Communication-Efficient Federated Learning for Wireless Edge Intelligence in IoT

TLDR

The proliferation of IoT devices generates vast data, yet privacy concerns deter central data collection, prompting the use of edge‑side federated learning enabled by SDN, though existing FedAvg algorithms require many rounds and high communication, especially with non‑IID data. This work introduces CE‑FedAvg, a communication‑efficient variant of FedAvg that incorporates distributed Adam optimization and novel compression to reduce training rounds. The authors evaluate CE‑FedAvg on MNIST and CIFAR‑10 under IID and non‑IID client distributions, varying client counts, participation rates, and compression levels. CE‑FedAvg converges up to six times faster and uploads three times less data than compressed FedAvg, remains robust to aggressive compression, and on Raspberry Pi edge clients achieves target accuracy 1.7 times faster in real time.

Abstract

The rapidly expanding number of Internet of Things (IoT) devices is generating huge quantities of data, but public concern over data privacy means users are apprehensive to send data to a central server for machine learning (ML) purposes. The easily changed behaviors of edge infrastructure that software-defined networking (SDN) provides makes it possible to collate IoT data at edge servers and gateways, where federated learning (FL) can be performed: building a central model without uploading data to the server. FedAvg is an FL algorithm which has been the subject of much study, however, it suffers from a large number of rounds to convergence with non-independent identically distributed (non-IID) client data sets and high communication costs per round. We propose adapting FedAvg to use a distributed form of Adam optimization, greatly reducing the number of rounds to convergence, along with the novel compression techniques, to produce communication-efficient FedAvg (CE-FedAvg). We perform extensive experiments with the MNIST/CIFAR-10 data sets, IID/non-IID client data, varying numbers of clients, client participation rates, and compression rates. These show that CE-FedAvg can converge to a target accuracy in up to 6× less rounds than similarly compressed FedAvg, while uploading up to 3× less data, and is more robust to aggressive compression. Experiments on an edge-computing-like testbed using Raspberry Pi clients also show that CE-FedAvg is able to reach a target accuracy in up to 1.7× less real time than FedAvg.