Statistical Precoder Design in Multi-User Systems via Graph Neural Networks and Generative Modeling

This letter proposes a GNN-based framework for statistical precoder design that leverages model-based insights to compactly represent statistical knowledge, resulting in efficient, lightweight architectures. The framework also supports approximate statistical information in FDD systems obtained through a GMM-based limited feedback scheme in massive MIMO systems with low pilot overhead. Simulations demonstrate the superiority of the proposed framework over baseline methods, including stochastic iterative algorithms and DFT codebook-based approaches, particularly in systems with low pilot overhead.