Summary
Federated learning lets multiple devices train a shared AI model without sharing their data directly, but when data is very different across devices or wireless connections are unreliable, the process slows down or becomes less accurate. The authors introduce a new method called federated random walk averaging (FedRW) that carefully chooses the order of device updates and resends data packets if errors occur, improving learning even with bad connections and diverse data. They also optimize how devices communicate and update the model within delay limits to speed up learning. Their experiments show this new approach achieves higher accuracy and converges faster than existing methods, especially when device data varies a lot.
What this means in practice
- •For wireless network engineers: Optimize communication parameters and routing for federated learning to improve model accuracy and latency in unreliable wireless conditions.
- •For mobile ai developers: Implement federated learning on mobile devices with diverse data and unstable connections to achieve better model training quality and speed.
Authors
Changheng Wang, Xianchao Zhang, Zhiqing Wei, Lingzhu Zhao, Zhongming Yang, Zhiyong Feng
Abstract
In wireless federated learning (FL), data heterogeneity and multiple local updates induce client drift, degrading model convergence. It is further affected by unreliable wireless links, as transmission errors may invalidate model updates. To address these challenges, we propose a federated random walk averaging (FedRW) framework, which is a variant of federated averaging (FedAvg) that mitigates data heterogeneity by updating models along random walk (RW) paths and aggregating them at the server. Model parameters are transmitted in packets with retransmission support to improve training quality by mitigating wireless errors along RW paths. Meanwhile, wireless transmission delays hinder the exploration of FedRW. To this end, we formulate a joint optimization problem that integrates learning, RW path selection, and transmission parameter tuning, aiming to minimize the training loss under delay constraints. By deriving an upper bound on the expected convergence of FedRW over unreliable wireless networks, we reduce the problem to a general form agnostic to task type and model architecture. A distributed solution is then proposed, in which the server or clients optimize packet size and maximum number of retransmissions locally, and efficiently select reliable and expandable next-hop nodes via a resilience-aware beam search with dynamic pruning. Simulation results show that FedRW achieves 2.26%-9% higher accuracy than state-of-the-art baselines under high data heterogeneity. Furthermore, the jointly optimized FedRW yields at least 2.78% higher accuracy and faster convergence compared to baselines.