Maximum Likelihood Channel Estimation for RIS-Aided Communications with LOS Channels

A reconfigurable intelligent surface (RIS) reflects incoming signals in different ways depending on the phase-shift pattern assigned to its elements. The most promising use case is to aid the communication between a base station and a user when the user has a line-of-sight (LOS) channel to the RIS but the direct channel is blocked. The main challenge is to estimate the channel with limited resources because non-parametric estimation methods require a pilot length proportional to the large number of RIS elements. In this paper, we develop a parametric maximum likelihood (ML) channel estimation framework for estimating the LOS channel to the RIS. We demonstrate that the proposed algorithm can accurately obtain the channel and preferred RIS configuration using only a few pilots. A key novelty is that the RIS configurations used during pilot transmission are selected to progressively improve the estimation accuracy.