The Evaluation of Reliability Based on the Software Architecture in Neural Networks

Software reliability is one of the key quality attributes. To avoid reworks after developing software, mentioned attribute of software must be evaluated correctly. Therefore, estimation and prediction of reliability during system development are very important. In this article a method to predict the failure probability in whole system by the neural networks has been presented. The model of this neural network varies based on the architectural style of software. An evaluation model has been implemented for one architectural style of software as a case study.