Research Of Digital Twin-Based Routing Optimization Technology In Software-Defined Network

With the expansion of the application scope of Software Defined Network(SDN),the network environment and communication requirements are becoming more complex and changeable.Traditional routing algorithms faces great challenges in ensuring the packet loss rate and delay of network traffic because of its less decision-making information and low flexibility.The Digital Twin enabled SDN has the real-time digital image of the network,which can more comprehensively perceive the state of the network environment and analyze network data information,and then adjust the routing strategy in real time according to the dynamic changes of the network environment,so as to improve the transmission quality of traffic in the complex network.However,the current Digital Twin construction process is usually based on complete full mapping,which will bring a lot of resource overhead to SDN.Especially when the network is busy,the overhead may even affect the normal operation of network services.Therefore,this thesis proposes a variable granularity digital twin construction technology.On this basis,the routing optimization technology based on Digital Twin is designed to improve the routing efficiency in complex network environment in a low-cost and highly adaptive way.The concrete research content are as follows.(1)In view of the problems faced by constructing Digital Twin in SDN,such as high latency requirements,high computational overhead,and difficult resource coordination,this thesis proposes a variable granularity Digital Twin(VGDT)idea and construction framework for SDN.In this thesis,based on the real-time load of the network,using the available communication and computing resources in the network,considering the delay and integrity of digital twin construction,a multi-node resource collaborative optimization model for VGDT is established.Then,this thesis uses a hybrid coding genetic algorithm to solve the optimization model,and obtained the optimal mapping data granularity and virtual twin placement location for Digital Twin under limited network resources.Simulation experiments show that,under the condition of limited network resources,compared with the existing model,VGDT has reduced the delay of digital twin construction and obtained higher model validity.(2)In view of at the problems of less decision-making information and low flexibility of current SDN routing algorithm,the SDN routing optimization technology based on VGDT is designed.Firstly,the technical framework of route optimization based on VGDT is designed,which mainly collects network data through VGDT,models the network environment,and provides more data support for routing decision.On this basis,VGDT assists the controller to realize network state perception,network data analysis and routing algorithm update,so as to improve routing efficiency in complex network environment.Then,this thesis designs a VGDT-DRL routing algorithm combined with deep reinforcement learning.This algorithm uses the historical network state information provided by VGDT to perceive the changes of network environment and realize the dynamic upgrade of routing strategy in complex network environment.At the same time,VGDT-DRL used asynchronous model updating and invalid action masking to improve the training efficiency.Simulation results show that under the limited network resources,the routing algorithm based on VGDT has low delay and packet loss.