Probabilistic Client Selection for Asynchronous Wireless Federated Learning

To solve the straggler problem in federated learning (FL), in this paper we propose an asynchronous wireless FL scheme where each client keeps local updates and is probabilis-tically selected to transmit local model to the server at arbitrary times. We first derive the (approximate) expression for the convergence rate based on the probabilistic client selection. Then, an optimization problem is formulated to tradeoff the convergence rate of asynchronous FL and mobile energy consumption by joint probabilistic client selection and bandwidth allocation. We develop an iterative algorithm to solve the non-convex problem globally optimally. Experiments demonstrate the superiority of the proposed approach compared with the traditional schemes.