Advancing THz Communications: Hybrid Far- and Near-Field Channel Estimation

Realizing the potential of Terahertz (THz) ultra-massive MIMO systems requires overcoming significant channel estimation hurdles. In our latest paper, “Multi-User Hybrid Far- and Near-Field THz UM-MIMO Channel Estimation via UNet-OAMP,” we tackle the complex challenge of coexisting far-field and near-field users under severe multi-user interference.

Our key idea is to combine the interpretability of OAMP-based model-driven estimation with the representation power of UNet-based deep denoising. The proposed UNet-OAMP framework improves estimation accuracy while maintaining practical complexity.

🔹 Hybrid far-/near-field THz channel modeling

🔹 Multi-user estimation under non-orthogonal pilots

🔹 LMMSE-enhanced OAMP framework

🔹 UNet-based nonlinear denoising

🔹 Improved channel-estimation accuracy over strong baselines

This work provides a scalable and robust solution for reliable THz UM-MIMO communications.

Read the full paper for a detailed breakdown of our methodology and results. 

Multi-User Hybrid Far- and Near-Field THz UM-MIMO Channel Estimation via UNet-OAMP | IEEE Journals & Magazine | IEEE Xplore

J. Cao, S. Tarboush, H. Sarieddeen, N. Kouzayha and T. Y. Al-Naffouri, "Multi-User Hybrid Far- and Near-Field THz UM-MIMO Channel Estimation via UNet-OAMP," in IEEE Communications Letters, vol. 30, pp. 1771-1775, 2026, doi: 10.1109/LCOMM.2026.3686240