User Clustering And Power Allocation In Non-Orthogonal Multiple Access Systems

Mobile communications have profoundly changed people's lives.In order to cope with the explosive growth of traffic,massive device connections and the emerging new business and new scenarios in the future,the millimetre-wave multiple input multiple output(mm Wave MIMO)mobile communication system based on non-orthogonal multiple access(NOMA)technology arises at the right moment.The highly directional channels of mm Wave MIMO systems facilitate the use of NOMA transmission for multiple users sharing the same beam but with different distances to the BS,thereby significantly increasing the number of user connections and system capacity.However,the decoding order of multiple users clustered over a group is determined by the channel gain and the beam gain.Thus,it is vital to cluster multiple users into different groups(where each group consists of highly correlated users),while optimizing the power allocation to maximize the system throughputs and energy efficiency.This thesis focuses on the user clustering and power allocation schemes in the downlink mm Wave MIMO-NOMA system,and the main research works of this thesis are as follows:1.The user clustering in static and dynamic scenarios are studied respectively.In the static scenario,considering the practical scenario where users are physically clustered,the correlation of mm Wave channels is utilized,and the machine learning framework based on affinity propagation(AP)is adopted to realize user clustering.In the dynamic scenario,considering the increasing number and the movement of users in the cluster,a semi-supervised learning algorithm based on simplified constrained seed k-means based user clustering algorithm is designed to reduce the computational complexity.The simulation results show that the performance of the AP based user clustering algorithm is significantly better than the traditional user clustering algorithms,and there is no need to specify the number of clusters,the simplified constrained seed k-means based user clustering algorithm achieves similar performance as AP based user reclustering algorithm,and its computational complexity is significantly lower than the AP based reclustering algorithm.2.In the case of user clustering,the power allocation of the downlink mm Wave MIMO-NOMA system includes inter-cluster and intra-cluster power allocation.It is difficult to jointly solve the inter-cluster and intra-cluster power allocation.Therefore,by fixing the inter-cluster power,a inter-cluster power allocation algorithm based on the grey wolf optimization(GWO)is designed.The simulation results show that the the designed algorithm can effectively improve the system sum rate compared to the FTPA algorithm.3.Energy efficiency optimization is very important for millimeter wave communication systems.In this thesis,considering the requirements of quality of service for users in the cluster and the priority of different users,the weighting sum energy efficiency optimization model for the users in the cluster is established.The power allocation coefficient based on weighting sum energy efficiency is solved iteratively by the successive convex approximation(SCA)method.The simulation results show that the weighted sum energy efficinecy based power allocation can not only improve the user weight sum energy efficiency for the users in the cluster,but also control the energy efficiency of a single user and ensure user fairness.