The Future of Industrial Communication Automation Networks in the Era of the Internet of Things and Industry 4.0

©istockphoto.com/Besjunior

MARTIN WOLLSCHLAEGER, THILO SAUTER, and JÜRGEN JASPERNEITE

ith the introduc- in technology that allow interconnec- networks in automation. Moreover, we tion of the Internet tion on a wider and more fine-grained will point out the need for harmoniza- of Things (IoT) and scale. The purpose of this article is to tion beyond networking. cyberphysical system review technological trends and the (CPS) concepts in indus- impact they may have on industrial Recent Trends in trial application scenar- communication. We will review the im- Automation Technology ios,W industrial automation is undergo- pact of IoT and CPSs on industrial au- The core of distributed automation sys ing a tremendous change. This is made tomation from an industry 4.0 perspec- tems is essentially the reliable exchange possible in part by recent advances tive, give a survey of the current state of information. Any attempt to steer pro of work on Ethernet time-sensitive net- cesses independently of continuous hu-

Digital Object Identifier 10.1109/MIE.2017.2649104 working (TSN), and shed light on the man interaction requires, in a very wide Date of publication: 21 March 2017 role of fifth-generation (5G) telecom sense, the flow of information between

1932-4529/17©2017IEEE march 2017 ■ IEEE industrial electronics magazine 17 also included the physical layers, To facilitate information exchange, a multitude such as in bus systems like controller area network (CAN), PROFIBUS, of industrial communication networks evolved or INTERBUS, to name a few. Figure 1 reviews the timeline of this evolution over the years. and marks milestones in various technology fields relevant for the evolution of communication in automation. some kind of sensors, controllers, and in Germany, the term has quickly be- Things changed around the millenni- actuators [1]. After the introduction come a buzzword on a global scale um, when Internet technology became of steam power to relieve workers from [10]. As a kind of response with similar popular, and IT became widely used hard manual labor and the invention of goals, the Industrial Internet initiative and a part of everyday life. In automa- mass production based on division of (IIC) originated in the United States, al- tion, this stimulated a new wave of Eth- labor, the introduction of automation though it should be noted that the term ernet-based networks that borrowed technology was what is today often was coined much earlier [11]. basic technology from the IT world. The called the third industrial revolution [2]. From a communications perspective, lack of genuine real-time capabilities To facilitate information exchange, a IoT and CPSs rely largely on mobile Inter- in standard Ethernet, however, pre- multitude of industrial communica- net, i.e., telecommunication networks, vented the development of one single tion networks evolved over the years, which have not played a major role in Ethernet solution for automation pur- starting from the 1980s. It is noteworthy industrial communication so far. In addi- poses and led, again, to the emergence that these developments, in many cas- tion, they require solely Internet-based of dedicated solutions [13], [14]. Ap- es, picked up and accommodated new communication, which has not been proaches interfering with the Ethernet technologies emerging in other fields, possible in industrial automation, either. standard, such as PROFINET-isochro- primarily in the information and com- Both information technology (IT) and nous real time (IRT) or EtherCAT, were munication technology (ICT) world. telecom networks could not cope with much scarcer and typically appeared Ethernet, wireless networks, or web the automation-specific needs for de- later to meet particularly demanding technologies are examples of this cross- terministic, reliable, and efficient com- low-latency requirements, especially fertilization. These new technologies munication. This seems to be changing in motion control applications. Never- created new opportunities for making now. On the one hand, ongoing work on theless, real-time Ethernet (RTE) is still information exchange more compre- Ethernet TSN promises hard real-time a vivid research area [15]–[17]. hensive. Consequently, automation sys- capabilities. This is seen as a real game The third evolution phase dealt tems could grow more complex, too. changer for real-time automation net- with wireless networks. The foremost The latest trends influencing auto- working. On the other hand, the telecom advantage of using wireless connec- mation technology are the IoT, CPS, industry has discovered industrial auto- tions in industrial automation is that and the emerging tactile Internet. For mation as a promising application field devices and machines can be moved the latter, [3] mentions industrial au- of their products and seems determined and connected with greater ease and tomation as a key, steadily growing to consider the needs of automation in there are no restricting cables. More application field. These concepts are the development of 5G networks. Both than ever, basic ICTs were adopted, not entirely new and emerged in a con- developments, together with unified and mainly from the IEEE 802 protocol fam- text of ICT several years ago. Recently, semantic information modeling based ily. All practically relevant wireless however, they are penetrating industri- on web standards, might indeed change networks in use today build on stan- al automation and changing the angles the structure of industrial networks, and dards devised for computer networks, from which people look at automation they might be the prerequisite for actu- such as IEEE 802.11 [18], or wireless systems [4]–[6]. Moreover, they sup- ally implementing industrial IoT (IIoT) personal area networks (WPANs) like port recent trends, such as achieving a and CPSs. IEEE 802.15.1 or IEEE 802.15.4 [19]. The higher degree of interconnection, cog- upper layers are, in many cases, con- nitive automation, and shifting infor- A Brief History of Industrial sistent with wired networks to retain mation collection and processing into Communication compatibility, and the main challenge cloud-based applications [7]–[9]. In the early days of industrial com- is again to ensure real-time and reli- Applying the ideas of CPSs and IoT munications, dedicated automation ability capabilities [20], [21]. Typically, to the industrial automation domain networks called fieldbus systems were wireless networks are being used for led to the definition of the Industry 4.0 developed from scratch to overcome subsystems in otherwise wired net- concept, where 4.0 alludes to a fourth the limitations caused by the parallel work infrastructures [22], [23]. Indus- industrial revolution enabled by In- cabling between sensors, actuators, trial automation is a rather conserva- ternet technologies to create smart and controllers and to close the com- tive domain, and the higher reliability products, a smart production, and munication gap on the lower levels of wired networks often outweighs the smart services. Originally developed of the automation pyramid [12]. This flexibility of wireless links. Wireless

18 IEEE industrial electronics magazine ■ march 2017 sensor networks in their pure form, though a vibrant research field [24], [25], are therefore not widely used in

automation practice. esearch Such was the situation until about

one or two years ago. Industrial com- 5G dvanced R munications was a mixture of fieldbus 2020 systems, Ethernet-based approaches, NET: A NET: TSN T: highwayT: addressable and some wireless solutions [26], [27], A P all of them struggling with the legacy net TE+ Inter of four decades of history in a market y 4.0 L

with life cycles of plants that are in the Ind. S: manufacturingS: messaging specification; 4G:

range of decades. The recent adoption Industr E

of IoT and CPS concepts in the auto- N: wireless local area network; WWW: World AN A LT mation world, however, changes the T 2010 scenery again. They put the old and 4G: 6loWP ISA 100.11a UWB HAR still valid quest for integration of infor- A CPS

mation flows in automation into a wid- Wirel. IEC61784-2 HSP

er context [28]. The idea that every- OPC UA ZigBee T 3G: TS: universal mobile telecommunications system; UWB: ultrawide thing in automation is connected and, AP UMTS M y e.g., that individual products or work- SO 3G: EtherCA link us/TC P xRa pieces are parts of this ecosystem is IEC61784 OFINET wer Fle not new, it was introduced with agent- net/IP PR Bluetooth Po Modb 2000 IoT based distributed manufacturing sys- LIN P: manufacturing automation protocol; MM EN50325 IEC61158 WLAN tems years ago [29]. Nevertheless, Ether GSM recent advances in communication EN50254 TTP technology allow interconnection on a 2G: FF EN50170 ControlNet wider and more fine-grained scale [30]. OFIBUS: process field bus; FIP: factory instrumentation protocol; HAR R viceNe t On the application side of the automa- SDS De WWW

tion pyramid, the other big trend is to ISA SP50 ASi move the business logic into cloud- LON . EIB based applications [7], [8]. This is in 1990 FIBUS O line with IT trends and has much to do Sercos Comp PR AY with new business models of software FIP Ubiq. OW solution providers on the one hand INTERBUS CA N MMS HA RT PR

and the wish to make IT-related costs N: controller area network; P T: EthernetT: for control automation technology. us

smaller and more predictable on the P: transport control protocol; TTP: trime triggered protocol; U CA C net Bitb customers’ side. P-NET MAP The big difference with respect Inter us

to the previous waves of evolution 1980 in industrial communication is that Modb ISO/OSI the technological driving force is consumer electronics. So far, the pre- dominant roots of industrial commu-

nication were instrumentation and IT. net

This seems to change. The work on OFINET: process field net; Ether R Ether Ethernet TSN originated in the stan- P: simple object access protocol; T A dardization of audio video bridging NE T 1970 : global system for mobile communication; ISO: International Organization for Standardization; LTE: long-term evolution; WL

(AVB) [31]. In addition, the interest M : high-speed packet access; LIN: local interconnect network; LON: local operating network; MA of the telecom industry stems from AR PA A

extending their business as mobile SP ks Internet providers, which draws from ia l net ends

developments in consumer electron- Tr Si: actuator/sensor interface; EIB: European installation bus; CA Industr Communication Computer Networ ics [3]. After all, one of the appeal- Mobil e Inter IT gency Network; GS S: smartS: distributed system; P Y: processY: data highway; SO ing features of the IoT concept is the D A The milestones in the evolution of industrial communication and related technology fields. 2G: second generation; 3G: third generation; 4G: fourthE 1 — The milestones evolution AR 4G: industrial the of in generation; generation; communication third related technology and 3G: second fields. generation; 2G: promise to use everyday Internet- R OW R P band; S Wide A Web; remote transducer; H Projects A enabled devices like smartphones or FIGU

march 2017 ■ IEEE industrial electronics magazine 19 of effective runtime solutions with engi- One of the appealing features of the IoT concept is neering approaches and network management is required. the promise to use everyday Internet-enabled devices The Future of Industrial Ethernet as end points for accessing industrial data. Today, RTE has become a standard in the industrial automation domain. Un- tablets as end points for accessing in- a private interenterprise cloud with lim- fortunately, there is currently no single dustrial data [32]. ited access to organize the flow of the standard but many different mutually Figure 2 shows the complexity of product and the information related to incompatible implementations. The ex- communication in industrial automa- the product and its production process- isting RTE solutions are based on Fast tion systems. There are application es. Finally, other enterprises or custom- Ethernet and can be divided into three domains with different requirements, ers may access selected information us- classes, which differ in the achieved re- e.g., regarding real time, mobility, safe- ing the Internet or a public cloud. al-time performance and the necessary ty and security, explosion protection, Overall, there is a tremendous extensions of the IEEE 802 standards, availability, and so forth. Typically, up change in specific technologies used as shown in Figure 3 [33]. In class A, the to a supervisory control and data ac- for networking in an industrial context. real-time services are realized above quisition (SCADA) system, industrial However, what has not changed over the transport layer with cycle times communication solutions are heteroge- time is that application relations exist in the range of 100 ms. Modbus-Inter- neous but are optimized to fulfill these bundling different categories of com- face for Distributed Automation (IDA), requirements. They include fieldbuses, munication relations with specific sets Ethernet/industrial protocol (IP), and industrial Ethernet, and industrial wire- of requirements, like hard time bound- Foundation fieldbus (FF) high-speed less networks. Today, with the adoption aries, isochronous communication, low Ethernet are implementations that be- of ICTs, local manufacturing clouds are jitter, high availability, and, of course, long to this class. These were the earli- increasingly used. Devices may be con- low cost. From an end user’s point of est implementations of industrial Ether- nected directly to this cloud. Manufac- view, the underlying network tech- net. They build on the entire transport turing execution systems (MESs) and nologies are of less interest, as long as control protocol (TCP)/IP suite and enterprise resource planning (ERP) they meet these requirements. This is use best-effort bridging. In class B, the solutions are also being integrated especially challenging when new ap- real-time services are realized directly into the cloud, as is enterprise ICT. The plication relations and application flex- on top of the media access control partners along the value chain may use ibility are introduced. A combination (MAC) layer by using approaches like

Value Chain

Enterprise Public Cloud Enterprise Private Interenterprise Cloud

Enterprise Internet Enterprise IT Public and Private ERP Value Networks Chain Manufacturing Cloud Enterprise MES

Mobile Devices SCADA Mobile Devices Customer

Industrial Customer Intrinsic IT Industrial Wireless Safety Customer Real Time Inbound Mainstream Outbound Logistics Processes Logistics

FIGURE 2 — The complexity of communication in industrial automation systems.

20 IEEE industrial electronics magazine ■ march 2017 Class AClass B Class C

IEEE 802.1 TSN Add-Ons to IEEE 802-Standards

Real-Time Performance

Real-Time Real-Time Client/Server Real-Time Client/Server Client/Server Streams with Streams with Applications Streams Applications Applications Priorities Scheduling

vice TCP/IP TCP/IP TCP/IP inal De

rm IP IP IP Te

Ethernet Ethernet Ethernet with Priorities with Scheduling

Best-Effort Bridging with (Best-Effort) Bridging Best-Effort Bridging mediate Priorities with Scheduling System Inter

FIGURE 3 — The classification scheme for RTE. prioritization and virtual local area network (VLAN) tagging to separate real- There is a tremendous change time data from best-effort traffic. The achievable cycle time using Fast Ether- in specific technologies used for networking net is in the range of 10 ms. An exam- ple protocol in this class is PROFINET in an industrial context. Real Time. These RTE approaches still use standard Ethernet and best-effort proceeded in different steps. The IEEE Furthermore, the maximum number of bridging with priority support. Class C 802.3 Residential Ethernet Study Group seven hops in a line topology is a big is the most powerful class, where the was formed in 2004 to explore the need limitation for a lot of industrial automa- real-time communication capabilities for an Ethernet specification for resi- tion applications. With the completion are achieved by modifications of the dential applications, with major con- of the AVB standard in 2012, it became Ethernet MAC layer, like strict commu- tributions from companies like Broad- clear that streaming data can also be nication scheduling and high-precision com, Nortel, Pioneer, Samsung, NEC, control data, like that used in automo- clock synchronization according to and Gibson Brands. This activity was tive and industrial applications [35]. IEEE 1588 in the terminal devices as merged into the IEEE 802.1 AVB Task Therefore, the task group changed its well as in the bridges [34]. The achiev- Group in 2005, where key players like name to TSN to better reflect the new, able cycle time lies below 1 ms. Repre- Intel, Broadcom, Marvell, and Samsung enlarged scope of its standards due to sentatives for this group are PROFINET are involved. The most distinctive fea- the expanded target applications. IRT, EtherCAT, time-triggered Ethernet, ture of AVB is the ability to guarantee The IEEE TSN Working Group cur- Sercos III, and Ethernet Powerlink. Here, upper time bounds to all priorities of all rently aims to improve the reliability compatibility with the classical Ether- data streams, which is an improvement and real-time capabilities of standard net standard was finally abandoned to in comparison with standard Ethernet Ethernet (IEEE 802.3, IEEE 802.1D). In achieve higher performance. [31]. Because of the usage of a nonpre- particular, it addresses five essential In the meantime, the development emptive scheduler, however, the worst- shortcomings of the AVB standard of standard Ethernet in the direction case latency is not better than best- that are nevertheless crucial require- of a real-time communication system effort bridging according to IEEE 802.1. ments for industrial automation:

march 2017 ■ IEEE industrial electronics magazine 21 Architecture (UA)], automotive commu- Although the work on TSN has not yet been nications (BroadR-Reach) or entertain- ment and infotainment systems already completed, the potential for industrial explore the capabilities of TSN. automation applications is appealing. The Role of 5G Networks in Industrial Automation ■■ reduced latencies and accurate real-time support will be included di- Digital transformation is the core of determinism rectly in the Ethernet standard. Most the fourth industrial revolution, and 5G ■■ independence from physical trans- of the standards shown in Table 1 are network infrastructures will be key sup- mission rates focused on enhancements of the bridg- porting assets. In the next decade, the ■■ fault tolerance without additional ing functionality. manufacturing industry is expected to hardware While standardization is still in prog- evolve toward a distributed organization ■■ support for higher security and ress, several manufacturers are already of production, with connected goods safety able to show preliminary implementa- (products with communication ability), ■■ interoperability of solutions from tions of the new functionalities enabled low-energy processes, collaborative different manufacturers. by TSN. Against the backdrop of indus- robots, and integrated manufacturing Table 1 shows the progress of the trial automation, [36] demonstrated and logistics. These concepts are no- standardization of the seven current that these prototypic TSN functions tably embodied under the Industry 4.0 TSN-related standards. In total, 60 IEEE ascertain a much higher determinism paradigm and led to several application standards are correlated under TSN, than comparable state-of-the-art com- scenarios defined by a working group including the 13 security-associated ponents. However, the benefits of TSN of the German Plattform Industrie 4.0 standards resulting in several different come with several challenges, like high- [39]. One driving application scenario enhancements for standard Ethernet er configuration efforts, which could be is to form a network of geographically on layer 2 of the open systems intercon- solved, e.g., by autoconfiguration mech- distributed factories with flexible ad- nection (OSI) model. Inspired by the al- anisms [37] or with the help of software- aptation of production capabilities and ready known time slot procedure from defined networking (SDN) [38]. sharing of resources and assets to im- PROFINET IRT, the time-aware shaper Although the work on TSN has not prove order fulfillment. Among other (IEEE 802.1Qbv) prioritizes different yet been completed, the potential for things, a reliable wide-area communi- transfer queues in switches. By doing industrial automation applications is cation is needed for this use case. As a so, a guaranteed data transfer rate and appealing. Contrary to previous devel- result of these transformations, vertical latency can be accomplished even in opments toward RTE requirements and industries will have enhanced technical high-load traffic situations. From a net- best practices for low latency from the capacity available to trigger the devel- work structure viewpoint, the external automation domain are taken into ac- opment of new products and services. or built-in Ethernet switch becomes count in the standardization of Ether- A vertical in this context represents a the most important element, as it must net itself, rather than putting them on system of end-user entities belonging to implement all the novel traffic manage- top of the existing standard. This can a certain industry. They reside on top of ment strategies. With respect to the be expected to make a huge difference the networked structure, using end-to- classification scheme for RTEs shown in the technology acceptance. That is end communication services provided in Figure 3, TSN belongs to class C. The why user communities for industrial by the 5G network. The network offers a difference to existing class C RTE solu- automation networks (e.g., PROFINET), horizontal communication within a ver- tions, however, is that the necessary middleware solutions [like OPC Unified tical structure and across them.

TABLE 1 — THE IEEE STANDARDIZATION PROGRESS OF THE OPEN TSN STANDARDS (SEPTEMBER 2016). STANDARD NAME PROGRESS IEEE 802.1AS-Rev Timing and Synchronization for Time-Sensitive Applications in Bridged Local Area Networks Working group IEEE 802.1Qbv Enhancements for Scheduled Traffic Sponsor ballot IEEE 802.1Qcc Stream Reservation Protocol (SRP) Enhancements and Performance Improvements Task group IEEE 802.1Qbu Frame Preemption Sponsor ballot IEEE 802.1CB Frame Replication and Elimination for Reliability Sponsor ballot IEEE 802.1Qch Cyclic Queuing and Forwarding PAR* approval IEEE 802.1Qci Per-Stream Filtering and Policing PAR approval *PAR: Project Authorization Request.

22 IEEE industrial electronics magazine ■ march 2017 In Europe, the next generation of networks is seen as a combined sys- In the next decade, the manufacturing industry tem covering both wired and wireless communication solutions, from private is expected to evolve toward a distributed and public communication providers, offering virtualized and physical com- organization of production. munication functions. The heterogeneity of such an approach is obvious, infrastructure. This mapping defines capacity of current wireless technolo- as providing a set of suitable applica- the configuration of the highly manage- gies through self-organizing 5G tech- tion services to the end users will be a able infrastructure components. It will nology and prepare for future sce- core requirement. It is also important be deployed and changed on demand at narios where up to 100 sensors can to consider end users’ requirements. runtime, always promising a resource- be operated per cubic meter without In this context, end users are also re- efficient network configuration. compromising the availability of ro- garded as verticals. It should be noted that 5G is more bots or connected machines. For the vertical factories of the fu- than mobile Internet and, therefore, Adoption of Internet technology ture, such requirements have been dis- more than the simple extension of to- will enable easier integration of work- cussed and put down in a white paper day’s telecom networks. 5G needs to flows through standardized interfac- [40]. An analysis of the corresponding integrate different enabling technolo- es. Each of the workflows may have requirements shows that latency (be- gies (e.g., mobile, wired, satellite, and different requirements with respect to low 5 ms), reliability, and density (up to optical), spectrum regulatory frame- bandwidth, latency, and availability, 100 devices/m2), along with tight con- works (e.g., licensed and unlicensed), and, as a result, the cost for network- straints on territory and/or population and enabling capabilities (e.g., IoT, ing should be linked to the needs. In coverage, are the most important per- CPSs). To cope with the increasing di- addition, end-to-end communication formance targets 5G needs to achieve versity of wireless IoT systems in man- may require the integration of pub- for supporting all possible services of ufacturing, there is the need for novel lic cellular networking technologies the five investigated sectors. Moreover, capabilities to ensure the same level (such as 5G) with private networks with universal availability of instanta- of reliability as offered in wired to- (such as picocells or meshed net- neous communications, high level of pologies. Given the nondeterministic working topologies). The concepts of guaranteed quality of service (QoS), behavior of the wireless medium, new network virtualization, SDN, and dis- and cost levels appropriate to meet cus- challenges arise to manage the spec- tributed cloud resource management tomers’ expectations, 5G will pave the trum, in particular in environments can be leveraged to give the factory way for new business opportunities. where the number of wireless applica- operator a unified view on the net- Furthermore, the requirements from tions and devices are increasing. Com- work. There is the opportunity for the the different verticals have been inte- pared to other industries, the wireless 5G community to extend the manage- grated into an overall vision of 5G [41]. industrial Internet has one of the most ment capabilities beyond the network- A common structure has been devel- stringent requirements in terms of la- ing aspects and include networked oped consisting of different layers with tencies and reliability, in particular for services for security, data analytics, specific levels of abstraction (Figure 4). use in time-critical closed-loop com- and cloud/edge computing. Intelligent orchestration platforms will munication scenarios. Open questions emerge from 5G networks and 5G archi- still exist to manage Harmonization Above tecture is expected to accommodate a ■■ coexistence of different wireless the Networks wide range of use cases with advanced protocols and systems The previous sections showed that requirements, especially in terms ■■ coexistence of different wired communication infrastructures in au- of latency, resilience, coverage, and protocols tomation systems are complex and are bandwidth. Another major challenge is ■■ interoperability between communi- getting even more complex and even to provide end-to-end network and cloud cation systems more heterogeneous. Coming back to infrastructure in the form of slices over ■■ seamless engineering, adaptive dur- the starting point of this article, i.e., the same physical infrastructure to fulfill ing operation (based on previously the transfer of information between vertical-specific requirements as well collected real-life data). different entities in an industrial auto- as mobile broadband services in paral- For enabling the coexistence of mation system, it is evident that com- lel. Such a slice can be seen as a logical wireless technologies, new protocols munication networks alone will not network structure with components are needed to manage the coopera- be sufficient. The actual transport of providing application functions and ap- tion of technologies working in the information is only one aspect. Equal- plication relations among them. They same frequency band or to spread the ly important are the description and contain QoS requirements for the single usage over multiple frequency bands modeling of information as well as defi- relations. A slice will be dynamically in a coordinated and adaptive way. nitions how to access it. These aspects mapped to a (flexibly changing) network The main objective is to increase the are not fully covered in communication

march 2017 ■ IEEE industrial electronics magazine 23 Infrastructure Layer Multiservice Control Layer Business Function Layer Network Function Layer Business Service Layer

Northbound Southbound Interfaces Interfaces 3 Service Network Function Repository Core Cloud 2 Service Cloud Optical Core Networ k (e.g., Automotive) Vertical 2 Services 3 1 Optical Slice Service Vertical Function Repository Satellite Edge Cloud Optical Metro Network 3 Service Wireless N: radio access network. 2 . RA Vertica l Service [41] Vertical 1 Services Optical Access Network (e.g., Factory of the Future) 1 Service Vertical 2 Function Repository 3 Service 2 Service RAN and Wireless Backhaul 1 Mobile Broadband Services Service

Vertical 1 Satellite Network Function Repository E 4 — The integrated architecture 5G vertical and broadband mobile for services

Vertical-Centric Services Network-Centric Services R FIGU

24 IEEE industrial electronics magazine ■ march 2017 standards, let alone treated in a harmonized way. They must be tackled above the actual networks. Business Functions Years ago, a generic communication model was established consisting of the three layers of networks, middleware, and application [42]. This Service Service general approach is still valid and … even more than before (Figure 5). The MES::Calculate Application Functions and Energy Consumption uppermost layer is built by the appli- Sensor::Calibrate Information Models cation functions that need to be inter- … connected, the middle layer covers Services *** Services the (partly application-agnostic) com- Subscribe … Connect munication services and the middle- Browse Middleware and Communication Services, ware management services, and the Generic Information Models lowest layer contains the transport- Read Write … Create oriented protocols guaranteeing the Services required QoS. Native APIs Application functions and informa- Transport-Oriented Protocols tion models are the building blocks AN for actual business functions. In the T 5G IR IIoT TSN CAN OFINET OFIBUS terminology of service-oriented ar- WLAN 6loWP EtherC AT PR chitectures, the higher-level business PR *** *** *** services are orchestrations of application functions. The application func- FIGURE 5 — The three levels of industrial communication: application, communication services tions are more and more exposed via and middleware, and transport protocols. API: application programming interface. services. Pushed by the Industry 4.0 idea, these services and the related information models are under definition for different application domains Intelligent orchestration platforms will emerge and contexts. from 5G networks. Because the application functions should be applicable to different resources, they cannot rely on specific systems on the other hand. Depend- Moreover, such gateways could be more communication functions directly. ing on these demands, varying along than just protocol and data conversion Generic communication services are the functional hierarchy, the existence units, they will have to act as smart en- required. They will be provided by the of one single system fulfilling all the tities controlling and representing the middle layer. This layer benefits from requirements is doubtful. In addition, underlying automation (sub)system. Internet technologies such as web ser- the resource capabilities of the con- Last but not least, they can also secure vices or recent IoT-specific protocols. nected components need to be consid- access to parts of the system [32]. In addition, IT technologies adopted ered [45], [46]. The heterogeneity will rather infor automation purposes, such as OPC As described in the “The Role of 5G crease. This calls for harmonized ser- UA, will be used here [43], [44]. Which Networks in Industrial Automation” vice layers and for network function protocols will be used depends on the section, new technologies like 5G com- virtualization (NFV) at these lower lay- level of functional hierarchy according munication will penetrate this level as ers. But also for these legacy systems, to IEC 62264-1, the resource capabili- well. Will Ethernet and 5G mobile com- services and data models on the higher ties, the QoS requested from the appli- munication replace all existing indus- levels should be consistent with the new cation, and so forth. trial communication systems? Likely networks, so that there is some form of The transport-oriented technologies they will not. There will still be wired uniform data exchange at the higher like fieldbuses, industrial Ethernet, and or wireless legacy systems, which will networking layers, irrespective of the industrial wireless approaches provide be in operation for years and maybe underlying communication systems. a communication system guaranteeing decades to come. To interconnect all IT in automation finally seems to be the application demands regarding reli- of these, proxies and gateways will be mature enough to meet this challenge. ability, availability, real-time behavior, needed [47], even though one of the and so forth on one hand, and the flex- aims of especially the development of Conclusions ibility and (self-)adaptability of future Ethernet-based automation networks Industrial communication systems un- industrial automation and production was to eliminate the need for gateways. derwent a long evolution with many

march 2017 ■ IEEE industrial electronics magazine 25 in 1991 and a Habilitation degree in 2001 Network harmonization on a logical level for research in automation and control system. He is a full professor at Tech- is one of the challenges to face. nische Universität (TU) Dresden, Ger- many, for industrial communications and director of the Institute of Applied influences from technologies from out- to attach less critical data points. On Computer Science, Faculty of Computer side the actual automation domain. The top of mobile networks, however, the Science at TU Dresden. His research fundamental requirements always re- same higher-level protocols could be topics are industrial communication mained unchanged: to exchange infor- placed, as in the case of Ethernet. systems and automation networks, in- mation about industrial processes in a What does this development mean formation modeling, middleware con- timely, reliable, and possibly uniform for research and education? This, of cepts, management of heterogeneous way. The absence of an optimum tech- course, depends largely on the view- networks, life-cycle management, and nology to meet these goals inspired point. From an application point of semantic descriptions. He is actively engineers and stimulated a multitude view, industrial communication needs involved in standardization activities at of diverse and incompatible solutions, to fulfill the requirements, nothing the German Commission for Electrical, and the calls for unified approaches else. The specific technology is mostly Electronic & Information Technologies were heard but not heeded. Even when out of scope of end users. They will rely of DIN and VDE and with the Interna- the technology basis shifted toward on service providers guaranteeing QoS tional Electrotechnical Commission ICT standards, the variety of solutions for the intended application, regard- Technical Committee 65. remained and even got worse. Is there less if this is provided by networks Thilo Sauter (thilo.sauter@tuwien a hope that things will finally change they own themselves or by public or .ac.at) received a Ph.D. degree in elec- back to the desirable? Perhaps there is. private networks. Network harmoni- trical engineering from TU Wien in For the first time, the requirements zation on a logical level, by defining 1999. He is a tenured associate pro- of industrial automation applications generic communication services and fessor of automation technology at are taken into account in the develop- adequate information models, is one of Technische Universität Wien, Vienna, ment of a new ICT standard. Ethernet the challenges to face. Austria, and was the founding direc- TSN has the potential to satisfy even From a communication provider tor of the Center for Integrated Sensor demanding requirements without the view, however, there might be still the Systems at Danube University Krems, need of dedicated add-ons tailored to need to further optimize or even de- Wiener Neustadt, Austria. His profes- the demands of automation. On top of velop specific industrial technologies, sional expertise includes integrated such a novel network that can equally especially when harsh application re- circuit design, smart sensors, and accommodate both real-time and best- quirements need to be met. The adop- automation networks with a focus on effort traffic, existing high-level auto- tion of IoT technologies and concepts real-time, security, interconnection, mation protocols could be placed for in automation will grow substantially. and integration issues. He is an IEEE backward compatibility or a middle- These technologies need to be evalu- Fellow and an Administrative Com- ware such as OPC UA that serves the ated and need to be further tailored to mittee member of the IEEE Industrial needs of automation as far as function- industrial automation needs. For 5G, Electronics Society and the IEEE Sen- ality and data models are concerned similar tasks can be seen. The inte- sors Council. He has been working in but builds on established and wide- gration of end users representing the fieldbus standardization with Interna- spread Internet technology for the com- verticals is promising. One of the main tional Electrotechnical Commission munication services. challenges of future industrial com- Technical Committee 65 for more The massive interest of telecom munication will be the management than 20 years and had leading posi- industries in industrial applications is of complexity and heterogeneity. NFV tions in several national and interna- without precedence and a direct con- and the use of SDN could be enabling tional research projects concerned sequence of the adoption of IoT and technologies to provide a flexible with industrial communication and CPS scenarios. Contrary to the devel- network topology and its monitoring enterprise integration. opment of Ethernet TSN, the possible and management to meet the require- Jürgen Jasperneite (juergen application of 5G networks in automa- ments of the end users along the life .[email protected]) received a tion is not an expression of a steady cycle of an enterprise. Dr.-Ing. degree in electrical engineer- evolution but indeed rather disrup- ing and information technology from tive. Nevertheless, it is unlikely that Biographies the Otto-von-Guericke-University of 5G will be able to satisfy all stringent Martin Wollschlaeger (martin Magdeburg, Germany, in 2002. He is automation demands for real time and [email protected]) stud- a full professor of computer networks completely replace dedicated indus- ied electrical engineering at Otto-von- at the Ostwestfalen-Lippe (OWL) trial automation networks. Rather, it Guericke University of Magdeburg, Ger- University of Applied Sciences and might work as a kind of backbone or many, where he received a Ph.D. degree the founding director of the University

26 IEEE industrial electronics magazine ■ march 2017 Institute for Industrial Information tion via Ethernet in industrial automation en- [31] J. Imtiaz, J. Jasperneite, and L. Han, “A perfor- Technologies as well as the Fraunhofer vironments,” in Proc. 2014 IEEE Emerging Tech- mance study of Ethernet audio video bridging nology and Factory Automation (ETFA), pp. 1–8. (AVB) for industrial real-time communica- Anwendungszentrum Industrial Auto- [14] J. Jasperneite, J. Imtiaz, M. Schumacher, and K. tion,” in Proc. 14th IEEE Int. Conf. Emerging mation in Lemgo, Germany. He is one Weber, “A proposal for a generic real-rime Eth- Technologies and Factory Automation (ETFA), ernet system,” IEEE Trans. Ind. Informat., vol. 5, 2009, pp. 1–8. of the main initiators of the Centrum no. 2, pp. 75–85, May 2009. [32] M. W. Condry and C. B. Nelson, “Using smart Industrial IT, which is Germany’s first [15] F. Tramarin and S. Vitturi, “Strategies and ser- edge IoT devices for safer, rapid response with vices for energy efficiency in real-time Ether- industry IoT control operations,” Proc. IEEE, science-to-business center in the field net networks,” IEEE Trans. Ind. Informat., vol. vol. 104, no. 5, pp. 938–946, May 2016. of industrial automation. He initiated 11, no. 3, pp. 841–852, June 2015. [33] M. Schumacher, J. Jasperneite, and K. We- [16] S. Fuchs, H. P. Schmidt, and S. Witte, “Test and ber, “A new approach for increasing the per- the SmartFactoryOWL, which is a re- on-line monitoring of real-time Ethernet with formance of the industrial Ethernet system search and demonstration factory for mixed physical layer for industry 4.0,” in Proc. Profinet,” in Proc. IEEE Int. Workshop Factory 21st IEEE Int. Conf. Emerging Technologies and Communication Systems (WFCS), May 2008, pp. ICT-based automation technologies Factory Automation (ETFA), 2016, pp. 1–4. 159–167. operated by Fraunhofer and the OWL [17] R. Schlesinger, A. Springer, and T. Sauter, “Au- [34] G. Cena, I. Cibrario Bertolotti, S. Scanzio, A. tomatic packing mechanism for simplification Valenzano, and C. Zunino, “Synchronize your University. His current research inter- of the scheduling in Profinet IRT,” IEEE Trans. watches, part II: Special-purpose solutions for ests include distributed real-time sys- Ind. Informat., vol. 12, no. 5, pp. 1822–1831, Oct. distributed real-time control,” IEEE Ind. Elec- 2016. tron. Mag., vol. 7, no. 2, pp. 27–39, June 2013. tems, especially in the domain of intel- [18] F. Tramarin, S. Vitturi, M. Luvisotto, and A. [35] L. Lo Bello, “Novel trends in automotive net- ligent automation. Zanella, “On the use of IEEE 802.11n for indus- works: A perspective on Ethernet and the IEEE trial communications,” IEEE Trans. Ind. Infor- Audio Video Bridging,” in Proc. 19th IEEE Int. mat., vol. 12, no. 5, pp. 1877–1886, Oct. 2016. Conf. Emerging Technologies and Factory Auto- [19] E. Toscano and L. Lo Bello, “Comparative as- mation (ETFA), 2014, pp. 1–8. References sessments of IEEE 802.15. 4/ZigBee and 6LoW- [36] K. Kellermeier, L. Wisniewski, A. Biendarra, [1] T. Sauter, S. Soucek, W. Kastner, and D. Diet- PAN for low-power industrial WSNs in realistic C. Pieper, and H. Flatt, “Performance evalu- rich, “The evolution of factory and building au- scenarios,” in Proc. Ninth IEEE Int. Workshop ation of PROFINET RT devices in TSN based tomation,” IEEE Ind. Electron. Mag., vol. 5, no. 3, Factory Communication Systems (WFCS), May backplanes” (in German), in Proc. Kommuni- pp. 35–48, Mar. 2011. 2012, pp. 115–124. kation in der Automation (KommA), Nov. 2016, [2] M. Guarnieri, “The roots of automation before [20] S. Vitturi, F. Tramarin, and L. Seno, “Industrial pp. 151–158. mechatronics,” IEEE Ind. Electron. Mag., vol. 4, wireless networks: The significance of timeli- [37] L. Dürkop, J. Jasperneite, and A. Fay, “An anal- no. 2, pp. 42–43, June 2010. ness in communication systems,” IEEE Ind. ysis of real-time Ethernets with regard to their [3] G. Fettweis, H. Boche, T. Wiegand, E. Zielinski, Electron. Mag., vol. 7, no. 2, pp. 40–51, June automatic configuration,” in Proc. IEEE World H. Schotten, P. Merz, S. Hirche, A. Festag, W. 2013. Conf. Factory Communication Systems (WFCS), Häffner, M. Meyer, E. Steinbach, R. Kraemer, [21] S. Girs, A. Willig, E. Uhlemann, and M. Björk- 2015, pp. 1–8. R. Steinmetz, F. Hofmann, P. Eisert, R. Scholl, man, “Scheduling for source relaying with [38] D. Schulz, “Network models for the industrial R. Ellinger, E. Weiss, and I. Riedel. (2014, Aug.). packet aggregation in industrial wireless net- intranet,” in Proc. 21st IEEE Int. Conf. Emerging The tactile Internet. International Telecommu- works,” IEEE Trans. Ind. Informat., vol. 12, no. 5, Technologies and Factory Automation (ETFA), nication Union. Geneva, Switzerland. [Online]. Oct. 2016. 2016, pp. 1–10. Available: http://www.itu.int/dms_pub/itu-t/ [22] G. Cena, A. Valenzano, and S. Vitturi, “Hybrid [39] Plattform Industrie 4.0. (2016). Aspects of the oth/23/01/T23010000230001PDFE.pdf wired/wireless networks for real-time commu- research roadmap in application scenarios. [4] A. W. Colombo, S. Karnouskos, Y. Shi, S. Yin, nications,” IEEE Ind. Electron. Mag., vol. 2, no. Federal Ministry for Economic Affairs and and O. Kaynak, “Industrial cyber–physical sys- 1, pp. 8–20, Mar. 2008. Energy. Berlin, Germany. [Online]. Available: tems [scanning the issue],” Proc. IEEE, vol. 104, [23] T. Sauter, J. Jasperneite, and L. Lo Bello, “To- http://www.plattform-i40.de/I40/Redaktion/ no. 5, pp. 899–903, May 2016. wards new hybrid networks for industrial au- EN/Downloads/Publikation/aspects-of-the- [5] A. J. C. Trappey, C. V. Trappey, U. H. Govinda- tomation,” in Proc. 14th IEEE Int. Conf. Emerging research-roadmap.pdf rajan, J. J. Sun, and A. C. Chuang, “A review Technologies and Factory Automation (ETFA), [40] 5G-PPP. (2015, Oct.). White paper on factories- of technology standards and patent portfo- 2009, pp. 1–8. of-the-future vertical sector. [Online]. Available: lios for enabling cyber-physical systems in ad- [24] M. Ehrlich, L. Wisniewski, and J. Jasperneite, https://5g-ppp.eu/wp-content/uploads/2014/ vanced manufacturing,” IEEE Access, vol. 4, pp. “State of the art and future applications of 02/5G-PPP-White-Paper-on-Factories-of-the- 7356–7382, Oct. 2016. industrial wireless sensor networks,” in Proc. Future-Vertical-Sector.pdf [6] R. Venkatesan, M. V. Raghavan, and K. S. Sai Kommunikation in der Automation (KommA), [41] 5G-PPP. (2016, Feb.). 5G empowering vertical Prakash, “Architectural considerations for a Nov. 2016, pp. 80–87. industries. [Online]. Available: https://5g- centralized global IoT platform,” in Proc. IEEE [25] F. Pramudianto, J. Simon, M. Eisenhauer, H. ppp.eu/wp-content/uploads/2016/02/ Region 10 Symp., May 2015, pp. 5–8. Khaleel, C. Pastrone, and M. Spirito, “Prototyp- BROCHURE_5PPP_BAT2_PL.pdf [7] J. Weinman, “The economics and strategy of ing the Internet of Things for the future factory [42] T. Sauter, “The continuing evolution of inte- manufacturing and the cloud,” IEEE Cloud using a SOA-based middleware and reliable gration in manufacturing automation,” IEEE Comput., vol. 3, no. 4, pp. 6–11, July–Aug. 2016. WSNs,” in Proc. 18th IEEE Int. Conf. Emerging Ind. Electron. Mag, vol. 1, no. 1, pp. 10–19, Mar. [8] D. Georgakopoulos, P. P. Jayaraman, M. Fazia, Technologies and Factory Automation (ETFA), 2007. M. Villari, and R. Ranjan, “Internet of Things 2013, pp. 1–4. [43] J. Imtiaz and J. Jasperneite, “Scalability of OPC- and edge cloud computing roadmap for manu- [26] S. Vitturi, P. Pedreiras, J. Proenza, and T. Sau- UA down to the chip level enables ‘Internet of facturing,” IEEE Cloud Comput., vol. 3, no. 4, pp. ter, “Guest editorial special section on com- Things,’” in Proc. 11th IEEE Int. Conf. Industrial 66–73, July–Aug. 2016. munication in automation,” IEEE Trans. Ind. Informatics (INDIN), 2013, pp. 500–505. [9] O. Chenaru, A. Stanciu, D. Popescu, V. Sima, G. Informat., vol. 12, no. 5, pp. 1817–1821, Oct. [44] A. Faul, N. Jazdi, and M. Weyrich, “Approach Florea, and R. Dobrescu, “Open cloud solution 2016. to interconnect existing industrial automation for integrating advanced process control in [27] K. F. Tsang, M. Gidlund, and J. Åkerberg, systems with the industrial Internet,” in Proc. plant operation,” in Proc. 23rd Mediterranean “Guest editorial industrial wireless networks: IEEE 21st Int. Conf. Emerging Technologies and Conf. Control and Automation (MED), June Applications, challenges, and future direc- Factory Automation (ETFA), 2016, pp. 1–4. 2015, pp. 973–978. tions,” IEEE Trans. Ind. Informat., vol. 12, no. 2, [45] S. Prüter, F. Golatowski, and D. Timmermann, [10] R. Drath and A. Horch, “Industrie 4.0: Hit or pp. 755–757, Apr. 2016. “Adaptation of resource-oriented service tech- hype?” IEEE Ind. Electron. Mag., vol. 8, no. 2, [28] J. Wan, S. Tang, Z. Shu, D. Li, S. Wang, M. Im- nologies for industrial informatics,” in Proc. pp. 56–58, June 2014. ran, and A. V. Vasilakos, “Software-defined 35th Annu. Conf. IEEE Industrial Electronics [11] T. Sauter, “Public discussion: Future trends industrial Internet of Things in the context of (IECON), 2009, pp. 2399–2404. in factory communication systems,” in Proc. industry 4.0,” IEEE Sensors J., vol. 16, no. 20, pp. [46] T. Bangemann, M. Riedl, M. Thron, and C. Died- Sixth IEEE Int. Workshop Factory Communica- 7373–7380, Oct. 2016. rich, “Integration of classical components into tion Systems (WFCS), May 2006. [29] A. Bratukhin and T. Sauter, “Functional analy- industrial cyber-physical systems,” Proc. IEEE, [12] T. Sauter, “The three generations of field-level sis of manufacturing execution system distri- vol. 104, no. 5, pp. 947–959, May 2016. networks—Evolution and compatibility is- bution,” IEEE Trans. Ind. Informat., vol. 7, no. 4, [47] T. Sauter and M. Lobashov, “How to access fac- sues,” IEEE Trans. Ind. Electron., vol. 57, no. 11, pp. 740–749, Sept. 2011. tory floor information using Internet technolo- pp. 3585–3595, Nov. 2010. [30] Y. Zhang, C. Qian, J. Lv, and Y. Liu, “Agent and cy- gies and gateways,” IEEE Trans. Ind. Informat., [13] P. Danielis, J. Skodzik, V. Altmann, E. B. Sch- ber-physical system based self-organizing and vol. 7, no. 4, pp. 699–712, Nov. 2011. weissguth, F. Golatowski, D. Timmermann, and self-adaptive intelligent shopfloor,” IEEE Trans. J. Schacht, “Survey on real-time communica- Ind. Informat., vol. PP, no. 99, p. 1, Oct. 2016.

march 2017 ■ IEEE industrial electronics magazine 27