An economic model for cloud service composition based on user's preferences

Cloud computing services can be categorized as computing services and application services. The composition of cloud services indicates the combination of application and computing services at application level or computing level. Different economic models have been constructed to model end users long-term behaviors for cloud service composition. This paper proposed a quality of service based composition of cloud service approach. The problem of cloud service composition is modeled as a problem of multi-objective optimization (MOP). This work extends a previously developed algorithm, Non-dominated Sorting Genetic Algorithm (NSGA-II), to solve the service composition optimization problem. The proposed approach improved the crossover operator to adapt the algorithm convergence and used the adaptive crowding distance (ACD) parameter for a better distribution of the Pareto-optimal front. Results in different experiments show that the improved NSGA-II algorithm is better than the traditional one.