Near-capacity MIMO multiuser precoding with QRD-M algorithm

In this paper, we propose two extensions to the vector perturbation based Multiple Transmit Multiple Receive (MIMO) Multiuser (MU) precoder originally proposed in [1]. First, we propose a suboptimal QRDM (QR Decomposition-M) algorithm to solve the closest lattice point search problem in the vector perturbation based method. The QRD-M algorithm not only avoids the random complexity of the sphere encoder in [1], but also provides flexible tradeoff between performance and complexity. Second, we propose to jointly optimize the linear prefilters and nonlinear perturbation vectors by taking into account the impact of perturbation in the optimization process. In contrast, the prefilters used in [1] and [2] are optimized for the linear precoding structure, not the nonlinear vector perturbation structure. Finally, we verify the proposed QRD-M algorithm and the joint optimization algorithm with numerical simulations.