DFrFT-Based Near-Field Channel Estimation for Extremely Large-Scale MIMO Systems

To deal with channel estimation in the near-field area, previous studies have introduced orthogonal matching pursuit (OMP)-based algorithms, which are generally with high computational overhead. To tackle this issue, a novel approach based on discrete fractional Fourier transform (DFrFT) called DFrFT-clean is proposed in this letter. Instead of employing OMP, our method identifies the peaks of the DFrFT of the processed received signal to estimate the parameters of each path component of the channel. Additionally, it implements a cleaning procedure to mitigate the interference of strong paths on weaker paths. Simulation results show that the proposed DFrFT-clean method can significantly reduce the computational complexity while maintaining comparable or even favorable normalized mean square error (NMSE) performance compared to the benchmark algorithms.

I. Introduction

New spectrum and ultra-massive multiple input multiple output (MIMO) are two highly promising avenues in 6G to meet the increasing demands for high data rate transmission [1]. Ultra-massive MIMO, referring to the deployment of larger antenna arrays comprising hundreds or even thousands of antennas, can lead to higher spectral efficiency and energy efficiency [2]. The utilization of new spectral bands, e.g., Terahertz (THz), can provide wider bandwidth and lower latency.