What question did this study set out to answer?

This research aims to evaluate the effectiveness of federated learning approaches for intrusion detection in UAV networks under privacy constraints.

Federated Learning for Intrusion Detection in UAV Networks

View Full Paper

Ask AI

Bookmark

Authors

Randomized trial assesses performance of federated learning for intrusion detection in UAV networks, indicating strong privacy benefits.

This research aims to evaluate the effectiveness of federated learning approaches for intrusion detection in UAV networks under privacy constraints.
Constructed a federated intrusion detection system using the UAVIDS-2025 benchmark with $K=176$ non-IID clients.
Trained a multilayer perceptron across three federated optimizers: FedAvg, FedProx, and SCAFFOLD.
Analyzed communication footprints and tested performance on a second dataset (T-ITS) to evaluate transfer trends.
FedAvg achieved $93\%$ accuracy, $93\%$ macro-F1, and AUROC $0.99$, performing within 3 percentage points of centralized models (96\%).
FedProx yielded $86\%$ accuracy and $84\%$ macro-F1, while SCAFFOLD saw $82\%$ accuracy and $79\%$ macro-F1 but struggled with classwise performance.

Berke Ozsoy et al. (2026) studied this question.

View Full Paper

Ask AI

Bookmark