Current browse context:
cs.IT
Change to browse by:
References & Citations
Computer Science > Machine Learning
Title: Communication-Efficient Device Scheduling for Federated Learning Using Stochastic Optimization
(Submitted on 19 Jan 2022 (v1), last revised 4 May 2022 (this version, v2))
Abstract: Federated learning (FL) is a useful tool in distributed machine learning that utilizes users' local datasets in a privacy-preserving manner. When deploying FL in a constrained wireless environment; however, training models in a time-efficient manner can be a challenging task due to intermittent connectivity of devices, heterogeneous connection quality, and non-i.i.d. data. In this paper, we provide a novel convergence analysis of non-convex loss functions using FL on both i.i.d. and non-i.i.d. datasets with arbitrary device selection probabilities for each round. Then, using the derived convergence bound, we use stochastic optimization to develop a new client selection and power allocation algorithm that minimizes a function of the convergence bound and the average communication time under a transmit power constraint. We find an analytical solution to the minimization problem. One key feature of the algorithm is that knowledge of the channel statistics is not required and only the instantaneous channel state information needs to be known. Using the FEMNIST and CIFAR-10 datasets, we show through simulations that the communication time can be significantly decreased using our algorithm, compared to uniformly random participation.
Submission history
From: Jake Perazzone [view email][v1] Wed, 19 Jan 2022 23:25:24 GMT (2292kb,D)
[v2] Wed, 4 May 2022 19:36:34 GMT (2297kb,D)
Link back to: arXiv, form interface, contact.