Sparse Representation-Based Robust Channel Estimation in XL-RIS-Assisted Systems

The extremely large reconfigurable intelligent surface (XL-RIS) is an architecture that shows potential in expanding the transmission region and meeting the high transmission requirements. With the deployment of XL-RIS, users transmit signals through the spherical-wavefront near-field channel, which presents challenges in channel estimation with low pilot overhead. Additionally, XL-RIS is a non-stationary system whose elements are prone to failure due to accidental damages or blockages, thereby further complicating the channel state. To address these issues, this paper proposes a two-stage sparse representation-based channel estimation (SRCE) scheme to jointly recover the near-field channel state and diagnose RIS element failures with low pilot overhead. The first stage utilizes a proposed double-domain filter (DDF) method to eliminate part of the received noise based on the sparse representation in the angle and polar domain. Then, a robust failure-aware double sparsity orthogonal matching pursuit (FA-OMP) algorithm is proposed by iteratively reconstructing the channel state, as well as eliminating the perturbation invoked by RIS element failures. The efficiency and robustness of our proposed scheme are validated through simulations.