摘要
针对软件多模块化发展的趋势以及当前软件可靠性测评把软件当成一个整块进行处理的缺点,本文提供了一种Markov控制转换多模块软件的可靠性测评方法。方法的主要思想包括建立基于软件体系结构的可靠性模型,并依据该模型把系统级的可靠性指标优化分配到模块级,实现在模块级对整个软件系统可靠性的监控,并在系统级的测评中进行确认。文中针对系统级测评缺少失效数据的情况,重点阐述了如何基于最大熵原理求解系统可靠性参数分布函数,以实现对系统可靠性指标的验证。
Due to the software multi-modules developing trend and the disadvantage of current software testing methods to treat the software system as a monolithic block, a reliability testing method is presented for the modular software with Markov transfer of control. The main idea of the testing method is that we can monitor and control the software reliability in the module level via allocating the reliability target of software system to the software modules based on the reliability model optimally,and validate the system reliability target by testing the software in system level again. According to the maximum entropy principle, the problem of how to estimate and demonstrate the system reliability with few failure data is solved emphatically.
出处
《计算机科学》
CSCD
北大核心
2008年第9期270-273,共4页
Computer Science
基金
上海市科技攻关项目(06DZ150003)
上海申通地铁集团课题(KY-07-025)
东华大学青年教师基金(112-10-0044056)资助
关键词
软件可靠性
可靠性测试
经验贝叶斯方法
最大熵原理
Software reliability,Reliability testing,Empirical bayesian statistics,Maximum entropy principle
作者简介
覃志东博士,讲师,研究方向为实时计算与可信性计算;
王洪亚博士,副教授,研究方向为实时计算;
李继云博士,副教授,主要研究方向为实时计算、图像处理;
乐嘉锦教授,博士生导师,主要研究领域为数据库、可信性计算。