Intelligent Resource Allocation for RIS-Assisted Cell-Free Massive MIMO Systems

Cell-free massive multiple-input multiple-output (mMIMO) systems distribute a large number of access points (APs) over a wide geographic region, improving coverage, capacity, and user experiences notably. Similarly, reconfigurable intelligent surface (RIS) allows a significant leap forward in signal quality and energy efficiency by adding additional flexibility and intelligent control over the propagation environment of wireless systems. However, optimal resource allocation in RIS-integrated cell-free mMIMO systems faces substantial challenges due to the presence of a large number of APs and passive reflecting elements (PREs) at RIS while simultaneously providing network coverage to aerial and ground users. To efficiently navigate the high-dimensional optimization space, this paper proposes a novel framework for optimal resource allocation in RIS-assisted cell-free mMIMO systems employing deep learning (DL) techniques. In particular, we train a deep neural network offline to optimize the transmit powers at APs and the phase shift of PREs at RIS. The trained model is tested online to perform real-time optimization with low computational complexity. Simulation results demonstrate the effectiveness of the proposed data-driven resource allocation scheme while comparing it with the state-of-the-art baseline scheme.