QoS-Based Web Service Composition by GA Using Consumer Decision-Making Function

There are various approaches for QoS-based web service composition using GA (Genetic Algorithm). Most of them use the sum of the values of QoS properties times their weights as the fitness function of GA, to evaluate the QoS of service compositions. However, by using this function, the service composition calculated by GA has shortcomings. For example, response time of the produced service composition is low, good for QoS, and availability is also low, bad for QoS sometimes. It influences users’ QoS experiences. The problem is the function just considers overall QoS while ignoring the equilibrium of QoS properties’ contributions to overall QoS. To deal with this problem, referring to the theory of customer behavior, we use the Cobb-Douglas function indicating users’ decision-making behavior as the fitness function in GA. In this way, the service composition not only has high overall QoS, but each QoS property’ contribution to overall QoS is more balanced. Keywordsweb service composition, QoS, GA, Cobb-Douglas function