Development of 5G Cyber-Physical Production System

The cyber-physical production system (CPPS) was developed for the interconnection between operational technology (OT) and information and communication technology (ICT) among the machines and decentralized production control. However, OT and ICT are now getting more complex in the modern industry. To enhance industrial management and production process, the CPPS based on the 5G architecture under production environments can provide a more comprehensive solution in the context of Industry 4.0. For this purpose, a 5G CPPS model to improve the interoperability of CPPS in ICT and OT is proposed to build a virtualized Industry 4.0 asset management shell system. First, real-time task artifacts and 5G communication modules are bound in a certain form of virtualization. Second, task artifacts are released to the production line in a real-time scheduling. Third, an order is placed on a production line and split into multiple tasks for production process. Single-machine single-workpiece and multi-machine and multi-workpiece processing cases are used for testing in the production line. Experimental results show that the proposed 5G CPPS can efficiently achieve multi-machine work piece processing.


Introduction
The continuous development of communication technology has accelerated the arrival of the Internet of Things era. With the continuous improvement of the computing power of embedded systems and the ubiquitous connection, life has become more convenient.
Conceptually, interconnected embedded systems are also called "cyber-physical systems" (CPS), whose definition is the unification of computing and physical processes. In principle, CPS means that embedded devices and networks must be unified in the physical process of system operation monitoring. There is a feedback loop mechanism in the CPS system, which can make the physical process affect the calculation, or the calculation affects the physical process. At present, the CPS concept has been widely used in industry 4.0 development and intelligent manufacturing system platform development.
CPS is used in the production environment to control and execute the manufacturing process. This system is called the cyber-physical production system (CPPS). The next-generation communication technology in the form of 5G brings new functions and new possibilities to CPPS.

Motivation
With the widespread use of standardized communication interfaces and improved interoperability of the underlying connectivity layer, 5G brings flexibility, variability, sustainability, and optimal utilization of available resources. These new features and functions are of great help to the 1 3 implementation of smart manufacturing. Using 5G technology to support the physical part of CPPS means that enterprises can use 5G to deploy a large number of diversified and highly networked devices to build 5G private networks, such as machines, sensors, robots, AGVs and others. However, it still has a space to develop advanced technology for practical applications in industry.

Building CPPS in the 5G Environment
To build a 5G CPPS, the CPS must meet relevant conditions. First, the 5G infrastructure must ensure that the CPS is highly adaptable. Second, CPS requires a comprehensive description of the 5G infrastructure to efficiently gatAsset her the data and information needed to effectively collect the current and target state of the infrastructure configuration.
To achieve this goal, the proposed approach aims to use virtualization in a dual sense. First, a virtual representation of all necessary assets is represented by an asset management shell to facilitate unified description. Second, following the information and communication technology (ICT) paradigm, the virtualization of functional and operational technology (OT) components is exploited to increase the flexibility and variability of the system. In addition to above conditions, the general issue of interoperability needs to be considered.
Interoperability can be divided into four levels that can be applied in the context of CPPS. First, technology interoperability supports the general ability to exchange data between technology components and systems. This fundamental interoperability is addressed by industry standards for ISO-OSI hardware and software interfaces with communication protocols at all levels. Second, in syntactic interoperability, it can recognize and transfer individual semantically evaluable units of data and extract them for further processing. Third, these mantic interoperability describes the ability of interpretation for the extracted information unit to be semantically correct. Fourth, the interoperability of organizations and processes refers to the ability to effectively and efficiently organize interactive processes.

Problems and Solutions
5G networks can be configured to meet application needs, but its configuration is not dynamic at runtime due to system uncertainty. How to make CPPS components aware of current network conditions? How to dynamically adjust the 5G network to ensure the safety and reliability of the process? It is necessary to consider the entire system with multiple layers and all its distributed components, including virtualized functions implemented in communication, computing, control, software services and process requirements. Therefore, this study proposes concepts of technical and architectural features applicable to Industry 4.0 Asset Management Shell (AAS) technology and 5G to outline next-generation CPPS architectures to overcome the outlined obstacles.

5G Architecture and Network
5G is a standardized fifth-generation mobile network defined by 3GPP [1]. 5G has improved the physical layer in terms of bandwidth, speed, and latency. It is mainly achieved through technical features such as millimeterwave radio, small cells, massive multiple-input, and multiple-output (MIMO) antennas, beamforming, and fullduplex communication [2].
Unlike previous generations of mobile communications, the breakthrough in the architecture of 5G is the extensive use of virtualization technology, which will bring considerable benefits to users. For example, from the cloud radio access network (RAN) in 5G, virtualization technology is widely used with SDN, NFV, and even integrated computing and storage virtualization applications. The typical architecture of a 5G communication system is depicted in Fig. 1 [3].
The baseband unit (BBU) and the remote radio unit (RRU) basically constitute the front haul part of the 5G network, while the backhaul connects to the core network and the Internet. The central office can provide multiple BBUs for resource pooling. The BBU is to provide the core functions and processing of the 5G network, and the RRU is the antenna access point of the terminal equipment. These virtualization-based capabilities allow the deployment of many 5G cells and network slices ranging in size from 10 m to 30 km [4], enabling the deployment of multiple parallel mobile networks on a unified hardware infrastructure. Each network slice can be configured according to the specific requirements of any of the three main characteristics of 5G: Moreover, 5G TSN can bring real-time capabilities to the network. It can realize the hard real-time network configuration required for PLC-based process control applications and non-real-time critical applications, such as AGV navigation or many robot controls, all benefit from 5G connectivity.

CPPS Architecture and Components
CPS was originally evolved from embedded systems. It has higher computing power than the original embedded system and allows communication through the Internet by adding connection ports [5]. Figure 2 shows the structure of CPPS and its main modules. It can achieve complex internal functions and logic capabilities so that it can be offloaded to external scalable infrastructure [6].

Aspects of Virtualization
Cognitively, virtualization is defined as the act of creating virtual versions of things, including virtual computer hardware platforms, storage devices, and computer network resources. This research is focused on the virtualized representation of industrial production resources in the form of assets, as well as the virtualization of functions typically provided by the physical capabilities and properties of the assets.
In the past, ICT development has mainly paid attention on cloud virtualization. Hardware virtualization is one of its main applications. It is mainly used to provide scalable computing and storage services. Other virtualization applications include virtual local area networks (VLANs) implemented through software-defined networking. Furthermore, network functions virtualization allows further virtualization of specific functions required for network functions, such as 5G software-defined radio [7]. Due to the continuous convergence of ICT and OT, OT has also started a similar development. An example is the software programmable logic control (PLC) to implement hard real-time reliability. However, it continued to evolve and promote its use in less critical applications [8]. Processes can also be virtualized by creating a formalization of process steps that are digitally represented. In principle, the entire production process can be virtualized.

Virtual Representation, Digital Twin, and Asset Management Shell
Conceptually, digital representations are often referred to as digital twins. However, no clear definition was described for what a digital twin will look like, what it will do, and how it should be implemented. Nonetheless, we can in principle correspond to different interpretations of digital twins by several concepts [9]. One of the most relevant concepts is the Asset Management Shell (AAS) defined in the Reference Architecture Model Industry 4.0 (RAMI 4.0) [10]. It forms Industry 4.0 components around each relevant asset in Industry 4.0, as shown in Fig. 3 [11]. Although assets exist as physical components in the real world, AAS is a defining term and characteristic in the RAMI information environment [12].
AAS has the structure defined in RAMI and is complemented by the meta-information model in RAMI. Since AAS is an important part of Industry 4.0 components, it and related concepts must be standardized for interoperability reasons. The communication between AAS is compliant with Industry 4.0 (Industry 4.0 language), thus conceptually defining the message structure and interaction protocol for the communication between Industry 4.0 components and AAS [13,14].
In the Industry 4.0 system, data exchange requires components to perform. However, in some cases, not all assets have to do this, so the use of the management shell can be provided as passive (information available on external servers), passive or semi-active (using API), and active, to be fully compatible with Industry 4.0 communication [15].
In addition to the most important basic asset information, AAS contains submodels [16]. It can support descriptions of properties, parameters, variables, and functions of assets, containing multiple submodels to implement different functions. Mandatory and optional submodels are specific to asset classes complement the mandatory or optional base submodels. Finally, free submodels can be defined and designed according to existing standard dictionaries. For example, the IEC 61360 Common Data Dictionary (CDD) and ecl@ss specifications provide parameters for fieldbus profiles and variables for OPC UA support specifications. Additionally, it provides a reference implementation of the AAS registry for registering and providing AAS and submodel information [17].
The "Asset Administration Shell Packet Explorer" can be used to create and edit AAS and submodels, and export them to AASX format to import and exchange AAS information. The AASX format is similar to the AMLX format, which is an interchange format for AML files [18].

Bridging the Gap Between CPPS
In the description of the 5G and CPPS architecture, several key points can be understood: first, the connection between 5G-supported CPPS must consider the underlying computing and communication infrastructure. Second, through 5G and the option of deploying private networks, organizations can deploy their own 5G infrastructure in place. Third, the BBU unit in the 5G network provides core functions by providing various SDN and NFV functions. They can also provide computing power based on their size, or connect directly when needed (for example, 40Gbit fiber) to compute components to achieve low latency. This means that CPPS can use BBU and its integrated computing components as resources, as shown in Fig. 4.

Asset and Resource Virtualization
To achieve asset and resource virtualization, all relevant aspects of the entire system need to be virtualized. Accordingly, OT and ICT assets must be represented digitally, and one or more sub-models can be created to express the characteristics of the assets in AAS. The asset virtualization for CPPS with AAS is shown in Fig. 5.
The concept of virtualization mentioned here can be implemented step by step. Assets do not have to be fully realized as physical simulation models or perfect twins to reflect all aspects of asset status in real-time. You only need to capture the important features required for the envisaged use case, as shown in Fig. 4. The amount of information can be gradually expanded according to the needs and regulations defined by the digital twin to meet the needs of the application.

Dynamic Function Virtualization
With the concept of hardware virtualization, the functions of OT hardware must continue to be virtualized. The core functions provided by proprietary hardware in the past are gradually being replaced by the software. The architectural features of CPS that externalize service functions can be combined with the capabilities of next-generation networks. This enables CPPS to be linked with RT key services, providing functions such as control or sensing, or extended functions such as AI-supported applications. Figure 6 shows the gradual evolution of the CPPS realized by integrating the communication plane into the CPPS into the future target architecture.

CPPS in 5G Production Lines
The CPPS architecture in 5G production line is shown in Fig. 7. The real-time task artifacts and 5G communication modules are bound in a certain form of virtualization so that they become virtual task artifacts and information carriers at the same time. At this time, the task artifacts are connected with the entire manufacturing network in series, and they can be sensed at each manufacturing node in a similar production line.

3
In real-time scheduling, during the entire production process after task artifacts are released to the production line, task artifacts have been continuously banning the right to use these two types of resources, logistics equipment and processing equipment.
After an order is placed on a production line, it is split into multiple tasks, and each task contains an artifact, which is the virtualized task artifact we mentioned earlier, and the task artifact is the object to be perceived. Define the perceived information flow as (1): 1. OD, represents the order number of the order where the task artifact is located. 2. P n , represents the type of task artifact, P n ∈ P. 3. ID, which represents the number of the workpiece. The serial number of the workpiece assigns a unique identification number to multiple workpieces of the same type in an order. 4. X = (X 1 , X 2 , … X n ) represents the constraint set of task artifacts, such as order delivery date, priority set, urgent order, etc. 5. V = (V 1 , V 2 ,…, V n ) represents the process route of the task workpiece, and V n represents local n processes. There is a corresponding relationship between the type of processing equipment used in the process and the process, which is defined as (2): (1) T = OD, P n , ID, X, V, E � , C, T, S } represents the logistics equipment used between adjacent processes, E 2 n+1 ∈ E. Among them, E 2 1 and E 2 n+1 respectively represent the logistics equipment used for outbound and inbound tasks. 7. C = {C 1 , C 2 ,…, C n } represents the numerical set of various size parameters of the task workpiece, which is customized by the user when placing an order. 8. T = {T 1 , T 2 ,…,T n } represents the estimated processing time of each process, and the calculation of the estimated time is related to the size parameter. 9. S = {S a 1 , S b 2 ,…,S 2 n } represents the state of each process, and Szn represents the nth process (that is, the current state of process V n ) is z. The status of all processes of the task workpiece can be subdivided into 5 states: waiting for transportation, in transportation, waiting for processing, in processing, and processing completed. The process state z occupies 5 bits of memory space when storing, as shown in Fig. 8. Now suppose that a task workpiece has 4 processes. Process . According to the above description, the processing flow of a single task workpiece as shown in Fig. 9 is thus constructed.

Model Test
The 5G CPPS model that applies AAS to dynamic infrastructure management has been implemented in a production-line environment. Figure 8 describes the hardware settings and maps them to the functional modules of the architecture CPPS concept.

CPPS Integrated 5G Hardware Construction
The first implementation phase is to create the required underlying service infrastructure to collect, manage, and provide the data and information needed for all infrastructure components in the form of AAS. To this end, a repository and registry system for specific types of infrastructure assets in the form of AAS has been implemented by current non-normative and emerging normative specifications [19].
AAS and sub-models can be created and edited using AASX Package Explorer provided as open-source software. This setting uses the implementation of the AAS registry and repository according to the meta-information model defined in I4.0 and implements the AAS API according to the specifications in I4.0. AAS uses the AASX format to import. Alternatively, services and CPS components can be connected through custom middleware to create a service bus [20], which integrates its AAS registry and maps the self-description of services and components to the AAS format as a simple resource description sub-model. The self-description is generated Fig. 8 Storage of process status in memory Fig. 9 The processing flow of a single task workpiece  [21].
In Fig. 10, the BBU can be used as a computing device to deploy containerized services (such as Docker containers) [22,23]. Due to resource constraints, only functions that do not require resources are deployed to the BBU. Under this assumption, the experimental setup is expanded into a settop box for service deployment, including a computing unit for data processing and model training, and a flash storage cluster for persistence [24]. The test platform can provide industrial applications that use distributed architecture, computing, and control.

Operation Interface
As shown in Fig. 11, the 5G virtual CPPS operation interface is designed to achieve four goals based on functional modules [25,26]. The first goal is to monitor and adjust the number of task artifacts. The second goal is to monitor and adjust the state of the process. The third goal is to monitor and adjust processing equipment resources. The fourth goal is to record production line data. These are realized through the process status module, workpiece processing flow module, and monitor module.
1. Process status module: the sequence of state execution is from S5 to S1, from bottom to top, and is executed layer by layer. The current status is displayed in decimal. All the process status toggle switches of the task workpiece form a binary display as "1 waiting for transportation", "3 in transportation", "7 waiting for processing", "15 in processing" and "31 processing completed" five states. 2. Workpiece processing flow module: this module is mainly used to adjust the number of effective task workpieces in the warehouse, the quantity out of the warehouse, and the number of processing equipment that can be used. 3. Monitor module: this module displays the number of task workpieces that have been put in, the number of out of the warehouse, and the number of processing equipment used in the completed processing process.

Operation Result
Through the 5G CPPS virtual operation interface, we operate two types of production line production cases,   including single-machine single-workpiece processing and multi-machine and multi-workpiece processing cases. Table 1 shows the status record of the single-machine single-workpiece processing operation mode used in the production of similar production lines. The workpiece is processed by a machine from the warehouse, and it takes 1 h from the completion of the processing to the final storage. Table 2 shows the status record of the multi-machine and multi-workpiece processing operation mode used in the production line. It takes 1 h for a variety of workpieces to be processed by multiple machines from the warehouse.
From the results of the two operations, 5G CPPS can achieve the efficiency of multi-machine, multi-workpiece processing, and multi-productivity under the condition that the process flow remains unchanged.

Conclusions
This study has outlined the 5G CPPS capabilities and potential from an architectural point of view. Importantly, AAS based on the expansion of the CPPS virtualization has been proposed working as a unified component to achieve interoperability between different ICT and OT domains. The actual CPPS operation case in the 5G production line is successfully established by defining the processing flow model at the task workpiece. Judging from the operating results, under the same process flow, the proposed 5G CPPS verifies the achievement of efficient multi-machine, multi-workpiece processing, and multi-productivity.

Conflict of interest
The authors have no relevant financial or non-financial interests to disclose. The authors have no competing interests to declare that are relevant to the content of this article. All authors certify that they have no affiliations with or involvement in any organization or entity with any financial interest or non-financial interest in the subject matter or materials discussed in this manuscript. The authors have no financial or proprietary interests in any material discussed in this article.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.  1  10  0  7  0  3  0  0  10  0  7  0  0  8  0  15  0  0  15  1  31  0  20  7  1