Modeling Allocation Problem for Software with Varied Levels of Fault Severity

The resource allocation problem considered in this chapter focuses on the objective of maximization of removal of fault content from a modular software. The allocation problem accounts the view wherein there is an ordering of the severity of the faults on continuity of levels varying from the lowest level to the highest level; lowest representing the easiest traceable and detectable fault in terms of time and resources; while the highest severity level indicating a long time gap and more resources for getting traced and rectified. For modeling the allocation problem, aid of Software Reliability Growth Models is taken. The proposed Software Reliability Growth model takes into consideration two important factors. First, the effect of time distribution of testing-resource function is considered as reliability growth curve depends strongly on this factor. Second, the growth model incorporates a novel idea of faults being categorized under varied levels (1-Level being simplest severity level and k-Level being hardest severity level) of fault severity. After the parameter estimation and analyses of the goodness of fit criterions; the proposed testing resource, varied severity fault modeling framework is used in resource allocation problem. The allocation problem is subjected to availability of resources and budget with an aspired level of reliability for each module. Also, for devising the optimal allocation problem, it is taken that the cost of removing the fault from each module is dependent on its severity. The formulated problem is a complex nonlinear programming problem and is solved by a meta-heuristic technique of genetic algorithm. Numerical illustrations are also taken in the chapter. Managerial implications, conclusion and limitations of the proposed allocation modeling are highlighted at the end of the chapter.