Every second counts in fast decision-making processes in production environments, but evaluation, analysis and transfer of more and more data to the cloud and its return to the point of origin often takes too long. Modern smart factory concepts therefore rely on edge and fog computing to complement the cloud. But how exactly does this work?
A public cloud solution requires a constant online connection and high bandwidths. Even a private cloud in your own data center has this shortcoming. Current cloud concepts, especially when it comes to controlling machines, therefore provide for not transporting all data to a distant cloud, processing it there and then transmitting the parameters determined. Instead, the data should be processed where it is needed and only later - and not to the full extent - transferred to the cloud. At the edge of the network, i.e. at the transitional step to the machines, the so-called edge computers are used. They now have so much processor power and memory capacity that they are suitable for standardized, fast evaluations of machine data.
Intelligent Edge means close to the process
The computing power from the edge to the cloud is scalable, allowing the tasks of individual processes to be adapted precisely to the circumstances. In the first step, tasks are trained on high-performance servers that still can be performed non-time-critically. The trained model can then be applied immediately in the real process on demand - close to the process and on the edge. In the future, as much raw data as possible from sensors and machines should be processed on the so-called 'Intelligent Edge' instead of in the cloud, because that way data can be filtered, processed and analyzed on the basis of trained models close to the point of origin and before the transition to the network. This ensures that the control loop is quick and efficient, close to the process. The required scalability is achieved through new technologies, such as the 'containerization' of complex functions or the creation of digital twins. This also makes so-called 'Software Defined Machines' possible, which even allow the outsourcing of powerful tools such as Edge Analytics, AI or Machine Learning.
Controlling Industry 4.0 Architectures - What's important for Edge Computing Solutions
The flexibility required for edge computing is achieved through embedded computing solutions based on the globally standardized SMARC 2.0 (Smart Mobility Architecture) and COM Express form factor specifications. The Kontron SMARC-sAL28 module with up to five TSN-enabled 1GB Ethernet ports is an example of a cost-effective solution with guaranteed latency and quality of service for controlling industry 4.0 architectures. Wherever the on-premise establishment of a private cloud is desired, industrial servers are being used. These are then connected to the edge devices - or to so-called Fog computers for higher computing power - and store the collected data in the local cloud. They also perform higher-level backup tasks or data preparation for the public cloud. These tasks require industrial computer platforms with a high degree of modularity and flexible storage options based on the latest Intel Core and Server Class Intel Xeon processors. The compact ZINC-CUBE-SKD embedded server and the KISS family from Kontron, for example, have been specially designed for CPU-intensive applications such as AI or machine learning. With edge computers, it is also important to be able to connect different sensors or PLC devices in order to obtain information from them for modern applications for data analysis and process optimization. The Kontron KBox series includes a broad portfolio of edge and fog computers that are ideal gateways in industrial environments with classic fieldbus interfaces or TSN network interfaces.
This increasing networking presents system integrators with new challenges
This increasing networking of industrial computer systems naturally presents system integrators with new tasks and challenges. For this purpose, S&T Technologies has developed the IoT Software Framework SUSiEtec, which acts as an 'adhesive' to securely connect devices to each other and to the cloud. SUSiEtec joins the IoT infrastructure - from the sensor or actuator to the edge computer and the embedded cloud, to the private or public cloud - like puzzle pieces and combines them into a complete package. In addition, SUSiEtec has the ability to integrate machine learning so that it can propose its own decisions on the basis of available data - real AI, that is! ????
Supporting open standards such as OPC UA and TSN , it even can solve interface problems in the medium term. The boards, gateways, modules and systems of IoT applications should be protected against piracy, reverse engineering and unauthorized changes with TPM 2.0 components and special security technologies. In addition to TPM chips, Kontron's Edge products feature CodeMeter technology from Wibu-Systems. The 'Trusted Hardware' thus secured enables standard technologies - such as Secure Boot - to be implemented with secure operating systems.
SUSiEtec works seamlessly with Microsoft IoT-Edge, which enables the use of Microsoft Azure as a public cloud solution and thus significantly accelerates response times in processes. Microsoft's Azure-IoT edge services ensure seamless scalability of computing power between computing resources on Intelligent Edge, Embedded Cloud, On-Premise Data Center or Public Cloud. Technically, all of these are containers that map software functions and can then be moved between the cloud and edge devices. This allows flexible decisions to be made - depending on the desired level of security and performance - as to where data preparation and analysis should take place.
Intelligent Edge opens up new business models
The scope of tasks for edge computers is constantly expanding with new technologies for hardware, software and connectivity. In the future, machines will be implemented in the software with their full range of functions. The usable range of functions will then be activated via the software, depending on the type and license - and of course this will result in completely new, exciting business models! Collect data, process it intelligently, optimize processes efficiently and make higher-level information platforms available: For modern smart factories, hybrid cloud architectures consisting of their own on-premise architecture with edge and fog computers and the use of a public cloud are often the ideal combination.
What are your experiences? Are your Edge computers already doing cloud tasks? We look forward to hearing from you!