Dual-Structure Regularized Sparse Bayesian Learning Channel Estimation for RIS Assisted Wireless Communication

Reconfigurable intelligent surface (RIS) technology, as a revolutionary potential key technology for 6G, presents innovative avenues for wireless communication technology. Channel estimation stands as a fundamental and pivotal aspect of RIS assisted wireless communication systems, however, it becomes a complex challenge due to the integration of a large number of passive RIS elements. In this paper, we exploit the dual-structure (DS) sparsity of multi-user equipment (UE) cascade channels and propose a novel channel estimation algorithm based on hyperparameter regularized Sparse Bayesian Learning (HRSBL). We divide the cascade channel estimation problem into two stages. In the first stage, the position index of the dual-structure sparse channel matrix is estimated. In the second stage, we estimate the position index of the first stage is used as the channel prior information, and then HRSBL is applied to estimate the cascade channel matrix. Simulations substantiate the feasibility of this algorithm, showcasing its ability to significantly reduce pilot overhead.