In this paper, we propose a communication network architecture for industrial applications that combines new 5G technologies with other existing communication technologies on the shop floor. This architecture connects private and public mobile networks with local networking technologies to achieve a flexible setup addressing many different industrial use cases. We show how the advancements introduced around the new 5G mobile technology can address a wide range of industrial requirements. We further describe relevant use cases and develop an overall communication system architecture proposal, which is able to fulfill not only technical requirements but also system requirements, which result from specific applications existing in today’s and future manufacturing scenarios.
In diesem Paper wird eine Architektur für Kommunikationsnetze für industrielle Anwendungen vorgestellt, die neue 5G-Technologien mit vorhandener Kommunikationstechnik auf der Feldbusebene kombiniert. Diese Architektur verbindet private und öffentliche Mobilfunknetze mit lokalen Funktechnologien, um einen flexiblen Aufbau zu ermöglichen, der in der Lage ist, viele industrielle Anwendungsfälle zu unterstützen. Es wird gezeigt, wie die Errungenschaften, die mit der neuen 5G-Technologie eingeführt werden, einen großen Bereich der industriellen Anforderungen erfüllen können. Weiterhin werden relevante Anwendungsfälle beschrieben und eine Gesamtsystemarchitektur vorgeschlagen, welche nicht nur die technischen, sondern auch die funktionalen Anforderungen, welche von den spezifischen Anwendungen heutiger und zukünftiger Herstellungsprozesse gestellt werden, erfüllen kann.
The growing digitalization and interconnection of manufacturing processes is leading to a closer gearing of companies, suppliers and customers. Production becomes increasingly flexible and the connectivity of the devices has to become likewise flexible. This is why 5th generation mobile networks (5G) are considered to be key enablers for new use cases towards the vision of Industry 4.0, such as advanced condition monitoring, predictive maintenance and product quality assurance. In this regard, the research initiative “5G: Industrial Internet (5G: II)”  of the German Federal Ministry of Education and Research addresses the requirements on a 5G communication network in industrial production. As a part of this initiative, the collaborative 5Gang research project  considers different use cases of future industrial production and their requirements on the communication network. The project scope covers not only local production sites but also opportunities arising from adding inter-site connections. 5Gang is a consortium of eight partners from industry and academia who bring in their experience covering technical, business and production process aspects.
In this paper, we describe a way towards a flexible architecture for industrial networking, which is able to support the heterogeneous networking technologies in today’s production facilities while benefiting from 5G technology at the same time. In Section 2, we summarize related work on 5G and establish relations to existing architectures. In Section 3, we present use cases along with their technical requirements on 5G systems, as well as system requirements together with exemplary uses cases, which we use to verify the developed system architecture. We detail our 5Gang architecture, its system components, functions and interfaces in Section 4 and describe in Section 5, how the system architecture integrates into existing reference architectures. Finally, we draw conclusions in Section 6.
2 Related work
The field of industrial information technology and industrial communication already plays an important role in today’s production facilities and is increasingly becoming a focus of interest from both academia and industry along with the course of the German “Industry 4.0” initiative. As we will show in the following sections, the fulfillment of requirements of manufacturing use cases necessitates the use of 5G systems, as all three main capabilities of 5G are addressed, either by the sheer number of participants in the network (massive Machine Type Communications (mMTC)), by demanding real-time capabilities (Ultra-Reliable and Low Latency Communications (URLLC)) or by demanding very high bandwidth (Enhanced Mobile Broadband (eMBB)).
There are already descriptions of 5G system architectures, partly covering manufacturing scenarios, which we summarize in the following. Also, industrial use cases and related requirements have been discussed in the literature already. We will provide a short overview in the following.
2.1 Benefits of 5G
The standardization of 5G is planned to be finished by 2020. Its first release, Release 15, was already approved by 3GPP in June 2018 and first commercial, pre-standard systems were deployed already in 2018. By the end of 2019, 3GPP Release 16 will be frozen. Although requirements might change under way, the overall targets are taking shape and can be considered as a basis for 5Gang. The United Nations organization ITU-R addresses three main capabilities that define the requirements for 5G :
eMBBwill provide up to 20 Gbit/s of data towards the end-users (100 Mbit/s per user) and 10 Gbit/s (50 Mbit/s per user) from the end-users towards the network. The user plane latency shall be below 4 ms.
URLLC, or critical Machine Type Communications (cMTC), requires a user plane latency of less than 1 ms.
Softwarization and virtualization: The same physical resources (communication, computing, storage) can be used for many use cases. A distributed cloud (DC) allows to move certain nodes of the mobile network closer to the end-user, which, in the case of a production environment, can even mean core network nodes. From an economic standpoint, this would not be reasonable with high-performing, specialized hardware. Here, the virtualization techniques reduce costs: Instead of installing expensive hardware in the factory, 3GPP has standardized the option to execute certain virtualized user plane functions inside the factory. This allows to keep data inside the factory, which is an important privacy aspect, and reduces network latency .
Software Defined Networking (SDN): SDN  enables network elements to be controlled by a central intelligent management platform, enabling dynamic and flexible information routing. Based on SDN, network slicing ,  dynamically reserves resources per application and provides the required QoS, improving system reliability and flexibility (see, e. g.,  for the condition monitoring use case).
Mesh network: In a mesh network , each node can have a direct connection to several neighbor nodes and cooperate with them to efficiently route data. Mesh networks are self-organizing, self-healing and provide a high robustness against link/node loss. Also, the communication range and resilience can be increased while enabling efficient device-to-device communication.
Mobility support: Moving workpiece carriers can be controlled or tracked not only inside the factory but also on their way between different production sites.
Energy efficiency: 5G allows operating times of 10 years for transmitting small volumes of data from battery-powered devices in an energy-efficient manner.
Security: SIM cards (or alternatives) provide a secure way to manage devices and restrict network access.
Economy of scale: The large 5G ecosystem will increase the volume of the radio modems leading to cheaper equipment, which would not be possible if every production solution used separately specialized hardware.
2.2 5G architecture descriptions
In the following, we provide an overview of the 5G architecture descriptions existing today and will investigate, how our proposed architecture aligns with them.
2.2.1 NGMN 5G architecture for verticals
The “5G White Paper”  of the NGMN consortium is one of the first that brought up 5G for verticals. It addresses verticals and envisions the use cases of massive sensor networks in the Internet of Things (IoT), which is termed “massive IoT”, of ultra-reliable communication, and of extreme real-time communication, e. g., for collaborative robots. All three use cases are also considered in the 5Gang project and require security, identity and privacy; real-time, seamless and personalized experience; responsive interaction and charging, as well as QoS and contextual behavior of the system. Beyond connectivity, 5G may offer services like transparency of connectivity, location, resilience, reliability, and high availability. We design our architecture, such that it integrates with and details the (virtualized) infrastructure resource layer and parts of the business enablement layer of the “5G architecture” in [14, sec. 5.4].
2.2.2 5GPPP 5G architecture
A 5G architecture has been developed by the EU-funded project “5GPPP” in . It primarily focuses on network slicing and the radio access. It also gives a detailed overview of the capabilities of each component and describes the infrastructure, into which the 5Gang architecture well integrates for the use in the factory of the future. The 5GPPP architecture is divided into resource/functional, network and service levels. The resource/functional level provides the physical resources for communication, computation and storage. In addition to the wireless and fixed access it consists of edge cloud and core/central cloud resources. Virtualization of the physical network is achieved by a network operating system and programmable network control units, such that network slices can be built on top. At the service levels, these slices are orchestrated in an end-to-end fashion using management functions of all levels. As our architecture enables network slicing by providing the physical network infrastructure and corresponding management and orchestration interfaces, it integrates well into the 5GPPP architecture.
2.2.3 3GPP 5G architecture
3GPP also proposed a first architecture of its 5G system in . It details the 5G components on the network side, e. g., base stations and the core network, assuming that 5G-enabled devices will have an anyhow natured 5G modem. In this paper, we embed the architecture of  into the factory of the future and abstract it as the “cellular backend”.
2.2.4 5G-ACIA 5G building blocks
The 5G Alliance for Connected Industries and Automation (5G-ACIA)  is a global initiative that brings together Operational Technology (OT) industry, Information and Communication Technology (ICT) industry, and academia to provide a common platform for discussing technical, regulatory and business aspects of 5G for the industrial domain. In their white paper , they provide an overview of 5G’s basic potential for manufacturing, outline use cases and requirements, and name important 5G building blocks. There exists a plethora of operational/functional requirements that go far beyond the technical requirements of industrial use cases. In addition to availability and reliability, the networks must feature maintainability, safety and integrity. For example, functional safety considers safety measures that prevent harming people or the environment and which must be integrated as native network services. Towards the success of 5G in the industrial domain, the 5G-ACIA promotes in  the integration of 5G-enabled industrial components with legacy communication technology, the operation of private 5G networks, seamless handovers between public and private 5G networks, and end-to-end network slicing across technologies, countries and network operators. 5G-ACIA particularly names wireless connectivity, edge computing, and network slicing as technologies that make 5G disruptive for the manufacturing industry. In this paper, we tackle many of the mentioned technologies and integrate them into our proposed architecture for industrial networking.
2.3 Requirements of industrial use cases
In the following, we briefly review existing technical requirements derived from industrial use cases on proprietary wireless communications and 5G systems.
2.3.1 Requirements derived within the ZDKI programme
Requirements of industrial use cases have been investigated in research projects of the German funding programme “Reliable wireless communication in industry” (German acronym: ZDKI). For example, in the “HiFlecs” project, requirements of different industrial applications have been grouped into three requirement profiles . In , two specific use cases of the “KoI” project and their quantitative requirements have been provided. The results in the “ZDKI” programme have been collected by the accompanying research project (German acronym: BZKI) in the technical group 1: “Applications, Requirements and Validation” and have been published in the report . A thorough compilation of different industrial applications and quantitative requirements has been provided in . The authors showed that the spectrum of applications in an industrial environment is very broad and consequently their requirements are very distinct. The authors conclude that in order to have a suitable system design, different technologies have to be combined purposefully with regard to the specific applications, which we attempt to realize in our proposed architecture, as well.
2.3.2 Requirements from 3GPP TR 22.804
In preparation of Release 16, 3GPP performed a study on communication for automation in Vertical Domains. The resulting report  names use cases and their requirements for the 3GPP 5G system. One major part is on the use case class “factories of the future” in general and on factory automation in particular, as well as on industrial security requirements. Within these sections, the report  addresses similar use cases as 5Gang, and the technical requirements of 5Gang and  are well-aligned such that this paper can be considered to detail the first approach of the principle architecture in .
3 Use cases and requirements
In the following, we review some exemplary use cases and technical requirements of 5G which have been developed by 3GPP and within our 5Gang project. All use case classes of , except for the “process automation” class, are considered and fulfilled when designing the overall network architecture. However, for brevity, we only introduce the ones, for which we exemplary will check that their system requirements are fulfilled by the proposed system architecture (cf. Section 4). Whereas the focus of 3GPP is on the use cases’ technical requirements, in Section 3.2 we derive and highlight functional system requirements, which are important for the design of our system architecture.
3.1 Example use cases and technical requirements
We now describe some use case classes of 3GPP and give a more application-specific view compared to . We only touch technical requirements of our use cases by highlighting some values and classifying them into the eMBB, URLLC and mMTC categories (see Section 2.1). Safety requirements are mostly omitted for the sake of brevity, as a detailed safety architecture is left as future work. At first, 5G can be used to extend the field of application in an innovative way while ensuring backward compatibility. A second use case class focuses on mobile robots and mobile material supply in a factory for mounting or distribution. The third use case class focuses on the production operations themselves. Applications like tracking and tracing systems, as well as flexible production without any wired connected machines are examples within this class. The fourth class addresses the control of the processes of the production and logistics.
3.1.1 Use case class 1: Infrastructure retrofit
The benefits of 5G for infrastructure retrofits are included in the use case “connectivity on the factory floor” of . Within 5Gang we divide this use case into three sub-problems: First, existing sensor and actor technologies can be enhanced by their integration with mobile communication technology modules. In this regard, 5G is planned to be more energy-efficient than, e. g., LTE and provides a wider coverage than Bluetooth Low Energy and the-like. This use case class contains all kinds of combinations of sensor stacks (AUTO-IDs, cameras, condition sensors, etc.) with 5G modules. Here, the potential for large-scale infrastructure retrofits and improved, wireless connectivity between retrofitted sensors and actuators is unlocked by both, 5G’s high bandwidth (eMBB) and ability for critical communication (cMTC or URLLC). Second, existing communication infrastructure is often very weak in many terms, e. g., in old production facilities without a proper IT infrastructure or in large production areas without Internet connection in the countryside. Here, 5G provides a more reliable network than LTE and can transfer critical data. The third case addresses information multiplexing, e. g., in a machine. The retrofit of communication systems in machines by 5G allows outsourcing of the multiplexing process by network slicing, so that configurations of the machine can be performed independently from the machine and its use case.
Within 5Gang, we extended this use case (class) by the process automation-related use case “decentralized measurement network”, e. g., devices, which measure the gas flow through pipelines in remote areas for central gas flow monitoring and regulation. Due to long distances between measurement spots, e. g., across countries, and decentralized equipment, 5G can provide connectivity to the sensors. In areas of no coverage, few sensors can act as 5G gateways, where the sensors are connected with each other using another mesh network technology. Since the pipeline infrastructure already exists, we classify it into retrofit. Here, a latency of the information multiplexing processes is essential, such that the 1 ms provided by 5G is the key potential for this use case (URLLC).
3.1.2 Use case class 2: Mobile robots
This class includes applications with different automated driven vehicles, like automated guided vehicles (AGVs), which transport material in a dynamic environment. In a well-planned production, the material flow follows a milk run principle with dynamic changes of the milk run route. In some cases, AGVs can be controlled in a master-slave relation, in which only the master AGV communicates with the production system and the slave-AGVs are connected to the master through their own network.
The most challenging scenario is “Distributed Indoor SLAM” (SLAM: Simultaneous Location And Mapping), in which AGVs or mobile production robots reliably navigate in a spatially varying production environment using high-precision navigation based on real-time maps, as shown in Fig. 1. Each AGV provides its own measurements for other devices to create the common map. The mapping needs to be timely synchronized (within less than 50 ms of latency) and shared among the devices, so that they can simultaneously locate themselves and plan further movements. The necessity of using 5G technology arises from the real-time requirement of the served application (URLLC). Furthermore, high bandwidth requirements exist (eMBB). Also, computing power, e. g., using a local (edge) cloud is needed for generating the timely synchronized maps in real-time.
3.1.3 Use case class 3: Inbound logistics for manufacturing, flexible and modular assembly area, plug-and-produce
Due to the transition to flexible manufacturing and customization of the production processes, companies increasingly ask for a dynamic order-change request system. This requirement demands reliable tracking of the desired product during the entire supply chain process. Hence, “Track and Trace” applications cover orders and material flow across different locations of the supply chain and include sub-use cases such as the following: A company-independent tracking via an Auto-ID combined with mobile communication technologies like 5G, as well as dynamic allocations for material in different production environments (different facilities). Attaching sensors to workpiece carriers provides a solution approach to track and trace each workpiece carrier independently when connected via a mobile network. This redesigned sensor-stack using 5G can be realized independently for each workpiece carrier across company locations, even on global logistic routings. As a result, the customer will have an overview of the exact location of his product and which modifications can still be ordered.
The second problem that requires a shared tracking and tracing system is the stocking of material in a shared warehouse. So far, the delivery of materials requires a manual check-in followed by the search for a suitable stocking place in a warehouse. To optimize this, a shared, digitalized, 5G-connected warehouse provides a solution approach.
In 5Gang another use case has been identified, which tracks the product quality along the production line. Additional processing of defective parts is costly and should be avoided. So, the use case aims to detect defects in parts at earlier stages through inline quality control. Corrective measures can be taken and necessary post-production can be initiated before final inspection is completed. A similar case occurs when tracking items within a logistics chain. Here, the transport conditions (e. g., cooling temperature) are monitored continuously. In the event of significant deviations, a re-order can be made before the items arrive. In the former example, tracking of items is performed within a production facility, whereas in the latter, the item is tracked around the world. In both example, sensors, placed on or near the item, on the workpiece or on the workpiece carrier, transmit their measured values (temperature, humidity, shock, vibrations) to a central point for evaluation. The benefit of using 5G technology arises from the high amount of participants in the network (mMTC).
3.1.4 Use case class 4: Massive wireless sensor network and process monitoring
In production, unexpected machine defects cause down-times, which result in high costs and delays in delivery, and the quality of products depends on the condition of machines which produce them. Automated supervision of machine conditions can prevent unplanned defects, enable planning of maintenance activities, and support continuous product quality control. Distributed sensing of condition values, e. g., noise recordings, offers a solution for anomaly detection. Collecting such data from several locally uncorrelated sources requires a large number of sensors. The 5G mMTC profile is optimized towards these requirements in order to build an Industrial Internet of Things (IIoT) network by supporting up to 1,000,000 devices per km2, and dedicated frequency bands allow for independence from the existing infrastructure. This use case requires high mobile broadband connectivity because sensor technologies like high resolution cameras need to be integrated into the machines. If the production process contains many different machines with various connected sensors and actors, the number of connected devices rises and a transmission technology like 5G is necessary to ensure a reliable IIoT. In order to realize services, such as predictive maintenance, condition monitoring or anomaly detection, sensor nodes need to be allocated to dynamic processes and to be mobile over different locations.
In a specific use case, audio data recorded by microphones shall be sent to a central evaluation unit (e. g., in a cloud), which detects anomalies in this data. The sensors have limited mobility and can be distributed across the entire production hall. They can change their positions occasionally (e. g., during reconfigurations or rebuilds) and they are required to have a long battery life, since they can be placed at locations that are difficult to access. The massive amount of wireless sensor nodes will require the 5G mMTC profile as an enabler. Furthermore, a proper device management must be installed to manage and update the wireless sensor participants and to control access rights.
3.2 Functional system requirements
The functional requirements on the system basically equal the main reasons for a plant manager to introduce the new network system. In principle, the new system must have a good overall return on invest within only a few years and shall be future-proof. The return typically comes either from higher production quantity at the same fixed effort (increased efficiency), from a higher degree of automation or from slimmed processes, which in turn increase production quantity, too. As we do not see the one and only killer use case, which will pay off the invest, the system needs to accumulate benefits from different use cases—and the collection of beneficial use cases will for sure differ from plant to plant. The invest includes effort for purchasing and operating the network, for retrofitting solutions, and necessary adoptions of existing systems and processes. A smooth transition from legacy solutions to new technology, e. g., 5G, is necessary, requiring interoperability across systems and technologies. Salient requirements are the following:
Flexible and Interoperable Infrastructure: The system shall support all use cases with one common and interoperable infrastructure, which can be deployed worldwide. User equipment surely will implement only subsets of the standard, which are appropriate for the specific application, as not all technical requirements of all use cases can be fulfilled at a time, e. g., high data rate with long battery lifetime.
Flexible Network Ownership and Operation: This means in particular:
Easy and fast deployment, operation and management in small and medium-size companies, which do not want to afford owning the network infrastructure and which want to use an operator instead, as well as in large-size companies, which might want to operate their own network infrastructure.
Private, sensitive production data (the key value and know-how of production) shall stay on-premise and private. Data shall be shared only through well-defined channels, e. g., across different plants.
Thus, private networks shall be supported in licensed (local/regional licensed spectrum or sub-licensed from MNO) and in unlicensed bands (2.45 GHz WLAN).
In case the manufacturer does not setup a private network, public infrastructure shall provide a local break-out without the operator being able to access the data.
System and network operation/management with minimum human/operator intervention and automatic interference alignment between neighboring plants (public-private, private-private).
5G gateways and sensor nodes shall be easy to install and maintain as costs scale with installation effort.
Existing production/process infrastructure shall be securely connected through their existing interfaces (fieldbus, Industrial Ethernet, etc.) and 5G modems in order to become a part of the overall network.
The infrastructure shall allow to build isolated sub-networks with well-defined transfer paths.
The system shall integrate existing infrastructure without adjustments or additional hardware/software for incremental retrofitting. This particularly includes the integration of different wired or wireless access technologies like (Industrial and classic) Ethernet, field bus networks, WLAN, RFID and Bluetooth.
The system shall support indoor and outdoor operation with handover. The typical size of an indoor shop floor is some 100 m in square.
All infrastructure shall be wirelessly covered. If some area is not covered by a base station, device-to-device or mesh network technologies shall be supported.
Machines and devices shall be managed remotely, e. g., by a cloud service.
The overall network shall support intra-site connections and (seamless) connectivity along the logistics routes.
Localization shall be supported with required accuracy ranging from 1 m to 1 cm in three dimensions.
The system shall manage non-cellular radio sub-systems, such that intra-system interference is avoided.
In order to extend coverage and to handle high numbers of devices, gateways shall support mesh network technologies or 5G device-to-device.
The system shall support user equipment to establish multiple connections to different base stations in order to balance data traffic and to increase reliability.
4 5Gang system architecture
The proposed architecture and its involved components, which are depicted in Fig. 5, have been designed to meet the requirements derived above while assuming 3GPP’s 5G will fulfill all their technical requirements in each flavor, eMBB, URLLC and mMTC. In the following, we present the architecture along with descriptions of its components.
Apart from the improvements and specifications mentioned in Section 2, our architecture covers not only the 5G context but also technologies already in use in today’s manufacturing facilities. This flexible combination of technologies to enable specific use cases forms, to the best of the authors’ knowledge, a unique characteristic. The network system is a system of systems, where typically functions are already mapped onto components. Hence, and for the ease of description, we combine components, the functions they provide and the respective interfaces into one description. As has been described in Section 2, softwarization and virtualization will make the alignment of component and function arbitrary in many cases, which backs the reasons.
4.2 5G RAN and core network
On a very high level the network consists of a core and a radio network. 5G will provide a new architecture  building on a completely new core network, 5G Core, and an optimized radio network, 5G RAN. Normally the network is operated by Mobile Network Operators (MNO) because they own the frequency spectrum. If a factory owns or leases the spectrum it can as well operate an own network. There is also the option to run only critical components as private infrastructure on-premise and connect it to a public network, which is managed by a MNO. In the scope of this paper we assume that the network supports public/private operations with all the respective requirements given in Section 3.2, as well as converge with other network technologies like WLAN, fieldbus and Industrial Ethernet by the use of the proposed Aggregation Point.
The 5G RAN is often called 5G New Radio (5G NR). It uses an optimized radio interface with a much higher bandwidth compared to the 4G RAN. For sensors with a very low bandwidth it can as well operate in a more energy efficient setup. For more stringent URLLC use cases, sensors and actuators are most likely be connected via a 5G NR interface or a wired network.
The 5G Core network will be built on cloud native software meaning among other advantages that the SW is better adapted for virtualization. This will enable the implementation of smaller software components for smaller factories in less powerful, smaller data centers.
4.3 Aggregation point (AP)
The aggregation point (AP) is the core of our proposed architecture. It is primarily responsible for providing IP-based connectivity of the networks with large numbers of sensors, controllers, and actuators, all of which could be using different wireless/wired technologies, to the core 5G network and/or to the cloud. The AP is thus to be located at the edge of the network and includes different wireless, as well as wired interfaces. The AP employs SDN solutions to manage the traffic within and between its connected networks. It comprises of an SDN-capable switch (e. g., Open vSwitch ) to facilitate the traffic management, and an SDN controller (e. g., Ryu, OpenDaylight) to obtain a centralized logical view of the network. A radio management system entity is also part of the AP. It manages a large number of connected wireless nodes while implementing self-organization network techniques to optimize the overall network performance. Regarding the functional system requirements derived in Section 3.2, the aggregation point is able to address the following demands:
The AP provides a gateway functionality, i. e., it allows the integration of heterogenous radio technologies while maintaining transparent routing.
It provides an interface between the fieldbus domain and the network domain, while considering the different network properties.
It offers different wireless interfaces, thus providing 5G-capability to existing hardware, i. e., in a brown-field scenario.
Being equipped with several wireless interfaces, the AP allows to provide seamlessness during handover procedures and to improve reliability and/or data rate by exploiting interface diversity.
4.4 IIoT edge gateway
The IIoT edge gateway primarily provides connectivity to networks with a large number of sensors, controllers and actuators, all of which can use different wireless/wired technologies. It is thus located at the edge of the 5G network close to the data provider and at the edge of the network leading from the machine to the cloud. The main task of the gateway is to collect, edit and reduce the data available (at the edge), thereby optimizing the transport by the network in terms of speed and costs, which distinguishes it from the transparent routing in the aggregation point. It interfaces sensors and actuators and provides them with further connectivity, e. g., to a (edge) cloud service or a service within the production environment/company network. Therefore, it contains a 5G modem and includes different wireless as well as wired interfaces like analog/digital I/Os (e. g., classic 4 to 20 mA/0 to 10 V), (Industrial) Ethernet and Ethernet-based fieldbuses (PROFIBUS, Sercos III, CAN, Modbus, etc.) or wireless (Bluetooth low energy, IEEE 802.11 WiFi). As a special case, a 5G device-to-device link is used, where sensors and actuators share the 5G connection of the gateway. Furthermore, the gateway contains storage and processing resources, which might offer edge cloud services to the connected devices. From a production point of view, the AP shall interface to the process, i. e., collect the operating and diagnostic data of the machine or its sensors together with a time stamp. The compute and storage resources provide maximum flexibility to the user: The data can be cached, pre-processed and analyzed on-site such that, e. g., only alarms are forwarded through the 5G interface. In addition, the type of transmitted data and the size of packets can be dynamically managed and different modes of link operation can be used in order to adjust connectivity costs: an online mode with permanent connection; an interval mode which transmits only regularly at specified periods; and a sleep mode where the device transmits only if needed. Furthermore, classic IT security mechanisms can be implemented using transport layer and end-to-end encryption. As special cases, the processed data can be sent through the mobile radio provider with the strongest signal (“unsteered roaming”). The attached devices are typically managed via a cloud service (device cloud). An extremely stripped-down version of the IIoT edge gateway could be wireless sensor/actor devices themselves, which build up a mesh network among each other.
4.5 Device cloud
The device cloud handles tasks such as user, device and access rights management and sends new settings to the device. Furthermore, it lets users further process data using a standardized API (e. g., RESTful), in an Enterprise Resource Planning (ERP) system or with data cloud services. Further functions are alerting, sending and receiving confirmation via SMS, field strength display of the current data connection, and creation of reports and their delivery to authorized recipients.
4.6 Non-cellular radio management
The radio management entity gathers context information about the radio environment and particularly about the link quality of the wireless links associated with the AP. This information can be used to apply self-organization mechanisms to the overall connectivity solution using the SDN capabilities of the AP in order to achieve a complication-free operation of different wireless technologies. With this, the functional requirement of providing interference alignment, derived in Section 3.2, is fulfilled. Moreover, utilizing several wireless interfaces, interface diversity can be exploited, which in turn leads to an improved reliability and/or to an increased data rate, which are functional requirements of many industrial use cases.
One example is shown in Fig. 2. A device, in this case a controller, is connected to the AP using two independent WLAN adapters. In a first step, only one wireless link is established using one WLAN interface, which occupies channel A. The SDN switch is configured in a way, such that traffic going through this interface is routed towards the cloud. The second WLAN interface is used for wireless monitoring. Both interfaces send context information about the current wireless status to the radio resource management. Using this information, if the utilization of a different channel would be beneficial, a second WLAN link is established using channel B. The SDN switch is reconfigured, so that the second wireless interface is now routed to the cloud. After this, the first wireless link can be shut down. This process is “make before break” and works seamlessly, i. e., without packet loss. It is also possible to improve this approach with multipath techniques such as Multipath TCP or parallel redundancy protocol (PRP). The intelligent Radio Management entity also enables to align interference between private and public networks, and between private networks, which increases the robustness of the communication and reduces human and operator intervention. The latter itself reduces costs for the plant owner.
4.7 SDN controller
The SDN controller is the centralized management unit of the network. It acts as a strategic controller in the SDN network, manages the flows inside this network and computes the route which a flow has to follow. In order to do so, it manages the flow-tables on SDN switches which are along this route. With the help of an SDN controller, it is possible to configure the network dynamically so that it meets the changing needs related to configuration, security and optimization. The most popular protocols used by SDN controllers to communicate with SDN switches are OpenFlow and Open vSwitch data-base (OVSDB). OpenDaylight and Ryu are two typical SDN controllers.
4.8 SDN switch
An SDN switch is a device, which receives, sends and forwards data packets in a network in order to meet specific requirements. It follows the rules in a flow-table, which is managed by the SDN controller via SDN protocols, like OpenFlow. There are some virtual SDN switches, e. g., Open vSwitch (OVS), to provide a switching stack for hardware virtualization environments. In case high switching capacity is needed, dedicated SDN hardware switches are available which are offering enhanced SDN performance.
4.9 Tracing of requirements
With the exception of localization, all functional system requirements of all use cases, and especially those of the exemplary ones shown in this paper, are fulfilled by the communication system architecture. Table 1 shows a summary of the requirements trace.
|Functional system requirements||Fulfilled by component|
|Integration of heterogeneous radio technologies, transparent routing||Aggregation Point|
|Interface between “fieldbus domain” and “network domain”||Aggregation Point|
|Retrofitting, Support of brown-field scenarios||Aggregation Point, IIoT Edge Gateway|
|Separation of private and public data transmissions||SDN Controller, SDN Switches, Private 5G RAN, Private 5G Core|
|Interference alignment, preventing intra-system interference||Non-cellular Radio Management|
|Seamlessness (Handover, Reliability)||Aggregation Point (Multipathing), SDN (Network topology)|
|Very high data rates (uncompressed video streams)||Aggregation Point (Multipathing)|
|IT-Security||Inherent cellular security mechanisms (SIM cards)|
|Remote Management of devices||Device Cloud|
5 RAMI 4.0 and IIRA integration
In the literature, there are similar reference architectures for productions already described and it seems sensible to compare them to the architecture proposed in this paper. For the vertical integration of production technologies, the two most-cited reference architectures for the Industrial IoT (IIoT)  are the Industrial Internet Reference Architecture (IIRA) of the Industrial Internet Consortium (IIC) and the RAMI 4.0 architecture of the German “Plattform Industrie 4.0’, for which we show ways of integration of our proposed architecture below.
5.1 RAMI 4.0
The “Reference Architectural Model Industrie 4.0” (RAMI 4.0) ,  has been designed as a future reference model for industrial production and automation to categorize and differentiate different architectural views that are related to each other. The RAMI4.0 (cf. Fig. 3) is structured as a three dimensional model comprised of the axes Hierarchy Levels, Layers and Life Cycle & Value Stream. The Hierarchy Levels represent the classical automation pyramid, which structures different layers of responsibility and aggregation from field devices over control hardware to higher-level applications (Manufacturing Execution System (MES), ERP etc.). The Hierarchy Levels enhance the classical pyramid by the categories Product and Connected World, which includes intelligent (communicating) products, as well as interconnecting enterprises and shop floor software technologies to cloud technologies. The industrial networking architecture presented in this paper covers all hierarchy levels, i. e., its offered functionality addresses every layer in the automation pyramid, as well as the connection to superordinate cloud services.
The RAMI4.0 Layers axis divides the solution into six functional levels, of which the communication layer is covered by the architecture presented in this paper. The last dimension of the RAMI4.0 addresses the Life Cycle and Value Stream of products. It divides the product development and the usage process into a type and an instance phase. Whereas the type phase refers mainly to product development including documentation, construction plans, etc., the instance phase refers to the usage phase of the product, in which data is collected during operation. The architecture presented in this paper covers run-time of the 5G system and hence the instance phase of a product life cycle, which is more challenging from a communication point of view.
The IIRA  of the Industrial Internet Consortium (IIC) follows ISO/IEC/IEEE 42010:2011 “Systems and Software Engineering—Architecture Description”  and contains an IIoT architecture framework, which in turn contains views on stakeholders, concerns and viewpoints as its architecture frame, and views and models as its architecture representation. Considered viewpoints are business, usage, functional and implementation, which help to find stakeholders of an IIoT solution. In general, the IIRA consists of three dimensions: Functional domains, system characteristics and cross-cutting functions (cf. Fig. 4). It considers functional domains, namely control, operations, application, business and information. As highlighted in Fig. 4, our proposed architecture covers all functional domains except the Business domain and all system characteristics except Safety. Our architecture addresses especially the Information functional domain as 5G allows efficient and high-performance communications. The 5G system allows information gathering and is part of the cross-cutting functions; however, it is limited to the functions Connectivity and Distributed Data Management contained therein.
Our use case analysis has shown that a transition from a legacy production factory to the factory of the future is a complex task. It is important to consider the diversity of use cases on the one hand, as well as the requirements imposed by plant owners and managers on the other hand. While 5G is expected to fulfill the technical requirements, such as data rate and latency, the process requires additional efforts to design and deploy an architecture, which takes into account interoperability of legacy communication technology, operation (business) models of the wireless infrastructure, and maintainability.
In this regard, we proposed a communication network architecture for industrial applications, which combines future 5G technology and non-cellular network technologies with existing technologies, such as fieldbus, on the shop floor. Through the introduction of IoT Edge Gateways, communication technology aggregation points and by leveraging the benefits of SDN, network slicing and intelligent radio resource management, the architecture is flexible enough to address many different industrial use cases. It is also designed in a way to reduce human intervention and provides the possibility to operate private and/or public cellular networks depending on the factory owner’s needs and capabilities. We have also illustrated that our proposed architecture fits well into the existing architectural frameworks and builds upon existing 5G architectures, which makes it universally suitable framework for future industrial networking.
5Gangs’s future work comprises the implementation of system components and the actual numerical evaluation of the overall system performance with regard to the identified requirements.
Funding source: Bundesministerium für Bildung und Forschung
Award Identifier / Grant number: 16KIS0725K
Funding statement: This work has been supported by the Federal Ministry of Education and Research of the Federal Republic of Germany (Foerderkennzeichen 16KIS0725K, 5Gang).
The authors alone are responsible for the content of the paper.
1. “5G: Industrielles Internet,” https://www.bmbf.de/foerderungen/bekanntmachung-1044.html, accessed 2019-02-13.Search in Google Scholar
2. “5Gang – 5G angewandt in der Industrie,” https://5gangprojekt.com/, accessed 2019-02-13.Search in Google Scholar
3. “Minimum requirements related to technical performance for imt 2020 radio interface(s),” https://www.itu.int/md/R15-SG05-C-0040/en, accessed 2019-02-13.Search in Google Scholar
4. E. Dutkiewicz, X. Costa-Perez, I. Z. Kovacs and M. Mueck, IEEE Network, vol. 31, no. 6, pp. 6–7, 2017.Search in Google Scholar
5. C. Bockelmann, N. Pratas, H. Nikopour, K. Au, T. Svensson, C. Stefanovic, P. Popovski and A. Dekorsy, “Massive machine-type communications in 5g: Physical and mac-layer solutions,” IEEE Communications Magazine, vol. 54, no. 9, pp. 59–65, 2016.10.1109/MCOM.2016.7565189Search in Google Scholar
6. Z. Dawy, W. Saad, A. Ghosh, J. G. Andrews and E. Yaacoub, “Toward massive machine type cellular communications,” IEEE Wireless Communications, vol. 24, no. 1, pp. 120–128, 2017.10.1109/MWC.2016.1500284WCSearch in Google Scholar
7. S.-Y. Lien, S.-L. Shieh, Y. Huang, B. Su, Y.-L. Hsu and H.-Y. Wei, “5g new radio: Waveform, frame structure, multiple access, and initial access,” IEEE communications magazine, vol. 55, no. 6, pp. 64–71, 2017.10.1109/MCOM.2017.1601107Search in Google Scholar
8. “Architecture enhancements for control and user plane separation of epc nodes; stage 2 (release 15),” https://portal.3gpp.org/desktopmodules/Specifications/SpecificationDetails.aspx?specificationId=3077, accessed 2019-02-13.Search in Google Scholar
9. Software-defined networking: The new norm for networks, Open Networking Foundation, 2012.Search in Google Scholar
10. “Architecture enhancements for dedicated core networks; stage 2 (release 13),” https://portal.3gpp.org/desktopmodules/Specifications/SpecificationDetails.aspx?specificationId=868, accessed 2019-02-13.Search in Google Scholar
11. “3gpp tr 23.711, enhancements of dedicated core networks selection mechanism,” http://www.3gpp.org/DynaReport/23711.htm, accessed 2019-02-13.Search in Google Scholar
12. H. Wu, G. T. Nguyen, A. K. Chorppath and F. H. Fitzek, “Network slicing for conditional monitoring in the industrial internet of things,” IEEE Software Defined Networks, IEEE Softwarization, Jan 2018.Search in Google Scholar
14. 5G white paper, NGMN Alliance, 2015.Search in Google Scholar
15. S. Redana, A. Kaloxylos, A. Galis, P. Rost and V. Jungnickel, “View on 5g architecture,” White paper of the 5G-PPP architecture WG (2016, July), 2016.Search in Google Scholar
16. J. Kim, D. Kim and S. Choi, “3GPP SA2 architecture and functions for 5G mobile communication system,” vol. 3, no. 1, pp. 1–8, http://www.sciencedirect.com/science/article/pii/S240595951730019X, accessed 2019-02-13.Search in Google Scholar
17. “5G-ACIA Website,” https://www.5g-acia.org/, accessed 2019-01-09.Search in Google Scholar
18. White Paper – 5G for Connected Industries and Automation – Second Edition, 5G-ACIA, 2018.Search in Google Scholar
19. C. Bockelmann, A. Dekorsy, A. Gnad, L. Rauchhaupt et al., “Hiflecs: Innovative technologies for low-latency wireless closed-loop industrial automation systems,” in 22. VDE-ITG-Fachtagung Mobilkommunikation, Osnabrück, Germany, May 2017.Search in Google Scholar
20. B. Holfeld, D. Wieruch, T. Wirth, L. Thiele, S. A. Ashraf, J. Huschke, I. Aktas and J. Ansari, “Wireless communication for factory automation: an opportunity for lte and 5g systems,” IEEE Communications Magazine, vol. 54, no. 6, pp. 36–43, 2016.10.1109/MCOM.2016.7497764Search in Google Scholar
21. L. Rauchhaupt, D. Schulze, A. Gnad and M. Krätzig, “Requirement profiles in zdki,” Accompanying Research – Reliable wireless communication in industry (BZKI), Tech. Rep., oct 2016.Search in Google Scholar
22. I. Aktas, A. Bentkus, F. Bonanati et al., “Wireless technologies for industrie 4.0,” Frankfurt am Main, 2017.Search in Google Scholar
23. 3GPP, “Study on Communication for Automation in Vertical domains (CAV),” https://portal.3gpp.org/desktopmodules/Specifications/SpecificationDetails.aspx?specificationId=3187, accessed 2019-02-13.Search in Google Scholar
24. “System architecture for the 5g system (5gs),” https://portal.3gpp.org/desktopmodules/Specifications/SpecificationDetails.aspx?specificationId=3144, accessed 2019-02-13.Search in Google Scholar
25. B. Pfaff, J. Pettit, T. Koponen, E. J. Jackson, A. Zhou, J. Rajahalme, J. Gross, A. Wang, J. Stringer, P. Shelar et al., “The design and implementation of open vswitch,” in NSDI, vol. 15, 2015, pp. 117–130.Search in Google Scholar
26. S. Jeschke, C. Brecher, T. Meisen, D. Özdemir and T. Eschert, “Industrial internet of things and cyber manufacturing systems,” in Industrial Internet of Things, Springer, 2017, pp. 3–19.Search in Google Scholar
27. P. Adolphs, “Rami 4.0 – an architectural model for industrie 4.0,” Platform Industrie, vol. 4, 2015.Search in Google Scholar
28. M. Hankel, “Industrie 4.0: The reference architectural model industrie 4.0 (rami 4.0),” Frankfurt am Main, Germany: ZVEI-German Electrical and Electronic Manufacturers’ Association, 2015.Search in Google Scholar
29. S.-W. Lin, B. Miller, J. Durand, R. Joshi, P. Didier, A. Chigani, R. Torenbeek, D. Duggal, R. Martin, G. Bleakley et al., “Industrial internet reference architecture,” Industrial Internet Consortium (IIC), Tech. Rep, 2015.Search in Google Scholar
30. Systems and Software Engineering, Architecture Description, ISO/IEC/IEEE International Standard ISO/IEC/IEEE 42 010:2011.Search in Google Scholar
© 2019 Karrenbauer et al., published by De Gruyter
This work is licensed under the Creative Commons Attribution 4.0 Public License.