Research On Precoding Techniques In FD-MIMO Downlink Systems

As one of the key technologies of the fifth generation mobile communication technology,massive multi-input multi-output(MIMO)technology has attracted the attention of experts and scholars because of its great performance.In massive MIMO systems,the number of antennas is huge.In theoretical analysis,it is usually assumed that the antenna is arranged in a one-dimensional linear structure.However,in practical applications,it is difficult for a base station(BS)to deploy the large-scale antenna arrays because of the large space required by this structure.In order to overcome this weakness,both academia and industry have proposed a scheme of changing linear array antenna into two-dimensional array antenna structure.This system,which arrays large antenna arrays in a 2-D array structure,is called the full-dimension(FD)-MIMO system.FD-MIMO system can deploy more antennas in a limited BS space.Moreover,because of the 2D array structure,beam forming control can be carried out in both vertical and horizontal directions,making full use of the degree of freedom in space.In the downlink system of frequency-division duplexing(FDD)FD-MIMO,the following steps are generally used in the communication process:The first,the BS sends training sequences to the user.After receiving the training signal,the user estimates channel state information(CSI);The second,based on some codeword selection criteria,the user selects the codeword,which is closest to the estimated channel.The index of this codeword is fed back to the BS through the feedback link.The third,the BS searches the codeword in the codebook according to the index to get the codeword.Finally,the BS and users use the codeword to transmit data based on the systems above.This paper proposes a new precoding design scheme.The specific work includes the following two aspects.First,a two-stage sequential codebook design scheme is proposed.In this scheme,codeword consists of inner codeword and outer codeword.The inner codeword is used to approximate the radiation paths,while the outer codeword is used to make the suitable linear combination of the selected inner codewords.The simulation results show that the proposed method can effectively improve the system performance.Secondly,in the FDD downlink massive MIMO systems,in order to obtain accurate CSI,a large number of training times are needed to be spent.In order to reduce this training time,a dimension reduction method is proposed.By using this method,total of M antennas can be merged into N antenna units,so the length of training sequences can be reduced from M to N.Moreover,the system throughput can be improved.For given number M,the optimal length number N is calculated based on different coherence time,number of users and signal-to-noise ratio.The dimension reduction method is also applied to FDD FD-MIMO downlink systems,and its effectiveness is confirmed based on simulations.