Channel Estimation for RIS Assisted Wireless Communications: Stationary or Non-Stationary?

Reconfigurable intelligent surface (RIS) is considered as a promising technology for future 6G communications. In RIS assisted communication systems, precise channel state information (CSI) is the prerequisite of efficient beamforming. Most existing channel estimation schemes rely on the spatial stationarity assumption. However, with the large array aperture of RIS, spatial non-stationarity effect naturally exists, which cannot be captured by most existing channel estimation schemes due to the spatial ambiguity effect of RIS and causes a significant decrease in the estimation accuracy. To tackle this problem, in this paper, we study the non-stationary channel estimation for RIS assisted communication systems. Specifically, we propose a two-dimensional group time block code (GTBC) based RIS decoupling scheme to convert the entire spatial non-stationary channel into a series of spatial stationary sub-channels. The proposed scheme is realized by artificially creating the time-domain relevance and exploiting it to recognize the spatial non-stationarity effect in the space domain. Based on the decoupled signals, a GTBC based dynamic orthogonal matching pursuit (GD-OMP) scheme is then proposed to update the dynamic codebook and estimate each path iteratively. Moreover, the computational complexity of the proposed GD-OMP scheme is analyzed. Finally, simulation results verify that the proposed scheme can capture the spatial non-stationarity effect and reach a higher channel estimation accuracy than existing schemes.