Channel Feedback for Reconfigurable Intelligent Surface Assisted Wireless Communications

Abstract

Reconfigurable intelligent surface (RIS) has recently received increasing interest due to the superiority of changing the wireless propagation environments intelligently. Channel feedback is essential in frequency division duplex (FDD) RIS-assisted wireless communications, since the downlink channel state information (CSI) has to be acquired by the base station (BS) for the joint beamforming at the BS and the RIS. In this paper, we first investigate the single-structured sparsity of RIS channel, which means that the angular-domain sparse channels of different users share the same non-zero element indices only in the column dimension, but not in the row dimension. Based on this channel characteristic, we then propose a dimension-reduced channel feedback scheme with a low overhead for channel feedback. Specifically, the downlink CSI can be decomposed into the structured CSI and unstructured CSI. The structured CSI of the multi-user channels can be fed back to the BS by only one user according to the single-structured sparsity, while the unstructured CSI can be fed back with a fairly low overhead by different users, respectively. Moreover, by utilizing the angle coherence time, the per-user overhead can be reduced further, while the near-optimal per-user rate can still be guaranteed.