Summary: This paper gives an overview of statistical models and methods for software reliability estimation. Most commonly used models for analyzing software failure data are classified into two groups and comparative features of those two groups of models are discussed. Difficulties in estimating the unknown parameters from standard debugging data are explained and an alternative experiment, called recapture debugging, which generates additional statistical information is described. Some methods for analyzing recapture debugging data are reviewed. It is suggested that further statistical research should be conducted not only on analysis of routinely collected data but also on the design of the debugging experiment with the goal of generating increased statistical information.
For the entire collection see [Zbl 0916.00007].