In this project, we aim to investigate the impact of wireless fading and noise on the performance of over-the-air computation (AirComp) empowered federated learning (FL) systems by considering uplink model aggregation and downlink model dissemination
jointly. We will jointly optimize each edge node transmit and receive equalization coefficients
along with the cloud server forwarding matrix to minimize distortions in the received global gradient vector at each edge node. Finally, we will rely on the MNIST dataset to evaluate the performance of the considered wireless FL system using the handwritten digit recognition task.