Simulated Annealing for Aiding Genetic Algorithm in Software Architecture Synthesis

Automatic synthesis of software architecture has already been shown to be feasible with genetic algorithms. A natural problem is to augment – if not replace – genetic algorithms with some other search methods in the process of searching good architectures. The present paper studies the possibilities of using simulated annealing for synthesizing software architecture. We start from functional requirements which form a null architecture and consider three quality attributes, modifiability, efficiency and complexity. Synthesis is performed by adding design patterns and architecture styles; the end result being a software architecture which corresponds to the quality attributes. It is concluded that simulated annealing as such does not produce good architectures, but it is useful for speeding up the evolution process by quickly fine-tuning a base solution achieved with a genetic algorithm.