Research On Nonlinear Tomlinson-Harashima Precoding For MIMO System

With the development of wireless mobile communications,precoding technology has always been of high research value and engineering application value because of its advantages of improving system capacity,transmission quality and reducing receiver complexity.The system performance of precoding technology is closely related to the channel state information acquired by the sender.However,in actual communication environment,perfect channel state information is difficult to obtain,the precoding technique under imperfect channel state information is also an important research direction.At present,5G mobile communications will soon be commercial,but energy penalty effect of linear precoding technology will rapidly reduce system performance,nonlinear precoding algorithms become a research focus.This thesis researches THP nonlinear precoding techniques under perfect channel state information and imperfect channel state information,the main research work and results are as follows:1.MIMO system model is studied deeply and the sources of inter-stream interference and inter-user interference are analyzed,then introduce a channel estimation model based on mean feedback.THP nonlinear algorithm in SU-MIMO system is studied,and its feasibility in Massive MIMO and imperfect channel state information is analyzed.Simulation verification is carried out.It is pointed out that the THP algorithm is suitable for Massive MIMO systems;however,it has bit error rate platform effect under imperfect channel state information,which can be weakened by increasing the number of transmit antennas.2.Focused on precoding algorithms in MU-MIMO systems.Due to the energy penalty of linear BD precoding in Massive MIMO and the performance of THP algorithm in MU-MIMO is not ideal,this thesis proposes a new multi-user non-linear precoding algorithm GMD-BD-THP by the combination of THP algorithm and GMD-BD algorithm which is different from the existing BD-GMD-THP algorithm.The proposed algorithm performs non-linear optimization for the equivalent sub-channel after BD processing,so it can be applied in Massive MIMO systems and improve the system performance.3.GMD-BD-THP algorithm proposed in this thesis is further improved.In block diagonalization part,different interference matrices are constructed to divide the multiuser MIMO channel into equivalent parallel single-user sub-channels and eliminate initially the inter-stream interference in equivalent sub-channel at the same time.Then using THP precoding technique to eliminate the inter-stream interference in equivalent sub-channel secondly,proposing a reconstructed interference matrix to perform second optimization precoding algorithm RISO-BD-THP which is greatly improved the bit error rate performance.