Super-Resolution Sparse Reconstruction Based Near-Field Channel Estimation for Terahertz Communication Systems

Due to the abundant spectrum resources, THz communication has received widespread attention and has been proposed as a fundamental technology for 6G. To ensure the wireless transmission performance, accurate channel state information (CSI) is essential for channel adaptive technology. In this paper, a near-field channel estimation method based on super-resolution sparse reconstruction is proposed for THz communication by utilizing the sparse characteristics of THz channels. Firstly, we utilize THz near-field signal transmission model to construct higher-order cumulant matrices of the received signals, and employ a subspace decomposition algorithm to estimate the CSI of user channel with the largest path gain. Subsequently, the path gains are estimated through directional beam pilot training. Finally, we use the obtained CSI to separate the channel with the maximum path gain from the received signal matrix and iteratively acquire CSI of all users to accomplish channel estimation. Simulation results show that the proposed algorithm outperforms traditional channel estimation methods in scenarios characterized by low signal-to-noise ratio (SNR) and constrained training overhead.