Low-Complexity Channel Estimation and Passive Beamforming for RIS-Assisted MIMO Systems Relying on Discrete Phase Shifts

Reconfigurable intelligent surfaces (RISs) are capable of enhancing the capacity of wireless networks at a low cost. In practical RIS-assisted communication systems, the acquisition of channel state information (CSI) and RIS reflection optimization constitute a pair of challenges. In this paper, a low-complexity channel estimation and passive beamforming design is proposed.
First of all
, we conceive a low-complexity framework for maximizing the achievable rate of RIS-assisted multiple-input multiple-output (MIMO) systems having discrete phase shifts at each RIS element. In contrast to existing solutions, the proposed arrangement partitions the channel training stage into several phases, where the RIS reflection coefficients are pre-designed and the effective superposed channel is estimated instead of separately training the source-destination and source-RIS-destination links. Based on this, the active beamformer can be designed at low complexity and the RIS reflection optimization is performed by selecting that one from the pre-designed training set which maximizes the achievable rate.
Secondly
, we propose novel techniques for generating the training set of RIS reflection coefficients. The theoretical performance of the proposed scheme is analyzed and compared to the optimal RIS configuration.
Finally
, our simulation results demonstrate that the proposed framework is more competitive than its existing counterparts when relying on imperfect CSI, especially for rapidly time-varying channels having short channel coherence time.