Soft Computing-Based System for Performance Modeling of Object-Oriented Programming (OOP) Software

Abstract

Performance is a vital attribute for most software, making performance investigation an essential software engineering task. The problem is that modern applications are challenging to analyse for performance. The advancement of Software engineering has seen the application of scientific approach at various stages of Software Development Life Cycle (SDLC). New technique is available for implementation known as soft computing. This approach is a synergistic combination of artificial intelligence methodologies to model and solve real world problems that are either impossible or challenging to be modeled mathematically. The complexity, the commercial constraints and the expectation for high quality software demand, measuring the quality performance of an Object-Oriented Software (OOS). This work presents a set of testing metric (criteria) which hold the potential for ensuring remarkable improvements in quality performance of Object-Oriented software (OOS). A Prediction Model focused on identifying and evaluating the quality of the performance model based on the data set of software design components. The software design properties are comprehensive elucidated and integrating concepts of fuzzy logic and software engineering. The system enables the performance of modeling of Object-Oriented software using Fuzzy Inference System (FIS) to accurately predict the quality Object Oriented Software based on quality performance criteria such as reliability, efficiency, functionality and maintainability. Empirical validation of the results using checking data set is made to prove the usefulness of the design metrics and design quality attributes on the software prediction models. The purpose of this work is to provide development tool to solves practical problems, and focus on identifying and evaluating the quality performance of software design components. Again, the results demonstrate that, despite the research including approaches explicitly aimed at object-oriented software, there are substantial challenges in providing realistic feedback on the performance of large-scale object-oriented applications accordingly.