Research On 5G Network Slicing Resource Allocation Management Technology

Mobile communication is about to enter the 5G era,5G mobile communication network can not only provide customized services for users,but also meet people's various communication needs,which puts forward higher requirements for 5G itself.In order to meet the needs of 5G's diversified business capabilities,5G is ready to adopt network slicing technology to achieve this goal.Network slicing divides a single physical network into multiple logical virtual networks,each of which can assume part or all of the network functions independently and flexibly provide one or more network services according to the needs of users.Network slices isolated from each other,which improved the dynamic and flexibility of network services.To achieve network sectioning technology,there are many technical bottlenecks problems at present,including:1.As an end-to-end logical network,network slices must have different kinds of virtual node mapping overhead.How to reduce the overall overhead in the process of node mapping is one of the challenges.2.In 5G multi-application scenarios,data flow often increases dramatically.Virtual resources in network slices may not meet the changing needs of users.It is also worth studying how to take slice resource scheduling into account when creating slices.3.How to make the slice have the ability of "breathing" can dynamically adjust the service ability when providing services to users,and can expand and shrink and migrate the functions with the increase of user needs,so as not to cause service interruption,which is also a problem to be solved.In this paper,we focus on the cross-domain mapping of network slices,the creation stage and resource management during the service process.We rely on the national high-tech research and development plan("863" plan)"5G wireless network virtualization key technology research and verification" project.The main research results are as follows:1.A two-step node mapping strategy PSO-M(Particle Swarm Optimization-Minimum Spanning Tree)is proposed to solve the problem of virtual node mapping overhead in network slicing.Firstly,the nodes are distinguished critically from the important virtual nodes(Major Point)and the common virtual nodes(Common Point).Considering the bandwidth resource between M-M nodes and M-C nodes on the basis of resource bidding,the bandwidth resource overhead between M-M nodes is the main overhead of the whole node mapping process,so a node mapping algorithm strategy based on particle swarm optimization algorithm is proposed between M-M nodes,which can improve the search ability rapidly.Considering the low cost of M-C node mapping,a fast optimization algorithm based on minimizing spanning tree is proposed to shorten the mapping time of nodes in slices.Compared with the proposed virtual node mapping strategy in 5G,this strategy considers the bandwidth mapping overhead between two different virtual nodes in the node mapping phase,and focuses on the bandwidth resource overhead between M-M nodes,thus realizing the effective use of virtual network resources.Simulation results show that this strategy can achieve more optimized virtual node mapping process with less bandwidth mapping overhead and faster mapping speed.2.In order to optimize the resource utilization in the network slice creation stage,a dynamic resource scheduling strategy based on improved greedy algorithm is proposed.In the process of slice scheduling and user scheduling,a two layer slice-user model combined with the improved greedy algorithm is proposed,which ensures the slices with maximum service weight and the highest priority users combined together to complete corresponding services.The global resource optimal problem is transformed into a smaller sub problem by means of the optimal combination of each slice and every user,which improves the network resource utilization and the user experience in the meantime.The simulation results show,compared with existing algorithms,the proposed method could effective improve the system remaining resource utilization,system average QoE level and system throughput.3.In In order to flexibly adjust the service capability of network slices,a network slice function migration mechanism aiming at load balancing is proposed.The mechanism simulates the virtual machine into particles based on particle swarm optimization algorithm.In the process of migration,all particles were divided into several subgroups,and particle swarm optimization algorithm was applied within and among groups.According to the historical optimal solution and the current global optimal solution,the particle location is updated,and the best target particles were determined by selecting the smaller particle size of the particle in real time.The mechanism not only improves the convergence speed,but also improves the accuracy of the algorithm.Compared with other migration methods,the proposed migration mechanism has the advantages of high accuracy and fast convergence.And it can also improve the efficiency of resource utilization,reduce the energy consumption of data center.