Channel Estimation of Correlated Channels in RIS-Assisted Wireless Communication Systems

Abstract

In Reconfigurable Intelligent Surfaces (RIS) aided communication systems, knowledge of the channel state information is critical in optimizing the reflection coefficients of the RIS. However, the existing methods proposed for the non-parametric channel models involve a large training overhead. To reduce the channel estimation overhead, the spatial correlation inherent in the communication systems due to small inter-element distances in the antennas and the RIS as well as the directionality of the antennas can be exploited to group a set of adjacent RIS elements to share a common training reflection coefficient, thus effectively reducing the number of training time slots needed for the channel estimation by a factor of group size. The study defines a correlated channel model for Uniform Linear Array (ULA) and Uniform Planar Array (UPA) antenna arrays and implements a channel estimation algorithm based on the Discrete Fourier Transform (DFT) reflection training pattern at the RIS. The effects of the correlation coefficients and the RIS size on the channel estimation performance are observed. The results show that even at low correlation, the grouping scheme is effective at reducing the training time.