A Novel Approach to Modularization Programming of Legacy Systems

The modularization of large legacy software systems has attracted a great deal of attention in recent years. Such modularization improves the maintenance and reuse of smaller, more manageable pieces of program source codes, and also provides insight into the overall structure of the software system. This paper develops a methodology to determine the modularization programming of a legacy software system. This methodology breaks programs into modules, or groups of related functions based on the observation that a module can be defined as a group of functions having high cohesion and low coupling. Furthermore, the determination of good alternatives to perform the desired attributes, with some fuzzy constraints, is crucial in building software systems. With known modules, a rule-based fuzzy representation of the module development problem is presented, while the tradeoff of the attributes of performance among modules is analyzed, using a fuzzy neural network approach. The approach taken to reach this solution is illustrated with simulation software.