摘要
软件可靠性分析是根据软件失效数据等信息,通过合理建模来对软件可靠性进行预计和评价.现有的基于随机过程的可靠性模型一般采用均值过程来描述软件失效数据,然而,软件失效数据的模型化实质上应该是使其成为某个随机过程的一个样本轨迹.文中建立了考虑软件不同失效过程偏差的软件可靠性模型,用NHPP过程表示失效过程均值函数的变化趋势,ARMA过程表示实际失效过程对均值过程的偏差序列.在两组公开发表的真实数据集上对模型的实验表明,新模型较之一些广泛使用的NHPP软件可靠性模型在拟合能力及适用性上有明显的提高,并且保持了较好的预测能力.
The software reliability model is one of the important approaches to predict and evaluate software reliability quantitatively. The software failure data to be analyzed should be considered as a particular realization of a stochastic process. In this paper, a new software reliability growth model considering warps between different software failure processes is proposed. The experimental results based on two real data sets show that the proposed model has better prediction and curve fitting abilities than that of some other conditional NHPP software reliability growth models. The measures used for comparison are mean square of fitting error, predicted error and relative error.
出处
《计算机学报》
EI
CSCD
北大核心
2010年第7期1263-1271,共9页
Chinese Journal of Computers
基金
国家"八六三"高技术研究发展计划项目基金(2007AA01Z142)
上海市科学技术委员会信息技术领域重点科技攻关计划项目基金(04DZ15032
06DZ15003)资助~~
关键词
软件可靠性模型
样本轨迹
非齐次泊松过程模型
随机过程
自回归滑动平均过程
software reliability model
sample realization
nonhomogeneous poisson process （NHPP）
stochastic process
auto regressive moving average process
