Federated Spectrum Learning for Reconfigurable Intelligent Surfaces-Aided Wireless Edge Networks

Increasing concerns on intelligent spectrum sensing call for efficient training and inference technologies. In this paper, we propose a novel federated learning (FL) framework, dubbed (FSL), which exploits the benefits of reconfigurable surfaces (RISs) overcomes unfavorable impact deep fading channels. Distinguishingly, endow conventional RISs with capabilities by leveraging fully-trained convolutional neural network (CNN) model at each RIS controller, thereby helping base station to cooperatively infer users who request participate in FL beginning iteration. To fully exploit potential RISs, address three technical challenges: phase shifts configuration, user-RIS association, wireless bandwidth allocation. The resulting joint learning, resource allocation, association design is formulated as an optimization problem whose objective maximize system utility while considering prediction accuracy. context, accuracy interplays performance optimization. particular, if trained CNN deteriorates, allocation worsens. proposed FSL framework tested using real radio frequency (RF) traces numerical results demonstrate its advantages terms utility: better can be achieved larger number reflecting elements.