Rotating Extremely Large-Scale MIMO with Generative AI for Vehicular Communications

Abstract

In this paper, we propose a novel concept of rotating extremely large-scale multiple-input multiple-output (XL-MIMO) systems based on a cell-free network, wherein the direction of each uniform planar array (UPA) is jointly controlled by a central processing unit to achieve full coverage for mobile user equipments (UEs) in vehicular communications. Using the near-field channel, we derive the expression for the achievable spectral efficiency (SE) of the considered systems for analysis. Furthermore, we formulate a joint optimization problem that maximizes the average SE of the rotating XL-MIMO system by optimizing the direction angle of each UPA. Then, we propose two low-complexity methods to enhance the SE performance of weak and strong UEs respectively. Moreover, we develop a method based on generative AI, employing a trained diffusion model to efficiently generate UPA direction angles in dynamic environments with mobile UEs.