Effect of Manpower Deployment and Bug Generation on Software Error Models.
Interim rept. 1 Jul 75-Jun 76,
POLYTECHNIC INST OF NEW YORK BROOKLYN DEPT OF ELECTRICAL ENGINEERING AND ELECTROPHYSICS
Pagination or Media Count:
Early software error models by Shooman and Jelinski-Moranda related the number of errors in a large software systems to the rate of error removal. Expressions for the number of remaining errors as the software undergoes debugging were formulated and additional assumptions were made to relate the number of residual errors to the operational system reliability. One of the key assumptions of the above models was that the sum of the errors removed and those remaining in the program is constant. This report adds a major refinement to the above models by introducing the possibility of error generation during debugging. In this refinement the error generation terms are modeled in several different ways proportional to the number of detected errrors, corrected errors, the number of remaining errors, or some function of these effects. The correction rate is assumed to be a function of the manpower deployed on the project, thus permitting the use of the model to investigate optimum manpower deployment strategies. The effects on the economics of debugging due to error growth have also been analyzed. Author
- Computer Programming and Software