
nication causes interference. Additionally, edge computing makes a full use of embed-
ded computing of terminal-side equipment, which provides
4) EDGE COMPUTING autonomy of equipment with a disperse treatment. Mean-
Edge computing denotes an open platform with many fea- while, the edge computing that is coordinated with cloud
tures such as networking, computing, storage, and appli- computing increases intelligence of the entire manufacturing
cation. Edge computing is performed at the network edge system. As edge intelligence advances, communication and
near to device or data source. Moreover, edge computing computing performances in distributed perception, decision-
provides intelligent services that meet key requirements of making and control will change significantly.
intelligent manufacturing for agile connection, real-time pro-
cessing, data cleaning, and privacy protection [73], [74]. V. DATA APPLICATION LAYER
Brito et al. [75] supposed that deployment of programmable The semantic association between manufacturing data is
fog nodes, which is attributed to inter-node Peer-to-Peer established by ontology model [78]. The essence of data
(P2P) communication and services orchestration without cen- application is to discover knowledge from data resources
tralized control, could increase flexibility, reliability and effi- and build the industrial value chain. The industrial big data
ciency. Pizoń et al. [76] argued that edge computing enables mainly include structured data and semi-structured data.
dynamic monitoring and control of manufacturing process. As data mining technology advances, data-driven innovation
Meanwhile they pointed out that edge computing requires will further promote intelligent manufacturing.
special IT ecosystem to build dynamic systems for transfer
of production data. Georgakopoulos et al. [77] presented a A. ONTOLOGY-BASED MANUFACTURING MODEL
manufacturing roadmap of IoT and edge cloud computing to Considering the large amount of manufacturing resources,
address elastic and virtual manufacturing resources, which resources concept may vary among different perspectives.
provides opportunities for real-time monitoring of production A formal description is presented by ontology for domain
KPIs and smart inventory management. knowledge. Namely, ontology is a semantic representation of
In the context of distributed computing based on hetero- related concepts and their relationship in intelligent manufac-
geneous networks, the concept of edge computing is closely turing. It is significant to construct the ontology for knowl-
related to distributed autonomy in industry. Edge computing edge sharing, reuse, and reasoning. Chungoora et al. [79]
combines disperse treatment with centralized upload. This presented a model-driven method with ontology to achieve
mode provides an effective use of network bandwidth and the interoperability and the knowledge sharing for manufac-
ensures autonomy, security and robustness of manufactur- turing system across multiple platforms in product lifecycle.
ing system. Fig. 5 shows the generic architecture of edge Usman et al. [80] supposed that a verifiable formal semantic
computing. The middle layer is edge computing platform. base could solve the problem of inappropriate interpretation
The lower layer controls manufacturing equipment with the of manufacturing concepts, and that formal manufacturing
characteristics of low latency, interoperability, autonomy, and reference ontology is a key component of future interoperable
adaptability. manufacturing system. Wang et al. [81] proposed a cloud-
based manufacturing semantic model which can be used
to obtain general task ontology construction and task sub-
ontology matching. Alsafi and Vyatkin [82] proposed an
approach for fast reconfiguration of modular manufacturing
systems, which is based on the reconfiguration agent with
ontology. The ontology-based model achieves standardized
terms and semantic knowledge for decision-making or other
applications in intelligent manufacturing [83].
The ontology-based modeling of manufacturing resources
provides a novel technical reference for construction of smart
factory. Due to the improvement of manufacturing system
configuration, the ontology-based model promotes event-
driven interoperability in smart factory. Ontology- based
applications are able to optimize scheduling for manufac-
turing resources and provide a semantic basis for consistent
FIGURE 5. The generic architecture of edge computing. description of interactions between different applications.
The modeling language for manufacturing ontology (e.g., the
The upper layer reduces pressure to the core network, Web Ontology Language) has rich semantic expression
optimizes data transmission, and provides support for appli- ability and support reasoning mechanism. Additionally,
cation extension. The edge computing integrates opera- the ontology-based modeling for intelligent manufactur-
tion technology with ICT, while its distributed control ing should align the following criteria: clarity, consistency,

6512 VOLUME 6, 2018

B. Chen et al.: Smart Factory of Industry 4.0: Key Technologies, Application Case, and Challenges

FIGURE 7. Open application architecture for active maintenance based on

manufacturing big data.

FIGURE 6. Ontology-based domain modeling method.

built models for bearing faults prediction based on historical
wind turbine data.
extendibility, coding minimum deviation, and ontology min- The regular maintenance in traditional mode just wastes
imum commitment. resource and time. In contrast, active maintenance based on
As shown in Fig. 6, the ontology-based modeling combines big data can provide necessary maintenance service. The
knowledge base with database to achieve semantic reasoning, active maintenance reduces machine downtime, optimizes
and this method decouples data processing from specific resource utilization, and increases production. In the mean-
applications. The data-centric architecture has both flexibility time, the maintenance cost is also reduced. The open appli-
and universality. The ontology-based manufacturing model cation architecture of active maintenance based on manufac-
provides a new research direction in fault diagnosis, equip- turing big data is presented in Fig. 7. The requirements for
ment health prediction, and active preventive maintenance in active maintenance of production facilities are as follows:
smart factory. (1) improve intelligence of workshop equipment; (2) coop-
erate with domain experts; and (3) express knowledge pro-
B. APPLICATIONS OF BIG DATA IN MANUFACTURING fessionally. The machine learning and statistical analysis
The big data in smart factory mainly include real-time are the mainstream approaches of active maintenance.
sensor data, machine log, and manufacturing process data, However, many factors should be considered. Such as uncer-
which have large volume, multiple sources, and spare value. tainty of production process, integration of different meth-
In the context of intelligent manufacturing, the applications ods and model applicability. Presently, active maintenance
of big data develop rapidly in industrial supply chain anal- of complex equipment still lacks the effective methods, thus
ysis and optimization, product quality control, and active further improvement is needed.
maintenance [84]–[87].
2) PRODUCT DESIGN OPTIMIZATION BASED ON
1) ACTIVE MAINTENANCE BASED ON BIG DATA MANUFACTURING BIG DATA
Proper operation of intelligent equipment is a guarantee With the development of IoT technology, it becomes conve-
of continuous production in the smart factory, where oper- nient to collect the product data. At present, the optimization
ation conditions have an important impact on equipment based on big data in manufacturing is effective on the stages
utilization and productivity. Data mining technology brings of logistics, warehousing, and selling. Analysis technology
new breakthroughs to failure prediction and active main- based on big data is used in product design, which takes
tenance. Wan et al. [88] proposed a manufacturing big advantage of knowledge discovery and trend prediction of
data solution for active preventive maintenance in manu- data. Graening and Sendhoff [92] proposed a shape mining
facturing environment, which combines a real-time active as a framework, based on engineering design data, which
maintenance mechanism with an off-line prediction method. was applied to the passenger car design. Williams et al. [93]
Regev and Benson-Karhi [89] proposed a segmented model developed a big data management infrastructure with seman-
for preventive maintenance of semiconductor manufac- tic technologies, which provides unified data access layer and
turing equipment, namely they presented both paramet- consistent approach to analytic execution. Bae and Kim [94]
ric and non-parametric models for preventive maintenance. proposed the apriori and C5.0 algorithms for data mining,
Hashemian and Bean [90] discussed the limitations of time- and mining results were attributed as knowledge to provide
based equipment maintenance methods and advantages of the suggestions for product design and marketing. However,
predictive and online maintenance techniques for early iden- if data receiver and feedback mechanism are added to the
tification of equipment failure. Xiong et al. [91] used data traditional product, the product itself will become the data
mining to identify bearing faults in wind turbines, and they source for design optimization without affecting its use.

VOLUME 6, 2018 6513

 B. Chen et al.: Smart Factory of Industry 4.0: Key Technologies, Application Case, and Challenges

In addition, the product logs, maintenance data, and user data • The physical resources layer was consisted of intelli-
will become the valuable resources to build product informa- gent equipment, conveyor equipment, packing products.
tion model. The scientific and reasonable product evaluation It was mainly responsible for execution of tasks such as
results can be obtained by data fusion and data mining, thus processing, monitoring, and assembling. The manufac-
these technologies represent the basis for design of a next- turing process information was the primary data source
generation product. for upper application.
Data mining helps designers to transform big data into • The cloud service layer contained the cloud platform
enlightening knowledge using selection, analysis and mod- (a service cluster system based on the Hadoop archi-
eling methods. The data fusion technology can analyze the tecture) which provided data storage and computing
same object based on multiple data resource. The data mining resource for data application. The ontology model of
technology is introduced in optimization of product design packing line was built on the cloud platform, and a
because it can deal with fuzzy and uncertain reasoning results relationship between objects was established in two
caused by a designer. However, the product optimization dimensions: structure and interaction. The Protégé was
design based on big data lacks the knowledge discovery used to build application-oriented ontology, and com-
tools. Therefore, we cannot take full advantage of potential plex constraints were expressed by semantic web rule
information from design examples. The decision-making, language. The manufacturing data were uploaded to the
which is based on existing knowledge, is less flexible in cloud platform to form a semantic data model. Based on
design optimization. The product design optimization based the knowledge base, the Jena tool was used to reason
on big data requires multidisciplinary knowledge. Therefore, the equipment operating mode. This layer provides sup-
the other interdisciplinarities associated with big data, such as ports for the fault alarm, the resource allocation, and the
deep learning and artificial intelligent, have become impor- scheduling optimization.
tant research directions in product design optimization. • The terminal layer mainly included end-user devices,
such as smart phones, desktop computers and electri-
VI. APPLICATION CASE STUDY cal boards, which were distributed in workshop, office,
From the view of intelligent manufacturing, manufacturing monitoring center and other regions. Terminal devices
equipment should be equipped with the abilities of edge com- were used to visualize the results of cloud processing,
puting, environment perception, and coordination between and they supported remote monitoring of operation and
equipment. The smart factory, which is a cyber-physical maintenance. Moreover, customers were able to check
production system (CPPS), integrates intelligent sensors, the order in real time using the intelligent terminal.
embedded terminal systems, intelligent control system, and • The network layer was used to connect layers within
communications facilities. The peer to peer interaction smart factory. According to the distributed control,
(e.g., person to equipment, equipment to equipment, service the connection between controller and actuator was
to service) is achieved by CPPS. Therefore, building of smart implemented by field bus, Modbus, and EtherCAT. The
factory should take into consideration manufacturing char- connection between equipment was achieved by
acteristics to meet rapidly-changing market needs [95], [96]. the combination of Ethernet and DDS, which formed
In the following, we use the laboratory prototype platform as the self-organized network. The connection between
an example to explore typical characteristics of smart factory. equipment and cloud platform was implemented by
The laboratory prototype platform, which represents a candy integration of Ethernet and OPC UA, which provided
packing production line, is shown in Fig. 8. data interaction.
The basic flow of candy packing is as follows. Firstly, cus-
tomers choose candy and purchase it online. Then, the order
information is sent directly to the manufacturing system.
Finally, the completed order is automatically transported by
logistical system of smart factory. The candy packaging line
has typical characteristics of smart factory, such as high
interconnection, dynamical reconfiguration, deep integration,
and so on.
In the experiment, we took into account factors such as
equipment availability, equipment performance, and qual-
ified rate of products. Next, we developed a cloud-aided
FIGURE 8. Prototype platform for the validation of key technologies. manufacturing system, which gave suggestions on schedul-
ing optimization of self-organized equipment. The six-month
According to the smart factory architecture, the prototype experiment was performed on the equipment of candy pack-
platform had four layers including the physical resources aging line. The smart meter is used to measure the total
layer, the cloud service layer, the terminal layer, and the electrical energy for the equipment in the lab. We calculated
network layer. The prototype layers were as follows. the related parameter every day, and the average was took

6514 VOLUME 6, 2018

B. Chen et al.: Smart Factory of Industry 4.0: Key Technologies, Application Case, and Challenges

The high-speed operation in workshop needs higher standards

of data acquisition. Namely, data acquisition is the basis of big
data analytics, where physical resource should: (1) support
fine-grained and efficient data acquisition, and achieve visi-
bility of manufacturing process; (2) integrate heterogeneous
data in a unified system by generic protocols (e.g., RFID,
ZigBee, and NFC); and (3) improve extensibility of controller
for access to the core industrial networks.

B. DEEP INTEGRATION NETWORKS

The IIoT facilitates a deep integration of information and
industrialization. The advanced IIoT technology is important
for the implementation of the smart factory. The mature and
generic standards have not been formed yet in the field of
FIGURE 9. The overall effectiveness and influential factors of prototyping IWSNs, and the standardization process of IWSNs should be
platform. promoted continuously. In the complicated electromagnetic
environment, data transmission should meet the requirements
as the value of the month. As shown in Fig. 9, the Over- of reliability and real time equipment control. Due to the
all Equipment Effectiveness (OEE) of tested equipment is limited energy, energy efficiency is a key issue that affects
improved from 0.42 to 0.82 using the cloud-aided manu- the deployment of IWSNs. Moreover, because of the access
facturing system. Compared to the centralized scheduling, of large-scale devices, network security also becomes very
the self-organized scheduling with the cloud-aided manufac- important. As the information technologies have advanced
turing system has more remarkable effect. In general, the (e.g., NB-IoT, 5G, LTE-Advanced, and 3GPP), a signifi-
candy packing line is an example which was used to explore cant progress has been made in industrial wireless networks.
flexibility, efficiency, and transparency of smart factory. These progress has provided new solutions for key issues
VII. ISSUES AND CHALLENGES
of IWSNs, such as reliability, real-time performance, energy
Due to the rapid progress of manufacturing, smart factory efficiency, and security strategy.
should be flexible and reliable, and satisfy the high qual- Development of information technologies has brought
ity standards. The technological breakthroughs bring many opportunities to the intelligent manufacturing. The OPC
opportunities for the implementation of intelligent manufac- UA-based interaction facilitates the coordination between
turing. However, there are still some issues and challenges. intelligent agents because a multi-agent system can solve
problems in a parallel way. The OpenFlow-based SDN tech-
A. INTELLIGENT REQUIREMENTS OF EQUIPMENT nology provides more flexible solution for network config-
Due to the foundation place of the underlying equipment, uration, which enhances network management ability. The
it is crucial to monitor and control the underlying man- D2D technology makes communication between devices
ufacturing resource for reconfiguration of production line, more efficient and expands network capacity. The edge com-
dynamic scheduling, and information fusion in smart fac- puting equips terminal system with decision-making ability
tory. Therefore, it is necessary to improve the intelligence and autonomy. Additionally, high transmission rate of data,
level of manufacturing equipment. The configurable con- low duty cycle, and IP network availability are the require-
troller and self-reconfigurable robots can provide potential ments of network layer which denote the foundation for
solutions to function expansion of manufacturing units. In the ubiquitous communication in smart factory.
context of hybrid production, the coordination and informa-
tion interaction among the multi-module manufacturing units C. KNOWLEDGE-DRIVEN MANUFACTURING
should be explored. The optimized combination of programs The large amount of manufacturing data provides a com-
should be made to enhance the workshop efficiency. The prehensive description of the smart factory, but manufactur-
intelligent equipment should be able to collect production ing data cannot be utilized directly due to high dimension,
information, provide compatible data interface, and support variable metric, and high noise. Consequently, it is impor-
generic communication protocol. In addition, the equipment tant to define the data semantic through the manufacturing
could perceive manufacturing environment and cooperate glossary. The domain ontology provides a potential solution
with other equipment in smart factory. Flexible manufactur- to data semantic for data application. Using the big data of
ing is a typical feature of smart factory, but there are still intelligent manufacturing, the active maintenance of equip-
many problems such as strong proprietary of production line, ment, the optimization design for manufacturing product,
dynamic scheduling, and tight coupling between functions and the optimization of production line is achieved in the
and devices. smart factory. The knowledge-driven manufacturing brings
In the context of intelligent manufacturing, the data opportunities to transformation from traditional industry to
generated by intelligent equipment is mostly unstructured. intelligent industry, meanwhile the data mining technology is

VOLUME 6, 2018 6515

 B. Chen et al.: Smart Factory of Industry 4.0: Key Technologies, Application Case, and Challenges

a serious challenge to enterprises. Moreover, the data-based [7] J. Zhou, ‘‘Intelligent manufacturing-main direction of ‘Made in China
optimization of product design needs to add data receiver and 2025,’’’ China Mech. Eng., vol. 26, no. 17, pp. 2273–2284, 2015.
[8] J. P. Holdren, T. Power, G. Tassey, A. Ratcliff, and L. Christodoulou,
feedback mechanism to the traditional product, and then the ‘‘A National strategic plan for advanced manufacturing,’’ US Nat. Sci.
product itself will become the data source. Consequently, the Technol. Council, Washington, DC, USA, Tech. Rep., 2012. [Online].
product will become a participant in the process of data col- Available: http://www.docin.com/p-391856652.html
[9] J. Wan, M. Yi, D. Li, C. Zhang, S. Wang, and K. Zhou, ‘‘Mobile services
lection, which provides technical information to the product for customization manufacturing systems: An example of industry 4.0,’’
designer. IEEE Access, vol. 4, pp. 8977–8986, 2016.
In order to achieve a knowledge-based intelligent manu- [10] N. Benkamoun, W. ElMaraghy, A.-L. Huyet, and K. Kouiss, ‘‘Architecture
framework for manufacturing system design,’’ Procedia CIRP, vol. 17,
facturing, the manufacturing entity should be able to provide pp. 88–93, 2014.
data collection, data fusion, and extraction of manufacturing [11] A. Radziwon, A. Bilberg, M. Bogers, and E. S. Madsen, ‘‘The smart fac-
resource characteristics. The smart factory should integrate tory: Exploring adaptive and flexible manufacturing solutions,’’ Procedia
Eng., vol. 69, pp. 1184–1190, Jan. 2014.
data resources (e.g., supply chains, product data, and logistic [12] J. Lin et al., ‘‘Application technology of cloud manufacturing for
data) into service platform, which provides product services aerospace complex products,’’ Comput. Integr. Manuf. Syst., vol. 22, no. 4,
pp. 883–896, 2016.
such as sales forecasting and quality analysis. In general, data [13] J. Wan, S. Tang, Q. Hua, D. Li, C. Liu, and J. Lloret, ‘‘Context-
mining and knowledge discovery provide a scientific decision aware cloud robotics for material handling in cognitive industrial Inter-
for planning and scheduling of manufacturing product. net of Things,’’ IEEE Internet Things J., to be published, doi: 10.1109/
JIOT.2017.2728722.2017.
[14] J. Wan, S. Tang, H. Yan, D. Li, S. Wang, and A. V. Vasilakos,
VIII. CONCLUSIONS ‘‘Cloud robotics: Current status and open issues,’’ IEEE Access, vol. 4,
The smart factory is an intelligent production system which pp. 2797–2807, 2016.
[15] Y. Lyu and J. Zhang, ‘‘Big-data-based technical framework of smart
utilizes the integration of manufacturing and services. It inte- factory,’’ Comput. Integr. Manuf. Syst., vol. 22, no. 11, pp. 2691–2697,
grates communication process, computing process, and con- 2016.
trol process to meet the industrial demands. In this paper, [16] D. Zhang, J. Wan, C.-H. Hsu, and A. Rayes, ‘‘Industrial technologies and
applications for the Internet of Things,’’ Comput. Netw., vol. 101, pp. 1–4,
we proposed the hierarchical architecture of smart factory Jun. 2016.
according to the newly research. Then the main issues of [17] Z. Shu, J. Wan, D. Zhang, and D. Li, ‘‘Cloud-integrated cyber-physical
key technologies involved in the physical resource layer, the systems for complex industrial applications,’’ Mobile Netw. Appl., vol. 21,
no. 5, pp. 865–878, 2016.
network layer, and the data application layer in the smart [18] M. Fiasché, G. Ripamonti, F. G. Sisca, M. Taisch, and A. A. Valente, ‘‘Man-
factory are analyzed, and some application cases were dis- agement integration framework in a shop-floor employing self-contained
cussed to explore potential solutions for key technologies. assembly unit for optoelectronic products,’’ in Proc. IEEE 1st Int. Forum
Res. Technol. Soc. Ind. Leveraging Better Tomorrow (RTSI), Turin, Italy,
Finally we established verification platform by means of the Sep. 2015, pp. 569–578.
proposed key technologies, which showed that the OEE ration [19] B. Piranda and J. Bourgeois, ‘‘A distributed algorithm for reconfigura-
is improved. The compound talents and multi-field coopera- tion of lattice-based modular self-reconfigurable robots,’’ in Proc. 24th
Euromicro Int. Conf. Parallel, Distrib., Netw.-Based Process. (PDP),
tion is required for the implementation of smart factory. The Heraklion, Greece, Feb. 2016, pp. 1–9.
research of key technologies not only refers to the integration [20] A. Valente, ‘‘Reconfigurable industrial robots—An integrated approach to
of information technology as a main body, but also includes design the joint and link modules and configure the robot manipulator,’’
in Advances in Reconfigurable Mechanisms and Robots II, vol. 36. Cham,
the traditional disciplines such as control theory, mechanical Switzerland: Springer, 2016, pp. 779–794.
technology, and materials and energy. As the big data tech- [21] S. Wang, C. Zhang, and J. Wan, ‘‘A smart factory solution to hybrid
nology advances, the data-based virtual manufacturing mode production of multi-type products with reduced intelligence,’’ in Proc.
IEEE Inf. Technol., Netw., Electron. Autom. Control Conf., Chongqing,
will improve product quality, increase production efficiency, China, May 2016, pp. 848–853.
and reduce energy consumption. Furthermore, the intelligent [22] S. Wang, J. Wan, D. Zhang, D. Li, and C. Zhang, ‘‘Towards smart fac-
manufacturing based on big data will lead to the revolution of tory for Industry 4.0: A self-organized multi-agent system with big data
based feedback and coordination,’’ Comput. Netw., vol. 101, pp. 158–168,
traditional industry. Jun. 2016.
[23] W. Dai, V. Vyatkin, J. H. Christensen, and V. N. Dubinin, ‘‘Bridging
service-oriented architecture and IEC 61499 for flexibility and interoper-
REFERENCES
ability,’’ IEEE Trans. Ind. Informat., vol. 11, no. 3, pp. 771–781, Jun. 2015.
[1] M. Brettel, N. Friederichsen, M. Keller, and M. Rosenberg, ‘‘How virtu- [24] M. G. Valls, I. R. Lopez, and L. F. Villar, ‘‘ILAND: An enhanced middle-
alization, decentralization and network building change the manufacturing ware for real-time reconfiguration of service oriented distributed real-time
landscape: An industry 4.0 perspective,’’ Int. J. Mech., Ind. Sci. Eng., vol. 8, systems,’’ IEEE Trans. Ind. Informat., vol. 9, no. 1, pp. 228–236, Feb. 2013.
no. 1, pp. 37–44, 2014. [25] M. Short and K. Burn, ‘‘A generic controller architecture for intelli-
[2] F. Li, J. Wan, P. Zhang, D. Li, D. Zhang, and K. Zhou, ‘‘Usage-specific gent robotic systems,’’ Robot. Comput.-Integr. Manuf., vol. 27, no. 2,
semantic integration for cyber-physical robot systems,’’ ACM Trans. pp. 292–305, 2011.
Embedded Comput. Syst., vol. 15, no. 3, 2016, Art. no. 50. [26] L. Morales-Velazquez, R. de J. Romero-Troncoso, R. A. Osornio-Rios,
[3] J. Wan, D. Zhang, Y. Sun, K. Lin, C. Zou, and H. Cai, ‘‘VCMIA: A novel G. Herrera-Ruiz, and E. Cabal-Yepez, ‘‘Open-architecture system based
architecture for integrating vehicular cyber-physical systems and mobile on a reconfigurable hardware–software multi-agent platform for CNC
cloud computing,’’ Mobile Netw. Appl., vol. 19, no. 2, pp. 153–160, 2014. machines,’’ J. Syst. Archit., vol. 56, no. 9, pp. 407–418, 2010.
[4] J. Wan, H. Yan, D. Li, K. Zhou, and L. Zeng, ‘‘Cyber-physical systems [27] W. Lepuschitz, A. Zoitl, M. Vallée, and M. Merdan, ‘‘Toward self-
for optimal energy management scheme of autonomous electric vehicle,’’ reconfiguration of manufacturing systems using automation agents,’’ IEEE
Comput. J., vol. 56, no. 8, pp. 947–956, 2013. Trans. Syst., Man, Cybern. C, Appl. Rev., vol. 41, no. 1, pp. 52–69,
[5] Europe 2020: A Strategy for Smart, Sustainable and Inclusive Growth, Jan. 2011.
European Commission, Brussels, Belgium, 2010. [28] M. Adamski and J. Tkacz, ‘‘Formal reasoning in logic design of recon-
[6] H. Lasi, P. Fettke, H.-G. Kemper, T. Feld, and M. Hoffmann, ‘‘Industry figurable controllers,’’ IFAC Proc. Volumes, vol. 45, no. 7, pp. 1–6,
4.0,’’ Bus. Inf. Syst. Eng., vol. 6, no. 4, pp. 239–242, 2014. 2012.

6516 VOLUME 6, 2018

B. Chen et al.: Smart Factory of Industry 4.0: Key Technologies, Application Case, and Challenges

[29] H. J. Kim, Y. K. Lee, Y. A. Jeon, and J. Y. Son, ‘‘ICT convergence- [52] P. Leitão, V. Marík, and P. Vrba, ‘‘Past, present, and future of indus-
based application service development to support the re-configurability trial agent applications,’’ IEEE Trans. Ind. Informat., vol. 9, no. 4,
of door trim assembly line,’’ in Proc. Int. Conf. Inf. Commun. Technol. pp. 2360–2372, Nov. 2013.
Converg. (ICTC), Jeju, South Korea, Oct. 2016, pp. 1126–1128. [53] A. Girbea, C. Suciu, S. Nechifor, and F. Sisak, ‘‘Design and implemen-
[30] E. Järvenpää, N. Siltala, and M. Lanz, ‘‘Formal resource and capability tation of a service-oriented architecture for the optimization of industrial
descriptions supporting rapid reconfiguration of assembly systems,’’ in applications,’’ IEEE Trans. Ind. Informat., vol. 10, no. 1, pp. 185–196,
Proc. IEEE Int. Symp. Assembly Manuf. (ISAM), Fort Worth, TX, USA, Feb. 2014.
Aug. 2016, pp. 120–125. [54] A. Fernbach, W. Kastner, S. Mätzler, and M. Wollschlaeger, ‘‘An OPC UA
[31] D. Gyulai, B. Kádár, A. Kovács, and L. Monostori, ‘‘Capacity management information model for cross-domain vertical integration in automation sys-
for assembly systems with dedicated and reconfigurable resources,’’ CIRP tems,’’ in Proc. IEEE Emerg. Technol. Factory Autom. (ETFA), Barcelona,
Ann., vol. 63, no. 1, pp. 457–460, 2014. Spain, Sep. 2014, pp. 1–8.
[32] K. K. Goyal and P. K. Jain, ‘‘Design of reconfigurable flow lines using [55] Z. Shu et al., ‘‘Traffic engineering in software-defined networking: Mea-
MOPSO and maximum deviation theory,’’ Int. J. Adv. Manuf. Technol., surement and management,’’ IEEE Access, vol. 4, pp. 3246–3256, 2016.
vol. 84, nos. 5–8, pp. 1587–1600, 2016. [56] J. Liu, J. Wan, B. Zeng, Q. Wang, H. Song, and M. Qiu, ‘‘A scalable
[33] D. Zhang, Y. Qian, J. Wan, and S. Zhao, ‘‘An efficient RFID search and quick-response software defined vehicular network assisted by mobile
protocol based on clouds,’’ Mobile Netw. Appl., vol. 20, no. 3, edge computing,’’ IEEE Commun. Mag., vol. 55, no. 7, pp. 94–100,
pp. 356–362, 2015. Jul. 2017.
[34] R. Y. Zhong, Q. Y. Dai, T. Qu, G. J. Hu, and G. Q. Huang, ‘‘RFID- [57] Y. Li and M. Chen, ‘‘Software-defined network function virtualization:
enabled real-time manufacturing execution system for mass-customization A survey,’’ IEEE Access, vol. 3, pp. 2542–2553, 2015.
production,’’ Robot. Comput.-Integr. Manuf., vol. 29, no. 2, pp. 283–292, [58] J. Liu, Q. Wang, J. Wan, J. Xiong, and B. Zeng, ‘‘Towards key issues of
2013. disaster aid based on wireless body area networks,’’ KSII Trans. Internet
[35] D. Zhang, Z. He, Y. Qian, J. Wan, D. Li, and S. Zhao, ‘‘Revisiting unknown Inf. Syst., vol. 7, no. 5, pp. 1014–1035, 2013.
RFID tag identification in large-scale Internet of Things,’’ IEEE Wireless [59] N. Bizanis and F. A. Kuipers, ‘‘SDN and virtualization solutions for the
Commun., vol. 23, no. 5, pp. 24–29, Oct. 2016. Internet of Things: A survey,’’ IEEE Access, vol. 4, pp. 5591–5606, 2016.
[36] V. Coskun, B. Ozdenizci, and K. Ok, ‘‘A survey on near field commu- [60] J. Wan et al., ‘‘Software-defined industrial Internet of Things in the con-
nication (NFC) technology,’’ Wireless Pers. Commun., vol. 71, no. 3, text of industry 4.0,’’ IEEE Sensors J., vol. 16, no. 20, pp. 7373–7380,
pp. 2259–2294, 2013. Oct. 2016.
[37] S. Choudhury, P. Kuchhal, R. Singh, and Anita, ‘‘ZigBee and Bluetooth [61] Flowgrammable: Driving the Next SDN Generation. Accessed:
network based sensory data acquisition system,’’ Procedia Comput. Sci., Jul. 15, 2015. [Online]. Available: http://flowgrammable.org/
vol. 48, pp. 367–372, Jan. 2015. sdn/openflow/
[38] M. Chen, Y. Ma, Y. Li, D. Wu, Y. Zhang, and C.-H. Youn, ‘‘Wearable 2.0: [62] The ONF is an Operator Led Consortium. Accessed: Jun. 21, 2017.
Enabling human-cloud integration in next generation healthcare systems,’’ [Online]. Available: http://www.opennetworking.org
IEEE Commun. Mag., vol. 55, no. 1, pp. 54–61, Jan. 2017. [63] K. Lin, W. Wang, X. Wang, W. Ji, and J. Wan, ‘‘QoE-driven spec-
[39] R. A. Gupta and M.-Y. Chow, ‘‘Networked control system: Overview trum assignment for 5g wireless networks using SDR,’’ IEEE Wireless
and research trends,’’ IEEE Trans. Ind. Electron., vol. 57, no. 7, Commun., vol. 22, no. 6, pp. 48–55, Dec. 2015.
pp. 2527–2535, Jul. 2010. [64] S. Das, G. Parulkar, and N. McKeown, ‘‘Why OpenFlow/SDN can suc-
[40] T. Sauter, ‘‘The three generations of field-level networks—Evolution ceed where GMPLS failed,’’ in Proc. Eur. Conf. Exhib. Opt. Commun.,
and compatibility issues,’’ IEEE Trans. Ind. Electron., vol. 57, no. 11, Amsterdam, Netherlands, Sep. 2012, pp. 1–3, paper Tu.1.D.1.
pp. 3585–3595, Nov. 2010. [65] Z. Shu, J. Wan, D. Li, J. Lin, A. V. Vasilakos, and M. Imran, ‘‘Security
[41] X. Li, D. Li, J. Wan, A. V. Vasilakos, C.-F. Lai, and S. Wang, ‘‘A review of in software-defined networking: Threats and countermeasures,’’ Mobile
industrial wireless networks in the context of industry 4.0,’’ Wireless Netw., Netw. Appl., vol. 21, no. 5, pp. 764–776, 2016.
vol. 23, no. 1, pp. 23–41, 2017. [66] M. Chen, J. Wan, S. Gonzalez, X. Liao, and V. C. M. Leung, ‘‘A survey
[42] C. Zou, J. Wan, M. Chen, and D. Li, ‘‘Simulation modeling of cyber- of recent developments in home M2M networks,’’ IEEE Commun. Surveys
physical systems exemplified by unmanned vehicles with WSNs navi- Tuts., vol. 16, no. 1, pp. 98–114, 1st Quart., 2014.
gation,’’ in Proc. 7th Int. Conf. Embedded Multimedia Comput. Technol. [67] L. Kong, D. Zhang, Z. He, Q. Xiang, J. Wan, and M. Tao, ‘‘Embracing
Service, Gwangju, South Korea, Sep. 2012, pp. 269–275. big data with compressive sensing: A green approach in industrial wireless
[43] J. Niu, L. Cheng, Y. Gu, L. Shu, and S. K. Das, ‘‘R3E: Reliable reactive networks,’’ IEEE Commun. Mag., vol. 54, no. 10, pp. 53–59, Oct. 2016.
routing enhancement for wireless sensor networks,’’ IEEE Trans. Ind. [68] B. Holfeld et al., ‘‘Wireless communication for factory automation: An
Informat., vol. 10, no. 1, pp. 784–794, Feb. 2014. opportunity for LTE and 5G systems,’’ IEEE Commun. Mag., vol. 54, no. 6,
[44] F. Dobslaw, T. Zhang, and M. Gidlund, ‘‘End-to-end reliability-aware pp. 36–43, Jun. 2016.
scheduling for wireless sensor networks,’’ IEEE Trans. Ind. Informat., [69] J. Liu et al., ‘‘High-efficiency urban traffic management in context-aware
vol. 12, no. 2, pp. 758–767, Apr. 2016. computing and 5G communication,’’ IEEE Commun. Mag., vol. 55, no. 1,
[45] D. Zhang, J. Wan, Q. Liu, X. Guan, and X. Liang, ‘‘A taxonomy of pp. 34–40, Jan. 2017.
agent technologies for ubiquitous computing environments,’’ KSII Trans. [70] N. K. Pratas and P. Popovski. (2013). ‘‘Low-rate machine-type commu-
Internet Inf. Syst., vol. 6, no. 2, pp. 547–565, 2012. nication via wireless device-to-device (D2D) links.’’ [Online]. Available:
[46] Y. Gao, W. Dong, L. Deng, C. Chen, and J. Bu, ‘‘COPE: Improving energy https://arxiv.org/abs/1305.6783
efficiency with coded preambles in low-power sensor networks,’’ IEEE [71] M. Bagaa, A. Ksentini, T. Taleb, R. Jantti, A. Chelli, and I. Balasingham,
Trans. Ind. Informat., vol. 11, no. 6, pp. 1621–1630, Dec. 2015. ‘‘An efficient D2D-based strategies for machine type communications
[47] J. Lee, L. Kim, and T. Kwon, ‘‘Flexicast: Energy-efficient software in 5G mobile systems,’’ in Proc. IEEE Wireless Commun. Netw. Conf.
integrity checks to build secure industrial wireless active sensor networks,’’ (WCNC), Doha, Qatar, Apr. 2016, pp. 1–6.
IEEE Trans. Ind. Informat., vol. 12, no. 1, pp. 6–14, Feb. 2016. [72] C.-H. Yu, K. Doppler, C. B. Ribeiro, and O. Tirkkonen, ‘‘Resource shar-
[48] J. Luo, J. Hu, D. Wu, and R. Li, ‘‘Opportunistic routing algorithm for relay ing optimization for device-to-device communication underlaying cellular
node selection in wireless sensor networks,’’ IEEE Trans. Ind. Informat., networks,’’ IEEE Trans. Wireless Commun., vol. 10, no. 8, pp. 2752–2763,
vol. 11, no. 1, pp. 112–121, Feb. 2015. Aug. 2011.
[49] L. Sun, P. Ren, Q. Du, and Y. Wang, ‘‘Fountain-coding aided strat- [73] W. Shi, J. Cao, Q. Zhang, Y. Li, and L. Xu, ‘‘Edge computing: Vision and
egy for secure cooperative transmission in industrial wireless sensor challenges,’’ IEEE Internet Things J., vol. 3, no. 5, pp. 637–646, Oct. 2016.
networks,’’ IEEE Trans. Ind. Informat., vol. 12, no. 1, pp. 291–300, [74] A. V. Dastjerdi, H. Gupta, R. N. Calheiros, S. K. Ghosh, and R. Buyya.
Feb. 2016. (2016). ‘‘Fog computing: Principles, architectures, and applications.’’
[50] Y. Zou and G. Wang, ‘‘Intercept behavior analysis of industrial wireless [Online]. Available: https://arxiv.org/abs/1601.02752
sensor networks in the presence of eavesdropping attack,’’ IEEE Trans. [75] M. S. De Brito, S. Hoque, R. Steinke, and A. Willner, ‘‘Towards pro-
Ind. Informat., vol. 12, no. 2, pp. 780–787, Apr. 2016. grammable fog nodes in smart factories,’’ in Proc. IEEE Int. Workshops
[51] W. Dong and X. Liu, ‘‘Robust and secure time-synchronization against Found. Appl. Self Syst., Augsburg, Germany, Sep. 2016, pp. 236–241.
sybil attacks for sensor networks,’’ IEEE Trans. Ind. Informat., vol. 11, [76] J. Pizoń and J. Lipski, ‘‘Perspectives for fog computing in manufacturing,’’
no. 6, pp. 1482–1491, Dec. 2015. Appl. Comput. Sci., vol. 12, no. 3, pp. 37–46, 2016.

VOLUME 6, 2018 6517

 B. Chen et al.: Smart Factory of Industry 4.0: Key Technologies, Application Case, and Challenges

[77] D. Georgakopoulos, P. P. Jayaraman, M. Fazia, M. Villari, and R. Ranjan, JIAFU WAN (M’11) has been a Professor with
‘‘Internet of Things and edge cloud computing roadmap for manufactur- the School of Mechanical and Automotive Engi-
ing,’’ IEEE Cloud Comput., vol. 3, no. 4, pp. 66–73, Jul. 2016. neering, South China University of Technology,
[78] P. Shvaiko and J. Euzenat, ‘‘Ontology matching: State of the art and future since 2015. He has directed 16 research projects,
challenges,’’ IEEE Trans. Knowl. Data Eng., vol. 25, no. 1, pp. 158–176, including the National Key Research and Develop-
Jan. 2013. ment Project, the National Natural Science Foun-
[79] N. Chungoora et al., ‘‘A model-driven ontology approach for manufactur-
dation of China, the High-level Talent Project
ing system interoperability and knowledge sharing,’’ Comput. Ind., vol. 64,
of Guangdong Province, and the Natural Science
no. 4, pp. 392–401, 2013.
[80] Z. Usman, R. I. M. Young, N. Chungoora, C. Palmer, K. Case, and Foundation of Guangdong Province. Thus far, he
J. A. Harding, ‘‘Towards a formal manufacturing reference ontology,’’ Int. has published over 140 scientific papers, includ-
J. Prod. Res., vol. 51, no. 22, pp. 6553–6572, 2013. ing over 80 SCI-indexed papers, over 20 IEEE transactions/journal papers,
[81] T. Wang, S. Guo, and C.-G. Lee, ‘‘Manufacturing task semantic model- 12 ESI Highly Cited Papers, and 4 ESI Hot Papers. His research inter-
ing and description in cloud manufacturing system,’’ Int. J. Adv. Manuf. ests include cyber-physical systems, industry 4.0, smart factory, indus-
Technol., vol. 71, nos. 9–12, pp. 2017–2031, 2014. trial big data, industrial robot and Internet of Vehicles. His research
[82] Y. Alsafi and V. Vyatkin, ‘‘Ontology-based reconfiguration agent for intel- results have been published in several famous journals, such as the IEEE
ligent mechatronic systems in flexible manufacturing,’’ Robot. Comput.- TRANSACTIONS ON INDUSTRIAL INFORMATICS, the IEEE COMMUNICATIONS SURVEYS
Integr. Manuf., vol. 26, no. 4, pp. 381–391, 2010. AND TUTORIALS, the IEEE Communications Magazine, the IEEE TRANSACTIONS
[83] X. Chang, R. Rai, and J. Terpenny, ‘‘Development and utilization of ON INTELLIGENT TRANSPORTATION SYSTEMS, the IEEE Network, the IEEE
ontologies in design for manufacturing,’’ J. Mech. Des., vol. 132, no. 2, WIRELESS COMMUNICATIONS, the IEEE SYSTEMS JOURNAL, the IEEE SENSORS
p. 021009, 2010. JOURNAL, the IEEE INTERNET OF THINGS JOURNAL, and the ACM Transac-
[84] J.-F. Wan, D. Li, H.-H. Yan, and P. Zhang, ‘‘Fuzzy feedback scheduling
tions on Embedded Computing Systems. He is a Senior Member of CMES
algorithm based on central processing unit utilization for a software-based
computer numerical control system,’’ Proc. Inst. Mech. Eng., B, J. Eng.
and CCF. He is the General Chair of the 2016 International Confer-
Manuf., vol. 224, no. 7, pp. 1133–1143, 2010. ence on Industrial IoT Technologies and Applications and the 7th EAI
[85] Z. Song, Y. Sun, J. Wan, and P. Liang, ‘‘Data quality management International Conference on Cloud Computing (CloudComp 2016). He is
for service-oriented manufacturing cyber-physical systems,’’ Comput. currently an Associate Editor for the IEEE ACCESS and on the Editorial
Elect. Eng., vol. 64, pp. 34–44, Nov. 2017, doi: 10.1016/j.compeleceng. Board of PLOS One. He is also a Managing Editor for the International
2016.08.010.2016. Journal of Autonomous and Adaptive Communications Systems and the
[86] W. Yuan, P. Deng, T. Taleb, J. Wan, and C. Bi, ‘‘An unlicensed taxi iden- International Journal of Arts and Technology and a Leading Guest Edi-
tification model based on big data analysis,’’ IEEE Trans. Intell. Transp. tor for several SCI-indexed journals, such as the IEEE SYSTEMS JOURNAL,
Syst., vol. 17, no. 6, pp. 1703–1713, Jun. 2016. the IEEE ACCESS, the Computer Networks (Elsevier), the Mobile Net-
[87] Y. Xu, Y. Sun, J. Wan, X. Liu, and Z. Song, ‘‘Industrial big data for works and Applications, the Computers and Electrical Engineering, and
fault diagnosis: Taxonomy, review, and applications,’’ IEEE Access, to be the Microprocessors and Microsystems. According to Google Scholar, his
published, doi: 10.1109/ACCESS.2017.2731945.2017. published work has been cited over 3800 times (H-index = 31). His SCI
[88] J. Wan et al., ‘‘A manufacturing big data solution for active preventive
other citations (sum of times cited without self-citations) reached 1104
maintenance,’’ IEEE Trans. Ind. Informat., vol. 13, no. 4, pp. 2039–2047,
(H-index = 19) times according to Web of Science Core Collection.
Aug. 2017.
[89] I. Regev and D. Benson-Karhi, ‘‘Segmentation models for the duration
of expected preventive maintenance in semiconductor fabs,’’ IEEE Trans. LEI SHU (SM’16) is currently a Lincoln Profes-
Semicond. Manuf., vol. 29, no. 3, pp. 223–229, Aug. 2016. sor with the University of Lincoln, U.K., and a
[90] H. M. Hashemian and W. C. Bean, ‘‘State-of-the-art predictive main- Distinguished Professor with Nanjing Agricultural
tenance techniques,’’ IEEE Trans. Instrum. Meas., vol. 60, no. 10, University, China. He is also the Director of the
pp. 3480–3492, Oct. 2011. NAU-Lincoln Joint Research Center of Intelligent
[91] J. Xiong, Q. Zhang, J. Wan, L. Liang, P. Cheng, and Q. Liang, ‘‘Data
Engineering. He has authored over 350 papers in
fusion method based on mutual dimensionless,’’ IEEE/ASME Trans.
related conferences, journals, and books in the
Mechatronics, to be published, doi: 10.1109/TMECH.2017.2759791.2017.
[92] L. Graening and B. Sendhoff, ‘‘Shape mining: A holistic data mining area of sensor networks. His main research field is
approach for engineering design,’’ Adv. Eng. Informat., vol. 28, no. 2, wireless sensor networks. He was a recipient of the
pp. 166–185, 2014. GlobeCom 2010 and ICC 2013 Best Paper Award
[93] J. W. Williams et al., ‘‘Semantics for big data access & integration: and the IEEE Systems Journal 2017 Best Paper Award. He has served as a
Improving industrial equipment design through increased data usability,’’ TPC member for over 150 conferences, such as ICDCS, DCOSS, MASS,
in Proc. IEEE Int. Conf. Big Data (Big Data), Santa Clara, CA, USA, ICC, GlobeCom, ICCCN, WCNC, and ISCC. He has also served as the
Oct./Nov. 2015, pp. 1103–1112. Co-Chair for over 50 international conferences and workshops, such as
[94] J. K. Bae and J. Kim, ‘‘Product development with data mining techniques: IWCMC, ICC, ISCC, ICNC, and Chinacom, the Symposium Co-Chair for
A case on design of digital camera,’’ Expert Syst. Appl., vol. 38, no. 8, IWCMC 2012 and ICC 2012, the General Co-Chair for Chinacom 2014,
pp. 9274–9280, 2011. Qshine 2015, CollaborateCom 2017, and Mobiquitous 2018, and as the
[95] S. Wang, J. Wan, D. Li, and C. Zhang, ‘‘Implementing smart factory Steering Chair and the TPC Chair for InisCom 2015. He has been serving as
of industrie 4.0: An outlook,’’ Int. J. Distrib. Sensor Netw., vol. 2016, an Editor-in-Chief for the EAI endorsed Transactions on Industrial Networks
Jan. 2016, Art. no. 3159805, doi: 10.1155/2016/3159805.
and Intelligent Systems and as an Associate Editor for the IEEE TRANSACTIONS
[96] S. Wang, J. Wan, M. Imran, D. Li, and C. Zhang, ‘‘Cloud-based smart man-
ON INDUSTRIAL INFORMATICS, the IEEE Communications Magazine, the IEEE
ufacturing for personalized candy packing application,’’ J. Supercomput.,
to be published, doi: 10.1007/s11227-016-1879-4.2016. SYSTEMS JOURNAL, and the IEEE ACCESS.

BAOTONG CHEN received the B.A. degree in PENG LI received the bachelor’s and master’s
mechanical engineering from the Nanyang Insti- degrees in process automation from the Technical
tute of Technology, China, 2014. He is currently University of Dresden, Germany, in 2013. He is
pursuing the Ph.D. degree with the School of currently a Research Assistant with the Institute
Mechanical and Automotive Engineering, South Industrial IT, Ostwestfalen-Lippe University of
China University of Technology, China. His Applied Science, Lemgo, Germany. His research
research interests include cyber-physical systems, focuses on machine learning, data mining, infor-
industrial wireless networks, Internet of Things, mation modeling, and their application in the
and embedded control systems. cyber-physical-production-systems.

6518 VOLUME 6, 2018

B. Chen et al.: Smart Factory of Industry 4.0: Key Technologies, Application Case, and Challenges

MITHUN MUKHERJEE (S’10–M’16) received BOXING YIN received the B.A. degree from the
the B.E. degree in electronics and communication School of Electrical Engineering and Automation,
engineering from the University Institute of Tech- Zhejiang University of Science and Technology,
nology, Burdwan University, India, in 2007, the China, in 2016. He is currently pursuing the M.S.
M.E. degree in information and communication degree with the School of Jiangxi Electrical Engi-
engineering from the Indian Institute of Engineer- neering and Automation, University of Science
ing Science and Technology at Shibpur, Shibpur, and Technology of China, China. His research
India, in 2009, and the Ph.D. degree in electrical interests include industrial big data and machine
engineering from IIT Patna, Patna, India, in 2015. learning.
He is currently a specially assigned Researcher
with the Guangdong Provincial Key Lab of Petrochemical Equipment Fault
Diagnosis, Guangdong University of Petrochemical Technology, China. His
research interests include wireless sensor networks, wireless communica-
tions, energy harvesting, and fog computing. He was a recipient of the
EAI WICON 2016 and IEEE SigTelCom 2017 Best Paper Award. He was a
Guest Editor of the IEEE TRANSACTIONS ON INDUSTRIAL INFORMATICS, the IEEE
ACCESS, the Mobile Networks and Applications (ACM/Springer), and the
Sensors. He has been serving as a Special Issue Editor for the EAI endorsed
Transactions on Industrial Networks and Intelligent Systems.

VOLUME 6, 2018 6519