ANFIS Approach for Optimal Selection of Reusable Components

In a growing world, the development of modern software system requires large-scale manpower, high development cost, larger completion time and high risk of maintaining the software quality. Component- Based Software Development (CBSD) approach is based on the concept of developing modern software systems by selecting the appropriate reusable components or COTS (Commercial Off-The-Shelf) components and then assembling them with well-defined software architecture. The proper selection of COTS components will really reduce the manpower, development cost, product completion time, risk, maintenance cost and also it addresses the high quality software product. In this paper, we develop an automated process of component selection by using Adaptive Neuro-Fuzzy Inference Systems (ANFIS) based technique by using 14 reusable components' parameters as a first time in this field. Again, for increasing the accuracy of a model, Fuzzy- Weighted-Relational-Coefficient (FWRC) matrix is derived between the components and CBS development with the help of 14 component parameters, namely, Reliability, Stability, Portability, Consistency, Completeness, Interface & Structural Complexity, Understandability of Software Documents, Security, Usability, Accuracy, Compatibility, Performance, Serviceability and Customizable. In the recent literature studies reveals that almost all the researchers have been designed a general fuzzy-design rule for a component selection problem of all kinds of software architecture; but it leads to a poor selection of components and this paper suggests adoption of a specific fuzzy-design rule for every software architecture application for the selection of reusable components. Finally, it is concluded that the selection of reusable components through ANFIS performs better than the other models discussed so far.