TechnologyJuly 19, 2022
The rising role of Industrial Edge Computing in the IIoT
In this 2022 special report, industry experts provide their perspective on the role of the Industrial Edge in the IIoT. With edge computing processing data where it’s generated, using solutions that connect the physical and digital worlds, plants are leveraging data to optimize workflows, save resources and improve quality.
Industrial Edge computing offers a distributed computing framework that is bringing factory and enterprise applications closer to data sources such as IoT devices or local edge servers. This proximity to data at its source can deliver strong business benefits: faster insights, improved response times and better bandwidth availability.
In this special report on Industrial Edge, the Industrial Ethernet Book has reached out to industry experts to gain their insights into the rise of the Industrial Edge and particularly its impact on the IIoT.
Value of open ecosystems
Effective connections and interaction on the automation level.
Benjamin Homuth, Director of PLCnext Technology for PHOENIX CONTACT Electronics GmbH, told IEB that “open ecosystems are the key to a networked future”.
“In a world where everything needs to be interconnected in order to obtain the necessary data to optimize processes, automation systems have to be open to improve interoperability and collaboration between various providers and experts. This collaboration between various experts usually takes place at the edge level or in the cloud level,” Homuth said.
“Automation providers provide a cloud connection to the various cloud platforms or a connection to other automation providers. The key to this, however, is the connection and interaction on the automation level. Many software experts are able to optimally utilize and process the data originating from an automation application, but they do not possess the necessary automation know-how to obtain this data from an existing automation application. In order to overcome this obstacle, an open edge or automation platform is required so as to integrate the software or at least data connectors into an automation platform.”
He added that the challenge here is that there needs to be an easy way to route the data to the cloud or the edge, plus the process data information including the data semantics must also be transferred.
Delivering technology benefits
Looking at the specific benefits the industrial edge and cloud computing provides, Homuth said that the possibilities presented by industrial edge and cloud applications give rise to new business models and potential based on the data generated in the automation application. Capturing this data and making it as easy as possible to process the data and transfer it to the cloud presents a challenge not just in existing plants, but also in new plants to some extent.
“There are already many solutions on the market, for example, for data transfer and the processing or analysis of data on the edge or in the cloud. What is critical here is how these solutions can best get at the automation application data without any knowledge of or intervention in the existing plant,” Homuth said. “As a standard interface, OPC UA offers many advantages for data exchange when forwarding data from the sensors all the way to the cloud. However, to fully exploit all the possibilities of edge and cloud applications, providers of automation solutions must find a way to enable the implementation of software for data analysis, edge software stacks, or cloud connectors.”
With the PLCnext Technology Ecosystem, Phoenix Contact provides an open control platform and various possibilities for integrating third-party software on all levels. Installation based on the PLCnext Control hardware platform even makes an edge layer optional. When a PLCnext Control device is used, the PLC level already offers all possible functions, such as the seamless integration of third-party software, the aggregation and preprocessing of data, transmission including semantics via OPC UA, and direct connection to cloud systems.
The optional use of PLCnext Control on the edge layer simplifies interoperability with other automation providers and existing applications, and enables the bundling of data from larger installations.
Homuth added that, with PLCnext Control, know-how can be pooled from different areas. Various programming languages, such as IEC 61131-3, C++, C#, Python, or Java, can be used to program an application and can be combined, if required. The ability to integrate open-source software and easy-to-integrate apps from the PLCnext Store allows the controller to be tailored to the specific application. The programmer has an almost unlimited array of options at their disposal when creating the application.
The application program can be implemented using native implementation or encapsulated in a docker container. If required, the code can also be executed in real time and at cyclic intervals (in accordance with IEC 61131). In many applications, the use of AI and especially machine learning on the edge or in the cloud offers significant potential for optimizing production plants. By using different methods of implementation on a PLCnext Control device, it is possible to also integrate machine learning algorithms and execute them together with a control program. In the case of ML applications with fast response times or in the event that the bandwidth is not sufficient, execution on the edge or directly in the control level is an advantage over sending all the data to the cloud.
Edge application targets
Homuth said that it is evident that more and more small and medium-sized companies are using public cloud solutions with direct connection to the automation application. It is only larger companies with their own IT departments that are able to operate a private cloud, e.g., as an on-prem installation. Both versions offer different advantages and disadvantages and the companies in question need to weigh up which option will be best suited to their applications.
With both versions, an edge layer offers the option of bundling data from larger installations or connecting to an existing application. The data can then be analyzed, processed, and visualized on the company’s own servers or in partnership with major cloud providers. With Proficloud, Phoenix Contact offers another option for collecting data from the automation application, especially for small and medium sized applications.
Transmission to the cloud is implemented quickly and without any additional programming by means of a Proficloud connector that is preinstalled on the PLCnext Control device. In Proficloud, the user can record, evaluate, and visualize the data. This provides an easy way to exploit the potential of cloud applications, without having to select the right cloud components from the wide and complex range offered by the major cloud providers.
“The networking of automation technology enables the transfer of data to a cloud, which “increases the need for cybersecurity measures to ensure the security of the plant and to protect against unauthorized access and manipulation,” Homuth added. “Cybersecurity incidents are increasing at a rapid pace and can, in a worst-case scenario, damage the image of a company and result in financial losses due to production downtime. Along with the transfer of data to a cloud application, including the semantics of the process data, the proper and adequate implementation of these security measures presents a challenge for many automation users.”
With PLCnext Technology, devices and applications can be professionally installed and protected in accordance with the IEC 62443 cybersecurity standard. The PLCnext Control device is already certified in accordance with standard IEC 62443 and therefore provides the basis for the implementation of cybersecurity in machines and plants. So, with the possibilities afforded by PLCnext Technology, including the appropriate security, it is possible to benefit from all the advantages of edge and cloud applications.
Hardware, software & service
Easy-to-manage, long-term platform-level solutions.
Hermann Berg, Head of IIoT for Moxa, said that the key components of any Industrial Edge and Cloud Solution are hardware, software and (cloud) service. Robust hardware for industrial environments has been around for a while and continues to improve and get more cost effective. The most relevant advances of recent years however have been made on the software side.
“A robust and long-term support operating system (OS) and an execution environment for software containers are the infrastructure for modern edge applications. Examples are Moxa Industrial Linux, a 10-year support robust Linux OS, and container environments like docker, Microsoft’s Azure IoT Edge or AWS Greengrass,” Berg told the Industrial Ethernet Book recently.
“This should integrate seamlessly with a device lifecycle management or fleet management system that is often provided in the form of a Software-as-a-Service (SaaS) cloud platform.”
Berg said that all together this provides an easy-to-manage long-term platform where data engineers, scientists and software developers can quickly add and manage their apps without ever going on site to fix or improve the software on the IIoT gateway or edge computer.
Industrial Edge Benefits
IIoT projects have to be easy, fast and scalable to succeed. They have to be fast, so engineers and data scientists can quickly determine, if the data collected is “good enough” to create the desired outcome.
The platform should be easy to use, so internal resources can focus on developing and strengthening digital core competencies: making best use of data related to their own equipment, processes, material, and deliverables. And the platform has to scale quickly, once proof of concepts (PoC) and first deployments show the value of the new software. So, from the start hardware, software and services should be selected that are suitable for large-scale rollouts.
Berg said that, initially most companies focus on simple data collection to increase transparency: What is the status of my equipment? What is the performance of an important process, e.g., measured by indicators like overall equipment effectiveness (OEE)? This already allows for a quicker reaction to unforeseen issues and better short-term and long-term planning.
Another benefit of these open platforms is the fact that third-party applications and services can be easily combined with different hardware options. Moxa has started to establish partnerships with different types of partners to take advantage of the easy integration: https://moxa-europe.com/iiot-partners/.
“That way a stronger and stronger IIoT and edge infrastructure serves as the basis to offer the tools that allow workers and managers to successfully manage an increasingly volatile, uncertain, complex, and ambiguous production environment,” he added.
Solutions for industrial applications
Berg said there are multiple ways to implement such a scalable IIoT platform with the option to add and manage your own software apps from the cloud.
A high-end solution could be based on Microsoft Azure which provides a seamless integration of the Azure IoT Edge software modules into their cloud-based Azure Container Registry. Azure IoT Hub and their device and module twin framework is a perfect foundation for managing both edge devices and edge software containers from the cloud at scale. Increasingly third-party software applications are available through the Azure Marketplace and Azure allows for a rich integration with the cloud back-end and the rest of the IT stack of a company.
A more simple and cost effective solution, outlined by Moxa and qbee in recent project called “IIoT on a Shoestring Budget”, was centered on a very simple, but extendable open-source Python-based IIoT edge software that made very efficient use of GitHub, optionally Docker Hub, and fleet management software from our IIoT partner qbee.io (https://github.com/MEUIIOT/moxaiiot-uc2100-qbee-io).
In both cases, devices in the field can be easily monitored, upgraded and troubleshooted from remote. In particular, the software running on the devices can be managed and continually adapted and improved at scale without ever visiting the respective sites.
Potential industrial applications
When asked what are specific application areas are the newest Industrial Edge and Cloud solutions targeting, Berg responded that companies today seem to be looking less for closed third-party point solutions, rather they are prepared to build their own end-to-end data infrastructure on top of and integrated with their existing IT and OT environments.
“Based on this new and growing infrastructure they develop their own data capabilities complementary to their existing traditional core competence.
hose digital capabilities can then be used to saving costs, avoiding outages and delays, and generally providing better service to the customers,” he said.
“One visible sign of this increased digital ownership has been the surge of After Sales Service Portals. Machine and plant builders along with other vendors and users of high value industrial assets in segments as diverse as marine, mining, energy, and transportation have applied industrial IoT technology to provide more, better, and more digital service to their customers during the pandemic,” he added.
Since the beginning of 2022, Berg said that a strong and growing challenge is the efficient use of energy in Europe. Keeping operations up, while dramatically reducing gas and electricity consumption might become the single biggest challenge in 2022 and 2023 for many large industrial companies in Europe that currently depend on Russian gas or oil. IoT connectivity will be the basis of many solutions developed to address it.
Up until recently connectivity and security issues represented a significant hurdle during the deployment of IIoT solutions. As IIoT infrastructure matures and deployments scale up, new challenges arise.
“Building one integrated infrastructure for IIoT application means that data will be used by more than one application,” Berg said. “As a result, a common data model across applications has to be agreed between the development teams. The common data model becomes part of the IIoT infrastructure, so developers and data scientist can focus on the task of developing code and models based on the available data, rather than solving infrastructure problems.”
This helps to address the more commercial challenges that determine whether a good business idea will turn into a success in the real world or not: speed, scale, resources, and cost. Easy access to quality data allows developers to quickly develop, test, and deploy their apps at scale. Resources can be focused on core competencies, while the infrastructure is handled by other teams or outsourced.
“Separating IIoT edge infrastructure from the actual apps is the foundation for more valuable insights and more automation and thus better and faster service at lower cost,” he said.
Compatibility, scalability, security and artificial intelligence.
According to Georg Stöger, Director Training and Consulting for TTTech Industrial, four key technology trends are enabling the latest generation of Industrial Edge and Cloud Solutions.
“In addition to the wide adaption of open technologies and standards for compatibility and scalability, and the use of IT technologies such as virtualization and containerization for system management, we consider the integration of OT systems into IT security as a third key technology trend to further enable IIoT and OT-to-cloud solutions,” Stöger told IEB recently.
“With the rising threat to industrial systems from increasingly professional and state-sponsored cyberattacks, the security aspect has risen in importance in recent months. Fortunately, these trends are not opposing each other; in fact, the security architecture of open standards such as OPC UA tends to be up-to-date and sophisticated, and flexible application management using virtualization and containerization can – if done right – also contribute to security,” he added.
Nevertheless, the challenges to achieve secure and efficient system management in a cloud-enabled industrial infrastructure are tremendous and will likely become even more so with the fourth trend, the use of artificial intelligence at the Industrial Edge.
This trend will blur the line between cloud and edge further, offering even more flexibility to optimize existing applications and create new functionality that could not be effectively implemented in a fully cloud-centric architecture.
Benefits of the Industrial Edge
When asked about the specific benefits that industrial edge and cloud computing provides, and the potential impact on manufacturing, Stöger replied that each have their separate benefits but there are also substantial the benefits of combining industrial edge and cloud.
“With field level interfaces connected to the industrial edge, access to real-time process data is possible. This is essential for use cases which require a lot of storage and processing power, such as powerful digital twin models,” Stöger said.
He added that the process data for creating these models needs to be sent to the cloud. However, the same data might also be used immediately at the edge, e.g. for rapid diagnostics to avoid machine malfunctions that could affect product quality. Not all data may be needed in the cloud; users could send only the data required for analysis or storage to the cloud, thus saving bandwidth and cost, while processing the full data set at the edge.
An industrial edge computing platform such as TTTech Industrial’s Nerve, needs to support not only this flexible management of process data, but also the management and execution of multiple functions at the edge, ideally supporting various technologies for efficient use of hardware resources: virtualization and containerization of applications make it possible to maintain and update applications, devices, and software remotely. This saves hardware cost and enables remote maintenance as well as easier handling of devices (e.g. via a web-based management system).
Industrial application solutions
One key benefit is that Industrial Edge computing brings IT technologies to the shop floor and OT data into the data center. Applications using OT data can be run close to the data source (e.g. the machines on the shop floor); this allows data to be processed with lower latencies and can help mitigate network bottlenecks, but also security issues.
OT data made available to the IT infrastructure can also be used to provide additional services to customers, e.g. in connection with predictive maintenance and service packages, decreasing the risk of production standstills. Remote maintenance is based on real-time data as well – it helps to reduce travel cost and allows issues to be resolved more quickly and without customers having to wait for a technician to arrive.
“At a time of severe supply chain issues with electronics, virtualization can become a very interesting benefit of industrial edge computing. With this technology, which is well established in the data center, a single powerful IPC can host multiple applications and even entire virtual computers simultaneously. This saves cost and simplifies system management. However, specialized virtualization technologies (real-time hypervisor) are required to address the specific needs of some industrial applications. Our edge computing platform Nerve uses Linux Foundation’s open hypervisor ACRN™ and Intel® Time Coordinated Computing (Intel® TCC) to improve hard real-time virtualization,” Stöger said.
“The application areas that our customers and project partners are currently focusing on include condition monitoring of machines and field equipment, remote access for service, and predictive maintenance. These applications all need reliable access to real-time OT data provided by field I/O and controllers to industrial edge devices. Cloud analytics can enhance the picture by showing an aggregated picture across production sites and countries,” he said.
Machine learning for optimization of parametrization, anomaly detection, and digital twins are newer applications that make use of edge computing technology. They also need large amounts of real-time data for training, which has to be made available reliably and securely.
Machine learning builds on often large samples of data and enables machines to make predictions and decisions based on “examples of previous experience with this kind of process”.
This is an important step towards smart factories and more autonomous production. Digital twins are a digital representation of an asset such as a machine or a factory. They are trained on real-world data and allow companies to build and test parameters in the digital world before rolling them out in the physical world. This helps optimize processes and machines before they are built or reconfigured, saving resources and adaptation costs.
Addressing engineering challenges
Stöger said that a major challenge often encountered by automation engineers is networking. IT and OT networks are typically managed in very different ways, which makes sense in a traditional federated architecture with “closed” and tightly controlled production systems.
With more openness, more internet-connected applications, remote access, remote maintenance, and increasing frequency of software updates and patches, gateways and firewalls become a liability.
“This challenge is addressed by integrated industrial edge platforms that support secure tunneling, forwarding, and other mechanisms to provide the necessary flexible connectivity for multiple applications and devices at the edge, while maintaining full control over all traffic that enters or exits the factory network through the firewall. Thus, the automation engineer is not required to become an IT networking expert to address network performance and security,” Stöger said.
Edge computing can also address software security issues. Less secure applications (e.g. legacy software) can be containerized and isolated so that they can only access the intended resources within the edge. As the applications are running completely separately from each other, the system is only partially affected if one of the applications/devices is successfully attacked and compromised. Edge computing devices should support role-based access control which limits the risk of unauthorized access to critical data and resources.
Solutions that connect the physical and digital worlds.
According to Arun K. Sinha, an engineer at Opto 22, key technology trends that are enabling the edge-to-cloud solutions include Internet protocols and technologies being driven into systems at the edge, where the physical world and the digital world connect.
“Layers of complexity are being removed from the communication process between physical assets and cloud solutions. Modern IIoT system architectures are being flattened, streamlined, optimized, and secured,” Sinha said. “Edge computing systems must easily and securely access the cloud through the open, standards-based communication technologies the internet is based on.”
He added that most devices on an OT (Operational Technology) network today communicate using the request-response model. Edge-to-cloud architectures pose a problem of scale, and with this model network traffic can quickly become an issue. The trend is towards a model that is more effective in IIoT applications, which is publish-subscribe, or pub-sub. In the pub-sub model, a central broker becomes the clearinghouse for all data. Because each client makes a single lightweight connection to the broker, multiple connections are not necessary.
“Industrial edge and cloud computing ideally create a cohesive system of devices and applications able to share data seamlessly across machines, sites, and the enterprise to help optimize production and discover new cost-saving opportunities. The benefits include making it easier to connect industrial equipment with computer networks, software, and services, both on premises and in the cloud,” Sinha said.
The impact on manufacturing companies as well as OEMs is that they can now take advantage of analytics tools from software companies such as AWS, Microsoft, Google and others to improve their manufacturing processes and streamline their business. Manufacturers can then start to identify real opportunities for operational efficiency improvement and meaningful revenue generation. To foster such business benefits, data from the physical world of machines and equipment must be available to the digital world of the internet and information technology systems, quickly, easily, securely and continuously.
Sinha added that new software and communication tools have emerged onboard automation systems that help facilitate edge-to-cloud connectivity, industrial IoT, digital transformation, and Industry 4.0 for manufacturing, process and OEMs. These are unique in that they combine traditional automation and control with technology more common in IT and software development tools, data formats and communication methods.
MQTT with Sparkplug B is a secure, lightweight, open-source publish-subscribe communications protocol with a data payload designed for mission critical applications. Node-RED is a low-code, open-source IoT programming language for managing data transfer across many devices, protocols, web services, and databases. RESTful APIs allow software or cloud systems to get data from remote edge devices, without the layers of complexity and conversions that exist in traditional industrial automation systems.
Further, cybersecurity features commonly found on IT devices such as user authentication, configurable firewalls, encrypted communications and even LDAP are critical for edge devices when used in industrial applications involving cloud connectivity.
Industrial Edge and cloud solutions
Sinha said that one of the most exciting use cases for industrial edge and cloud solutions is predictive maintenance of machinery. Most analytics tools using anomaly detection, artificial intelligence or machine learning executed their algorithms in the cloud. The data is first gathered and pre-processed at the edge, and then communicated to the cloud for the heavy computational resources required.
“Remote monitoring of distributed assets is another application area where IIoT solutions such as MQTT with Sparkplug can play a key role,” he added. “Where field signals are distributed over large geographic areas or multiple sites, edge devices can facilitate data transmission to networked applications and databases, improving the efficiency and security of local infrastructure or replacing high-maintenance middleware such as Windows PCs. IoT enabled edge controllers can act even as IoT gateways to legacy control systems. All disparate devices can share data across the organization through the cloud with a unified namespace.”
Machine builders and OEMs are also utilizing these solutions to provide value add services to their end users. These include not only hosting but analyzing client data to provide proactive service measures as well as improve operational efficiency of their machine at the customer facility. MQTT is ideally suited for this application as it does not require a port to be opened by the customer IT department, as all data is communicated securely via device originated, outbound connections.
One common challenge that automation engineers face is one of connectivity. At the edge, things like sensors, circuits, relays, and meters are attached to industrial control systems used to operate equipment and machines. These sensors translate what’s physically happening (temperature, light, vibration, motion, flow rate, and so on) into an electrical signal like voltage or current, which is then interpreted by a controller.
Traditional PLCs typically are not capable of edge computing or communicating directly to cloud platforms. They typically use OT protocols for communication that are seldom internet compliant, and do not include information security standards like encryption and authentication.
Sinha added that another challenge is the large volumes of data that can be generated at the edge with industrial automation systems. Moving that much data onto existing network and internet infrastructures for cloud-based analytics and centralized management will clog networks, vastly increasing network and internet latency. For many industrial IoT applications, that is not acceptable, because real-time control and monitoring are mandatory.
For the industrial internet of things to reach critical mass, intelligence must be pushed to the network edge, where the physical world meets the digital world. Computing systems at the network edge must have the capability to collect, filter, and process data generated at the source, before it’s transmitted up to the IIoT. And at the same time these edge computing systems must be able to complete the local real-time process control and automation tasks of traditional industrial applications.