Abstract:

A new software reliability model (SRM) which combines maintenance policy and failure correlation is proposed to resolve the problem of current SRM. According to the characteristic of maintenance policy and six mode of state conversion, Markov renewal process of three class points including successful run and failed run as well as version update is used to establish the SRM respectively in discrete time and in continuous time. On this basis, a layer idea is introduced into the model: modeling dependent successive software runs at each layer and maintenance policy between two layers. Finally , some important reliability measurements including the distribution function of the time to failure, software reliability etc are concluded by applying this model. Through model comparison and analysis, the intrinsic consistency between the proposed model and other SRMs is validated, and the model itself can be also considered as a more flexible SRM framework.

Key words: maintenance policy, failure correlation, Markov renewal process, software reliability model