Research On Precoding In Multiple-User Multiple-Input Multiple-Output Systems

Mobile communication is faced with two major problems. One is how to provide high data transmission rate. The other is how to combat with multipath-induced fading. By introducing spatial dimension, multiple-input multiple-output (MIMO) technique delivers significant performance enhancements in terms of spectrum efficiency and link reliability without adding system bandwidth or transmission power. MIMO systems have been one of the most active research areas in the broad field of wireless communications. Further, MIMO has been already at the core of the fourth generation wireless communication technology.In MIMO systems, the transmitter obtains the channel state information (CSI) relying on the reciprocity principle or by the feedback from the receiver. Given the CSI, the transmitter precodes the symbols to simplify the receiver design and to improve the system performances. Precoder design is one of the key investigation issues in MIMO systems, and has been adopted by emerging wireless communication standard.Supported by IMF-Advanced Advanced-MIMO Research and Development in the item of Research, Development and Industrialization of IMT-Advanced in the National Key Science and Technology Project- New Generation Wideband Wireless Mobile Communication Network, this paper investigated precoding technique in Multiple-User MIMO system given perfect CSIT. The research is carried out aiming at how to increase the system capacity, how to meet the practical need and how to relax the requirement to the system antenna configuration. The paper's main contents and innovation are summarized as follows:1. Wide-sense stationary uncorrelated scattering MIMO channels are analyzed using Jake's model. The instantaneous mutual information and ergodic capacity is derived for MIMO channels with linear precoding based on outdated CSIT.2. Aiming at the problem that BD precoder distributes power strongly biased toward users in good channel conditions and SO algorithm achieves a comparatively low sum capacity, successive block diagonalization (SBD) precoding is proposed to solve the shortcomings mentioned above. SBD precoder defines user priority according to its channel state and desired rate. Users eliminate co-channel interference (CCI) using BD, without interfering the users with higher or equal priorities. Numerical results reveal that without adding complexity, SBD precoding algorithm meets the needs of the users fairly well, and achieves the fairness among users with the same priority.3. A generalized linear precoder and decoder design for MU-MIMO system broadcast channel is proposed under total power constraint. The precoder is designed with linear complexity, with no assumption to the noise distribution. As application scenarios vary, the precoder selects different weight matrix and subchannel dropping policy to optimize the performance. The proposed method well solved the problem of generalized design of linear precoder and decoder of MU-MIMO system under different application scenarios. Besides, the proposed twice water-filling algorithm further improved the practical capacity. The proposed vector index indicating subchannel-dropping order is designed with consideration of algorithm efficiency, as well as fairness among users.4. Aimed at the problems that the original proposed signal-to-leakage-and-noise ratio (SLNR) based precoder ignores the power distribution among users, and the SLNR is not very large, a novel power distribution method is proposed. Theoretical analysis and numerical simulations reveal that, through distributing the power according to the SLNR of subchannel, the proposed scheme further improve the total SLNR of the system, increases the system capacity and selects subchannels properly.5. A cross-layer linear precoder design of MU-MIMO broadcast channel is proposed based on convex optimization. The precoder takes network utility function as the objective function, which is decomposed to the optimization problem of physical layer and application layer through dual decomposition. The optimization problem of physical layer is transformed to convex optimization issue before each user's optimized precoding matrix is solved using the semi-definite programming software package. Theoretical analysis and numerical simulations reveal that the system performance is enhanced greatly through cross-layer optimization.