Improving Spectral Efficiency in RIS Enabled Massive MIMO Systems using Federated Learning

This paper investigates the use of Federated Learning (FL) to improve spectrum efficiency in Massive MIMO systems supported by Reconfigurable Intelligent Surfaces (RIS). Utilizing a Convolutional Neural Network (CNN) in conjunction with the FL architecture, the suggested method enhances the accuracy of channel estimate while maintaining data privacy. By enabling distributed learning, the FL-based model improves security and lowers communication overhead in contrast to conventional techniques that depend on centralized processing. The paper demonstrates the efficacy of FL-based channel estimation in enhancing spectral efficiency, dependability, and accuracy through a thorough performance analysis that contrasts it with traditional methods. The suggested model's superiority is further supported by experimental results, which demonstrate notable improvements in system performance and Mean Squared Error (MSE) reduction. These results demonstrate the promise of FL-integrated RIS-assisted Massive MIMO systems for upcoming intelligent connectivity applications and next-generation wireless communication networks, specifically Beyond-5G (B5G).