Comparison of two estimation methods of the mean time-interval between software failures

Effort-index-based software reliability growth models and performance assessment

“…[6][7][12][13] and was a simple modification of the NHPP to obtain an S-shaped growth curve for the cumulative number of failures detected. This model's software fault detection process can be viewed as a learning process that the software testers become familiar with the testing environments and tools as time progresses, these testers' skills gradually improve and then level off as the residual faults become more difficult to uncover [1,[6][7][12][13]. Because the original S-shaped model is for the analysis of fault isolation data, i.e.…”

“…Musa et al [2][3] and Ohba [6] showed that the effort index or the execution time is a better time domain for software reliability modeling than the calendar time because the shape of observed reliability growth curve depends strongly on the time distribution of the testing-effort. Recently, Yamada et al [10][11][12][13] and Huang et al [14][15][16] proposed a new and simple SRGM which describes the relationship among the calendar testing, the amount of testing-effort, and the number of software faults detected by testing. The test-effort index is measured by the number of CPU hours, the number of test runs, and so on.…”

“…significant test-effort is required, such as the number of test cases, human power, and CPU time. As usual, the test-effort during the testing phase and the time-dependent behavior of development effort in the software development process can be described by a Weibull-type consumption curve, see Yamada et al [10][11][12][13]. In fact, since actual testing-effort data represent various expenditure patterns, sometimes the testing-effort expenditures are difficult to be described by only an exponential or a Rayleigh curve.…”