In: J. Systems Software, 16, pages 121-128. 1991.
Abstract: Concurrent programming is witnessing a trend toward greater use of multiprocessing and distributed systems. However, the problem of reliability estimation for parallel programs has not yet been addressed. This article presents a methodology for software reliability estimation for parallel programs with modular structure. In deriving a model, the failure behavior of the parallel program is described by a stochastic Petri net. The reliability of the parallel program is calculated in terms of the relative frequencies of the failure events that are determined from simulation results of the stochastic Petri net model. An attractive feature of the proposed method is that the integration testing phase is not required for the estimation of reliability of the program. Therefore, the proposed method can be useful in an early phase of the software development process. An example is presented to illustrate the application of the proposed method.