This software reliability growth model tool is the only Frestimate component that is used exclusively during software testing or once there are observed failures to extrapolate current failure rate, MTBF, reliability and availability estimates into the future.
Reliability growth models are used late in the software development process. However, they are useful for validating predictions made earlier in the process. One common application of the reliability growth models is to establish "When To Stop" testing in order to meet a specific quantitative reliability objective. While predictor models use empirical data to forecast reliability, reliability growth models use actual defect data during a system level test.
The first step is to enter the number of observed software defects per day and the number of test/operation hours per day.
Next, Frestimate determines if the fault rate is increasing or decreasing. If it's decreasing then the models are able to estimate the reliability parameters. The total inherent defects is estimated by plotting the cumulative faults versus the cumulative fault rate. The y intercept of this graph is the estimated inherent defects. That estimate is used in several of the software reliability growth models. The slope of that graph as well as the x intercept are also used by the software reliability growth models. These are called the software reliability estimation parameters.
This shows the estimated software MTBF using various software reliability growth models.
This shows relative accuracy of each of the software reliability growth models for all data points as well as for a user selected number of recent data points.
This shows the software reliability estimation confidence bounds for each of the models for all data points as well as for a user selected number of recent data points.
This shows the estimated remaining software defects and time required to reach a reliability objective.