Musa-Okumoto 대수 포아송 실행시간 모형에 근거한 비용-신뢰성 최적정책

Cost-Reliability Optimal Policies Based on Musa-Okumoto Logarithmic Poisson Execution Time Model

It is of great practical interest to decide when to stop testing a software system in the development phase and transfer it to the user. This decision problemcalled an optimal software release one is discussed to specify the a, pp.opriate release time. In almost all studies, the software reliability models used are nonphomogenous Poisson process(NHPP) model with bounded mean value function. HNPP models with unbounded mean value function are more suitable in practice because of the possibility of introducing new faults when correcting or modifying the software. We discuss optimal software release policies which minimize a total average software cost under the constraint of satisfying a software reliability requirement. A numerical example illustrates the results.