Precoding Algorithms For Downlink Multi-user MIMO System

By configuring multiple antennas at the transmitter and receiver,the multiple input multiple output(MIMO)techonology significantly increases spectral efficiency of the wireless comunnication system without additional power and bandwidth.However,the signals received by different users,in the multi-user MIMO system,are not only affected by interferences from different kind of noise among antennas,but also disturbed by the other users.This differs from the single-user MIMO system.To eliminate the multi-user interferences,the precoding strategy at the transmitters can be applied in the downlinks,which also avoids the difficulties of single inter-cooperative communiation among individual users.During these days,a precoding algorithm namely block diagonalization(BD)has been the most widely applied in the real multi-user MIMO systems.The main idea of the BD algorithm is to find the orthorgnal basis of the interference matrix's zero space,thus to completly resist the interferences.However,the traditional BD algorithm is usually with high computation complexity,and can not achieve the maximum transimit diversity order,which limits its performance and application scenarios.In the foreseeable future,the scale of the antannas will be increasing exponentianly,which brings huge challenges to the performance of traditional precoding algorithms,e.g.,the computation complexity,bit error rate,and tranmission rate.Since that,it should be considered ahead to apply previous precoding algorithms to the larger scale MIMO systems,and furtherly discovering on algorithms with better performance are also significant.This thesis concentrates on the design of low-complexity BD precoding algorithm and its application on multi-user MIMO systems.Firstly,the key elements in the multi-user MIMO system is summarized.Then,the precoding algorithms on single-user and multi-user MIMO systems are studied,and two low-complexity BD algorithms were obtained by combining the existing algorithms.After that,this thesis proposed two newly block diagonalization based precoding algorithms with lower computation complexity.As for the first one,it employs the Givens transformation based QR decomposition(QR-Givens)to the BD algorithm,so as to generate the orthorganal matrix.With regard to the second one,it is designed based on the lattice reduction(LR)strategy,as well as theQR-Givens in the first algorithm,which futherly decreases the computation complexity and improves the performance of system.Finally,experiments were designed with simulated multi-user MIMO system,and both of the proposed algorithms were applied to.The result reveals that compared with the traditional BD algorithm,the proposed algorithms not only achieve higher efficiencies with lower complexity and higher capacity,but also obtain better BER performance.