软件可靠性增长模型的不确定性量化研究

发布时间：2018-04-28 02:26

本文选题：可靠性 + 不确定性　；参考：《软件学报》2017年07期

【摘要】：数量激增的软件系统被开发出来为用户提供了极大的便利,但也给系统开发带来了极大的不确定性.故障调试过程中的故障检测率(fault detection rate,简称FDR)是不规律变化的,且通常被描述为白色噪音.白色噪音具有马尔可夫性,但在实践中,噪音出现非马尔可夫性是普通现象,而噪音呈现马尔可夫性仅仅是例外.在许多真实情况下,白色噪声的理想化假设是不足的:真正的不规则因素总是非马尔可夫相关性的.提出了一个新的模型来量化调试过程相关的环境不确定性因素.基于广泛应用于软件故障检测过程的非齐次泊松过程(non-homogeneous Poisson process,简称NHPP)模型,将环境的不确定性考虑为任意分布和时间相关性的噪声.通过与一些现有模型的比较,新的框架表现出更接近实际观测数据的特征.除了常用的关注故障数的平均值以外,还提供了公式以计算其累积密度函数(CDF)和概率密度函数(PDF),从而获得调试过程的完整统计信息.
[Abstract]:The rapid increase in the number of software systems developed provides great convenience for users, but also brings great uncertainty to system development. Fault detection rate (FDR) in fault debugging is irregular and is usually described as white noise. White noise has Markov nature, but in practice, the appearance of non-Markov noise is a common phenomenon, and the noise appearance of Markov is only the exception. In many real cases, the idealized assumption of white noise is inadequate: true irregularity is always non-Markov correlation. A new model is proposed to quantify the environmental uncertainties associated with the debugging process. Based on the non-homogenous Poisson process (NHPP) model, which is widely used in software fault detection process, the uncertainty of environment is considered as the noise of arbitrary distribution and time dependence. Compared with some existing models, the new framework is more similar to the observed data. In addition to the commonly used mean values of the number of faults, formulas are provided to calculate the cumulative density function (CDFF) and the probability density function (PDF), so as to obtain the complete statistical information of the debugging process.
【作者单位】：北京航空航天大学计算机学院;
【基金】：航空科学基金(2013ZC51023)~~
【分类号】：TP311.53

【相似文献】