Research On Non-cooperative Game Model Of Multi-user Collaboration In Cloud Market

Cloud providers and cloud users are the two most important types of participants in the cloud market.The relationship and behavior of the two is related to the operation of the cloud market.The cloud provider is more concerned about the profit issue,and he mainly focuses on revenue and cost.Therefore increasing resource utilization and minimizing costs are two ongoing battles for him.The cloud user wants to complete his requests with a lower price and better quality of service,so he concentrates on Price of Service(PoS)and Quality of Service(QoS).Thus how to reasonably request a strategy is an inevitable problem for cloud users.There is a need for an appropriate model and mechanism within cloud providers,within cloud users,as well as between the cloud provider and the cloud user to coordinate their relationships and behaviors.The paper establishes a cloud market model for cloud users to use cloud services collaboratively.The main work and contributions are as follows:The paper first optimizes the cloud composite service concurrent request strategy among multiple cloud users under single cloud provider.Since the cloud provider and the cloud user handle cloud composite services differently,so the paper decomposes the composite service request into atomic service requests,and then distributes them to the server of the cloud providers who undertake the corresponding atomic services.The paper designs a utility function about PoS and QoS for the cloud user's request strategy and tries to maximize the utility of the cloud user.Since the cloud user's utility interacts and depends on each other,and he can chose request strategy actively instead of passively,therefore the utility optimization problem is modeled as a non-cooperative game model.The existence of the Nash equilibrium solution of the game problem is proved by the equivalent variational inequality problem,and the solution is found by iterative proximal algorithm(IPA),where the cloud user can get the optimal request strategy under current situation.A series of experiments are conducted to verify that of the IPA algorithm can effectively improve the efficiency of cloud users and the resource utilization of the cloud provider.At the same time,the paper compare the impact of different parameters on the utility of the cloud provider and cloud users.The paper then researches on the cloud service concurrent request strategy among multiple cloud users under multiple cloud providers.Since cloud service market is no longer a monopoly model and cloud providers attract cloud users through different QoS and PoS level.Thus,the selection problem has become a many-to-many problem,and the original one-to-one or one-to-many work is no longer applicable.The paper design a request strategy for the cloud user that requires a selection on cloud providers as well as on time slots,and each cloud user expects to maximize his own utility.Therefore the utility optimization problem is modeled as a multi-user non-cooperative game model,and a Nash equilibrium solution of the formulated problem is theoretically and algorithmically found.The experimental results show that cloud users can be effectively improved and converged and the requested amount of cloud providers varies with the two dimensions of service price and service processing capability.