An error-detecting approach for fault tolerance parallel recomputing with parallel digital terrain analysis

摘要

In recent years, the research of parallel digital terrain analysis has become a hot spot. Using parallel computing technology to solve data-intensive computing problems has become a new trend in digital terrain analysis (DTA). On the other hand, with the development of hardware technology and new applications, how to ensure the reliability of the results of parallel computing is one of key problems. We can improve the system's ability to provide the right service by properly adopting fault tolerance technology. A parallel error-detecting approach based on parallel recomputing technology is presented and implemented by the combination of redundant process and multi-threads technology. Adopting a parallel comparison between the results of the process and the ones of its copy process, on the same data block, it can improve the efficiency of fault tolerance parallel recomputing. Furthermore, considering the relationship between the error-detection of the results and the recomputing, a modified scheme is proposed to make them to be executed concurrently. According to the error-detecting analysis of slope algorithm from DTA, it proves the effectiveness of the error-detecting approach based on fault-tolerant recomputing and achieves minor overhead.