The Design of the Multi-user MIMO Nonlinear Precoding System

Abstract

This article mainly studied the non-linear multi-user MIMO system precoding technique in view of the precoding technology. On the basis of reviewing and summarizing predecessors' work, combining with the actual situation, it studied nonlinear precoding in TH (Tomlinson - Harashima) precoding. And it also proposed TH precoding scheme based on the quantitative channel state information (CSI) to achieve the average sum rate. And it proved that the nonlinear TH projected can achieve performance better than the linear force based on perfect CSI and quantitative CSI zero (ZF) precoding. In order to reduce the interference between the parallel data flow, we can use precoding matching transmission channel both in the transmitter data flow (precoding) and the processing of received signal (equality). Therefore, common widely used linear precoding scheme has low complexity based on forced to zero (ZF) and minimum mean square error (MMSE) criterion. Despite the very low complexity, linear precoding scheme also has capacity loss problems. And both in the transmitter or receiver, nonlinear precoding processing can provide alternative methods which can improve the performance of linear precoding. The implementation method of high complexity in practice includes linear precoding combined with decision feedback equalization (DFE) (1), vector perturbation (2), Tom Lin Senyuan island (TH) precoding (3) and ideal dirty paper coding (4) these four methods. Vector perturbation has put forward multi-user MIMO channel model which can achieve rate close to full capacity. It is a linear precoding technique with superior performance, but this approach requires joint selection which will be sent to all receivers. It is a multidimensional integer least-squares problem of grid signal vector perturbation, the complexity of its existence is much higher than TH precoding. It is a compromise which provide a good choice between performance and complexity, and received much attention in recent. Like many precoding scheme, the main problems of TH precoding system are availability of channel state information (CSI) in the transmitter. In frequency division duplex (FDD) system, the transmitter can't estimate the convey information and CSI from receiver to transmitter through feedback channel. This article will focus on the implementation of the FDD system TH precoding scheme, using the results of the study in the quantification of the CSI at the transmitter (5) widely used in the multi-input multi-output (MIMO) system. This paper designed quantitative CSI ZF precoding based on a multi-user space TH. Assuming that quantitative CDI is only in the transmitter, the calculation of feedforward filter and feedback filter is based on available only quantitative CDI on the transmitter side. In addition, the proposed scheme system settings in the system the number of user K can be less than or equal to transmitting antenna number ~ T N. It suggested to quantification the function of the number of feedback bits based on the CSI - TH precoding scheme to realize the analysis of the average rate of combined quantitative characterization of the sum of the rate of average loss rate of CSI for each user. The derived upper limit of TH precoding closely tracked the real rate of loss and convergence speed than the upper limit for the number of feedback bits in the ZF precoding.