RFID Application in A Multi Agent Cyber Physical Manufacturing System
RFID Application in A Multi Agent Cyber Physical Manufacturing System
RFID Application in A Multi Agent Cyber Physical Manufacturing System
Article
RFID Application in a Multi‐agent Cyber Physical
Manufacturing System
Maryam Farsi 1,*, Christina Latsou 1, John Ahmet Erkoyuncu 1 and Geoffrey Morris 2
1 School of Aerospace, Transport and Manufacturing, Cranfield University, Cranfield MK43 0AL, UK;
[email protected] (C.L.); [email protected] (J.A.E.)
2 Cryogatt Systems Limited, Kings Langley, WD4 8FR, UK; [email protected]
* Correspondence: [email protected]; Tel.: +44‐123‐475‐0111
Received: 2 October 2020; Accepted: 27 October 2020; Published: 29 October 2020
Abstract: In manufacturing supply chains with labour‐intensive operations and processes,
individuals perform various types of manual tasks and quality checks. These operations and
processes embrace engagement with various forms of paperwork, regulation obligations and
external agreements between multiple stakeholders. Such manual activities can increase human
error and near misses, which may ultimately lead to a lack of productivity and performance. In this
paper, a multi‐agent cyber‐physical system (CPS) architecture with radio frequency identification
(RFID) technology is presented to assist inter‐layer interactions between different manufacturing
phases on the shop floor and external interactions with other stakeholders within a supply chain. A
dynamic simulation model in the AnyLogic software is developed to implement the CPS‐RFID
solution by using the agent‐based technique. A case study from cryogenic warehousing in cell and
gene therapy has been chosen to test the validity of the presented CPS‐RFID architecture. The
analyses of the simulation results show improvement in efficiency and productivity, in terms of
resource time‐in‐system.
Keywords: cyber physical system; agent‐based simulation; RFID; cryogenic warehousing; complex
systems; supply chain; cell and gene therapy
1. Introduction
Over the last decade, the deployment of radio frequency identification (RFID) within different
supply chains has had a major influence on the traceability of materials and information. In the
manufacturing sector, RFID technology improves the trackability of production processes, and the
integrity of information flows on a shop floor. In warehousing, RFID enhances inventory
management and streamlines its interactions with other logistical components, through the improved
track and trace of materials and transporters. Moreover, RFID technology, together with connected
IT platforms, has a crucial impact on control systems within businesses. The application of such
digital technologies can ultimately lead to significant performance and productivity improvements
for businesses [1,2].
RFID readers use radio waves to capture information and can be categorised into two types:
passive and active. Passive RFID tags do not have an internal power source, and they are equipped
with an electromagnetic chip that is read by a battery‐powered RFID reader. These tags are relatively
small and have a long lifespan. However, such tags can be read at only short distances (i.e., a few
feet), which limits their use for certain applications. Active RFID tags have their own source of power.
These tags can be read at more than a hundred feet distance. However, such tags have a limited
lifespan due to battery constraints and are more expensive. Regardless of the RFID type, implanting
such technologies advances the level of maturity in digitalisation, automation and real‐time data
capture. Moreover, RFID technology increases information visibility, contributes to lean
J. Manuf. Mater. Process. 2020, 4, 103; doi:10.3390/jmmp4040103 www.mdpi.com/journal/jmmp
J. Manuf. Mater. Process. 2020, 4, 103 2 of 15
manufacturing, safety and security. Additionally, the technology can potentially reduce manual
handling and labour‐intensive activities, operation downtime, and human errors.
The cyber physical manufacturing system (CPMS) and cyber physical production system (CPPS)
can be described as an integrated IT platform for interactive computation, networking and physical
processes on a shop floor or a supply chain. The Internet of Things (IoT) provides the infrastructure
for CPS implementation. CPS uses several devices such as sensors, RFID readers, and mobile phones
to provide interactivity in different applications [3]. In the manufacturing industry, CPS can support
businesses to retain their competitiveness by enabling product development optimisation,
production system control, real‐time informed decisions, and deployment of manufacturing smart
systems [4,5]. In addition, within a supply chain, manufacturers are required to effectively collaborate
with various stakeholders to be able to mitigate risks emerging from market diversity and customer
requirements. In this regard, CPS can enable social manufacturing to allow mass customisation and
to improve logistics transparency based on ‘customised community space configuration’ [6].
Acquiring essential data and maintaining information integrity within manufacturing processes
and supply chains are challenging and time‐consuming tasks, due to the amount of paperwork and
manual audit procedures involved. In many small and medium enterprises (SMEs), the application
of automated techniques such as IoT, CPS and RFID is not mature and is at the early stages of proof‐
of‐concept. Although such advancements create new challenges for industries in terms of uncertainty
management [7], security [8], cloud computing [9] and big data management [10], evaluating the
impact of such technologies on supply chains’ and businesses’ performance is crucial for any
investment appraisal.
For a system of systems, such as supply chains and complex manufacturing systems, the Agent‐
Based Modelling (ABM) technique is a promising approach for the system design and system
simulation [11]. To explore and substantiate the RFID application in CPMS, this paper addresses the
following research question: “How can the ABM technique be applied to develop an RFID‐CPMS
architecture?”. Moreover, this study contributes to knowledge by extending the application of ABM
technique to develop a multi‐agent architecture for an RFID‐CPS. The context of this research is the
application of RFID technology in a CPMS. We have also constructed a number of Unified Modelling
Language (UML) diagrams to depict different perspectives and layers of the multi‐agent RFID‐CPS
in a cryogenic supply chain.
This paper is structured as follows: Section 2 provides the theory and background of RFID and
CPS architecture in manufacturing and supply chains. The research methodology to carry out this
study is presented in Section 3. The multi‐agent CPS architecture with the application of RFID is
developed in Section 4. This is followed by a case‐study in a cryogenic supply chain in this section.
Section 5 provides the discussion on the multi‐agent RFID‐CPS architecture. The concluding remarks
and future work are presented in Section 6.
2. Theory
With the rapid development of digital and information technologies within manufacturing
sectors and supply chains, the novel application of IoT, CPS and RFID data management has been
the focus of many studies in the past two decades. In manufacturing, CPPS is an emerging
information system that, together with IoT and cloud computing technologies, supports automation
and digital maturity [4,12]. In this regard, Ding et al. [6] presented an RFID‐enabled social
manufacturing system platform for real‐time data monitoring, dispatching and transportation. They
argued that such a platform could be deployed for data sharing and collaborative decision‐making
between different stakeholders.
IoT infrastructures and CPS frameworks for the application of RFID sensors for security and
traceability purposes have been studied previously in the context of smart cities [13,14], logistics [15],
small manufacturing [16], production [17], smart manufacturing [5,18], automotive industry [19], and
product‐service systems [20]. Suhail et al. [21] discussed the data provenance and security challenges
associated with an RFID‐based IoT. The data provenance challenges are defined as data storage,
processing, biding, interoperability and fault tolerance, whereas, the security challenges relate to data
J. Manuf. Mater. Process. 2020, 4, 103 3 of 15
3. Methodology
Figure 1. PRISMA flow diagram and research methodology.
J. Manuf. Mater. Process. 2020, 4, 103 5 of 15
The proposed multi‐agent RFID‐CPS architecture is presented based on the literature review
and the theories behind RFID and CPS, as presented in Section 2. To validate the multi‐agent
architecture, an industrial case study from a cryogenic warehouse company is considered. The agent‐
based simulation model of the case study with the embedded RFID was developed in the AnyLogic
software, as presented in Section 4.
4. Results
A cold supply chain refers to the supply and logistics of perishable products and materials. In
such a supply chain, evaluating the location and condition of transporters and shippers are vital. In
addition, implementation of a logistics cyber‐physical monitoring and control system optimisation
supports shipment decisions and improve logistics efficiency [34]. In this study, a cryogenic supply
chain in the cell and gene therapy (CGT) sector is studied to develop an RFID‐CPS cyber architecture.
In a cryogenic warehouse at the temperature of −190 °C, the advanced RFID tag and reader
technologies [35] can be implemented for the recording, monitoring and auditing of cryo‐materials
on the shop floor. In such an environment, advanced RFID tags can considerably enhance inventory
management, since the traditional labels can be easily frosted and become unreadable even after a
short time period after their liquid nitrogen storage.
In this study, a cryogenic supply chain that is comprised of the manufacturer, distributor and
the end‐user is considered as a case study. The advanced RFID system is designed and installed at
each stakeholder’s site. The system architecture for RFID implementation is illustrated in Figure 2.
According to this architecture, the system components are connected via the in‐house network. A
server PC, holding the system database, runs a web service that allows computers and mobile devices
to be connected to the same network, to view and manipulate the recorded data. RFID readers located
in the goods in and out areas are used to collect data on the arrival and dispatch of shippers,
respectively. Similarly, the RFID readers situated in the cryogenic storage area are used to capture
data for either storing cryo‐materials after shippers’ arrival or to retrieve cryo‐materials preparing
for dispatch.
Figure 2. Radio frequency identification (RFID) system architecture.
J. Manuf. Mater. Process. 2020, 4, 103 6 of 15
4.1. Multi‐Agent Cyber‐Physical Manufacturing System
In this research work, a hybrid method that employs object—and process‐oriented approaches
using Agent‐Based Modelling (ABM) and Discrete Event Simulation (DES) is proposed to model the
CGT enterprise architecture. Additionally, a built‐in fully integrated database to read real‐time data
provided from RFID PC servers at each site within a cryogenic supply chain is deployed. The UML
class diagram of the proposed multi‐agent simulation model is presented in Figure 3. It should be
noted that the CGT manufacturing system consists of multiple sub‐systems that are able to operate
simultaneously. In this work, these sub‐systems are referred to as manufacturing phases. Each
manufacturing phase entails various tasks that are described as goal‐directed activities with a start
and endpoint and an objective. Performing multiple tasks in different manufacturing phases can
potentially lead to parallel interactions within the manufacturing system. A common example of
parallel interactions is when operators with similar skills are required to perform a single task or a
sequence of tasks in more than one manufacturing phase at the same time. The system architecture
for the SCG supply chain consists of two main parts:
Part I—ABM‐DES hybrid system architecture: A system of multi‐agent discrete events,
inspired by [11], is developed to model a complex manufacturing system. Three levels of agents,
macro, meso and micro, are used to capture the dynamic behaviour of the manufacturing
system. The macro‐level agent models the global manufacturing system, comprising
manufacturing phases (meso‐level agent) and manufacturing components (micro‐level agent).
Meso‐level agents, modelled using the ABM approach, are employed to simulate the interactive
structure of manufacturing phases. As seen from Figure 3, agents at this level are created as a
single agent type always existing within the macro‐agent environment. Micro‐level agents,
including human and equipment resources, data, information, etc., are also created using the
ABM approach. Unlike meso‐level agents, micro‐agents are created as a population of agents of
the same type living in the same environment. Following Figure 3, the behaviour of each agent
at the micro‐level belongs to a specific resource type (static, moving or portable), and may be
characterised by specific capacity, rate and schedule. Function methods for describing algebraic
rules and events for scheduling one‐time or recurrent, concurrent or independent actions can
also be defined. Finally, the DES modelling approach is employed for describing the discrete
event states of each manufacturing phase modelled inside the meso‐level agents.
Part II—CPS‐RFID system architecture: A database element, representing an actual database,
is created to ensure communication with the agents described in Part I. The database element is
associated with a database file (MS Access database/MS Excel spreadsheet) that holds data
captured from the RFID system. An absolute file path is created, ensuring the connection
between the database element and the database file. Once this connection is established, a
database table is generated retrieving all the data from the database file. This ensures access to
the database. Database views, namely the result sets of queries on the data stored in the database
table, are also developed. Once manufacturing components, modelled at the micro‐level agent,
request access to a database view, a virtual table so‐called database view table is generated. The
data in this database view table are selected from the database table based on the requests made
via the database view. The database view (table) concept is employed to categorise RFID data
into different tables according to the location from which they are collected by the RFID system
within the cryogenic supply chain. This facilitates a structured input of data into the multi‐level
agents of the model when requested (see Figure 3).
J. Manuf. Mater. Process. 2020, 4, 103 7 of 15
Figure 3. Unified Modelling Language (UML) class diagram of multi‐agent cyber physical
manufacturing system.
4.2. Cyber‐Physical Architecture of IoT
An IoT data structure with three main layers of perception, transportation, and application is
considered to demonstrate the CPMS architecture with RFID application. The three layers of CPMS‐
RFID system architecture are illustrated in Figure 4.
Figure 4. Cyber physical architecture of Internet of Things (IoT).
J. Manuf. Mater. Process. 2020, 4, 103 8 of 15
4.2.1. Perception Layer
The perception layer is employed for data collection via RFID tags. In this layer, various physical
devices, including RFID tags and readers, sensors, router(s), antenna, cables, in‐site computers and
mobile devices such as laptops and tablets, are deployed. RFID tags are attached to the containers
(e.g., vials and bags) that hold the sample media, as well as the packaging items (e.g., boxes used for
the storage of vials, racks used for the storage of bags, and shippers) used for the transportation of
the material within the cryogenic supply chain. Additionally, different types of RFID readers are
designed and installed: (i) ‘shipper readers’ for scanning RFID tags attached to shippers; (ii) ‘close
proximity readers’ for close up reads of bags and racks to automatically update the location of stored
items without requiring manual data entry (typing), and (iii) ‘cold box readers’ for reading cryogenic
vial boxes. In the latter RFID reader type, each vial stored in the box can be read individually as each
vial slot provides a unique antenna.
The equipment used in the perception layer enables the collection of real‐time information from
the physical environment of the CPMS. The collected raw data include the complete track and trace,
and the history of each stored sample from an assignment into storage, through audit checks and
shipping events. The types of raw data gathered from the RFID readers situated in the cryogenic
supply chain include task identification, supplier’s and recipient’s information or user name/ID once
the RFID reading takes place within a site, date, time and location of RFID reading, order number,
delivery location, carrier information, sample types, batch number, type of container, quantity and
others. The perception layer playing a key role in the CPMS has a twofold purpose: firstly, to obtain
data from the RFID sensors; and secondly, to execute operations by the command of the application
layer. Raw data collected in the perception layer are sent to the network layer through the Internet.
4.2.2. Transportation Layer
The transportation layer is employed to connect all ’things’, including anything like an object or
a person for sharing and exchanging data. This layer supports the transferred real‐time information
through a wired or a wireless network from the perception layer to the application layer, and
secondly the storage of the data captured from sensors. In this layer, data can be transmitted using
local area network (LAN), the Internet, communications channels and private networks. In this
research work, the Internet was selected due to its wide applicability, global availability and
inexpensiveness. At each site within the cryogenic supply chain, there is a server PC that holds the
system database (Microsoft SQL) for storing and processing the collected RFID data. Each server PC
runs a web service that allows any device (laptop or tablet) connected to the same network to view
and process the stored RFID records. The SQL database is hosted on a virtual machine on Microsoft
Azure Cloud, enabling cross‐site communication and providing easier interconnection between
clients and manufacturers.
4.2.3. Application Layer
The application layer is the most interactive layer of the CPMS. It is employed to support a set
of business services and realises intelligent computation and resources allocation in screening,
selecting, producing and processing data. In this layer, the first step considers real‐time data
acquisition and processing, as seen in Figure 4. A Microsoft Excel file with real‐time data is initially
obtained. These data are collected from the RFID system as described in the perception layer and
stored in the Azure SQL Database, as discussed in the transportation layer. This file is obtained via
the Google Drive API and updated automatically at specific time intervals which can be configured
by users via the Azure and Google Drive APIs. The data from this file are then processed
automatically, by employing a set of macros with Excel Visual Basic for Applications (VBA), in order
to create timestamps and calculate the time required for performing various tasks within the
cryogenic supply chain. The next step focuses on multi‐agent simulation model development that
follows the multi‐agent simulation technique discussed earlier in this work. Thus, a database to read
real‐time data obtained from the Excel file and a simulation model that retrieves the data from the
J. Manuf. Mater. Process. 2020, 4, 103 9 of 15
database and reflects the complex behaviour of manufacturing systems within a cryogenic supply
chain are developed. Considering a cryogenic supply chain with the manufacturer, distributor and
end‐user, the input and output information flows to and from each location where RFID devices are
implemented are seen in the UML composite structure diagram in Figure 5. The distributor acting as
a cryogenic warehouse comprises three phases: Phase I—Receipt and Inventory, Phase II—Storage
and Monitoring and Phase III—Distribution.
Figure 5. UML composite structure diagram of a cryogenic supply chain.
Manufacturer and end‐user consider distribution and receipt tasks, respectively. The types of
collected sensing data, obtained from the database to enable real‐time simulation, include tags ID,
date and time stamps recording after each tag is scanned by an RFID reader and process time required
for each task to be carried out. After performing the model simulation, real‐time data analysis and
decision‐making are introduced. In these steps of the application layer, the data obtained from either
part of the simulation (each block in Figure 6) or the entire model are analysed. These data may
include resource and space utilisation, lead times, time‐in‐system, system throughput, work‐in‐
progress (WIP) and inventory size. Relevant data analytics can then be carried out to provide
informed decisions in terms of resource planning, time‐in‐system, WIP, dispatch planning, inventory
control and space layout planning. Finally, application interfaces—dashboard in Cloud transforms the
developed simulation model into a decision‐support platform allowing day‐to‐day operations. The
developed simulation model can be stored in the Cloud and hence connected with operational data
and set‐up experiments providing decision‐makers access to leverage simulation insights.
Additionally, multiple users at different locations can remotely access the model, run simulations
and visualise the results through interactive dashboards. Users are also able to perform various
experiments and compare the results obtained in order to achieve better scenario management. A
great benefit of the Cloud solution in the proposed cyber‐physical architecture is the high‐
performance computing capabilities that they can provide by allowing the execution of complex
multi‐run experiments.
4.3. Result: A Case Study in Cryogenic Supply Chain
To demonstrate the applicability and validity of the proposed CPS‐RFID architecture, a
manufacturing system at a cell and gene therapy (CGT) cryogenic warehouse is studied. The
manufacturing system is characterised by labour‐intensive processes and repetitive tasks. The
J. Manuf. Mater. Process. 2020, 4, 103 10 of 15
multiple parallel executions of interdependent tasks within different manufacturing phases increase
the complexity of these systems. A dynamic simulation model in AnyLogic software (version 8) is
developed to implement the presented CPS‐RFID solution employing the agent‐based technique.
Simulating such a complex system following the proposed technique can simplify the complicated
structure of the model. Additionally, RFID implementation can improve the efficiency and
productivity of manufacturing systems in terms of system throughput, time‐in‐system and resource
utilisation. The proposed CPS‐RFID solution was applied to model the case study in order to identify
the benefits of the RFID implementation within the selected manufacturing system.
4.3.1. Outline of the System
The manufacturing system at the studied CGT cryogenic supply chain consists of three sites, a
manufacturer, a distributor (cryogenic warehouse) and an end‐user (patient). In this case study, the
focus is on distributor role, who is responsible for receiving cryogenic material from the
manufacturer, storing and monitoring the material, and dispatching it when requested from the end‐
user. The flow of material and information in the studied warehouse are modelled within three
manufacturing phases of the system: (i) Receipt and Inventory; (ii) Storage and Monitoring; and (iii)
Distribution. The tasks’ execution within the cryogenic supply chain is explained with the help of the
UML activity diagram shown in Figure 6.
Figure 6. Case study: UML activity diagram for a cryogenic supply chain with RFID.
RFID devices are implemented into the cryogenic warehouse, capturing information from three
main warehouse areas: goods in, cryogenic storage area and goods out. Data produced by the RFID
devices located in these areas are obtained from tasks including ‘arrivals checking’, ‘storing material’,
‘picking material’ and ‘dispatching, as highlighted in Figure 6. Considering the implementation of
the RFID system, the UML use case diagram of the studied cryogenic supply chain shown in Figure
7 represents the interactions of users in the cryogenic warehouse with manufacturers and end‐users.
J. Manuf. Mater. Process. 2020, 4, 103 11 of 15
The exchange of information between the three sites, i.e., manufacturer, distributor and end‐user, is
carried out with the help of a unique identifier, an Autonomous System Number (ASN).
Figure 7. Case study: UML use case of cell and gene therapy (CGT) cryogenic supply chain.
4.3.2. Numerical Analysis
To develop the simulation model, data were collected for the CGT cryogenic storage processes
of the cryogenic warehouse in the UK. Following the ABM‐DES simulation method, the micro‐level
agents and the corresponding data are summarised in Table 1.
Table 1. Input data—micro‐level agents.
Moreover, the cycle time distribution input for different processes is presented in Table 2. The
tasks which will be affected by the RFID implementation are indicated in by an asterisk (*).
Table 2. Input data—cycle time distributions.
4.3.3. Results: Model Validation
Validation of the proposed technique is accomplished using the data from the case study. Real
data on the RFID cycle times for a five‐week period were collected from the shop floor of the case
study company. Average cycle times taken for each test procedure carried out within a trial are
J. Manuf. Mater. Process. 2020, 4, 103 12 of 15
summarised in Table 3. According to these data, the continuous probability distribution for each
RFID task is also found.
Table 3. RFID input data: Cycle times.
The cycle time distributions are implemented in the simulation model, and the average time‐in‐
system for the three manufacturing phases for the ‘RFID state’ is obtained. The results are compared
with the corresponding computational data obtained for the state ‘Without RFID’, and also the
reductions in the average time‐in‐system are seen in Table 4.
Table 4. Average time‐in‐system for the states ‘Without RFID’ and ‘With RFID’.
From the computational results, it is seen that the RFID implementation can offer a significant
reduction in the time required for storing and picking cryogenic material, as the system acts as an
electronic witness. This eliminates the need for an extra person to act as a ‘second witness’.
Additionally, the system automates current processes by reducing the paperwork and
documentation before dispatching the cryogenic material. Thus, RFID implementation reduces the
time taken in filling out and making copies of forms, making resources simpler and safer. According
to the simulation results, the efficiency and productivity of the cryogenic warehouse can be
improved. The reduction observed in the cycle times (i) decreases the time‐in‐system, as shown in
Table 4; (ii) can reduce the resource utilisation as the human and equipment resources are available
for more time compared to the state without RFID; and (iii) can increase the number of deliveries
from the manufacturer and orders from the end‐user as operators are able to receive, store and
dispatch a greater amount of materials.
5. Discussion
CPS modelling and design are being studied in the literature using different techniques and
approaches such as mathematical algorithms [9], ontology‐based approaches [23], DES methods [9]
and the agent‐based technique [5,12,25]. However, the existing literature has suggested that hybrid
system design and engineering approaches are a suitable approach for the specification and analysis
of CPS models due to the limitation of individual methods [36]. In the context of complex
manufacturing and supply chain systems, mathematical modelling becomes complicated and
computationally expensive due to the complexity and the high level of interactivity of sub‐systems
[11]. To tackle this challenge, the DES method provides a multi‐agent simulation approach for CPS
design and is a popular approach to simulate manufacturing processes. However, it limits the CPS
architecture to a set of individual layers and functions rather than taking the heterogeneous nature
and interactivity of CPS architecture. In this regard, ontology‐based and agent‐based approaches
provide the flexibility to create a multi‐layer system engineering architecture for CPS design. Such
bottom‐up approaches require detailed and comprehensive information about complex systems and
CPS architecture. The ABM technique takes an object‐oriented programming approach to create a
multi‐layer system engineering architecture with multi agents for CPS design. This study proposed
a hybrid multi‐agent CPS architecture for complex manufacturing and supply chain systems. Such
systems consist of several manufacturing phases and components that operate simultaneously and
J. Manuf. Mater. Process. 2020, 4, 103 13 of 15
interact with each other over time. The concurrent processes and procedures can be labour‐intensive
with various types of manual tasks and quality checks. Therefore, the system performance is affected
by human error and near misses significantly. To minimise such impact, a multi‐agent cyber‐physical
system architecture with RFID technology is presented to assist inter‐layer interactions between
different manufacturing phases and among different stakeholders within a supply chain. In this
context, the application of the multi‐agent approach for CPS design has been previously studied for
individual products [12,25], individual RFID functions [26], RFID control systems [5,33] and a set of
products and machines with embedded RFID tags at a shop floor [29]. In this study, the proposed
UML class diagram in Figure 3 presents the multi‐agent CPS‐RFID architecture. An agent‐based
simulation technique in the AnyLogic software was implemented to model the architecture.
Moreover, the cyber‐physical architecture of the IoT infrastructure is presented in Figure 4. A case
study from a UK cryogenic supply chain in the CGT sector was selected to test the architecture and
evaluate the impact of CPS‐RFID on the supply chain performance. A number of UML diagrams are
illustrated to present the developed multi‐agent CPS‐RFID architecture for the case study (see Figures
5–7). The performance is analysed based on the time‐in‐system for different manufacturing phases,
as presented in Table 4. It is found that integrating RFID tags with appropriate IT platform is an
effective approach in cryogenic warehouse management and can enable users to write and read
information to and from the tags more effectively, during the cryogenic storage procedures. The
outcome from this research shows that such capability has a clear impact on the performance of the
supply chain by minimising the time‐in‐system by ~49.4%.
6. Conclusions
This paper extends the application of the agent‐based approach and proposes a multi‐agent CPS
architecture for complex manufacturing and supply chain systems with embedded RFID technology.
The proposed architecture is composed of a multi‐layer agent at three different layers, called a macro‐
level agent for global manufacturing supply chain systems, meso‐level for manufacturing phases,
and micro‐level for manufacturing components. RFID‐based CPMS has been implemented at
multiple levels within the global supply chain. A multi‐layer data structure provides an appropriate
IoT platform for the CPS‐RFID system. A case study from the cryogenic supply chain has been
selected to test the validity of the proposed architecture. Several UML diagrams are developed to
present the different layers in the cyber‐physical architecture of IoT. The proposed RFID‐CPMS
architecture shows that the multi‐agent simulation model using the hybrid ABM‐DES approach,
together with the cyber‐physical architecture of IoT infrastructure, is an effective approach to deploy
a multi‐agent CPS‐RFID system architecture.
The further work will be focused on identifying different uncertainties within the multi‐agent
CPS‐RFID architectures and developing an integrated DT‐based CPS within the complex
manufacturing and supply chain systems. Currently, the authors are working on studying the impact
of RFID‐based CPMS on resource utilisation and throughput. Further research can also be conducted
to quantify this impact in the context of sustainability to evaluate the cost of goods and energy
consumption.
Author Contributions: Conceptualization, M.F. and C.L.; methodology, M.F. and C.L.; software, M.F.;
validation, M.F. and C.L.; formal analysis, M.F. and C.L.; investigation, M.F., C.L. and J.A.E.; resources, M.F.,
C.L. and J.A.E.; data curation, M.F., C.L.; writing—original draft preparation, M.F., C.L.; writing—review and
editing, M.F., J.A.E. and G.M.; visualisation, M.F., C.L. and G.M.; supervision, M.F. and J.A.E.; project
administration, M.F. and J.A.E.; funding acquisition, J.A.E. All authors have read and agreed to the published
version of the manuscript.
Funding: This research is funded by the Innovate UK [Grant No. 104515].
Acknowledgments: The authors would like to acknowledge the support form Cryogatt Systems Limited,
ThermoFisher UK and Cell and Gene Therapy Catapult UK.
Conflicts of Interest: The authors declare no conflict of interest.
J. Manuf. Mater. Process. 2020, 4, 103 14 of 15
References
1. Evdokimov, S.; Fabian, B.; Günther, O.; Ivantysynova, L.; Ziekow, H. RFID and the Internet of Things:
Technology, Applications, and Security Challenges; Foundations and Trends® in Technology. Inf. Oper.
Manag. 2011; Volume 4: No. 2, pp 105‐185. doi:10.1561/0200000020
2. Farsi, M.; Bailly, A.; Bodin, D.; Penella, V.; Pinault, P.L.; Nghia, E.T.; Sibson, J.; Erkoyuncu, J.A. An
Optimisation Framework for Improving Supply Chain Performance: Case study of a bespoke service
provider. Procedia Manuf. 2020, 49, 185–192, doi:10.1016/j.promfg.2020.07.017.
3. Greer, C.; Burns, M.; Wollman, D.; Griffor, E. Cyber Physical Systems and Internet of Things in Industry.
Mart Grid Cyber‐Phys. Syst. Progr. Off. Eng. Lab. 2019, 2839–2840, doi:10.6028/NIST.SP.1900‐202.
4. Choi, S.; Kang, G.; Jun, C.; Lee, J.Y.; Han, S. Cyber‐physical systems: A case study of development for
manufacturing industry. Int. J. Comput. Appl. Technol. 2017, 55, 289–297, doi:10.1504/IJCAT.2017.086018.
5. Tran, N.H.; Park, H.S.; Nguyen, Q.V.; Hoang, T.D. Development of a smart cyber‐physical manufacturing
system in the Industry 4.0 context. Appl. Sci. 2019, 9, 3325, doi:10.3390/app9163325.
6. Ding, K.; Jiang, P.; Su, S. RFID‐enabled social manufacturing system for inter‐enterprise monitoring and
dispatching of integrated production and transportation tasks. Robot. Comput. Integr. Manuf. 2018, 49, 120–
133, doi:10.1016/j.rcim.2017.06.009.
7. Wang, Y.; Qian, J. Measuring the uncertainty of RFID data based on particle filter and particle swarm
optimization. Wirel. Netw. 2012, 18, 307–318.
8. Jing, Q.; Vasilakos, A.V.; Wan, J.; Lu, J.; Qiu, D. Security of the Internet of Things: Perspectives and
challenges. Wirel. Netw. 2014, 20, 2481–2501.
9. Hosseinian‐Far, A.; Ramachandran, M.; Slack, C.L. Emerging trends in cloud computing, big data, fog
computing, IoT and smart living. In: Dastbaz M., Arabnia H., Akhgar B. (eds) Technology for Smart Futures.
Springer:Cham, Germany doi:10.1007/978‐3‐319‐60137‐3_2.
10. Mourtzis, D.; Vlachou, E.; Milas, N. Industrial Big Data as a Result of IoT Adoption in Manufacturing.
Procedia CIRP 2016, 55, 290–295, doi:10.1016/j.procir.2016.07.038.
11. Farsi, M.; Erkoyuncu, J.A.; Steenstra, D.; Roy, R. A modular hybrid simulation framework for complex
manufacturing system design. Simul. Model. Pract. Theory 2019, 94, 14–30, doi:10.1016/j.simpat.2019.02.002.
12. Wang, S.; Zhang, C.; Wan, J. A Smart Factory Solution to Hybrid Production of Multi‐Type Products with
Reduced Intelligence. In Proceedings of the 2016 IEEE Information Technology, Networking, Electronic
and Automation Control Conference, Chongqing, China, 20–22 May 2016; pp. 848–853,
doi:10.1109/ITNEC.2016.7560481.
13. Mora‐Mora, H.; Gilart‐Iglesias, V.; Gil, D.; Sirvent‐Llamas, A. A computational architecture based on RFID
sensors for traceability in smart cities. Sensors 2015, 15, 13591–13626.
14. Farsi, M.; Daneshkhah, A.; Hosseinian‐Far, A.; Jahankhani, H. (Eds.) Digital Twin Technologies and Smart
Cities; Springer International Publishing: Cham, Switzerland, 2020; doi:10.3390/s150613591.
15. Gładysz, B. An assessment of RFID applications in manufacturing companies. Manag. Prod. Eng. Rev. 2015,
6, 33–42.
16. Sánchez, B.B.; Alcarria, R.; Martín, D.; Robles, T. TF4SM: A framework for developing traceability solutions
in small manufacturing companies. Sensors 2015, 15, 29478–29510, doi:10.3390/s151129478.
17. Yu, Q.; Sriram, P.K.; Alfnes, E.; Strandhagen, J.O. RFID Integration for Material Management Considering
Engineering Changes in ETO Industry. In IFIP International Conference on Advances in Production
Management Systems; Springer: Cham, Switzerland, 2017; doi:10.1007/978‐3‐319‐51133‐7\_60.
18. Damjanovic‐Behrendt, V.; Behrendt, W. An open source approach to the design and implementation of
Digital Twins for Smart Manufacturing. Int. J. Comput. Integr. Manuf. 2019, 32, 366–384,
doi:10.1080/0951192X.2019.1599436.
19. Segura‐Velandia, D.; Neal, A.; Goodall, P.; Conway, P.; West, A. Industrie 4.0 implementations in the
automotive industry. Adv. Transdiscipl. Eng. 2016, 3, 319–324.
20. Zhang, X.; Ming, X.; Liu, Z.; Qu, Y.; Yin, D. A framework and implementation of Customer Platform‐
connection manufactory to service (CPMS) model in product service system. J. Clean. Prod. 2019, 230, 798–
819, doi:10.1016/j.jclepro.2019.04.382.
21. Suhail, S.; Hong, C.S.; Ahmad, Z.U.; Zafar, F.; Khan, A. Introducing Secure Provenance in IoT:
Requirements and Challenges. In 2016 International Workshop on Secure Internet of Things (SIoT),
Heraklion, Greece, 26–30 September 2016; pp. 39–46, doi:10.1109/SIoT.2016.011.
J. Manuf. Mater. Process. 2020, 4, 103 15 of 15
22. Zhang, D.; He, Z.; Yuming, Q.; Wan, J.; Li, D.; Zhao, S. Revisiting Unknown RFID Tag Identification in
Large‐scale Internet of Things. Time 2016, 16, 1536–1284, doi:10.1109/MWC.2016.7721738.
23. Tao, M.; Zuo, J.; Liu, Z.; Castiglione, A.; Palmieri, F. Multi‐layer cloud architectural model and ontology‐
based security service framework for IoT‐based smart homes. Futur. Gener. Comput. Syst. 2018, 78, 1040–
1051, doi:10.1016/j.future.2016.11.011.
24. Esmaeilbeigi, M.; Chatrabgoun, O.; Hosseinian‐Far, A.; Montasari, R.; Daneshkhah, A. A low cost and
highly accurate technique for big data spatial‐temporal interpolation. Appl. Numer. Math. 2020,
doi:10.1016/j.apnum.2020.03.009.
25. Hentout, A.; Hamdania, A.; Kachouane, H.; Messous, M.A.; Bouzouia, B.; Senouci, S. Multi‐Agent Control
Architecture for RFID Cyber‐Physical Robotic Systems. In Proceedings of the 2016 Global Information
Infrastructure and Networking Symposium (GIIS), Porto, Portugal, 19–21 October 2016; pp. 1–6,
doi:10.1109/GIIS.2016.7814851.
26. Wang, S.; Zhang, C.; Liu, C.; Li, D.; Tang, H. Cloud‐assisted interaction and negotiation of industrial robots
for the smart factory. Comput. Electr. Eng. 2017, 63, 66–78, doi:10.1016/j.compeleceng.2017.05.025.
27. Lu, S.; Xu, C.; Zhong, R.Y.; Wang, L. A RFID‐enabled positioning system in automated guided vehicle for
smart factories. J. Manuf. Syst. 2017, 44, 179–190, doi:10.1016/j.jmsy.2017.03.009.
28. Murofushi, R.H.; Tavares, J.J.P.Z.S.; Towards fourth industrial revolution impact: smart product based on
RFID technology. IEEE Instrum. Meas. Mag. 2017 ;20:51‐6, doi: 10.1109/MIM.2017.7919135
29. Mihoubi, B.; Bouzouia, B.; Tebani, K.; Gaham, M. Hardware in the loop simulation for product driven
control of a cyber‐physical manufacturing system. Prod. Eng. 2020, 14, 329–343, doi:10.1007/s11740‐020‐
00957‐w.
30. Lu, S.; Xu, C.; Zhong, R.Y.; Wang, L. A passive RFID tag‐based locating and navigating approach for
automated guided vehicle. Comput. Ind. Eng. 2018, 125, 628–636, doi:10.1016/j.cie.2017.12.026.
31. Liu, C.; Jiang, P.; Jiang, W. Web‐based digital twin modeling and remote control of cyber‐physical
production systems. Robot. Comput. Integr. Manuf. 2020, 64, 101956, doi:10.1016/j.rcim.2020.101956.
32. Park, K.T.; Lee, J.; Kim, H.J.; Noh, S.D. Digital twin‐based cyber physical production system architectural
framework for personalized production. Int. J. Adv. Manuf. Technol. 2020, 106, 1787–1810.
33. Wang, C.; Jiang, P.; Ding, K. A hybrid‐data‐on‐tag‐enabled decentralized control system for flexible smart
workpiece manufacturing shop floors. Proc. Inst. Mech. Eng. Part C J. Mech. Eng. Sci. 2017, 231, 764–782,
doi:10.1177/0954406215620452.
34. Mejjaouli, S.; Babiceanu, R.F. Cold supply chain logistics: System optimization for real‐time rerouting
transportation solutions. Comput. Ind. 2018, 95, 68–80, doi:10.1016/j.compind.2017.12.006.
35. Morris, G. G. Cryogatt Systems Limited. RFID Tag and Reader Patents (Granted) for Cryogenic
Technology. U.S. Patents: 9418265 B2(16/08/2016),9501734 B2(22/11/2016), 9589225 B2(7/03/2017), 9619678
B2(11/04/2017), 9697457 B2(04/07/2017), 2016026911 B2(28/01/2016). European Patents: 2866938
B1(28/06/2013), 2873030 B1(11/07/2013), 2948247 B1(23/01/2014), 2965266 B1(07/03/2014), 3287192
A1(23/01/2014), 3633556 A1(07/03/2014),
36. Wan, K.; Hughes, D.; Man, K.L.; Krilavičius, T. Composition challenges and approaches for cyber physical
systems. In Proceedings of the 2010 IEEE International Conference on Networked Embedded Systems for
Enterprise Applications, Suzhou, China, 25–26 November 2010.
Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional
affiliations.
© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access
article distributed under the terms and conditions of the Creative Commons Attribution
(CC BY) license (http://creativecommons.org/licenses/by/4.0/).