Improvements in reliability-growth modeling

The Duane model has been used for a number of years for graphically portraying reliability growth. There are, however, a number of inherent limitations which make it unsatisfactory for the regular monitoring of reliability growth progress. These limitations are explored and a new model is presented based on variance stabilizing transformation theory. This model retains the ease of use while avoiding the disadvantages of the Duane model. Computer simulations show that the new model provides a better fit to the data over the range of Duane slopes normally observed during a reliability growth program. Computer simulations also show that the new model provides higher values of instantaneous Mean Time Between Failures (MTBF) than the Duane model. Twelve published software reliability datasets are used to illustrate the use and application of the new model. The performance of both the Duane and the new model is compared for these software reliability datasets. The results of the performance comparison are consistent with the computer simulations.