Robust Symbol-Level Precoding via a Symbol-Perturbed Zero-Forcing Structure

The present work studies symbol-level precoding (SLP) for multiuser multi-input single-output (MISO) downlink channels and with imperfect channel state information (CSI) at the transmitter. SLP has gained significant interest in recent years because it is capable of enhancing symbol-level performance, such as symbol error probability (SEP), in a direct manner. Lately it has been shown that under a design formulation of SEP-constrained total power minimization, SLP can be regarded as a zero-forcing precoding scheme with appropriate symbol perturbations. That result is obtained under the perfect CSI assumption. In this paper we illustrate that, for the imperfect CSI case and under a worst-case SEP-constrained total power minimization formulation, SLP can also be regarded as a symbol-perturbed ZF scheme. Following such observation, we further establish an algorithm customized for our robust SLP formulation. Numerical results are presented to demonstrate the performance of our robust SLP design.