Research Of Hybrid Precoding Technology For Millimeter Wave Massive MIMO Systems

The design of hybrid precoding is a key technology for massive multi-input and multi-output(MIMO)millimeter-wave wireless communication,which has an important application prospect in the fifth generation(5G)mobile communication system.Machine learning is an important research direction in the field of artificial intelligence,and its theories and methods can be used to solve many complex problems in the field of information and communication technology.This thesis aims at solving the problems of fully/partially connected hybrid precoding method,and tries to use the latent factor model(LFM)and the improved artificial fish swarm(IAFS)algorithm in the field of machine learning to solve these problems separately.Firstly,the survey of three hybrid precoding methods,namely,fully connected,partially connected and hybridly connected structure in massive MIMO system were presented.In order to solve the problems such as the performance improvement of the massive MIMO hybrid precoding method based on orthogonal matching pursuit(OMP),the design and search of candidate codebook,a fully connected hybrid precoding method based on the LFM was proposed in this thesis.The proposed method utilized the LFM optimization idea,and used a hybrid precoding matrix(i.e.the product of digital and analog precoding matrix)to approximate the full-digital precoding matrix.The gradient descent method was used to design the hybrid precoding matrix by minimizing the Euclidean distance between the hybrid precoding matrix and the full-digital precoding matrix,which can avoid the design and search of candidate codebook in the OMP method.Next,the performance of fully connected structure for related massive MIMO precoding method was simulated with computer.The results show that the spectrum efficiency of the LFM method is better than that of the OMP method,and is closer to that of the full-digital precoding method.The energy efficiency performance of the LFM method exceeds that of the full-digital precoding method,and it is also better than that of the OMP method under the condition of a low number of radio frequency chains.In view of that massive MIMO hybrid precoding method based on successive interference cancelation(SIC)is suitable for the millimeter-wave massive MIMO system with the number of data streams equal to the number of radio frequency chains,and it has good energy efficiency.A more general partially connected hybrid precoding method based on IAFS algorithm with variable step size was proposed in this thesis.Its core idea was based on the characteristics of partially connected structure,converting the spectrum efficiency optimization based on matrix variables into the vector variables,and then the IAFS algorithm was used to solve the spectrum efficiency optimization problem.Moreover,the proposed method does not require the number of data streams to equal to the number of radio frequency chains.Then,the performance of partially connected structure for the related massive MIMO hybrid precoding method was simulated with computer.The results show that under the condition of low signal-to-noise ratio,the spectrum efficiency of IAFS method is close to that of the OMP method with fully connected structure,and its energy efficiency performance is better than that of the OMP method with fully connected structure and the SIC method with partially connected structure.Finally,the conclusion and the future works are presented.