Development of Software Reliability Assessment Simulator and Its Calibration Method

为了测量大型软件系统的可靠性程度，迄今为止，许多研究者提出了许多软件可靠性评估模型。在这种情况下，我们开发了几个仿真模型来评估基于时间序列数据的软件可靠性。特别给出了软件可靠性模拟器中控制参数的标定方案。对控制参数进行了标定，以使实际观测数据与模拟器生成数据之间的平均误差平方和最小。给出了用该方案进行标定和软件可靠性评估的几个数值实例。我们确认这些结果符合我们的目标。

In order to measure the degree of reliability of large scale software systems, many software reliability assessment models have been proposed by many researchers so far. In such a situation, we have developed several simulation models to assess software reliability based on the time series data. Especially, we provide a calibration scheme of control parameters included in the software reliability simulator. The control parameters are calibrated so as to minimize the mean sum of squared errors between the actually observed data and the generated ones by the simulator. We showed several numerical examples of the calibration and software reliability assessment by the proposed scheme. We confirmed that these results met our aim.