Software Implemented Checkpointing Fault Tolerance In On-board Computer

On-board computers work in the complex outer space of high radiation, which cosmic rays, Van Allen radiation belt and other particles are important effects of the working state of on-board computers, in bad conditions these factors will lead to high occurrence of failure and errors. Therefore, the design of highly reliable on-board computer needs to provide hardware fault tolerance support and the software support. Nowadays on-board hardware fault-tolerance approach mostly refer to redundancy, implemented by design several redundant object. These redundant object perform the same task, and output the correct results through voting. Software approaches include software fault-tolerant redundancy and fault recovery techniques, among them checkpointing is an important technique.Double-mode redundant system can well meet the high reliability requirement of on-board computer. Double-mode redundant system involve the comparison of several computing nodes, making the intermediate result comparison and error handling become the key of double-mode redundant. In this paper we focus on software method, by setting voting points onto the critical task and comparing the intermediate result or the application data. Furthermore, we use checkpoint recovery method to recovery the system after failure, by fetching the data from the file system and replacing it with the current state.This study is design an checkpoiting fault-tolerance system based on double-mode redundant system, which selectμC/OS-II and MicroBlaze processer as resource. We use Xilinx Virtex 6 to build the hardware platform and develop fault-tolerant system supporting checkpoiting ,and verify the system's effective and reliable. And we develop fault-tolerance strategy in light ofμC/OS-II's embedded operator system features. By adding fault tolerance module toμC/OS-II kernel, we ensure the reliability of this double-mode redundant system.