Research On 5G Core Network Slicing Algorithm Based On Joint Optimization Of Resource Allocation And Routing

As one of the core technologies of 5G,network slicing is a technology that provides different business functions on the same physical network.Network slicing technology can realize the dynamic and flexible allocation of the underlying physical network resources,can improve the resource utilization rate of the entire network,improve the quality of service,and enhance the flexibility and elasticity of the network.The traditional network architecture can only provide a single service,and cannot achieve various customized service requirements.It cannot flexibly schedule various underlying physical network resources.The ITU has defined three major application scenarios for 5G,including enhanced mobile broadband(eMBB),ultra-reliable low-latency communication(uRLLC),and large-scale massive machine type communication(mMTC).If a dedicated physical network is constructed for each business scenario,it will inevitably lead to complexity and cost increase in network operation and maintenance.In order to adapt to the diversified business scenarios of 5G in the future,network slicing is a key technology to support multiple services on a physical network at the same time.Therefore,it is of great significance to study the routing and resource allocation algorithms in 5G network slicing.The main research works are made in the thesis as follows:(1)Addressing the issue of routing and resource allocation in 5G network slicing,SDN and NFV can effectively improve the flexibility of deploying and managing service function chains(SFCs).Therefore SDN and NFV is applied to the routing and resource optimization of network slicing,and a joint optimization algorithm for network slicing routing and resources based on SDN and NFV is proposed.First,according to the different business requirements of the three major application scenarios of 5G uRLLC,eMBB,and mMTC,the underlying physical nodes are divided into three virtual subnet node sets according to their function types.Then,according to the needs of different service function chains in different application scenarios,the mathematical model of mixed integer linear programming based on SDN and NFV network slice is formed,and a Lagrange based dual decomposition algorithm converts the mathematical model of network slicing into node and link sub-problems,and then solves the mapping scheme for the decomposed sub-problems.The simulation results show that compared with the previous algorithms,the algorithm proposed in this paper has better performance in terms of resource utilization,acceptance rate and average execution time.(2)In view of the traditional low-latency network slicing routing and resource allocation algorithms,the convergence speed is slow,difficult to perceive the dynamic network in real time,and cannot be performed dynamically and flexiblely.so the deep reinforcement learning algorithm is introduced in the routing and resource allocation in low-latency network slicing.In the process of mathematical modeling of routing and resource allocation of low-latency service function chains,for the deployment of dynamic and uncertain service function chains,real-time network slices that are being served in the physical network,and set node and link resource restrictions and service function chain delay restrictions in the model.Designed the reward value function for the service success of the service function chain,and proposed the algorithm of routing and resource allocation based on deep reinforcement learning.Compared with other algorithms,the algorithm of routing and resource allocation of deep reinforcement learning in this paper improves the deployment success rate of the service function chain,and reducing the average delay of network services.