Dictionary Learning-Based Channel Estimation for RIS-Aided MISO Communications

In this letter, we study the channel estimation problem for the reconfigurable intelligent surface (RIS)-aided multi-input single-output (MISO) system. By exploiting the channel sparsity, compressive sensing (CS) based sparse channel estimators can be applied to the system to reduce the training overhead. However, these existing sparse channel estimators adopt predefined dictionaries when formulating the sparse matrix recovery problem, which will cause grid mismatch issues and estimation performance degradation. Hence, in this letter, we formulate the channel estimation problem as a joint dictionary parameter learning and sparse signal recovery problem, in which the dictionary parameter can be optimized to adapt to the channel measurements, thereby improving the robustness of sparse channel representation and estimation performance. Then, we propose an iterative re-weighted algorithm to solve this non-convex problem efficiently. Simulation results show that the proposed algorithm outperforms other benchmark schemes significantly.

Automated summary

This letter addresses the channel estimation problem for RIS-aided MISO systems by optimizing dictionary parameters and sparse signal recovery to enhance the representation and estimation performance of sparse channels.