A Study On Soft Reinforcement Techniques For The Critical Path Of The On-board Operating System

Nowadays, more and more embedded real-time systems are used in every walk of life, bringing great convenience for us. But these systems are becoming increasingly complex and usually have severe requirement on the non-functional characteristics, such as usability, reliability, security, et al. Meanwhile, because of time constraints, the design of real-time systems becomes a challenge. It can solve the problems above efficiently to use RTOS, with improving the development efficiency, reinforcing the system’s non-functional characteristics and simplifying the design of the system. Therefore, the RTOS is becoming more and more popular while developing an embedded system. As a typical embedded system, on-board applications suffer from the problems above much more seriously. So, the RTOS must be used widely in on-board applications as well.However, Single Error Effects(SEUs) induced by space high energy particles have a serious influence on the reliability of on-board applications. The transient error induced by SEUs may cause an error branch, a deadlock or an error calculation result in the program. Consequently, we must take measures to cope with SEUs. Comparing with using anti-radiation devices, the usage of the technology roadmap based on “COTS + soft reinforcement” can enjoy the advantages of high performance, low power consumption, low cost and fast development, becoming the current research focus. The soft reinforcement is a software fault tolerant technology. It makes use of software to implement hardware fault tolerance, implementing the error detection and recovery in the software level. However, the current study on soft reinforcement technology is mainly used in the application level, in spite of having gained a lot of research achievements. The study on soft reinforcement method for operating system is just beginning, and there are only very few research outcomes. So we intend to explore the soft reinforcement methods for operating system.Firstly, this paper illustrated the concept of fault tolerance and the classification of fault tolerant technology oriented transient errors. Then the paper gave a comprehensive summary of software implemented fault tolerant technology oriented transient errors. It also studied the characteristics of on-board operating system and the research status of operating system fault tolerance.Secondly, we studied the general functional mechanisms of the on-board operating system and selected an embedded real-time operating system SYS/BIOS as a specific object of our study. A comprehensive source code analysis was carried out in this paper, summing up the internal implementation of the mechanisms.On the basis of the previous study, this paper presented a soft reinforcement framework for on-board operating system. It included two mechanisms, internal detection and external detection, both of which comprised several soft reinforcement methods. The soft reinforcement methods in each mechanism coordinated organic unity. The two mechanisms within the framework could complement and cooperate with each other to achieve the soft reinforcement for the on-board operating system.Finally, the functional mechanisms of SYS/BIOS was fine customized so that it had minimal and self-contained set of functions. The soft reinforcement framework was implemented in the customized SYS/BIOS kernel by modifying the kernel layer source code, for recompiling and generating a soft reinforced operating system. At last, we verified the effect of the soft reinforcement through fault injection experiments, then analyzed and evaluated the experimental results.