Federated Parameter-Efficient Adaptation for Interference Mitigation at the Wireless Edge

TL;DR

This paper introduces a federated, parameter-efficient adaptation method using LoRA for interference mitigation in wireless networks, significantly reducing communication costs while maintaining high performance.

Contribution

It adapts Low-Rank Adaptation (LoRA) to temporal CNNs for federated interference mitigation, enabling efficient, high-performance model updates with minimal communication overhead.

Findings

Fed-LoRA reduces communication by up to 20x compared to full model federated learning.

Local LoRA improves BER by 12.8% over frozen backbone.

Fed-LoRA performs comparably to local adaptation and outperforms full-model FedAvg in heterogeneous environments.

Abstract

Dense wireless deployments face co-channel interference from heterogeneous sources that vary across base stations (gNBs in 5G). While centralized DNN-based approaches to interference mitigation have shown strong performance, deploying and adapting these models across distributed gNBs via federated learning (FL) requires transmitting full model updates each round, resulting in a cost that scales poorly with network density. Parameter-efficient fine-tuning (PEFT) reduces this burden by training and communicating only a small fraction of parameters. While traditionally applied to large foundation models, we adapt Low-Rank Adaptation (LoRA) to temporal convolutional neural network architectures for interference suppression, placing low-rank adapters on the dilated convolutional layers. This placement enables LoRA to learn local interference-specific temporal patterns, while the frozen backbone retains the shared signal extraction capability. These lightweight adapters (5.1\% of backbone parameters) are federated via FedAvg, reducing per-round communication by up to 20$\times$ compared to federating full model updates. We evaluate various PEFT strategies across simulated distributed gNBs with non-IID interference environments. Results show that local LoRA achieves 12.8\% average BER improvement over the frozen backbone, while Fed-LoRA achieves comparable performance (12.6\%). Fed-LoRA outperforms local adaptation on data-starved nodes where federated knowledge transfer compensates for limited samples, all while avoiding the catastrophic degradation observed with full-model FedAvg under heterogeneous conditions.