A STUDY ON PRACTICAL SOFTWARE RELIABILITY ASSESSMENT MODELING AND ITS APPLICATON TO PROJECT MANAGEMENT

软件可靠性评估对于有效地开发高质量的软件产品是非常重要的。本文对软件可靠性的定量度量和评估进行了较深入的研究。该技术是基于软件可靠性增长模型，其特征在于随机过程，如非齐次Poissn过程。通过使它们所基于的假设更加现实，这里讨论的odels被设计为描述软件开发测试阶段的软件故障检测过程或软件故障发生过程。在此基础上，进一步讨论了软件最优发布问题和测试过程控制问题.本文主要研究了以下十个方面的内容，并取得了一些有益的研究成果：（1）总结了现有的软件可靠性增长模型。(2)不完善调试环境下的软件可靠性评估。(3)基于Gompertz曲线的随机软件可靠性增长建模。(4)具有故障可检测性的软件可靠性增长模型。(5)基于随机微分方程的软件可靠性增长建模。(6)量化软件可用性。(7)推广了一种延迟S形软件可靠性增长模型。(8)基于测试工作量和初始故障内容先验信息的软件可靠性度量与评估。(9)预测最佳软件发布时间。(10)制定和解决最佳测试工作量分配问题。

Software reiability assessment is very important in developing a quality software product effciently. This research investigates the quantitative measurement and assessment of software reliability more practically than before. The techniques are based on the software reliability growth models which are characterized by stochastic processes such as a nonhomogeneous Poissn process. By making the assumptions on which they are based more realistic, the odels discussed here were designed to describe a software fault-detection process or a software failure-occurrence process during the testing phase of software development. Further, based on the above research results, optimal software release problem and testingprocess control were discussed. That is, we studied the following ten usbjects and culd obtain the useful research results :(1) Summarizing existing software reliability growth models.(2) Software reliability assessent in imperfect debugging environment.(3) Stochastic software reliability growth odeling by using a Gompertz curve.(4) Software reliability growth modeling with fault detectability.(5) Software reliability growth modeling by using stochastic differential equations.(6) Quantifying software availability.(7) Generalizing a delayd S-shaped software reliability growth model.(8) Software reliability measurement and assessment with testing-effort expenditures and prior-information on initial fault content.(9) Predicting optimal software release times.(10) Formulating and solving optimal testing-effort allocation problems.