Software Testability Model,Metric And Its Application

Testability is an important quality characteristic of software, it refers to the probability that a piece of software will fail on its next execution during testing (with a particular assumed input distribution) if this software includes a fault. Based on the research of testability, we can distribute test resource more reasonably, and reduce the complexity of software test, and further, we can use it to direct software designing, so as to enhance the reliability and the dependability of software to a new level.After the analysis of several testability models this paper finds that different models fit for different phases of software test. During system design phase, we can use DRR model to go on testability design. By redefining the cardinality of set, we establish the relationship between the domain/range ratio and the testability metric. PIE model can help us doing white box test in unit test phase. Based on the Fault/Failure model and conditional probability, a formula is found to get testability metric,with which we prove there is a max value in software testability metric. Through the research of the frontal two models, we present a new macroscopical testability model(M model) used in system test phase. By the intuitionistic conception, we define the testability metric with the ratio of ease faults to the total faults and predict the total faults in S curve. This paper also gives a limit model of test process, with which we can know how to calculate the reliability of software. By combination of the M model and the reliability, we can settle the problem of how to reassign the test resource and when to release the software.At the last of the paper a testability detect tool's frame is put forward. We also give a particular design of PIE model.