A Feature-Driven Crossover Operator for Product Line Architecture Design Optimization

Abstract

The Product Line Architecture (PLA) design is a multi-objective optimization problem that can be properly solved in the Search Based Software Engineering (SBSE) field. However, the PLA design has specific characteristics. For example, the PLA is designed in terms of features and a highly modular PLA is necessary to enable the growth of a software product line. However, existing search based design approaches do not consider such needs. To overcome this limitation, this paper introduces a feature-driven crossover operator that aims at improving feature modularization. The proposed operator was applied in an empirical study using the multi-objective evolutionary algorithm named NSGAII. In comparison with another version of NSGAII that uses only mutation operators, the feature-driven crossover version found a greater diversity of solutions (potential PLA designs), with higher feature-based cohesion, and less feature scattering and tangling.

Search based software engineering for software product line engineering

“…Colanzi and Vergilio [30,31,32,33,34] formulate Product Line Architecture (PLA) optimisation as a multi-objective SBSE problem, focusing on architectural objectives such as extensibility and modularity. Guizzo et al [52,53] extend this work by using design patterns in the optimisation process.…”

OPLA-tool

“…The proposed operators are: MoveMethod, MoveAttribute, AddClass, MoveOperation and AddComponent. Furthermore, focused in feature modularization, the approach has the Feature-driven Operator [4] and Feature-driven Crossover [3]. MOEAs return several solutions called non-dominated solutions, which are solutions that dominate other solutions in at least one objective function and are not worse than the same solution in any other objective function.…”