Different Approaches for QoS-Aware Web Services Composition Focused on E-Commerce Systems

• Pedro Felipe do Prado Universidade de São Paulo
• Luis H. V. Nakamura Universidade de São Paulo
• Julio C. Estrella Universidade de São Paulo
• Marcos J. Santana Universidade de São Paulo
• Regina H. C. Santana Universidade de São Paulo

Resumo

In this paper four different algorithms are proposed to solve the QoS-aware Web Services Composition (QWSC) problem in six different search-space sizes and a realistic deadline (a point not covered in many related works). Differently from some related works, statistical techniques are adopted in this paper in order to ensure more precise results from the algorithms. Two new algorithms are proposed: Greedy Heuristic and Double Hybrid Genetic Algorithm. The results of these algorithms were compared with an Exhaustive Search algorithm that always guarantees the global optima and a heuristic algorithm called Utility Function. The results obtained showed that the design of experiments and the performance evaluation can be used to determine which algorithms has better performance according to the different search-space sizes and the established deadline, it is also possible to determine which genetic operators are better suited for the QWSC problem.

Palavras-chave: Quality of service, Web services, Mercury (metals), Algorithm design and analysis, Heuristic algorithms, Search problems, Genetics