Symbol-Level Precoding for Average SER Minimization in Multiuser MISO Systems

TL;DR

This paper proposes a novel symbol-level precoding method for multiuser MISO systems that minimizes average SER using a double-space alternating optimization algorithm, outperforming existing schemes.

Contribution

It introduces a new DSAO algorithm for optimizing symbol-level precoding with a block transmission scheme for practical QAM demodulation.

Findings

Significant SER reduction compared to existing SLP schemes

Effective optimization of transmit signal and rescaling factor

Enhanced performance across full SNR range

Abstract

This paper investigates symbol-level precoding (SLP) for high-order quadrature amplitude modulation (QAM) aimed at minimizing the average symbol error rate (SER), leveraging both constructive interference (CI) and noise power to gain superiority in full signal-to-noise ratio (SNR) ranges. We first construct the SER expression with respect to the transmitted signal and the rescaling factor, based on which the problem of average SER minimization subject to total transmit power constraint is further formulated. Given the non-convex nature of the objective, solving the above problem becomes challenging. Due to the differences in constraints between the transmit signal and the rescaling factor, we propose the double-space alternating optimization (DSAO) algorithm to optimize the two variables on orthogonal Stiefel manifold and Euclidean spaces, respectively. To facilitate QAM demodulation instead of affording impractical signaling overhead, we further develop a block transmission scheme to keep the rescaling factor constant within a block. Simulation results demonstrate that the proposed SLP scheme exhibits a significant performance advantage over existing state-of-the-art SLP schemes.