US20250330221
2025-10-23
Electricity
H04B7/0619
The patent application discusses a system designed to enhance uplink performance in Massive Multiple-Input-Multiple-Output (MIMO) wireless networks using a split Radio Access Network (RAN) configuration. This system includes a distributed unit (DU) that generates sounding reference signal (SRS)-based beams and a remote unit (RU) that generates demodulation reference signal (DMRS)-based beams. The RU performs digital beamforming by combining these beams to optimize uplink communication.
The focus is on improving combined beamforming, channel information transfer, and reference signal compression in massive MIMO uplinks within wireless RANs using a split fronthaul architecture. This configuration allows for the deployment of lightweight radio units across large areas, where processing is centralized, often in cloud-based systems.
In split fronthaul systems, the DU designs beams from SRS data before the RU receives signals, leading to potential beam aging issues due to channel changes over time. The invention addresses this by allowing the RU to generate up-to-date DMRS-based beams, which are then combined with SRS-based beams from the DU. This approach leverages both the freshness of DMRS data and the noise-reduced processing of SRS data.
The system can integrate artificial intelligence (AI) and machine learning (ML) to dynamically adjust beamforming responsibilities between the DU and RU based on various performance indicators. Factors such as user priority, channel conditions, and fronthaul load are considered to optimize load balancing and improve network performance without manual intervention.
This technology is particularly beneficial for distributed MIMO deployments in cellular networks, providing improved uplink performance through combined beamforming. The use of AI/ML further enhances network efficiency by optimizing resource allocation, ensuring robust communication even in complex deployment scenarios.