Date of Award

2016-01-01

Degree Name

Doctor of Philosophy

Department

Computer Science

Advisor(s)

Ann Q. Gates

Abstract

Critical software systems that have failed due to the software errors are well documented. As our dependency on computer-based systems increases and such systems become more complex, software verification becomes even more important. Enhancing and improving the verification and defect correction techniques used in software engineering for the development of software systems is of utmost importance to keep pace with our increasing reliance on software.

Pairwise testing has emerged as an effective technique for software system-level testing that have large combinations of inputs, although a drawback is the lack of support for defect location. This research aims to increase the efficacy of the pairwise testing technique through a new approach called Box-Fusion that combines the capabilities of structural source code analysis with the ease of pairwise test input generation. The Box-Fusion approach generated annotated control-flow graphs that can support the identification of untested code, location of faults, and selection of test cases for regression testing. To evaluate the Box-Fusion approach, a case study was conducted using two software implementations: the Simple LTL Generator that builds Linear Temporal Logic (LTL) formulae with atomic propositions and the Prospec Algorithm that can generate LTL formulae from more than 31,000 possible input combinations. The case study evaluates the following four propositions: the Box-Fusion approach can determine the areas of code that require additional verification; the code defects identified by the pairwise test are located in the execution paths identified by the approach; analyzing the annotated control flow graphs can narrow the location of the defect; and the regression tests can be optimized through the Box-Fusion approach. The case-study results demonstrate the value added by the Box-Fusion approach. The approach can be expanded to other black-box techniques and can be adapted to support educational models on software verification.

Language

en

Provenance

Received from ProQuest

File Size

132 pages

File Format

application/pdf

Rights Holder

Omar Ochoa Ochoa

Share

COinS