Efficient Dictionary Learning Based Channel Estimation in Reconfigurable Intelligent Surfaces Assisted Communications

Due to the large amount of reflective elements, it is challenging to acquire accurate cascaded channel state information in reconfigurable intelligent surface (RIS) aided communication systems. This paper investigates cascaded channel estimation for multi-RIS-aided communication networks, aiming to improve estimation accuracy and robustness with low training overhead. To effectively reduce training overhead, we conduct joint estimation of the sparse cascaded channel in presence of multiple RISs. Traditional sparse channel estimation methods often rely on predefined angle dictionaries, leading to mismatches with time-varying channel angles and reduced performance. Hence, in this paper, we propose an iterative re-weighted joint channel estimation algorithm to achieve more accurate cascaded channel estimation, where both the joint sparse channel coefficient matrix and the angle dictionary matrix are alternately optimized. Simulation results show that the proposed algorithm outperforms the benchmark algorithms in terms of the channel estimation and signal recovery, especially with low training overhead.