Tools and techniques for testing-based software reliability estimation

Mei-Hwa Chen, Purdue University

Abstract

In this dissertation we point out some fundamental problems with the time-domain models which use the notion of "time" between failures or the number of faults within a certain time duration, and present our coverage enhanced time-domain reliability models. The coverage enhanced models extend the existing time-domain models by using the notion of useless testing effort. A key feature of our model, unlike the existing models, is that it takes into account the structure of the software under development. This feature not only distinguishes our model from all the existing reliability models, but it is also the basis of our claim that structure-based reliability models are likely to provide more accurate reliability estimates than the existing time-domain models. We have developed a new tool, T${\cal E}$RSE, to benchmark existing and new models. Its feature of displaying graphical results enables users to select a good model. The main difference between T${\cal E}$RSE and other existing tools is that it provides a rich source of data for investigating effects of varying model parameters on reliability estimates and offers a means through which users can evaluate new models.

Degree

Ph.D.

Advisors

Mathur, Purdue University.

Subject Area

Computer science

COinS