D2D-Enabled Data Sharing for Distributed Machine Learning at Wireless Network Edge

Abstract

Mobile edge learning is an emerging technique that enables distributed edge devices to collaborate in training shared machine learning (ML) models by exploiting their local data samples and communication/computation resources. To deal with the “straggler's dilemma” issue faced in this technique, this letter proposes a new device-to-device (D2D)-enabled data sharing approach, in which different edge devices share their data samples among each other over D2D communication links, in order to properly adjust their computation loads for increasing the training speed. Under this setup, we optimize the radio resource allocation for both D2D-enabled data sharing and distributed training, with the objective of minimizing the total training delay under fixed numbers of local and global iterations (for training). Numerical results show that the proposed D2D-enabled data sharing design significantly reduces the training delay, and also enhances the training accuracy when the data samples are non-independent and identically distributed (non-IID) among edge devices.