Energy Efficiency Maximization in IRS-Aided Cell-Free Massive MIMO System

Then, an unsupervised learning Recently, cell-free massive multiple-input multiple-output based approach is proposed to tackle the EE maximization (MIMO) has emerged as a promising technology to effectively problem. Specifically, we model a two-stage deep neural alleviate inter-cell interference [1]. In this system, a large number network (DNN) and design a reasonable loss function, and of distributed access points (APs) are linked to the central train this DNN in an unsupervised manner to learn the optimal processing unit (CPU) through the backhaul link and provide beamforming and phase shifts. At last, simulation results show services to all the users without cell boundaries. However, the that compared with the traditional genetic algorithm (GA) and large-scale deployment of APs will bring some problems, such the proposed iterative optimization algorithm, the unsupervised as high deployment cost and power consumption. To address learning based approach can achieve better EE performance these difficulties, a promising technique called intelligent with extremely low running time.