Developing Assessment Models for Software Reliability at Late Design Phase

Doctoral Thesis / Dissertation from the year 2021 in the subject Computer Science - Software, Savitribai Phule Pune University, formerly University of Pune (Department of Computer Science), course: Ph.D, language: English, abstract: In this thesis report the pertinence of the models (Neural Network model and Hellinger Net model) for better software reliability prediction considering the parameters and software metrics affecting the software design in a real environment is described. And a method of software defect detection and software reliability assessment using NN model and Intelligent Water Drop (IWD) Technique is presented. Built on a Neural Network (NN), two models are developed which predicts the software reliability in a more accurate manner. There are two kinds of hybrid models developed. One uses IWD with NN and another is IWD with Spiking Neural Network (SNN). For both, the modelling feature selection technique and learning algorithm is implemented and the data representation methods and some metrics associated with software reliability models are discussed. Various datasets containing metrics values with software failures are applied to the proposed models. These datasets are acquired from variety of software projects.