On Group-Level Precoding for Multi-Subcarrier MU-MIMO Systems

This paper investigates group-level multi-user precoding schemes for multi-antenna multi-subcarrier systems. Existing literature often presupposes an individual precoder for each tone, uniformly applying standard narrowband precoding techniques to each subcarrier. Such solutions, however, are unfeasible in practice due to unrealistic hardware cost and computational complexity. Motivated by realistic industrial protocol that allocates one common precoder for multiple subcarriers, this paper investigates a more viable sum-rate maximization paradigm employing group-level precoding. We first introduce two innovative schemes for assessing spectral efficiency of subcarrier blocks, that takes into account the fluctuating signal-to-interference-and-noise-ratios (SINRs) across all tones. Our formulation shows that these group-level precoding strategies present highly nonconvex challenges. To address these difficulties, we adopt successive convex approximation (SCA) methodology, which resolves the original nonconvex challenge via iteratively convexifying subproblems. Moreover, we devise low-complexity methods utilizing gradient projection, obviating the necessity for numerical solvers. Numerical experiments affirm the convergence and efficacy of our proposed algorithms. Notably, our 5G link-level simulations reveal that, compared to traditional methods, group-level precoding not only ensures a more uniform distribution of SINRs across subcarriers but also significantly improves throughput performance.