Research On Network Slicing Resource Trading Algorithm Based On Competitive Game

With the development of fifth-generation mobile cellular network technology（5G）and cloud technology,diversified new services continue to emerge,and different industries,enterprises or individual users have put forward various service quality requirements for the network.5G networks need to be more flexible and dynamic to meet the diverse business needs of users and vertical industries.Therefore,network slicing technology came into being.Through network slicing,operators can build a virtualized,shared,and logically isolated network slice pool on a common physical network.Different slices can meet different service requirements of users.Users share the same physical network resources,but can enjoy different types of network services.In this resource-sharing network scenario,most existing researches focus on the technical aspects of resource scheduling and allocation.Researchers hope to maximize the benefits of operators by optimizing the resource composition of different slices.However,the economic business model of network slice resource transaction is rarely studied.In fact,network slicing has fundamentally reformed the interaction between different entities in the network resource trading market.For example,users can more accurately request the type of network service and pay the corresponding monetary compensation.Therefore,in the network resource market,resource transactions are attractive to all entities only when each entity can obtain benefits in a win-win situation.In this environment,research on resource transactions for network slicing is very important.Therefore,this thesis studies the economic business model in the wireless network slicing scenario.Specifically: First,this thesis designs a wireless network transaction system framework based on the Software Defined Network（SDN）architecture.Under this framework,this thesis introduces the consortium blockchain as the ledger of the transaction system,which ensures the security,fairness and privacy protection of network resource transactions.Then,this thesis focuses on the pricing and purchase interaction between multiple mobile virtual network operators（MVNOs）and users in network slicing resource transactions.We model this problem as a two-stage Stackelberg game to help both parties make trading decisions.We theoretically prove the existence of a Nash equilibrium for this model.At the same time,this thesis designs a multi-agent reinforcement learning algorithm based on deep reinforcement learning（DRL）to solve the optimal trading strategy of both parties in the game.Finally,several groups of experiments are carried out in the simulation environment.The simulation results of a single transaction show that the multi-agent algorithm proposed in this thesis can solve the Nash equilibrium of the game model.Then,through the comparison of multiple sets of simulation experiments,the two pricing schemes proposed in this thesis can create higher returns for both parties at the same time.Compared with the non-competitive pricing scheme,the competitive pricing scheme can enable consumers to purchase resources at a lower price and ensure that the MVNO can still have considerable benefits in the transaction.