Click the “blue words” above to follow us!
2025.09.13
Word count: 2798
Estimated reading time: 7 minutes
The leap in industrial production and industrial upgrading is closely dependent on the development of information technology, among which the key technology is embedded technology. This involves embedding computer systems into industrial products or manufacturing processes, providing industrial equipment and production systems with an intelligent brain, effectively enhancing the intelligence of equipment and production processes, and meeting the stringent requirements of industrial production for efficiency, reliability, real-time performance, and sustainability, ultimately achieving automation and intelligence of products or processes.
The overall process and trend of intelligent application development reflect characteristics of development from external to internal, and from vertical intelligence to horizontal intelligence. From the early days of a computing center, workstation, or server running management tools such as PDM, ERP, CRM (external intelligence, management tools), to technical tools like CAD, CAE, CAM, CAPP (external intelligence, management tools), moving into embedded computers (internal intelligence, achieving product intelligence), then relying on improvements in processor performance and algorithms (especially multi-core and edge computing) for vertical intelligence enhancement, and finally to the recent reliance on ubiquitous connectivity and cloud-edge collaboration for horizontal collaborative intelligence.
Such embedded systems have unique characteristics: they typically have stringent requirements for real-time performance, parallelism, reliability, and power consumption. On the other hand, with the rapid development of semiconductor technology, the trend in modern computer systems (including embedded systems) is the replacement of single-core processors with multi-core processors. At the same time, many industrial systems deploy multiple different applications on a single hardware platform, and these applications often have different criticalities, such as in embedded systems for aircraft and automobiles.
To build efficient and reliable systems on embedded platforms that meet different criticality requirements, designers often use virtualization technology to isolate the functions of different subsystems. Existing design and analysis technologies, as well as various platform tools, are still unable to effectively guide designers in efficiently and reliably building such systems on complex embedded platforms (especially in cloud, edge, and terminal collaborative platforms), and fully leverage the various heterogeneous computing resources and capabilities provided by multi-core and cloud-edge-terminal. At the same time, the integration of technologies such as cloud computing, the Internet of Things, artificial intelligence, and blockchain with industrial intelligence faces unprecedented challenges under the dual pressures of “complexity of functions and requirements” and “complexity of system architecture.” The development of embedded systems for industrial intelligence, as well as real-time operating systems for industrial applications, along with various middleware and platform tools surrounding these operating systems, provides great convenience for developing such systems, making real-time operating systems for industrial applications a major representative of industrial software.
From the previous analysis, it is clear that due to the limited resources of the hardware platform, a real-time operating system for industrial applications should support microkernel architecture and be customizable. From the perspective of real-time reliability required by industrial demands, it should support hard real-time capabilities to ensure safety (functional safety and resistance to attacks) and reliability (comprehensive fault tolerance).
At the same time, as an industrial embedded system, we hope it can boot quickly, and in the event of power loss or system crashes, it can quickly recover and return to the operational state prior to the issue. Therefore, Instant power on (quick boot) should also be a feature that a real-time operating system for industrial applications must possess.
On a technical level, research on real-time operating systems for industrial applications still faces many challenges. For example, under the trend of multi-core development, no simple and effective real-time scheduling algorithm has yet been found that can be easily implemented on operating systems as it is on single processors. There is also no simple and effective resource contention management protocol (such as priority inheritance, priority ceiling, etc.) similar to that on single-core systems. The fine-grained shared resources brought by multi-core systems have always posed a significant challenge for real-time systems that pursue predictability.
Modern industrial operating systems should not be limited to the closed embedded system requirements of a single product’s intelligence but should consider the new challenges brought by the industrial Internet environment under the trend of ubiquitous perception, ubiquitous connectivity, and ubiquitous intelligence in the integration of people, machines, and things. For instance, in a cloud, edge, and terminal collaborative environment, how to achieve optimized scheduling and arbitrary migration of tasks in edge terminals (a large number of dedicated embedded systems) where hardware and software are highly coupled, resources are limited, hardware is heterogeneous, operating systems vary or may not exist, and connectivity methods differ? Although there are many abstract theoretical studies, these issues have not been well addressed at the application level, especially in the context of industrial applications.
Another challenge comes from the industry’s entry barriers. Each industrial sector has its own specific industry requirements, necessitating good industry-oriented middleware or platform support; otherwise, a single operating system is unlikely to perform well across multiple industry applications.
Despite these significant challenges, we should also recognize the opportunities facing real-time operating systems for industrial applications. With the continuous deepening of industrial digitization and digital industrialization, the digital economy will usher in tremendous development opportunities. As the industrial Internet continues to promote the deep integration of the digital economy with the real economy, the characteristics of the digital economy have shifted from consumer connectivity to industrial connectivity, providing substantial application demands from the supply side.
At the same time, embedded systems used in industry (especially the embedded processors used) have characteristics that differ from consumer electronics like smartphones: they are specialized, pursuing stability and reliability. For a certain period, to ensure system stability and reliability, software and hardware remain relatively fixed and are not frequently updated for performance metrics (processing power and storage capacity). Although chip resources are limited, the bottleneck issues of high-end chips are relatively less pronounced. However, this does not mean that many industry-specific chips do not face bottleneck issues; it is just that overall, the updates of such systems and chips are relatively slow, which actually provides us with opportunities to study industrial embedded systems.
From the perspective of building an ecosystem for real-time operating systems for industrial applications, the following approaches can be referenced:
Inherit and be compatible with existing systems, considering that user adaptation to changes requires a process;
Rely on self-capabilities to support more chips and different hardware architectures (determined by the specialization and diversity of embedded processors), establish and rapidly update support for various industrial embedded chips, develop support for various industrial bus protocols and middleware, and create embedded system development platforms tailored to specific industries;
Open-source, relying on the open-source community to improve functionality, similar to Linux;
Offer the system for free to cultivate a large pool of technical talent and potential user groups familiar with the operating system;
Establish strategic partnerships with leading industry enterprises to enhance ecosystem development for specific applications;
Collaborate with educational institutions to establish joint laboratories and curriculum systems (similar to Intel‘s electronic design competitions, Google‘s competitions and curriculum systems), to cultivate technical talent and potential users starting from students.
The above viewpoints and opinions are intended to spark discussion. I hope industry peers and leading companies such as Huawei, Guoxin Chip Micro, and Dongtu Technology can work together to adapt to the current industrial development trends, seize the new opportunities for industrial development, and vigorously promote the research and development of domestic industrial software represented by industrial real-time operating systems and middleware, thereby building an independent and controllable industrial software system in China.
Deng Qingxu
Professor at the School of Computer Science and Engineering, Northeastern University, Vice Chairman of the Embedded Systems Committee of the Computer Society. His main research directions include real-time embedded systems, reconfigurable computing, and cyber-physical systems. He has led over 30 projects, including national 863 program projects, national natural science foundation projects, international cooperation funds, and national key research and development program topics in the fields of real-time embedded systems and the Internet of Things. His related achievements have been granted 14 invention patents, and he has received first-class provincial technology invention awards, first-class provincial science and technology progress awards, and second-class national science and technology progress awards. He has published over 140 papers in renowned domestic and international journals and conferences in the fields of real-time embedded systems, the Internet of Things, and reconfigurable computing. Several representative achievements have been published in top international conferences in related fields such as RTSS, RTAS, DAC, and in international journals such as ACM Transactions and IEEE Transactions. He has served multiple times as a program committee member for international conferences on embedded real-time systems and design automation systems such as RTSS and DAC.
(Author’s affiliation: School of Computer Science and Engineering, Northeastern University)
(This article is published with the authorization of the “Microcontroller and Embedded System Applications” magazine, originally published in the 2022 issue 1)
END
[Click the above👆🏻search term to view more content in the account]
Follow us for more exciting content
Share, like, and arrange to see?