Feng Liu

Low Complexity MMSE Vector Precoding Using Lattice Reduction for MIMO Systems

In this paper, a lattice-reduction-aided (LRA) minimum mean square error (MMSE) vector precoding (VP) is proposed for multiple input multiple output (MIMO) systems. Three schemes are provided for the perturbation vector design by Babais approximation procedures based on lattice reduction method to reduce the complexity. Performance and complexity analysis are provided. Simulation results show that the proposed schemes significantly outperform the conventional MMSE Tomlinson-Harashima precoding (THP) and the zero forcing (ZF) VP. Compared with the MMSE VP via closest-point search, our LRA approach provides a simple alternative with little performance loss.

Advanced Joint Transceiver Design for Block-Diagonal Geometric-Mean-Decomposition-Based Multiuser MIMO Systems

In this paper, we propose an advanced joint transceiver design for block-diagonal geometric-mean-decomposition (BD-GMD) based multiuser multiple-input-multiple-output (MIMO) systems. First, we use the lattice reduction (LR) method to design the BD-GMD-based advanced detection and precoding for an uplink and a downlink, respectively. Then, we exploit the vector perturbation (VP) technique to further improve the system performance of the multiuser downlink. To directly reduce the high complexity of VP by a sphere encoder, we provide an LR-based sphere encoder and LR-based approximations for perturbation symbols. Performance and complexity analyses are given to show the advantages of the proposed schemes. Particularly, the diversity-gain analysis shows some insights of the existing and proposed schemes. Simulation results verify the performance improvement as well as the theoretical analysis.

MMSE vector precoding with joint transmitter and receiver design for MIMO systems

The transmitter side (Tx) vector precoding (VP) shows excellent performance. However, joint transmitter and receiver (Tx-Rx) design might achieve further improvement. In this paper, we propose a novel joint Tx-Rx VP for multiple input multiple output (MIMO) systems. A unitary matrix is used at the receiver to partially equalize the channel. We design the joint Tx-Rx VP with the minimum mean square error (MMSE) criterion and deduce the general closed-form solution. Then we apply several methods including the singular value decomposition, QR decomposition and geometric mean decomposition (GMD) to specify the general joint Tx-Rx VP design. Moreover, by using the extended channel, we achieve an improved scheme to obtain further performance gain. Simulation results show that the specification with GMD method outperforms the other specifications and the Tx MMSE VP. The improved joint Tx-Rx MMSE VP with GMD is superior to other MMSE VP schemes.

Low complexity lattice reduction aided MMSE precoding design for MIMO systems

In this paper, we propose a low complexity lattice reduction aided (LRA) minimum mean square error (MMSE) precoding design for multiple input multiple output (MIMO) systems. We exploit the extended channel to implement the MMSE approach and obtain an optimal tradeoff between noise amplification and residual interference. Three schemes with linear and nonlinear processing are provided based on the lattice reduction method. Simulation results show the proposed schemes significantly outperform the conventional MMSE Tomlinson- Harashima precoding (THP) and the LRA zero-forcing (ZF) precoding. Interestingly, higher diversity order than the LRA ZF precoding is achieved. Moreover, our schemes perform not worse than the LRA MMSE vector precoding (VP) schemes.

Robust THP Transceiver Design for MIMO Interference Channel

This letter focuses on robust Tomlinson-Harashima precoding (THP) transceiver optimization for multiple-input multiple-output (MIMO) interference channel, where a bounded channel error model is assumed. Two optimization problems are formulated as minimizing the maximum per-user mean square error (MSE) and sum MSE with per-transmitter power constraint, respectively. Since both optimization problems are not convex, a sub-optimal alternating optimization method is proposed to iteratively update the transmit feedback matrix, precoding matrix and receive equalization matrix of THP transceiver until convergence. We show that the two problems of three matrix variables can be formulated as semidefinite program (SDP) when any two of three matrices are fixed, thus making the problems computationally tractable. Simulation results demonstrate that the proposed THP transceivers can improve the performance over the linear transceivers and are robust to the imperfect channel state information (CSI).

Robust Beamforming Design for Max–Min SINR in MIMO Interference Channels

This letter considers the max-min signal-to-interference-plus-noise ratio (SINR) problem for single stream multi-input multi-output (MIMO) interference channels with imperfect channel state information (CSI). Assuming bounded CSI error, we recast this problem as a worst-case fairness problem with linear matrix inequalities (LMI) constraints. Due to its nonconvexity, we propose an iterative algorithm in which the transmit and the receive beamformers are obtained alternately to solve the worst-case max-min fairness beamforming design problem. The beamformers generated by the proposed algorithm monotonically improve the min-SINR utility and guarantee to converge to a local optimal solution. Simulation results demonstrate the convergence behavior and the robustness of the proposed algorithm.

Non-linear transceiver design for MIMO interference channel with statistical CSI error

This study focuses on robust Tomlinson–Harashima precoding transceiver design for multi-cell multiple-input–multiple-output interference channel with statistical channel state information error. The objective is formulated to minimise the maximum per-user mean square error (MSE) with per-transmitter power constraint. Since the problem is not jointly convex, an approximated method is proposed. The authors optimise the receiver matrix to be minimum MSE form as well as derive the optimisation problems of transmitter matrix and feedback matrix to be second-order cone programming, respectively. By use of the results, the three matrices are updated iteratively. The convergence of the proposed method is also proved. Simulation results show that the proposed scheme outperforms the robust linear transceiver in both MSE and bit error ratio.