Research On Feedback Precoding Scheme With Limited CSI For Massive MIMO System

With the rapid development of communication technology,the high-speed data services of wireless communication and the access requirements of mobile users have shown explosive growth.Massive MIMO technology has been widely used to improve system capacity and communication quality.However,precoding matrix dimension and algorithm complexity also increase rapidly with a large-scale increase in the number of antennas in the Massive MIMO system.Therefore,how to design a precoding feedback scheme for Massive MIMO system under the limitation of feedback bandwidth has important research significance.This paper researches on feedback precoding scheme with limited bandwidth for Massive MIMO system as follows:Firstly,Massive MIMO channel model and its properties are introduced.Linear precoding techniques and nonlinear precoding techniques in common use are summarized.The codebook design methods and codeword selection principles are analyzed.Secondly,not only polarization characteristics and channel models of polarized MIMO system but also existing codebook design schemes are analyzed.The traditional codeword models are designed for the ideal dual-polarization MIMO system,and the effects of polarization leakage are ignored.The influence of polarization leakage and channel characteristics of non-ideal dual-polarized MIMO system are studied.The non-ideal dual-polarization channel is modeled in consideration of polarization leakage.A channel feature model based on multipath parameters is designed.Finally,the paper researches on the feedback precoding scheme with limited CSI for non-ideal dual-polarized Massive MIMO system,and designs a codeword model which matches the channel characteristics.Based on this,two improved feedback methods are given:multi-component merging method and dual-codebook reconstruction method.The feedback method based on multi-component merging design three feedback models according to the characteristics of different parts.Model 1 represents the feedback scheme of the traditional ideal channel model,while Model 2 and Model 3 consider the polarization leakage,and show improvement on information loss compared to the model 1.The feedback method based on dual-codebook reconstruction is based on the obvious features of three sub-matrices decomposed by the channel matrix.Two codebooks are designed for weighted combination and both codebooks have codeword models matched similar block diagonal structure,to maximize the use of channel space features.Finally,two improved feedback methods are simulated with MATLAB,string distance and bit error rate are used as performance indicators for simulation analysis.The simulation results show that two improved schemes proposed in this paper can achieve a good compromise between the overall performance and the feedback overhead.