Agent-Based Service Composition in Cloud Computing

In a Cloud-computing environment, consumers, brokers, and service providers interact to achieve their individual purposes. In this regard, service providers offer a pool of resources wrapped as web services, which should be composed by broker agents to provide a single virtualized service to Cloud consumers. In this study, an agent-based test bed for simulating Cloud-computing environments is developed. Each Cloud participant is represented by an agent, whose behavior is defined by means of colored Petri nets. The relationship between web services and service providers is modeled using object Petri nets. Both Petri net formalisms are combined to support a design methodology for defining concurrent and parallel service choreographies. This results in the creation of a dynamic agent-based service composition algorithm. The simulation results indicate that service composition is achieved with a linear time complexity despite dealing with interleaving choreographies and synchronization of heterogeneous services.