Upper Mid-Band (FR3) Massive MIMO

The upper mid-band — FR3, roughly 7–24 GHz — is the spectrum the 6G research community keeps pointing at: enough bandwidth to matter, propagation that isn’t terrible, and a regulatory window that is opening. Building a massive MIMO platform that exploits it is largely a systems-design exercise. How do you split beamforming, digital processing, linearization, and cancellation across a real RFIC + ASIC + SoC stack without paying for everything twice, and how do you do it at the array sizes FR3 demands?

This is the spine of what I work on at Samsung today, leading the platform-side effort across a global, cross-functional team from architecture through lab validation. The work spans hierarchical beamforming and reciprocity calibration for large arrays, AI-RAN testbed integration on a cuBB-based DU (see GPU-Based AI-RAN), and AI-based digital predistortion for the high-efficiency PAs we want to use in commercial radio units (see AI DPD for High-Efficiency PAs). The self-interference cancellation R&D that makes a lot of this practical lives in MIMO Self-Interference Cancellation.

Selected publications

• Upper Mid-Band Spectrum for 6G: Opportunities and Key Enablers — Samsung Research Blog (2026). Position piece on FR3 as the 6G band of choice, eXtreme-MIMO as the headline enabler, and the energy-efficient hardware innovations needed to make it real.
• Hierarchical Beamforming and Reciprocity Calibration for Hybrid MIMO Systems — U.S. Patent 12,401,405 (2025). Granted patent on a hierarchical beamforming + reciprocity-calibration approach for FR3-scale arrays.
• Reducing FDD MMU Form Factor With Active Cancellation — IEEE IMS 2022. Shrinking massive-MIMO unit form factor using active cancellation in FDD operation.
• 5G New Radio and Non-Terrestrial Networks: Reaching New Heights — IEEE ICC Workshops 2022. NTN-enhanced cellular and what it means for the underlying radio platform.
• Power Reduction for Massive MIMO Radios — U.S. Patent Application (2025). Power-efficiency techniques for massive-MIMO RUs.