ISCAS OpenIR  > 天基综合信息系统全国重点实验室
软件实现的控制流容错技术研究
陆庄伟
Major计算机应用技术
Supervisor刘立祥
2011-05-31
Degree Grantor中国科学院研究生院
Degree Level硕士
Place of Degree Grantor北京
Keyword软件容错 控制流检测 故障注入 单粒子翻转
Abstract随着世界各国在空间探测领域内的深入研究和扩展实验,空间探测领域使用的计算机系统所需完成的任务越来越复杂。但在空间环境中,由于空间辐射的影响,计算机系统很容易发生硬件故障,这些故障给计算机系统的可靠性带来了严重的影响。传统上,主要以硬件冗余的方式来保证计算机系统的可靠性,如使用抗辐射部件等。然而,抗辐射部件的价格十分昂贵,其性能也要比商业中广泛使用的COTS部件(Commercial-Off-The-Shelf,商用货件)落后许多。因此,COTS部件可以代替抗辐射部件来构造空间探测领域内使用的计算机系统。只是COTS部件的容错能力不足,需要用软件容错技术加以辅助,以提高其容错能力。硬件故障可以分为瞬时故障和永久性故障两大类,其中绝大多数故障均为瞬时故障,又经统计,33%~77%的硬件瞬时故障都导致了计算机程序发生控制流错误,因此,本文主要针对面向硬件瞬时故障的软件实现的控制流容错技术进行了研究,并通过对几种较为典型的控制流容错技术的检测原理及其优缺点进行分析,提出了一种基于插入块的控制流错误检测方法CFCIBControl flow checking based on inserted blocks)。该方法主要基于插入块模型来完成对程序控制流图的优化,同时又利用减法来完成签名值的计算。相比其他几种较为典型的控制流容错技术,CFCIB方法既提高了错误检测能力,又降低了运行开销,是一种优秀的控制流错误检测方法。最后,本文利用故障注入实验对CFCIB方法的错误检测能力和运行开销进行了验证,实验结果表明:CFCIB方法的平均错误覆盖率高达96.78%,其平均空间和时间开销分别为45%80%; With deep research and extended experiment in space exploration, computer which is used in space exploration needs to take on more and more complex tasks. However, radiation can easily cause hardware faults in computer system, and these faults bring great impacts on the reliability of computer system.Traditionally, redundant components can guarantee the reliability of computer system, such as radiation hardened component. But their prices are very expensive and their performance lags far behind COTS (Commercial off The Shelf). COTS can instead of radiation hardened component to construct computer system in space exploration. COTS is limited in fault tolerance, software implemented fault tolerance can increase its ability in fault tolerance.Hardware fault includes transient fault and permanent fault, and transient fault occupies a large proportion of hardware fault, besides, 33~77 percents of transient faults cause control-flow error in computer program. This paper is focused on software implemented control-flow checking methods. This paper proposes a new based on inserted-block control-flow checking method CFCIB, after analyzing checking principle, advantages and disadvantages of a few typical control-flow checking methods. CFCIB uses inserted-block model to optimize control-flow graph of program and subtraction to calculate signature. Compared with other typical control-flow checking methods, CFCIB enhances error-detecting ability and reduce running costs, it’s an excellent control-flow checking method.Finally, this paper verifies error-detecting ability and running costs of CFCIB by fault injection experiment. Simulation result shows that the average value of fault detecting rate is 96.78%, the average space overhead is 45% and the average time overhead is 80%.
Subject计算机应用其他学科
Language中文
Content Type学位论文
URIhttp://ir.iscas.ac.cn/handle/311060/10825
Collection天基综合信息系统全国重点实验室
Recommended Citation
GB/T 7714
陆庄伟. 软件实现的控制流容错技术研究[D]. 北京. 中国科学院研究生院,2011.
Files in This Item:
File Name/Size DocType Version Access License
陆庄伟_软件实现的控制流容错技术研究.p(934KB) 开放获取LicenseApplication Full Text
Related Services
Recommend this item
Bookmark
Usage statistics
Export to Endnote
Google Scholar
Similar articles in Google Scholar
[陆庄伟]'s Articles
Baidu academic
Similar articles in Baidu academic
[陆庄伟]'s Articles
Bing Scholar
Similar articles in Bing Scholar
[陆庄伟]'s Articles
Terms of Use
No data!
Social Bookmark/Share
All comments (0)
No comment.
 

Items in the repository are protected by copyright, with all rights reserved, unless otherwise indicated.