Resource Mapping For Sliced Mobile Communication Network

In recent years,with the rapid development of artificial intelligence and Internet of Things technology,a huge number of new mobile applications and new business scenarios have been spawned,posing more and higher requirements to 5G and B5G(Beyond 5G)systems,such as ultra-reliable low latency,massive Connections,etc.,and bringing unprecedented challenges to future mobile communication systems.In order to address these challenges,industry and academia are generally hoping for innovation in network architecture,where network slicing is considered to be the most crucial technological innovation in the next generation of mobile communication systems.Considering the difference of slice resource requirements and the resource characteristics of the underlying physical network,research on slice resource mapping in resource-constrained networks becomes a very challenging problem.The first part of this thesis studies the slice resource mapping problem in the access network environment where MEC is introduced.In order to reduce latency and the overhead of the backhaul link,it is expected that 5G will deploy a large number of MEC servers at the edge of the mobile network.On the basis of SDN/NFV and other technologies,the partial slicing network functionalities can be moved down to the mobile access network,thus improving the performance of slicing delay and capacity.Therefore,the access network slice mapping problem in the MEC environment must consider not only the resources of the base station but also the resources of the MEC server.In view of this,this thesis defines the access network slice resource mapping cost function,which takes into account the resources of the base station and the MEC severs.The optimization model is then constructed with the objective of minimizing the total cost of the slice map.Considering the NP-hard nature of the problem,this thesis uses a classical genetic algorithm to solve the problem.Finally,the performance gain of the scheme in terms of mapping cost,mapping failure rate,resource utilization,convergence,etc.is verified by computer simulation.The second part of this thesis studies the dynamic slice resource remapping problem in the MEC empowered access networks.Considering the characteristics of slice traffic changing continuously in its life cycle,this thesis studies the remapping of access network slice resources.Firstly,the service degradation function is introduced to describe the relationship between the slice service quality and the resource possession,and then the slice migration cost is introduced to describe the impact of slice migration on the quality of service(QoS).Based on these two factors,this thesis establishes an optimization model with the objective of minimizing the total service degradation in the process of resource remapping.Since the problem is indeed nonlinear mixed integer programming,it is very tough to directly solve it.Therefore,this thesis uses the dominant variable decision method,sequential quadratic programming,and simplex algorithm to solve it.Numerical results show that compared with the Semi-Dynamic and Remapping schemes,the proposed scheme can achieve a good tradeoff between service degradation and migration cost.The third part of the thesis studies mobile core network slice resource mapping based on virtualization technology.Taking into account the resource isolation characteristics of the slice,the core network slice is abstracted into a VNF chain,and then an optimization model aiming at minimizing the mapping cost is established.Then the model is decomposed into two parts: node computing resource and link bandwidth resource mapping respectively.The first part adopts dynamic programming method,and the second part uses minimum cost maximum flow algorithm to solve the original problem.Finally,the correctness and effectiveness of the proposed algorithm are verified by computer simulation.In this thesis,the new concept of mobile communication network slicing is studied,and the mobile access network and core network slicing resource mapping models are established,which provide some insights for future investigations of resource mapping problem of network slicing.