Reliability Analysis of Component Software Based on Testing Data Transformation

We propose an approach of component-based software reliability analysis which includes the benefits of both time domain, and structure based approaches. This approach overcomes the deficiency of existing black box and white box techniques that fall short of addressing repair, and internal system structures simultaneously. Our solution adopts a method of testing data transformation to cover both methods, and is expected to improve reliability prediction. This paradigm starts with consideration of component-based software testing process in view of the assumption of NHPP models. It accounts for software structures by the way of modeling the testing process. According to the testing model it builds the mapping relation from the testing profile to the operational profile of component-based software which enables the testing data transformation to build the reliability dataset required by non-homogeneous Poisson process (NHPP) models. At last an example is evaluated to validate and show the effectiveness of this approach.