MAB-Based Channel Scheduling for Asynchronous Federated Learning in Non-Stationary Environments

TL;DR

This work addresses the challenge of stale updates and unequal participation in asynchronous federated learning over non-stationary wireless channels. It casts channel scheduling as online MAB problems, deriving AoI-regret bounds for two non-stationary regimes and proposing two algorithms: M-Exp3 for extremely non-stationary channels and GLR-CUCB for piecewise-stationary channels. To mitigate update imbalance, it introduces an adaptive channel matching mechanism that blends marginal contribution with fairness, aided by server-side gradient buffering. Empirical results on CIFAR-10/100 demonstrate faster convergence and reduced AoI variance compared to random scheduling, highlighting practical gains for scalable, privacy-preserving learning in dynamic wireless environments.

Abstract

Federated learning enables distributed model training across clients without raw data exchange, but in wireless implementations, frequent parameter updates cause high communication overhead. Existing research often assumes known channel state information (CSI) or stationary channels, though practical wireless channels are non-stationary due to fading, user mobility, and attacks, leading to unpredictable transmission failures and exacerbating client staleness, which hampers model convergence. To tackle these challenges, we propose an asynchronous federated learning scheduling framework for non-stationary channels that aims to reduce client staleness while enhancing communication efficiency and fairness. Our framework considers two scenarios: extremely non-stationary and piecewise-stationary channels. Age of Information (AoI) quantifies client staleness under these conditions. We conduct convergence analysis to examine the impact of AoI and per-round client participation on learning performance and formulate the scheduling problem as a multi-armed bandit (MAB) problem. We derive theoretical lower bounds on AoI regret and develop scheduling strategies based on GLR-CUCB and M-exp3 algorithms, including upper bounds on AoI regret. To address imbalanced client updates, we propose an adaptive matching strategy that incorporates marginal utility and fairness considerations. Simulation results show that our algorithm achieves sub-linear AoI regret, accelerates convergence, and promotes fairer aggregation.

Figures (4)

• Figure 1: Asynchronous federated learning procedure in wireless network
• Figure 2: Regret comparison of different algorithms versus communication rounds
• Figure 3: Performance comparison on test accuracy.
• Figure 4: Performance comparison on AoI variance

Theorems & Definitions (9)

• Theorem 1
• Lemma 1
• Lemma 2
• Theorem 2
• Theorem 3
• Theorem 4
• Lemma 3
• Lemma 4
• Theorem 5