A Star-Chain Hybrid Decentralized Baseband Processing Architecture for XL-MIMO Systems

This paper proposes a novel star-chain hybrid decentralized baseband processing (DBP) architecture to enhance flexibility and efficiency in both uplink equalization and downlink precoding. Unlike conventional star or daisy-chain architectures, our hybrid approach delivers superior adaptability, robustness, and reduced fronthaul overhead while achieving higher performance. As a case study, we analyze downlink precoding under per-antenna power constraints, demonstrating how the hybrid DBP architecture enables distributed, spectrally efficient computation. Numerical results show a $50-70 \%$ reduction in fronthaul bandwidth requirements compared to centralized schemes while maintaining competitive sum-rate performance. These advantages make the proposed architecture a scalable and efficient solution for future large-scale MIMO systems.