RIS-Aided XL-MIMO Channel Estimation Based on Expectation-Maximization

Abstract

Intelligent reflecting surface (RIS)-aided extremely large-scale massive MIMO (XL-MIMO) is a promising technique for improving the spectrum efficiency in future 6G communications. However, channel estimation for the RIS-aided XL-MIMO system still faces challenges such as overhead and accuracy due to its large dimensionality. In this letter, an expectation-maximization (EM)-based channel estimation is proposed for the RIS-aided XL-MIMO system. By utilizing the properties of the polar-domain near-field channel and angular-domain far-field channel, the original hybrid-field channel is transformed into a common sparse structure to reduce computational complexity, in which the parameters are further modeled as an unknown Bernoulli-Gaussian (BG) distribution. The hybrid-field channel is estimated by iteratively updating the parameters. Simulations are performed and results demonstrate that the proposed EM-based method achieves better performance with the same pilot overhead.