Auxiliary Bayesian Learning Approach for Joint Channel Estimation and Data Detection With RIS-Assisted OFDM Systems
Chen Wei Ji, Jisheng Dai, Xueqin Jiang, Weichao Xu · IEEE Transactions on Wireless Communications · 2025
Joint channel estimation and data detection can enhance the performance of both tasks, improve communication efficiency, and reduce pilot overhead. However, this has been rarely explored in the context of reconfigurable intelligent surface (RIS)-assisted OFDM systems due to the complexity of handling multiple coupling effects between various data sequences and reflection coefficients. The recently proposed solution is tailored for purely phase modulations, whereas modern 5G/6G systems predominantly use quadrature amplitude modulation (QAM). To overcome this limitation, the paper proposes an auxiliary Bayesian learning approach for joint channel estimation and data detection with RIS-assisted OFDM systems, which integrates advanced techniques to offer robustness, improved performance, and reduced computational complexity. The novelties of the proposed method are threefold: i) present an auxiliary Bayesian learning framework to probabilistically link pilot and data information, effectively formulating the common sparsity pattern while avoiding potential modeling errors introduced by modulation schemes; ii) derive a hybrid approximate message passing (AMP) approach for efficient execution of Bayesian inference, with various approximation strategies seamlessly integrated to balance computational complexity and approximation accuracy; and iii) embed a novel rearrangement strategy to streamline multilayer message extraction, facilitating efficient message passing among sub-modules and subsequent parameter learning. Simulation results demonstrate its superiority.