Source：Micro-Electronics Magazine

Industrial edge computing continues to apply IT and data science technologies, and there is a trend of cooperation among OT, IT and data science teams. From the IT and OT perspective, the first task is to capture meaningful and high-quality data, which is then passed to data scientists for further analysis and processing, and ultimately used to optimize the plant's processes and schedules. In order to identify critical data, edge computing solution providers need to work closely with field personnel, interviewing plant managers and operators to understand the problems that the plant is trying to solve in order to assist customers in selecting critical data. In terms of information security, because of the frequent exchange of data between IT and OT systems, it is necessary to ensure security at the device and system level to avoid hacking into production line operations and staff safety.

IT/OT goes integrated

Traditional OT requires little computing power, as long as it satisfies the needs of production line stability and reliability. According to the observation of Wenzhong Wu, Associate Director of Open Instrument Business Center, IoT Strategic Solutions and Technology Division of ADLINK, under the development trend of Industry 4.0, edge computing has the capability of data filtering and processing, thus requiring the tandem and integration of IT and OT. With the introduction of IT's machine learning (ML) and data science technologies into the manufacturing field, large semiconductor and electronics manufacturers have started to establish data science teams internally, for example, for scheduling and process optimization within large semiconductor factories, there are professional teams to perform data processing and extraction.

The actual collaboration is more complex than performing data science work in the cloud in the IT domain, and must overcome challenges such as how to retrieve meaningful data from the OT side, or how to obtain good quality data.

In order to obtain richer production data, some factories not only retrieve data from the MES system of traditional machines, but also add more intelligent sensors to existing equipment to collect more information and provide data group processing and analysis. For example, the production work in semiconductor factories is easily affected by vibration, so factories use sensors to monitor the condition of floor slabs or use earthquake warning systems to predict the location and intensity of earthquakes to minimize the losses caused by earthquakes.

All data collection and analysis rely on the cooperation of IT, OT teams and data scientists. However, different roles have different professional languages, and the OT side needs to face heterogeneous data collected by different sensors, so team communication and data integration are challenges that industrial edge computing is facing.

Edge intelligence paves the way for Industry 4.0

At this stage, the factory is combining artificial intelligence (AI) software with the existing edge computing to establish factory-side edge intelligence. The first step of industrial edge computing is data collection, followed by data pre-processing, and now AI software is added at the edge to empower equipment through machine learning to achieve edge intelligence," said Eric Luo, director of iAT2000 Cloud Intelligence Product Division at NexAIoT.

When the data is collected and filtered at the edge, it needs to be uploaded to the cloud. When the data has an outlet for transmission, combined with the software capabilities, it can be used to further automate the entire factory from the cloud. Data filtering is one of the most important tasks in the process of data transfer in the factory. Edge computing solution providers need to understand the actual production conditions and the problems that customers want to solve in order to help them filter key data.

In addition to providing customers with the systems and equipment they need for edge computing, NexAIoT also emphasizes consulting services by interviewing factory managers and production line personnel to understand customers' needs for data collection/screening and immediate decision making at the manufacturing site. The factory wanted to check the PH of the water in the sink and use the change in PH to compare the amount of powder in the water to see if too much powder had fallen and the water needed to be changed. However, after interviewing, Luo Shi Yun found that if the factory wants to analyze the time to change the water, the actual indicator to be referred to should be the rate of water flow, because the more dust, the thicker the water will be, the slower the flow rate, and if the flow rate is too slow, then we know it is time to change the water.

Ensure device/system information security

When the system of industrial field goes to the integration of IT and OT, OT needs to find data from the IT system for analysis and comparison, and the data exchange brings information security risk at the same time. In industrial edge computing, the demand for information security is not only in the edge equipment in the field, but also when the data is collected at the edge, the data may be transferred to the external cloud or the factory's IT system, and the edge computer may need to find comparison data from the IT system, thus creating the demand for data exchange. This also means that the edge computing collects and analyzes data in a system that is not completely closed. Compared to traditional closed industrial control systems, contact with external networks exposes the IT system to the risk of virus and hacker infection. Therefore, IPC vendors will design corresponding information security products at the data transmission node to prevent hacking.

The information security risk in the manufacturing industry mainly comes from the equipment and network environment, and Mr. Lin Shih-Wei, the general manager of Taiwan in Moxa, explained that the focus of information security at the edge is to ensure that the production line is stable and the operators are working in a safe environment. The first is that once the system is connected to an external network, it may be subject to hacking attacks. Second, although the edge device may use a closed system and will not be exposed to information security risks, the operating system (OS) used by the edge computer may have OS vulnerabilities, and the information security risks brought by OS vulnerabilities may affect the production process.

The third source of risk is the storage device at the edge, for example, USB or files can be imported into the device, and the new files may carry viruses and cause device infection. The fourth is related to device security. In addition to ensuring system security, device security, including the design, production and actual use of the device, may have information security risks. Therefore, the international standard 62443 for network communication equipment information security, whose specifications cover a single device to the entire system, in addition to confirming system security, can also prevent a large number of devices in the factory to become information security vulnerabilities.

Production Expansion Drives Marginal Computing Development

Under the trend of manufacturing expansion, industrial edge computing has the opportunity to be adopted in large quantities and the penetration rate continues to rise. According to Wu's analysis, during 2021~2022, in the face of the COVID-19 epidemic, there will be a serious imbalance between supply and demand in the semiconductor and automotive markets, with a large amount of unmet demand in the market, resulting in massive production expansion in the semiconductor and automotive industries. At the same time, due to geopolitical influences, Taiwanese companies are returning from China and TSMC is setting up plants in the US/Europe, and the newly established production lines will definitely adopt more intelligent technologies. However, there is also a shortage of materials in the supply chain, so even if new factories and production lines wish to adopt industrial edge computing technology, there is still a shortage of materials and equipment, and the future development remains to be seen.

Overall, with the development of industrial edge computing, IT and OT are moving toward integration. Through the integration of heterogeneous data, factories are assisted by edge computing solution providers to understand the actual needs of the production site, and help factories to select and process critical production data for analysis by data science teams to optimize production efficiency with intelligent technology. The information security is controlled by the 62443 standard, from the device to the system, to ensure that the frequent data exchange and transmission of edge computing will not be exposed to information security risks, and to maintain the safe and stable operation of the production line.