Research on Service Composition Based on DQN in Multiple Clouds

Service organized all kinds of services to one for the user. In the cloud environment, the services and related QoSs (Quality of Services) in every cloud may be different. In this paper, how to compose those services together in the multiple clouds environment is the problem. We use a SCDQN (Service Composition based a Deep-Q-Network) method, and consider the execution time, privacy level, cost and availability of the service and the service request in the multiple clouds. Simulations show that SCDQN has a lower value in the execution time and cost while satisfying privacy level and availability requirement.