Research On Collaboration-based Clustering And Resource Allocation Methods In Ultra-dense Networks

For ultra-dense networks,using base station clustering for resource allocation is a competitive scheme,because in this scenario,the control complexity of the system is greatly reduced,and the same frequency interference is effectively suppressed.However,in this case,there is no interference coordination between clusters,which makes interference estimation complicated and reduces resource utilization efficiency.The purpose of this thesis is to meet the needs of users as much as possible and improve the system throughput by using distributed algorithms while reducing the algorithm complexity and signaling overhead.Firstly,this thesis introduces the system architecture under the current mobile communication standard,leads to the ultra-dense network architecture,and analyzes the problems of complex interference and difficult resource scheduling in the ultra-dense network.After analyzing the main clustering methods and resource allocation methods of the system in the current ultra-dense network scenario,the main problems and shortcomings of the current algorithms are discussed.The ultra-dense network scenario used in this thesis is built,including modeling by homogeneous Poisson point process,hard core Poisson process and Poisson cluster process,constructing path transmission loss by using current standards,and establishing user model considering user needs.Then,the mathematical model and common algorithms of base station clustering and power allocation are introduced.Then,in order to reduce signaling overhead,improve inter-cluster frequency coordination and reduce cluster edge interference,the overlapping cluster algorithm is introduced and analyzed.Firstly,an overlapping cluster clustering algorithm based on Kmeans is proposed by introducing the concept of overlapping clusters,After that,Gosper curve is introduced to partition the base stations in the cluster.After analyzing the resource distribution in hot areas and non-hot areas,an interference suppression method based on Gosper curve is proposed and analyzed.The feasibility and complexity advantages that it will bring to the subsequent frequency and power allocation are analyzed.Simulation results show that the interference suppression algorithm based on Gosper curve can improve the spectrum utilization and downlink capacity of the system.At the same time,it shows that the overlapping cluster algorithm and the interference suppression algorithm based on Gosper curve can improve the system throughput and user satisfaction.Finally,the distributed frequency and power allocation strategy based on priority is studied.Firstly,the power allocation in ultra-dense networks is modeled mathematically,and its observability and controllability are analyzed.It is proved that its power allocation is equivalent to interference estimation.Based on this,an algorithm of interference estimation using game theory algorithm is proposed and its feasibility is verified.On this basis,combined with clustering algorithm,an interference estimation based on adjacent cell information is proposed,and a distributed resource allocation algorithm based on interference estimation is proposed,which has less complexity and less signaling overhead.Simulation results show that the convergence speed of distributed algorithm is faster than that of global algorithm,interference estimation using game theory algorithm can improve system throughput and user demand,and distributed algorithm can be effectively combined with clustering algorithm.