Improved techniques for software testing based on markov chain usage models
Jesse H. Poore, Kirk Sayre · 1999
Markov chain usage models have been used successfully to aid in the testing of computer software. It is possible to extend the current Markov chain testing methodology. The additional results presented in this dissertation are: (1) The correct calculation for the probability of an arc appearing in a testing sequence. (2) The expected number of sequences to cover a state or an arc with associated variance. (3) A Bayesian single-use software reliability estimator. (4) A Bayesian single-action software reliability estimator. (5) A reliability estimator based on stratified sampling techniques. (6) A test of the convergence of the testing experience to the expected use of the software.