The Research Of Digital Twin Assisted Intelligent Edge Cooperation And Resource Allocation

Services in B5G/6G network propose the high-reliability,low-latency,super-connectivity and super-bandwidth demands on the network.By deploying computing and storage resources at the network edge,multi-access edge computing(MEC)can provide users with stable and fast network service.At the same time,through the cooperation between edge nodes,reasonable allocation of computing,storage and communication resources in the cooperative system can effectively save network power overhead and reduce network response time.However,limited by the distributed deployment characteristics of MEC and the low ability of edge nodes,the realization of efficient edge cooperation and resource allocation is faced with problems such as the lack of real-time awareness of dynamic collaboration information and the difficulty of joint optimization of multi-dimensional resources.Digital twin(DT)provides solutions to the above problems.Digital twin network(DTN)constructed by DT has the characteristics of visualization,low-cost verification,real-time data perception,etc.DTN can provide a visual test platform for multi-dimensional resource allocation scheme.At the same time,the virtual and real interactive links in DTN can realize realtime perception of network resource information.However,the objective data deviation between DTN and physical network has an impact on the resource allocation gain.In addition,the deployment of digital twin service(DTS)provided by DTN also causes different physical network resource costs.As an early attempt of the integration of DT technology and mobile communication network,this thesis mainly studies how to solve the problem of optimal allocation of cooperative resources in mobile edge systems with the assistance of DTN,and discusses the impact of data deviation between DTN and physical networks on resource allocation.In addition,this thesis also studies the DTN-assisted joint optimization of DTS deployment and result delivery.The main innovations of the paper are as follows:(1)Focus on the problem that unreasonable network resource allocation caused by the lack of edge node cooperation information and insufficient equipment capability,which increases network power consumption and response time,a DTN assisted intelligent edge collaboration and resource optimization allocation solution is proposed.Firstly,in order to provide data security guarantee and high-quality communication links to users,a DTN assisted edge cooperative node selection scheme is proposed.Secondly,in order to save the power consumption of the edge system and reduce the time for task solving,a DTN assisted task offloading scheme is proposed in combination with computing resource allocation.Then,a system value function characterized by power consumption and network delay is formulated.A mathematical optimization model is constructed to minimize the system value function.Afterwards,the task offloading process is modeled as Markov decision process(MDP).An algorithm based on decision tree and double deep Q network is proposed to solve the optimization model.The experimental results show that the proposed scheme can achieve the goal of reducing power consumption and response time.(2)Focus on the problem that data deviation between DTN and physical network reduces the gain of collaborative resource allocation,which then increases the network response time,a resource optimization allocation scheme based on edge-edge collaboration is proposed while the influence of data deviation is fully considered.Firstly,the data deviation models between DTN and physical network in computing resources,storage resources and communication resources are established.Secondly,in the DTN supporting edge-edge collaboration architecture,a DTN-assisted edge intelligent collaboration scheme under the influence of data deviation is proposed.Then,the mathematical expression of the time spends in the task solving of the edge system is derived.In order to minimize the time consumption,the optimal allocation of computing resources and communication resources is jointly considered.The cooperative node selection and task distribution are jointly modeled as a mixed integer nonlinear programming problem.Thereafter,by modeling the task distribution process as MDP,the greedy algorithm and deep deterministic policy gradient(DDPG)algorithm are respectively used to realize the selection of collaboration nodes and the optimal allocation of resources.The simulation results show that the proposed scheme can effectively reduce the network response time.(3)Focus on the problem that DTS differential deployment results in lower resource utilization,which increases the backbone network traffic overhead and network response time,a resource allocation scheme based on edge-cloud collaboration for joint DTS deployment and service result delivery is proposed.Firstly,in the DTN network architecture based on edge-cloud collaboration,the cost of DTS creation,update,deployment and invocation is constructed with network traffic overhead and time consumption as performance indicators.Then,the cost function used to measure system performance is formulated.Thereafter,in order to reduce the value of the cost function,a mathematical optimization model is built for joint communication and computing resource allocation.By describing the allocation process of communication resources and computing resources as MDP,a learning algorithm including decision tree and DDPG method is proposed to solve the optimization model.Simulation results show that the proposed scheme is effective in reducing network traffic overhead and network response time.