Convergence-Latency-Aware Adaptive Modulation and Resource Allocation in RIS-Assisted Wireless Federated Learning

2026-07-22

AI Summary

This work addresses the challenges of wireless federated learning in obstructed propagation environments, where unreliable transmission leads to high latency and degraded convergence performance—a trade-off between convergence and delay under modulation-related transmission errors that existing studies fail to model. Focusing on reconfigurable intelligent surface (RIS)-assisted wireless federated learning systems, this paper is the first to quantify the impact of symbol error rate on model loss decay by deriving a convergence upper bound. Building upon this analysis, a joint optimization framework integrating adaptive modulation and dynamic subchannel allocation is proposed. A hybrid alternating algorithm is developed to solve the resulting mixed-integer nonlinear programming (MINLP) problem. Extensive experiments on MNIST, CIFAR-10, and Speech Commands datasets demonstrate significant improvements over state-of-the-art methods, particularly achieving faster convergence and higher accuracy under complex learning tasks and harsh channel conditions.

Abstract

Federated learning (FL) over wireless networks suffers from significant training latency and degraded convergence due to unreliable wireless transmission, especially under blocked propagation environments. Although reconfigurable intelligent surfaces (RISs) can improve communication reliability, existing wireless FL studies rarely characterize the trade-off between learning convergence and communication delay under modulation-dependent transmission errors. In this paper, we consider a wireless FL system operating under RIS-assisted blocked-link propagation scenarios, and focus on adaptive modulation and sub-channel allocation for convergence-latency aware communication design. By characterizing the effect of symbol errors on uploaded local gradients, we derive a convergence-related upper bound that reveals the impact of symbol error rate (SER) on FL loss decay. Based on this result, we formulate a joint convergence-latency optimization problem, which is cast as a mixed-integer nonlinear programming (MINLP) problem, and solve it using a low-complexity hybrid alternating optimization framework. Extensive experiments on MNIST, CIFAR-10, and Speech Commands show that the proposed scheme consistently achieves faster convergence and higher test accuracy than existing adaptive communication schemes, especially in complex tasks and challenging wireless scenarios.

Problem

• federated learning
• wireless communication
• convergence-latency trade-off
• reconfigurable intelligent surface
• transmission errors

Innovation

• Reconfigurable Intelligent Surface (RIS)
• Federated Learning
• Adaptive Modulation
• Convergence-Latency Trade-off
• Symbol Error Rate (SER)