QoS-aware Service Composition Using Fuzzy Set Theory and Genetic Algorithm

The quality of service (QoS)-aware service composition problem is a lively topic of debate because of the fuzziness in the quality data and the user-oriented specific QoS requirements. The aim of this paper is to develop a model to select the most suitable service composition in a way that maximizes solutions expressed as functions over fuzzy/crisp QoS attributes, while satisfying user’s QoS requirements. In this paper, based on fuzzy set theory (FST) and genetic algorithm (GA), a triangular fuzzy genetic algorithm (TGA) is proposed for solving the service composition problem. The following set of strategies are presented: a quality model including crisp and fuzzy properties represented by triangular fuzzy numbers, a feasible method of normalizing the QoS matrix, aggregating formulas of each control structure for eight properties, a practicable method of defuzzification, a global best strategy with a fitness function which calculates the QoS priority vector and is considered as an objective evaluation criterion for selecting an optimal solution that meets user’s preferences best. Empirical comparisons with two algorithms on different scales of composite service indicate that TGA is highly competitive in regards to search capability, especially when the problem size is large. The results may be helpful to designers in selecting the best services for building a service-oriented system.