Resource Optimization Scheme Based On User Clustering In Millimeter Wave Non-orthogonal Multiple Access System

In the future,wireless communication will have features such as ultra-high speed and ultra-high number of connections,which will bring huge challenges to the design and deployment of wireless communication systems.Combining with large-scale multiple input multiple output(MIMO)technology,the millimeter wave system with a hybrid precoding structure has been highly valued.Although the hybrid millimeter wave system has abundant spectrum resources,it is not clear which access method should be adopted to meet the needs of future communications.In order to further improve the spectrum efficiency of the system,the power domain non-orthogonal multiple access(NOMA)technology is gradually applied to the millimeter wave massive MIMO system.The combination of the two requires that each beam serves multiple users at the same time.However,there is interference of user signals between beams and within beams.At the same time,multiple users sharing a beam will also cause loss of system array gain.This thesis studies the maximization of the array gain and multiplexing gain as well as the suppression of signal interference under the millimeter-wave massive MIMO-NOMA system in terms of the technical advantages of user clustering,hybrid precoding and power allocation,in order to achieve high spectral efficiency.The main research work of the thesis is as follows:(1)In the multi-cluster and multi-user millimeter-wave massive MIMO-NOMA system,the optimal cluster head selection scheme is proposed for the problem of signal power loss in the cluster caused by the reciprocity of the uplink and downlink channels of non-cluster-head users.First of all,from the angle error in the uplink and downlink channels of millimeter wave users,the formula of signal power loss in the cluster is derived,and the reason for the loss of system array gain is analyzed.Then,in order to maximize the array gains,it is mathematically analyzed that when the central user in the cluster acts as the cluster head,the signal power loss in the cluster will be minimized.Finally,in order to maximize the spectrum efficiency of the system without loss of generality,the optimal power distribution is obtained by introducing the iterative optimization power distribution algorithm of minimum mean square error detection.The simulation results show that the proposed scheme has higher system spectrum efficiency than the scheme where each cluster of strong users is used as the cluster head.(2)In the multi-cluster and multi-user millimeter-wave massive MIMO-NOMA system,aiming at the problem of inter-cluster signal interference in user signals,a joint hybrid precoding technology user clustering scheme is designed.Specifically,the cluster head selection algorithm based on distance metric learning and the user grouping strategy of joint hybrid precoding are designed.First,the central cluster head set is found through the K-means clustering algorithm.According to the channel vector of the users in the central cluster head set,the analog precoding is designed to maximize the array gain.Then,by taking the product of the analog precoding matrix and the channel vector of each user as the equivalent channel of the user,grouping is performed based on the equivalent channel correlation.Finally,in order to further reduce the inter-cluster interference and maximize the multiplexing gain,digital precoding is designed based on the Zero Forcing(ZF)algorithm.The simulation shows that the spectral efficiency performance of the designed scheme is close to the ideal situation.that is,there is no inter-cluster interference.Compared with the latest clustering scheme based on strong users,this thesis has better spectrum efficiency performance.