Key points are not available for this paper at this time.
The challenges of channel estimation in reconfigurable intelligent surfaces (RIS) are well known, in particular, its high demand on pilots and channel estimation resources. This paper explores gains in channel estimation, its efficiency, and achievable rates, by leveraging the differences in antenna correlations for different links, a phenomenon known ascorrelation diversity. Antenna correlations are a well-known feature of mm-wave and massive MIMO (multiple-input multiple-output) communication, and the existence of correlation diversity in multi-user systems is well-established in the literature. In RIS systems employing frequency-division duplex (FDD) transmission, we propose a novel joint transmit/RIS beamforming that exploits antenna correlation diversity via efficient channel training and pilots. Our pilot configurations are optimized according to the degrees of freedom (DoF), and sum-rates are optimized via beamforming. The proposed approach identifies and analyzes the common and disjoint eigenspaces of the antenna correlation matrices, and exploits them utilizingproduct superpositionand rate splitting. A key contribution of this work is reconciling the requirements of transmitter/RIS beamforming on the one hand, and features of product superposition in the presence of imperfect channel state information (CSI) on the other hand. Numerical results are presented to corroborate our findings.
Karbalayghareh et al. (Wed,) studied this question.