Research On Clustering-Based Resource Allocation Scheme In Ultra Dense Networks

With the growing demand of high-speed data services,future mobile communication is facing higher demands.Ultra dense networks(UDN)is a key technology in 5G by deploying a large number of smallcell base stations(SBS)with low transmission power to achieve intensive deployment of small cells and enhance the degree of spatial reuse of resource.At the same time,as the transmission power is low,the converge is reduced,the distance between the SBS and the user is reduced,and the transmission loss is also greatly reduced,which improve the system energy efficiency.As the network density increases,the inter-cell interference gets more severe,which seriously affects the system performance.Effective resource allocation can eliminate interference and improve system performance,which is one of the key technologies in the research of ultra-dense networks.Based on the analysis of the characteristics of UDN,resource allocation schemes of ultra-dense networks are studied in this thesis.Aiming at the problems in the existing scheme,the corresponding improvement methods are put forward.The main works are as follows:(1)To reduce the co-tier interference among small cells in UDN,a clustering-assisted resource allocation(CARA)scheme is proposed for the scene where each SBS serves single user,aiming at improving the system throughout of the networks.Firstly,interference graph corresponding to all SBSs in the network is constructed.SBS with the highest degree is selected as the CH for the first cluster,and then the SBS which is closest to the CH is selected as cluster member in turn,while ensuring that the sum of interference weight in each cluster does not exceed the pre-set threshold.Other clusters are generated in the same way.Secondly,sub-channel allocation is performed by CH within each cluster,which is divided into two stages,initial sub-channel allocation stage and sub-channel compensation stage.Each user is assigned one or two sub-channels.Finally,power allocation is designed to maximize the intra-cluster throughput within each cluster.Simulation results indicate that the proposed scheme distributes the SBSs into each cluster more evenly,and significantly improves the system throughput compared with the existing schemes in the same scenario.(2)To further improve the practicality of the scheme,in the environment where each SBS serves multiple users,an interference-limited clustering and resource allocation(ILCRA)scheme is proposed.Firstly,SBSs are aggregated into clusters,and the sum weight of interference in each SBS cluster can not exceed the pre-set threshold.Secondly,user equipments(UE)are clustered by an exhaustive graph coloring algorithm in each SBS cluster.Then,in each SBS cluster,sub-channels are assigned according to the throughput of UE clusters on each sub-channel by CH.Finally,in each SBS cluster,water-filling algorithm is performed to allocate power to UEs.Simulation results show that UE average throughput and spectral efficiency of the proposed scheme are higher than the existing schemes in the same scenario.