Software reliability modelling: achievements and limitations

An examination is made of the direct evaluation of the reliability of a software product from observation of its actual failure process during operation, using reliability growth modeling techniques described. Conclusions drawn from perfect working are examined. An assessment is also made of indirect sources of confidence in dependability. The author argues that one way forward might be to insist that certain safety-critical system should only be built if the safety case can be demonstrated to rely only on assurable levels of design dependability.<
>