Research On User Clustering And Power Allocation In Multiuser NOMA Downlink

Non-orthogonal multiple access,one of the key technologies in the fifth-generation mobile communication system(5G),is one of solutions to achieve massive user access.In the power domain,the size of power is used to distinguish different users so that multiple user signals can reuse the same time-frequency resource block,that is,the same sub-channel for communication,which can greatly improve the spectrum utilization.But at the same time,a new problem is also introduced,that is,how to allocate resources in a limited time-frequency domain.Resource allocation mainly includes two aspects: one is user clustering,that is,how to select users superimposed on the same time-frequency resource block;the second is power allocation,that is,how to assign users superimposed on the same time-frequency resource block Perform power distribution.The performance indicators for measuring user clustering algorithms and power allocation algorithms usually include system transmission rate,user fairness,and user quality of service(QoS).The research work of this thesis mainly includes:In the study of user clustering:In the single-cell scenario,in view of the poor performance of the edge users in the existing clustering scheme based on the similarity of user channel gains,and the serious interference of the intermediate users in the clustering scheme based on the difference in user channel gains,a hybrid is proposed.The clustering scheme,the simulation results show that although the transmission rate of the intermediate users is slightly reduced,the scheme can improve the low throughput of the edge users and improve the fairness of users.In the multi-cell clustering algorithm,for the case of multiple users in each cluster,a user clustering scheme based on the k-means algorithm is proposed.The simulation proves that the clustering is achieved compared with the traditional kmeans algorithm.The number of iterations required by the resulting conditions is significantly reduced,which reduces the computational complexity.Compared with the traditional random clustering and channel gain-based schemes,it also has advantages in system performance.In the study of power allocation problem:based on user clustering,the problem is decoupled into power allocation between user clusters and power allocation within user clusters.In a single-cell scenario,a water-flooding iterative power allocation scheme is adopted between user clusters,and the remaining power is allocated in the cluster based on the constraints of user QoS with the goal of maximizing user fairness.In the multi-cell scenario,user fairness and user QoS are the optimization goals,and the maximum fairness in a single cluster is derived to the maximum fairness among multiple clusters.The maximum fairness between clusters is achieved by adjusting the iteration method,and the maximum fairness between clusters is guaranteed.User QoS is the goal,and the remaining power is distributed fairly.Simulation proves that this scheme can effectively guarantee user QoS and fairness.