Modeling software testing costs and risks using fuzzy logic paradigm
Journal of Systems and Software
Hi-index | 0.00 |
This paper depicts Israel Aircraft Industry's (IAI's) real-life experience in measuring and optimizing system's quality costs and project duration associated with upgrading a transport helicopter's avionic system. Quality costs are composed of investments in systems appraisal as well as risk impacts and systems' failures costs. Our objective was to measure systems quality cost/times in a typical development project as well as suggest ways to optimize it in order to meet IAI's business objectives. Relevant data were gathered from historical records, a real-life “pilot project,” and domain experts. This data were aggregated and utilized to calibrate a Verification, Validation, and Testing (VVT) cost/time model for estimating key quality components. Optimized VVT strategies were generated intending to minimize the quality costs and/or project duration. Study findings indicate that over 60% of development cost was spent on quality. In addition, about 10% of the total development effort was spent on the Definition phase and, within that phase, only 1%–2% of the total development cost was invested in quality. Finally, the results of the optimization runs indicated a practical savings of about 15% in quality cost. Another optimized VVT strategy yielded over 20% improvement in project duration. Clearly our current VVT strategy is not optimal. First we should increase significantly the level of VVT performance during the Definition and Design phases. In addition, we should reduce the performance level of certain VVT activities which are not cost effective. © 2006 Wiley Periodicals, Inc. Syst Eng 9: 259–280, 2006