Neuro-Fuzzy Model to Estimate & Optimize Quality and Performance of Component Based Software Engineering

Abstract

Component Based Software Engineering (CBSE) constructs a quality software system by reusing existing components. For the construction of high-quality software system, reusability plays an important role. Software component should be designed and implemented in such a way that many different programs can reuse them. Reuse of software can increase the productivity and quality of software by reducing effort, time and cost which was elapsed in designing and developing reusable software component. In this paper, a Neuro-fuzzy model has been proposed that uses software component design patterns for analysis and Chidamber and Kemerer (CK) metric for evaluation, optimization and categorization of reusability for component based software. The work is divided into 2 phases. In the first phase, analysis and optimization of reusability are empirically evaluated with high precision value using CK metric and unsupervised Self Organizing Map (SOM) Neural Network. In the second phase, reusability is categorized as very low, low, medium, high and very high using a supervised Back propagation Neural Network (BPNN) and fuzzy inference rules applied on CK metric values. The proposed model may help a software designer to evaluate and optimize the reusability of components while designing software to make quality software system.