Software reliability modelling: a survey

The increasing dependence of society on systems containing software, and the growing criticality of the applications for which such systems are used, is leading to an increasing awareness of the need to make software reliable. It follows that it is necessary to have some means of measuring reliability in order to assess what has been achieved. Software reliability modelling is a set of statistical techniques for the estimation of the reliability of software. Although there are models which deal with the modular structure of software, the models described treat software as a `black-box´, which accepts inputs selected from a given `input space´ and maps these onto an `output space´. A failure occurs when a fault in the software leads to an incorrect mapping being performed, so that the behaviour of the system departs from what is required