Low-Complexity Nonlinear Hybrid Precoding Design: From Narrowband to Wideband Systems

Nonlinear hybrid precoding can approach capacity, but it is computationally demanding. This paper investigates low-complexity nonlinear hybrid precoding algorithms designed to maximize the sum-rate of massive multiple-input multiple-output broadcast channels (MIMO-BC). We propose a closed-form solution for analog precoding in narrowband systems, transforming the constant modulus constraint into an equivalent unconstrained problem with phase variables, which can be efficiently solved using the Fletcher-Reeves conjugate gradient method. Additionally, we introduce a nonlinear hybrid precoding technique for wideband systems, specifically tailored to delay-phase networks, which leverages channel characteristics to reduce computational complexity without relying on optimization. To further reduce the complexity of subcarrier-level digital precoding, we develop a resource block (RB)-level nonlinear precoding algorithm, where signals within each RB are independently precoded using a common precoder. Simulation results show that the proposed algorithms achieve near-optimal sum-rate performance while maintaining computational efficiency in both narrowband and wideband scenarios.