Study On Precoding Techniques For Multi-User MIMO Broadcast Channels

In recent years, multiple-input multiple-output (MIMO) wireless systems have receivedconsiderable attentions due to their potential high spectral efficiency and large systemcapacity, and will be one of the key techniques in future mobile communication networks. Inthe future mobile networks based on MIMO, both base stations and mobile terminals willhave multiple antennas to transmit or receive radio signals. When the channel stateinformation is known in the transmitter (i.e. base station), precoding is an effective techniqueto achieve capacity limit of MIMO broadcast channels. And it can cancel interferencesbetween multiple users or multiple antennas with low complexity of users’ receiver. Soprecoding technique for multiple-user MIMO broadcast channels is explored in depth in thisdissertation, including linear, nonlinear precoding techniques and other precoding schemesbased on limited feedback. The author’s major contributions are outlined as follows:The way of controlling transmit power in the maximal signal-to-leakage-plus-noise-ratio(SLNR) precoding method for multiuser MIMO downlinks，is not efficient to ensure eachuser’s available SLNR value, so a precoding scheme pursuing the goal that minimizes totaltransmit power under each user’s SLNR constraint is proposed. The goal problems can besuccessfully solved by using semidefinite relaxation (SDR) techniques, and power constraintcondition added in goal problems can efficiently reduce total transmit power of the basestation. Simulation results show that the proposed scheme, which satisfies large SLNRthretholds, has better bit error rate (BER) performance and lower total transmit power than themaximal SLNR precoding method.Because SLNR of each user is determined by its own precoding matrix and isindependent of other users’, the optimal precoding matrices for multi-user MIMO downlinksare obtained by solving the optimization problem that minimizes the transmit power intendedfor a user subject to this user’s SLNR constraint. Using the semidefinite relaxation (SDR)technique, these problems can be reformulated into the semidefinite programming (SDP) andbe solved effectively. Simulation results show that proposed precoding scheme is quitefeasible when each user has two receive antennas, and it has better BER performance than theoriginal maximal-SLNR precoding scheme when SLNR of each user satisfies large thresholdvalue.When the base station knows imperfect channel state information (CSI) with limitedestimation errors, the robust precoding scheme based on signal-to-interference-plus-noise–ratio (SINR) is less feasible. In this dissertation, a precoding scheme based on SLNR isproposed for MIMO downlinks. Through solving a series of semidefinte programming problems, the proposed design minimizes total transmitted power under each user’s SLNRconstraint. At the same time, it can dynamically allocate each user’s SLNR thresholdaccording to their channel states, so it is more feasible than the similar SINR-based precodingscheme. Simulation results show that the proposed precoding scheme is robust to channelestimation errors, and can achieve optimal solutions with higher probability. Moreover, it hasbetter BER performance than similar SINR-based precoding.Tomlinson-Harashima precoding (THP) combined with receiver beamforming for themulti-user MIMO broadcast channel is proposed. When the base station and user terminals allhave multiple antennas, each user’s receiver maximizes the system sum rate with receivebeamforming, and the base station use THP based on zero forcing criteria to cancel multiuserinterferences (MUI). A mathematic expression of the theoretic sum rate of ZF-THP system atasymptotical high SNR is attained. Then two ordering optimization methods is analyzed,which reveals that proposed scheme has more sum rate than traditional ZF-THP at low andmedium SNR.Different from traditional zero forcing (ZF) algorithm for maximizing system sum rate, anew precoding method based on block diagonalization (BD) for multi-user MIMO downlinksis proposed via minimum mean square error (MMSE) criterion, which is called BD-MMSE.BD-MMSE has better BER performance than ZF algorithm. At the same time, it can achievesignificant part of sum rates of ZF systems.