In this thesis, software release models under increasing correction cost of an error are considered. It is assumed that correction cost of an error is exponentially or polynomially increasing with test time. Underlying software growth models are Goel-Okumoto NHPP model or S-shaped reliability growth model. Four software release models are considered by the combination of cost assumptions and underlying growth models.
Total expected cost on the software life time is used as a criterion for optimization. Optimum software release time is determined by minimization of a total expected software cost. Existence and uniqueness of the optimal solution in each model are shown. Additionally, optimal software release policies minimizing a total expected software cost under the constraint of a software reliability requirement are discussed. The effects of unit cost parameters on the optimum solution are studied by numerical examples.
