Research On Massive MIMO Pre-coding Algorithm Based On Amend RZF

With the increase of users,intra-cell interference becomes a serious problem in massive multiple input multiple output(MIMO)systems.In the down link,pre-coding transmit signal on the base station(BS)can solve this problem effectively.Pre-coding algorithm includes linear and nonlinear pre-coding.Regularied zero-forcing(RZF)pre-coding algorithm has the advantage of low complexity and good performance.However,with the increase of the number of transmitting antennas,the operation of matrix inversion in RZF causes high complexity.Therefore,it is a research hotspot to decrease the complexity of RZF pre-coding algorithm.The thesis focuses on studying modified RZF pre-coding algorithm with lower complexity.The main idea is finding an approximation solution to matrix inversion.This thesis studies on RZF pre-coding based on truncated-series algorithm.The thesis uses truncated Neumann-series,Taylor-series and Kapteyn-series respectively to operate matrix inversion of RZF and get low-complexity RZF pre-coding.Then,this thesis optimizes coefficient in truncated Kapteyn-series and get faster convergence rate on the base of the same complexity.Further,the thesis studies RZF pre-coding based on Newton-iteration algorithm and uses it to estimate matrix inversion.Then the thesis optimizes Newton-iteration algorithm in two aspects:initial value and convergence rate of iteration.At last,the thesis proposes a low-complexity pre-coding scheme in different situation.According to results of simulations,it can be concluded that(1)in the long channel coherence period,using RZF directly gets the best performance and the lowest complexity;(2)in the short channel coherence period,two-iteration improved Newton-iteration and two-order truncated Kapteyn-series with optimize coefficient get 98%performance of RZF with lower complexity.(3)in the rest of situations,one-iteration improved Newton-iteration and three-order truncated Neumann-series get 70%performance of RZF and Newton-iteration with lower complexity.