Impact of Dynamic Metrics on Maintainability of System Using Fuzzy Logic Approach

Abstract

In software development life cycle, software maintainability is the most important and crucial phase as it requires more effort as compared to other phases. To minimize the cost of software maintenance, it is necessary to predict software maintainability during the early phases of software development life cycle. Most of the metrics available in existing literature to calculate maintainability factor were static, i.e., based on design patterns. Therefore, in this paper, authors propose a fuzzy logic model to measure maintainability of software that takes static and dynamic metrics set as input and maintainability as output. Static metrics were collected using CodeMR tool, i.e., an eclipse plugin whereas dynamic event tracing was done by AspectJ, an implementation of aspect oriented programming on an eclipse platform. Further, the proposed fuzzy model was applied on 15 java classes to calculate maintainability at compile time and run time based on static and dynamic metrics. Lastly, proposed model was validated using analytical hierarchy processing (AHP) technique and it is concluded that proposed fuzzy model gives satisfactory result in case of dynamic metrics. Hence, it is concluded that dynamic metrics are better predictor of maintainability factor as compared to static metrics that helps the software industry to measure maintainability of software early in advance.