Combinations of antipattern heuristics in software architecture optimization for embedded systems
Conference contribution, 2013
A large number of quality properties need to be addressed in nowadays complex embedded systems by architects. Evolutionary algorithms can help architects to find optimal solutions which meet these conicting quality attributes. Also, architectural patterns and antipatterns give the architect knowledge of solving design bottlenecks. Hence, antipatterns heuristics have been used as domain-specific search operators within the evolutionary optimization. However, these heuristics usually improve only one quality attribute and using them in multiobjective problem is challenging. This paper studies the extent to which heuristic-based search operators can improve multiobjective optimization of software architecture for embedded systems. It compares various combinations of heuristic-based operators in a real world automotive system case study.
Evolutionary multiobjective optimization (EMO)
Embedded system architecture design optimization
Domain-specific search operators