Federated Learning-based Resource Allocation in RSU assisted Moving Network

Motivated by the proliferation of smart applications and the ever accumulating concerns of data privacy, a distributed deep learning (DL) model named federated learning (FL) has been emerging. FL enables the model learning in a distributed manner without sending data from all users to a centralized hub. In this paper, we consider FL for resource allocation in roadside units (RSUs)-assisted moving network. In our proposed work, we investigate resource allocation in moving network with RSUs integrated along the roads that serve moving small cells (moSCs) deployed on trams travelling with deterministic mobility. The proposed algorithm trains the resource allocation model in a distributed manner, in which each RSU exploits its computational power and the training data of its associated moSCs to generate a shared model. We provide numerical results to validate our proposed algorithm.