Volume 16, Number 1

Coupling Metrics for Aspect Oriented Software

  Authors

Kelvin Mutunga Katonyi, John Gichuki Ndia and Geoffrey Muchiri Muketha, Murang'a University of Technology, Kenya

  Abstract

The Aspect Oriented Software (AOS) paradigm emerged as a response to the limitations of ObjectOriented Programming, specifically its inability to modularize cross-cutting concerns effectively. However, AOS have inherent complexity that keeps increasing as software is modified and most of the existing metrics have not been theoretically or empirically validated. This means we cannot rely on them for measurement of AOS complexity. This paper proposes four base metrics and two composite coupling metrics for analyzing the complexity of AOS. The metrics were derived using the Entity-Attribute-MetricTool (EAMT) model. The metrics were theoretically validated using Briand’s framework, and a tool was developed to automate the computation of these metrics. Theoretical results indicate that the proposed metrics are mathematically sound. A between-subjects experimental study was conducted to validate the proposed metrics and results indicate that the proposed metrics are strongly correlated with modularity, meaning they are important for modularity assessment in AOS-based software.

  Keywords

Aspects, aspect-oriented systems, model modularity, software metrics.