Research On User Clustering And Hybrid Precoding With Mda In Massive MIMO System

With the development of human industrial science and technology,terminals such as smart phones are gradually popularized,and people's demand for communication is also increasing.In the new generation of mobile communication technology(5G)applications,massive MIMO systems are widely used.Under this system,serious interference is likely to occur due to co-channel transmission among multiple users.User clustering and precoding are effective means.In recent years,some scholars have used manifold learning methods such as local linear embedding to achieve channel dimensionality reduction.This paper studies the downlink of the massive MIMO system and proposes improvements to user clustering and hybrid precoding methods based on manifold discriminant analysis.The specific work is as follows:First of all,for the cluster geometry model of massive MIMO systems,in order to reduce the interference between multi-user and make each user in the community have stronger separability between clusters,we use manifold discriminant analysis technology to reduce the dimension of massive MIMO user channel matrix,making the channel matrix keep the global channel characteristics as much as possible,at the same time we introduce the projection measurement and accordingly put forward a more effective user clustering algorithm,compared with the traditional user clustering solutions,it can better reflect space angle relationship between the user and the base station,and the spectral efficiency of the system can be obtained with lower computational complexity.Then,in order to further reduce interference between clusters,which caused by the clusters overlap,and reduce interference within the cluster,this paper proposes a MMSE hybrid precoding algorithm based on manifold conjugate gradient method based on user clustering.The analog precoding design of spatial separation is used to balance the self-transmission and inter-cluster interference,and then the digital precoding is designed according to the equivalent channel to eliminate intra-cluster interference.The simulation results show that the proposed algorithm can eliminate the interference between user clusters to a certain extent and make the system have higher spectral efficiency.Finally,taking inter-cell interference into account,in order to suppress interference between adjacent cells,this paper proposes a three-layer hybrid precoding scheme.Based on channel dimensionality reduction and user clustering,a three-layer hybrid precoding scheme is proposed from the point of view of reducing signal leakage.The problem of multi-cell precoding is transformed into the optimization problem of multiple single-cell by forcing the interference subspace of this cell to zero.Simulation results show that this solution has better inter-cell interference suppression effect in the scenario of multi-cell cooperative precoding.