Hybrid beamforming for MU-MISO communication via deep unfolding
Abstract
In hybrid beamforming design using the conventional gradient projection (GP) algorithm, it is common to use a fixed step size, which results in a slow convergence rate and unsatisfactory achievable rate performance. This paper employs a deep unfolding algorithm within a small fixed number of iterations to tackle the hybrid beamforming optimization problem. The optimal step size is obtained by combining the conventional GP algorithm with the deep learning technique, and every step in deep learning is explainable. Simulation results show that the proposed deep unfolding algorithm demonstrates a lower computational time and superior achievable rate performance than the conventional GP algorithm.
Fields of Study & Subjects
Millimeter-Wave Propagation and Modeling
Advanced Wireless Communication Technologies
Advanced MIMO Systems Optimization
Publication details
- Journal / source
- China Communications
- Volume
- 23
- Issue
- 2
- Pages
- 260–267
- Publication date
- 1 February 2026
- Language
- EN