Vaccinating Federated Learning for Robust Modulation Classification in Distributed Wireless Networks

TL;DR

FedVaccine, a novel FL model aimed at improving generalizability across signals with varying noise levels by deliberately introducing a balanced level of noise, is proposed, which identifies an optimal noise tolerance for DNN models, thereby regulating the training process and mitigating overfitting.

Abstract

Automatic modulation classification (AMC) serves a vital role in ensuring efficient and reliable communication services within distributed wireless networks. Recent developments have seen a surge in interest in deep neural network (DNN)-based AMC models, with Federated Learning (FL) emerging as a promising framework. Despite these advancements, the presence of various noises within the signal exerts significant challenges while optimizing models to capture salient features. Furthermore, existing FL-based AMC models commonly rely on linear aggregation strategies, which face notable difficulties in integrating locally fine-tuned parameters within practical non-IID (Independent and Identically Distributed) environments, thereby hindering optimal learning convergence. To address these challenges, we propose FedVaccine, a novel FL model aimed at improving generalizability across signals with varying noise levels by deliberately introducing a balanced level of noise. This is accomplished through our proposed harmonic noise resilience approach, which identifies an optimal noise tolerance for DNN models, thereby regulating the training process and mitigating overfitting. Additionally, FedVaccine overcomes the limitations of existing FL-based AMC models' linear aggregation by employing a split-learning strategy using structural clustering topology and local queue data structure, enabling adaptive and cumulative updates to local models. Our experimental results, including IID and non-IID datasets as well as ablation studies, confirm FedVaccine's robust performance and superiority over existing FL-based AMC approaches across different noise levels. These findings highlight FedVaccine's potential to enhance the reliability and performance of AMC systems in practical wireless network environments.

Figures (5)

• Figure 1: Architectural overview of FedVaccine. FedVaccine involves deliberate accommodation of a controlled level of noise and the incorporation of supplementary datasets through a queue structure aimed at mitigating bias in non-IID scenarios. Utilizing these locally fine-tuned models, the global model undergoes iterative updates through cluster-wise units, minimizing the information loss during the aggregation stage.
• Figure 2: The training and test outcomes of CNN o2016convolutional and GRU hong2017automatic models are presented. The figures in the initial row depict the original results, while the corresponding enlarged versions of each column in the first row are displayed in the second row.
• Figure 3: Visualization result after reducing feature dimensions through PCA. The figures arranged in the initial row depict the PCA outcomes corresponding to discrete SNR, while those in the second row illustrate the PCA results associated with the combined SNR range.
• Figure 4: The test performance outcomes under IID conditions are compared between FedAvg and FedVaccine in each SNR range. It is noteworthy that the discernible performance gap widens as the training datasets encompass higher SNR values, culminating in disparities of approximately 12% and 17% for datasets RML2016.10a and RML2016.10b, respectively.
• Figure 5: A comparative analysis of performance involving three distinct learning paradigms and a subset of FL models within three non-IID scenarios across two public datasets.