A novel approach for automatic remodularization of software systems using extended ant colony optimization algorithm

Context Software modularization is extremely important to streamline the inner structure of the program modules without influencing its core functionality. As the framework advances during the upkeep stage, the pristine design of the software package gets disintegrated and hence it is arduous to understand and maintain. There are many existing approaches being carried out to automatically remodularize using optimization techniques to ease the maintenance and improve the quality of the system. The outcomes are rather insufficiently optimal and depend on problem-specific operators, which in turn expands the time multifaceted nature to land at an answer. Apart from these limitations, the issues, such as time complexity, scalability and performance need to be addressed. Objective: In this paper, an efficient automatic software remodularization using extended Ant Colony Optimization (ACO) has been proposed to remodularize the software systems. Method: The proposed approach mainly includes two phases: optimised traversal of software system using ACO for finding the order of software files to be processed and remodularization of software system using the proposed approach of extended ACO. Results: We experimented our proposed approach on seven software systems. The performance is evaluated by using Turbo modularization quality (MQ) which supports Module dependency graph (MDG) that have edge weights. The time complexity of remodularized software system is evaluated based on number of Turbo MQ. Conclusion It can be concluded that when the performance has been compared with the subsisting methodologies, for example, Genetic algorithm (GA), Hill climbing (HC) and Interactive genetic algorithms (I-GM), the proposed approach has higher Turbo MQ value with lesser time complexity in the evaluated software systems.