AN ADAPTIVE CONCATENATED FAILURE RATE MODEL FOR SOFTWARE-RELIABILITY

This article introduces a software reliability model whose concatenate d failure rate function is motivated via considerations that reflect a n engineer's knowledge about the stochastic nature of software failure s. The model is adaptive (in a sense explained), has two parameters, a nd has characteristics that generalize those of existing models. A Bay esian approach for estimating the model parameters and fur testing hyp otheses about reliability growth is proposed. The prior distributions reflect structural considerations, and Markov chain Monte Carlo techni ques are used to implement the approach.