Abstract:Federated learning (FL) enables multiple devices to collaboratively train a global model while maintaining data on local servers. Each device trains the model on its local server and shares only the model updates (i.e., gradient weights) during the aggregation step. A significant challenge in FL is managing the feature distribution of novel, unbalanced data across devices. In this paper, we propose an FL approach using few-shot learning and aggregation of the model weights on a global server. We introduce a dynamic early stopping method to balance out-of-distribution classes based on representation learning, specifically utilizing the maximum mean discrepancy of feature embeddings between local and global models. An exemplary application of FL is orchestrating machine learning models along highways for interference classification based on snapshots from global navigation satellite system (GNSS) receivers. Extensive experiments on four GNSS datasets from two real-world highways and controlled environments demonstrate that our FL method surpasses state-of-the-art techniques in adapting to both novel interference classes and multipath scenarios.

What problem does this paper attempt to address?

### What problems does this paper attempt to solve? This paper aims to solve the challenges of federated learning (FL) in adapting to global navigation satellite system (GNSS) interference classification. Specifically, the researchers proposed an early - stopping method based on maximum mean discrepancy (MMD) to improve the model's adaptability to new interference categories and enhance the model generalization performance on imbalanced datasets. #### Main problems and solutions: 1. **Imbalanced data distribution**: - In federated learning, the data distribution on different devices may be unbalanced, leading to over - fitting or under - fitting of certain categories. To this end, the researchers introduced a dynamic early - stopping method based on MMD. By calculating the difference between the feature embeddings of the local model and the global model, the training rounds are adjusted to balance the category distribution. 2. **Adaptation to new interference categories**: - GNSS signals are vulnerable to multiple types of interference, which may exhibit different characteristics in different environments. To adapt to new interference categories, the researchers adopted few - shot learning (FSL) technology, enabling the model to quickly adapt to newly emerging interference types. 3. **Improvement of model generalization ability**: - By reducing the training rounds and focusing on adjusting difficult - to - classify categories, the researchers improved the model's generalization ability, reduced the training time, and at the same time improved the classification accuracy. #### Specific applications: - **Highway sensor stations**: In practical application scenarios, sensor stations are distributed on the highway, recording GNSS signals and detecting and classifying interference. Each sensor station is trained as a local model, and the weights are aggregated through the global model to enhance the re - identification ability of all sensor stations for new interference categories. #### Experimental verification: - The researchers used four GNSS datasets for experiments, including two datasets from real - world highway environments and two datasets from controlled environments. The experimental results show that the proposed FL method outperforms existing techniques in adapting to new interference categories and multipath scenarios, significantly improving the classification accuracy. ### Summary: This paper solves the problems of imbalanced data distribution and new - category adaptation in federated learning for GNSS interference classification by introducing an early - stopping method based on MMD and few - shot learning technology, and improves the model's generalization ability and classification accuracy.

Federated Learning with MMD-based Early Stopping for Adaptive GNSS Interference Classification

FedDGP: Disentangling Global and Personal Models for Federated Learning

Message Passing Based Wireless Federated Learning Via Analog Message Aggregation

Non-orthogonal Multiple Access Assisted Federated Learning for UAV Swarms: an Approach of Latency Minimization

Hierarchical Federated Learning with Adaptive Clustering on Non-IID Data

Decentralized Federated Learning under Communication Delays.

Federated Transfer Learning Aided Interference Classification in GNSS Signals

Optimizing Data-Driven Federated Learning in UAV Networks

Enhancing Federated Learning Convergence with Dynamic Data Queue and Data Entropy-driven Participant Selection

Federated Learning Based on Feature Decoupling and Domain Adaptation

A Distributed Aggregation Approach for Vehicular Federated Learning

FedSBS: Federated-Learning participant-selection method for Intrusion Detection Systems

FedSpace: An Efficient Federated Learning Framework at Satellites and Ground Stations

Federated Learning in Practice: Reflections and Projections

Delay-Aware Hierarchical Federated Learning

Federated Multi-Task Learning under a Mixture of Distributions

Federated Learning with Heterogeneous Data Handling for Robust Vehicular Object Detection

Federated learning on non-IID and long-tailed data via dual-decoupling

Federated Learning with Additional Mechanisms on Clients to Reduce Communication Costs

Cooperative Federated Learning over Ground-to-Satellite Integrated Networks: Joint Local Computation and Data Offloading

Multi-round decentralized dataset distillation with federated learning for Low Earth Orbit satellite communication