Space-Time Vector Perturbation Precoding

Vector Perturbation Precoding (VPP) has gained recognition as a reduced-complexity method to achieve high-rates in multiuser multiple-input multiple-output (MIMO) downlink. Here we propose Space-Time VPP (STVPP), where the nonlinear lattice-based precoding operations of standard VPP are extended into the time domain. This is done by introducing temporal precoding lattices for each user, and performing the perturbation step at the transmitter for all users jointly. Although the transmitter complexity is increased, we show that by employing root lattices for the temporal precoding the computational complexity at each mobile receiver remains O(1). We provide new bounds on the required transmit energy scaling, as well as the sum rate. We show that the optimal sum rates are approached for moderate to high signal-to-noise ratio (SNR). Simulations show that significant gains can be achieved by precoding over even a few time instants, and that the proposed bounds are tight.