Research On Key Technologies Of Dynamic Resource Trading For Wireless Network Slicing

The fifth-generation mobile network leverages network slicing to provide customized services such as enhanced mobile broadband,massive machine-type communications,and ultra-reliable low latency communications with service requirements such as gigabits per second data rate,extreme low latency in millisecond level and reliable mobile device connectivity.Although network slicing is a core concept of 5G networks,current deployments only support static resource slicing configurations rather than dynamic resource adjustments.That is,the study on dynamic resource slicing is in its early stages especially with the application of artificial intelligence methods.Secondly,from the perspective of the network slicing procedure,there is an urgent need new techno-economic models for slice operating,as there are few profitable 5G Io T slicing deployments available.Therefore,this dissertation focuses on resource trading for network slicing in the Beyond 5G,with emphasis on intelligence,security and trustworthiness.The core issue of this article is the dynamic trading transactions method for slicing the wireless network resources.Specifically,how to design resource trading models and pricing algorithms,as well as how to securely and reliably record all transactions and achieve consensus as soon as possible through incentive measures,require key breakthrough technologies and algorithms.Aiming at the resource transaction problem of network slicing,this paper proposes a dynamic game model,a federated decision-making model and algorithm under a hierarchical and distributed network architecture.This article proposes a distributed ledger and consensus algorithm based on blockchain for slice resource transactions to address issues such as secure and trustworthy recording of transactions,consensus,and incentives.Finally,simulations are used to validate the algorithm and to prove that it achieves desirable performance results.Specifically,the research results mainly include the following three aspects:(1)Focusing on the dynamic transaction of wireless network slice resources,and aiming at the two-layer network architecture composed of mobile virtual network operators(MVNOs)and end users,the pricing and purchase problem is expressed as a two-stage MLMF Stackelberg game.In this case,MVNO,as the leader,sets its unit price and user,as the follower,responds by determining its purchase volume.This dissertation theoretically proves the existence and uniqueness of Nash equilibrium(NE).In order to obtain the optimal dynamic pricing solution of MVNO,this dissertation transforms the optimization problem based on game into a stochastic Markov decision process(MDP)problem,and proposes a deep Q-Network(DQN)algorithm based on multi-agent competition.The simulation results show that the algorithm achieves convergence under competitive pricing scheme(CPS)and independent pricing scheme(IPS),and ensures high MVNO profit and user utility at an acceptable level.In terms of cumulative profits,the proposed competitive pricing algorithm based on Dueling-DQN outperforms Q-learning,premium pricing and under-cut pricing algorithms.(2)Focusing on the dynamic transaction of wireless network slice resources,and aiming at the three-layer network architecture composed of Infrastructure Provider(In P),MVNOs,and end users,a layered network slice resource transaction method based on federated strengthening is proposed.Different from the traditional business model,a hierarchy and intelligent resource transaction model is set up among a single In P,MVNOs,and end users.It emphasizes the dual-layer dynamic intelligent resource trading mechanism and model,which sets it apart from traditional commercial models by establishing a two-layer dynamic resource vertical trading model between infrastructure providers,MVNO,and end-users.The upper-level problem can be formulated as a Markov decision process,and a multi-agent federated reinforcement learning framework can be utilized to address the dynamic resource allocation problem between In P and MVNO.At a lower level,the transaction market between operators and Io T devices is modeled as a two-stage Stackelberg game.Operators set their unit prices and Io T devices set their purchase volumes.The running results of the algorithm in the simulation environment show that it can maximize the utility under different pricing schemes with high privacy protection,when the algorithm converges to the optimal solution.(3)In response to the credibility issue of network slicing resource transactions,this research focuses on transaction records,consensus,and incentive mechanisms based on distributed blockchain,which are necessary guarantee conditions for the correctness of transactions.Unlike the centralized third party(centralized database)method for recording transactions,which has potential security risks such as single point of failure and data tampering,decentralized blockchain provides a distributed ledger to record every transaction while supporting transparent transaction verification.This chapter focuses on studying the distributed consensus protocols and mechanisms necessary to guarantee the correctness of transactions.This dissertation considers the Unmanned Aerial Vehicle(UAV)as relay base station set up a network as a separate slice instance,which is operated and utilized by an Unmanned Aerial Vehicle Operator(UAVO)in a scenario where UAVO coexists with other MVNO slices.A new consensus method of network slicing is proposed to ensure the quality of service,security and privacy of real-time applications when taking UAV as relay base stations.This method selects mobile nodes in edge computing with practical experience as block verifiers.In order to calculate the experience of each edge node,this dissertation proposes a calculation model based on practical experiences considering the node historical consensus practice.In addition,an incentive scheme is proposed based on the Stackelberg game to encourage the honest cooperation nodes with mining rewards.We use a consortium blockchain in which mobile virtual network operators(MVNOs)sublet their idle spectrum to UAVOs.The simulation results show that the proposed scheme outperforms the baseline consensus scheme on throughput,delay and consensus time oberviously.