11.2.1 Energy

The global revenue for the smart energy segment amounted to 72 billion USD in 2015 (Frost & Sullivan, 2016c) and as depicted in Figure 11.1, the revenue is expected to show a CAGR, between 2015–2020, of 5.3% resulting in a market volume of approximately 93 billion USD in 2020. Leading technologies will be Advanced Metering Infrastructures (AMIs), Demand Response (DR), Distribution Grid Management (DGM), and Advanced Transmission Technology (ATT), while DGM will be the dominant segment with a 64% of the market share by 2020.

As highlighted in Chapter 14, the energy market is evolving to a more efficient, cleaner, and flexible ecosystem. For example, the aim of the Paris Agreement³, entered into force on November 4, 2016 with 116 partner ratifications, is to strengthen the global response to the threat of climate change. Energy generation accounts for 68% of the shares of global anthropogenic greenhouse gas (GHG) (International Energy Agency, 2016), therefore, it is necessary to shift to a cleaner and efficient energy production market. Renewable energy plants are being deployed all over the world, but nevertheless, one of the biggest challenges of integrating variable energy sources, like Photovoltaic (PV) or wind energy, is the difficulty in balancing the grid in real time. Moreover, renewable plants are erected where the resource (solar, wind, biomass) is available, and they are not always close to the consumer. The smart grid will facilitate the integration of variable and intermittent renewable resources, allow load adjustment and balancing, and distribute power over the network efficiently (ITU, 2012). The International Energy Agency (IEA) foresees that its share will reach at least a 26% increase in 2020 and IIoT technologies will change the utility business models. AMIs will allow a bidirectional power flow; hence, the customer will be able not only to consume but also to produce power, becoming a “prosumer” (World Resources Institute, 2016). Demand side management (DSM) will improve the energy grid from the consumption side, for example, by employing smart energy tariffs with incentives for using energy at a certain time of the day, or real-time control of distributed energy resources (Palensky and Dietrich, 2011).

11.2.2 Health care

As depicted in Figure 11.2, the global revenue in the health care market will grow from 86 billion USD in 2015 to 233 billion USD in 2020 and the projected CAGR is around 21% (Little, 2016).⁴ With a market share of 44% by 2020, the wireless health segment will be the most relevant one mainly driven by wireless sensors, handheld devices, and eHealth applications. The Organization for Economic Co-operation and Development (OECD) reported that in 2014, 9.945% of the world gross domestic product (GDP) was spent on health, up to 0.144% since 2005.⁵ Circulatory, digestive, cancer, and mental health conditions represent almost 60% of the current health spending and, likewise, chronic diseases account for 60% of the causes of death.⁶ The World Health Organization (WHO) and its member states endorsed health care as a cost-effective and secure approach to strengthen the health care systems (WHO, 2005), and governments are focusing on making them more efficient and sustainable health care (Frost & Sullivan, 2016b). For instance, European Union health care policies pursue making health care tools useful and widely accepted by involving health care professionals and patients in the strategy, design. and implementation.⁷

Devices such as heart rate monitors, pulse oximeters, blood pressure monitors, pedometers, smartwatches, smartphones apps, and so on, are being used to measure health conditions and activities. When this information is exchanged between the device and a health care platform, patients benefit not only from self-monitoring but the information could also be used for different purposes such as detection, prevention, treatment of diseases, supporting a rehab process, and so on. Seamless communication aids patients that need remote assistance, thus, reducing costs for them and the insurance system. This specific application of IoT technologies in the health care domain is further described in Chapter 16. The IIoT will help to improve access to comprehensive health care services, quality of medical services, decrease medical errors, and improve patients' quality of life. Moreover, real-time monitoring, control, and automation empower assisted living to provide personal safety and health care management at home. Additionally, one of the main benefits of health care is a patient's empowerment by providing more autonomy and increasing their treatment.

11.2.3 Manufacturing

As shown in Figure 11.3, the global revenue in the manufacturing market will grow from 39 billion USD in 2015 to 62 billion USD in 2020 and the projected average CAGR is 9.7% for the global market (Mordor Intelligence, 2017). The smart manufacturing domains include automotive, chemical and petrochemical, oil and gas, pharmaceuticals, aerospace, defense, mining, among others. The chemical and petrochemical industries hold the major share (23%) while the oil and gas market is expected to grow at a higher CAGR.

Within this major IIoT application domain, digital technologies will be used to move toward resource-saving and more efficient manufacturing. For example, Cyber-Physical Systems (CPSs) and prescriptive analytics will enable automated decision-making at the topological edge of a network to allow for timely maintenance measures and an extended lifecycle of machines while optimizing the overall production at the same time. Every digital device will be able to provide their real-time status (the so-called digital shadow), thus allowing other devices to react on this information. It is foreseen that the application of IIoT technologies in the manufacturing sector will lead to process optimization (possibly enabling efficient lot-size one productions) and supports the prioritization of workloads, and as a result will significantly reduce needed quality inspections, surveillance, and operational expenditures in the industrial manufacturing sector (Frost & Sullivan, 2014, 2016a). The development of a virtual factory will provide a holistic, scalable, and virtual representation of a manufacturing facility and allow synchronization, dynamic configuration and thus, enable cost savings of manufacturing facilities (Ghielmini, 2013).

However, these potential cost savings are also offset by risks. Downtimes in the production are very expensive, which is why reliability has been a top priority in automation technology over the last 40 years. Installations are also often operated over many years to decades without the need to install updates, as it is common (and required for at least security reasons) in IT infrastructures. Therefore, the continuous merging of OT and IT in this context faces both high potential and great challenges.

11.2.4 Smart Cities

The global smart cities value in 2015 was approximately 312 billion USD and is expected to reach 758 billion USD by 2020 with a CAGR of 19.4% (Markets and Markets, 2016) (see Figure 11.4). The building segment is projected to grow at the highest CAGR, on top of transportation, energy, and smart citizen services such as education, health care, and security. According to the United Nations (UN), urban areas represent approximately 70% of energy-related global emissions, and by 2050 more than half of the world's population will live in cities, mainly in African and Asian regions (United Nations, 2014). Nevertheless, energy efficiency and GHG emissions are not the only matters of concern. With urban population increasing, challenges such as security, balance public expenditure, transportation, health care, and education have to be considered. A city is a complex network of people and infrastructure that interacts, expands, and transforms continuously. Traditionally, the infrastructure and services of the cities are operated as verticals or domains, with little or no interaction: transportation, energy, health care, buildings, industry, and so forth.⁸

Each vertical is evolving to a smarter and more efficient version of itself, and cities must take advantage of those improvements. A smart city should be able to integrate the current infrastructure with ICT to operate more efficiently while improving the quality of life of its citizens. According to the International Telecommunication Union – Telecommunication Standardization Sector (ITU-T), a smart sustainable city is defined as “an innovative city that uses ICT and other means to improve quality of life, efficiency of urban operation and services, and competitiveness while ensuring that it meets the needs of present and future generations with respects to economic, social, environmental as well as cultural aspects.”⁹

The use of IIoT technologies will enable the efficient use of resources in urban areas; however, to become smarter, a city needs its municipality, industries, and society to participate. Some use case scenarios include Blackout Prevention, that is, the utility applies smart “self-healing” to reconfigure itself whenever there is a problem in the distribution network and, whenever there is an imminent cut-off of electric power inform, in advance, residential and industrial users to take appropriate measures; air quality monitoring, that is, collaborative sensing will help to determine contaminants before they reach a dangerous level and to identify the source and their impact on transportation, industry or energy generation industries; or smart parking, that is, buildings, streets, and parking lots are connected to determine available parking spaces to save time, make efficient use of resources (gas, diesel, and public spaces), and minimize stress as well as emitted pollutants. Further details can be found in Chapter 12.

11.2.5 Transportation

As shown in Figure 11.5, the global revenue for the smart transportation segment amounted to 10 billion USD in 2015 and its market is expected to show a CAGR between 2015–2025 of 18.7%, resulting in a market volume of 24.5 billion USD in 2020 (Zpryme Agency, 2015). The smart transportation ICT segment includes hardware, software, communications and networking, and sensors and Intelligent Electronic Devices (IEDs). Smart transportation or intelligent transport systems (ITSs) are those systems that enable connection, integration, and automation of the transportation network to improve experience for travelers and system operators (users) by enhancing vehicles and infrastructure (U.S. Department of Transportation, 2015). Therefore, the scope of smart transportation is not only limited to connected cars, but also to car/bike sharing systems, pay as you drive (users); smart roads, road pricing, parking systems, traffic management, backhaul communications, fleet management (infrastructure); connected car, automated vehicles, public transportation (vehicles), just to mention a few. Although in 2015 more than 69 million passenger cars were produced, the automobile industry is experiencing changes. City policies are discouraging private vehicles (McKinsey & Company, 2016) and today there are more than 80K car-sharing vehicles in operation with more than 6 million users (BCG Perspectives, 2016).

The use of IIoT technologies in the transportation industry will allow proactive maintenance and prevent failures through predictive analytics. Safer vehicles and roads will improve crash avoidance by developing vehicle-to-infrastructure cooperative systems. Advanced sensing technologies and high-bandwidth connectivity will enable real-time applications that interwork with different domains. For example, Advanced Traffic Management Systems (ATMSs) will improve the flow of vehicles, thereby decrease traffic commuting time and CO₂ emissions across urban areas. Moreover, fleet management (from rental cars to freight transport) will be supported by ubiquitous and affordable mobile communications as well as location systems to maximize customer service and productivity (GSMA, 2015). Finally, a transformation from current cars to driverless cars is expected and will be based on the IIoT (PWC, 2016).

11.3.1 Connectivity

As shown in Figure 11.6, overall connectivity between objects involve the first four layers of the OSI model. They build the basis for IIoT devices to connect with each other, and within single, noninterconnected use case domains, a crucial prerequisite is the use of interoperable wired or wireless links. Depending on the requirements of the application area, proprietary technologies, such as PROFIBUS¹⁰ or Modbus¹¹ fieldbus systems, might be used. However, two main trends can be observed: First, wired networks are in the transition to become mainly Ethernet based, with the IEEE Time-Sensitive Networking (TSN) standards as one notable development. Second, where possible low-power wireless technologies are applied for connectivity. For short ranges, mainly Bluetooth Low Energy (BLE), RFID technologies, such as Near Field Communication (NFC), and IEEE 802.15.4-based approaches like ZigBee are being deployed. For Low-Power Wide-Area Networks (LPWANs), LoRa, Sigfox, nWave, and Neul are popular technologies in unlicensed bands. At the same time, the 3rd Generation Partnership Project (3GPP) is standardizing LTE-M, NB-IoT, and EC-GSM-IoT for licensed bands.

As data exchange between multiple verticals is one particular characteristic of the IIoT, a joint routable network layer is a second prerequisite. Despite its specification of IPv6 already in the end of the twentieth century, many networks are still using IPv4. However, given the number of interconnected devices and their often-limited capabilities, the Internet Engineering Task Force (IETF) working group IPv6 over Low-Power Wireless Personal Area Networks (6LoWPAN) defined respective concepts to use IPv6 as the common networking layer, which will also be predominant in the IIoT context. Depending on the requirements, typical Transmission Control Protocol (TCP) and User Datagram Protocol (UDP)-based transports will be used on top of IPv6.

11.3.2 Communication

As indicated in Figure 11.7, in modern networks, the next and final OSI layer is the application. However, for interoperable data exchange, an additional layer is needed that we call “middleware” for convenience. First, the Hyper Text Transfer Protocol (HTTP), the Advanced Message Queuing Protocol (AMQP), the Message Queue Telemetry Transport (MQTT), the Constrained Application Protocol (CoAP), or WebSockets are typical examples of standard application-level protocols that are used within the IIoT context. Next, to enable standard Remote Procedure Calls (RPCs) and data exchange, mechanisms such as XML-RPC or the Simple Object Access Protocol (SOAP) are being used as well as Representational State Transfer (REST) concepts. Finally, depending on the use case and the geopolitical area, a variety of IIoT specific middleware systems are then deployed. The following three examples are currently under discussion within different IIoT verticals.

The oneM2M¹² (Swetina et al., 2014) alliance is a partnership of international standards bodies. While historically being focused on the telecommunications industry, the architecture aims to cover smart building, smart factory, and smart power grid use cases. In the current release 2.0 of the middleware specification, messages are allowed to be sent via HTTP, MQTT, CoAP, and WebSockets. Further, a number of Common Service Entities (CSEs) are defined that are often used in Machine-To-Machine Communication (M2M) (Wu et al., 2011) environments that can be invoked by Application Entities (AEs); and Network Service Entities (NSEs) provide respective services to the CSEs.

Developed within the international OPC Foundation, the Open Platform Communications Unified Architecture (OPC UA) (IEC, 2016) middleware is, as a successor to the former Object Linking and Embedding for Process Control (OPC) architecture, mainly being used in the automation industry. The two different communication types are either directly exchanging binary data using raw TCP sockets or exchanging XML data via SOAP and HTTP over TCP. Since June 2016, further application-level publish/subscribe protocols, such as AMQP, are being evaluated.¹³ The standards further define common base services such as (historical) data access, alarms and conditions, and programmability, and a common object-oriented meta model for describing exchanged information.

Finally, the Data Distribution Service (DDS) (Pardo-Castellote, 2003) was developed within the Global Information Grid (GIG) project and standardized by the Object Management Group (OMG). Based on the DDS Interoperability Real-time Publish-Subscribe Wire Protocol (DDSI-RTPS), the DDS Application Programming Interface (API) offers via TCP or UDP, access to a data-centric publish/subscribe system with potentially multiple hierarchical control domains. Since 2017, it has been further extended with RPC capabilities.¹⁴

11.3.3 Data Exchange

As shown in Figure 11.8, the aspect of data exchange is located on top of the afore discussed middleware layer. Following the European Telecommunications Standards Institute (ETSI) white paper on technical interoperability (van der Veer and Wiles, 2008), at least three different layers have to be considered. First, all aspects from the network layer up to the middleware have to be considered for technical interoperability. Due to the heterogeneity of the involved systems in IIoT environments, the application of a homogeneous set of protocols is unlikely. Therefore, implementations need to be abstracted from specific APIs. To implement such a Separation of Concerns (SoC) (Martin, 2003; 2012), a term coined by Edsger W. Dijkstra in 1974 (Dijkstra, 1982), numerous architectural design patterns can be applied. Examples are the classic Model View Controller (MVC) (Krasner et al., 1988), Entity, Boundary, Interactor (EBI) (Jacobson et al., 1992) or Data, Context and Interaction (DCI) (Coplien and Bjørnvig, 2010), in addition to more modern Microservices-based architectures (Newman, 2015; Fowler and Lewis, 2014; Thones, 2015).

Second, assuming the usage of appropriate design patterns or gateways to allow IIoT systems to exchange data with each other, this data has to either be serialized using the same syntax or unambiguous mapping rules have to exist. As indicated in Figure 11.9, to exchange information between distributed systems in general, world concepts are abstracted into an information model, often in form of a human-readable text document such as a Request for Comments (RFC). To transmit the information over a network, the derived data model is then serialized using a syntax such as the Extensible Markup Language (XML) or the JavaScript Object Notation (JSON). Within an application, this string is then deserialized again by functional code into an object for either document- or stream-based processing, depending on the size of the data and the way the recipient application is designed. To decrease transmission and deserialization time, more efficient serializations such as Protocol Buffers (protobuf) (Varda, 2008) can be used as well.

Finally, assuming the same syntax is being used, the meaning of the exchanged data has to be understood by each involved component. This in particular holds true in IIoT environments in which heterogeneous devices across multiple application domains shall negotiate interactions autonomously in order to further automate processes. Typically, a tree-based data model along with structure- and identifier-implied semantics are being used, for example, based on schema definitions known within the specific use case domain. This approach, however, does not scale with cross-domain IIoT-wide scenarios, as, due to their heterogeneous nature, it would result in involving n different approaches to encode information in a tree, which leads to a combinatorial problem of n² required conversions using functional code. A formal information model of types, properties, and relationships of objects within specific domains, also known as ontologies (see Figure 11.9), is needed instead. This would allow for semantic reasoning over the information to infer logical consequences such as transitivity, symmetry, or equality of specific data; in other words, it would allow machines to understand each other. One common approach is the Semantic Web (Berners-Lee et al., 2001), along with its canonical graph-based data model Resource Description Framework (RDF) (Cyganiak et al., 2014), ontology languages such as the Resource Description Framework Schema (RDFS) (Dan and Guha, 2014) and the Web Ontology Language (OWL) (Herman et al., 2012), and other related concepts. As indicated in Figure 11.10, the respective stack of technologies for semantic interoperability is located above the typical OSI model and middleware systems. Being mainly independent of the protocols and serializations used within different application areas, interoperability on this layer will probably be one of the most important aspects within the envisioned IIoT. For performance reasons, binary serializations, such as Header, Dictionary, Triples (HDT) (Fernández et al., 2011), can also be used.

11.4.1 Industrial Internet Consortium

The one initiative that stands out is the Industrial Internet Consortium (IIC)¹⁶ as it covers almost every vertical domain. While not being an SDO, the IIC is an open membership organization bringing together government, academia, and the industry. It aims at gathering use case requirements and coordinating standardization efforts across the whole IIoT ecosystem. One notable outcome of this effort is the Industrial Internet Reference Architecture (IIRA) (Industrial Internet Consortium, 2015). Its main purpose is to provide a common basis for heterogeneous stakeholders to design IIoT systems by presenting an architectural overview. Following the concepts and terminology introduced in ISO/IEC/IEEE 42010:2011 (ISO/IEC/IEEE, 2011), the overall architecture is decomposed into four different viewpoints. For each of them, concerns and models from different stakeholder perspectives are described in more detail. As the names imply, within the business viewpoint, commercial and regulatory aspects are identified; the usage viewpoint describes matters related to components or humans interacting with the IIoT system; the functional viewpoint focuses on the overall internal and external interactions and activities; and finally, the implementation viewpoint covers specifics for carrying out the described concerns in the other viewpoints.

From a technical perspective, the specification of the functional viewpoint provides the most in-depth details. In particular, the ongoing convergence between local control systems of traditional OT and the globally interconnected Information Technology (IT) is a major concern. As a result, a number of functional domains have been identified (see Figure 11.11) to form a concrete functional architecture based on interconnected CPSs. The control domain directly interacts with the physical system by sensing information and applying closed-loop logic through actuation. This domain further includes communication with external entities, data abstraction and analytics, and asset management. Within the operations domain, a number of functionalities are contained that are required to manage the systems under a single control domain. These functionalities include provisioning, deployment, monitoring, diagnostics, prognostics, and optimization. The information domain contains data and analytics functionalities that are complementary to those within the control domain. Its purpose is to transform, process, persist, distribute, and analyze data for systemwide, long-term optimizations. Next, the application domain holds both global use case specific logic and rules as well as interfaces for humans or applications to interact with the logic. Finally, within the business domain traditional functionalities such as enterprise resource planning (ERP), customer relationship management (CRM), or manufacturing execution system (MES) reside.

Another source for technical details on how to design IIoT systems is the description of the implementation viewpoint. It describes the architectural design patterns that are applied across use case domains. The most general abstraction is the three-tier architecture that defines three different tiers: the edge, platform, and enterprise tier. On the highest level, the enterprise tier mainly contains domain applications with their rules and controls, and the platform tier in the center holds more generic data aggregation and flow analytics functionalities. While outsourcing these activities to third parties to meet organization needs in an efficient manner (Hassan, 2011), communicating data to a centralized cloud over the Internet introduces latency and jitter. Therefore, functionalities can also be placed close to the devices within the lowest edge tier (also known as edge computing (Lopez et al., 2015)) to be connected to the potentially deterministic and real-time capable, access network. Missing in this view is the fog computing (Vaquero and Rodero-Merino, 2014; Yi et al., 2015) paradigm, which is further described in Chapter 4. To further reduce latency and at the same time to improve scalability, security, and data sovereignty, in this concept, functionalities of the Control domain are, depending on the resource constraints of the devices, either running on the device or on nodes that are attached directly to the devices. This architecture further allows nodes to actuate autonomously without dependency on the network and above all presents the basis to implement the aforementioned concept of CPSs. Finally, the decision where to place specific virtualized functionalities in a topology is always a tradeoff between available resources, network performance, data privacy, and device autonomy.

11.4.2 Plattform Industrie 4.0

In the mid-eighteenth century, the mechanical production was powered by water by over 80% in Great Britain (Minchinton, 1989). This changed gradually with the invention of the steam engine, until over 98% of the required power was supplied by steam in the beginning of the twentieth century. This development changed the way of production dramatically for the first time and is, therefore, called the first industrial revolution. After this, in the beginning of the twentieth century, the introduction of assembly line work changed the overall process of production again. While it took over 720 min to manufacture a Ford Model T in 1911, the time dropped below 90 min only 3 years later (Minchinton, 1989), thanks to this second industrial revolution.

Until 1968, dedicated controllers, relays, and fixed circuits were used to automate the production process in factories. As a result, the process to update such facilities was very time consuming, expensive, and error-prone. The invention of the Programmable Logic Controller (PLC), by Dick Morley, started the third industrial revolution. With its input/output (I/O) modules for field devices, it built the foundation of the modern five-layer automation pyramid. Multiple PLCs, remote terminal units (RTUs), and human machine interfaces (HMIs) are interconnected over Supervisory Control and Data Acquisition (SCADA) fieldbus systems, and on top of this a MES that monitors the most important key performance indicators (KPIs) and finally, the information can be integrated into the ERP system.

The current convergence between this OT and ICT toward self-managed CPS-based automation is called the fourth industrial revolution. The concept makes use of virtual representations of physical objects for smart factories of the future; also called Industry 4.0, based on the German term “Industrie 4.0.” As a union of the most relevant companies and associations in Germany, the “Plattform Industrie 4.0” aims at developing recommendations for the implementation of smart factories of the future. Hence, and in contrast to the IIC, the initiative is mainly focusing on modeling next generation manufacturing systems while focusing on the economic impact of interconnected cross-domain value chains.

The Plattform Industrie 4.0 specified the Reference Architecture Model Industrie 4.0 (RAMI) (Adolphs et al., 2015; Deutsches Institut für Normung, 2016). This three-dimensional layer model is the basis to systematically classify related technologies and builds upon standards defined by the International Electrotechnical Commission (IEC), namely, IEC 62890 (lifecycle management for systems and products used in industrial process measurement, control, and automation), IEC 62264 (enterprise control system integration), and IEC 61512 (batch control). Analog to the IIC IIRA, the RAMI defines six different layers and beginning in 2016, both initiatives agreed on a cooperation and started to map functionalities of both architectures.¹⁷ As shown in Figure 11.12, from the lowest to the highest level, the RAMI layers correspond to the IIRA domains as follows: The asset layer with the physical system, the integration layer with the control domain, the communication layer with the communication part of the control domain, the information layer with the information domain, the functional layer with the operations and application domains, and finally the business layer with the business domain.

The most important concept in this model is the so-called Asset Administration Shell (AAS) (Plattform Industrie, 2016), as depicted in Figure 11.13. It can be seen analog to the aforementioned control domain to implement a CPS. It wraps a software component around physical objects holding a digital representation, a so-called Digital Twin or Avatar, of the asset to which other systems communicate with. As such, it encapsulates an existing Asset/physical system to integrate it into Industrial Internet/Industry 4.0 environments. As neither the Plattform Industrie 4.0 nor the IIC has yet specified how these concepts should be implemented, the actual realization is an open and interesting research topic. Currently, mainly three different middleware approaches are under evaluation that were briefly described in Section 11.3.2: OPC UA, oneM2M, and DDS. As we learned in Section 11.3.3, however, the use of a single middleware in all IIoT use cases is both, unlikely and unnecessary. The challenging research question currently under active discussion in all relevant alliances is, how to achieve semantic interoperability to enable an Industrial Internet of autonomous CPSs.