GNSS Interference Detection and Affected Area Assessment Based on ADS-B

Global Navigation Satellite Systems (GNSS) are fundamental to modern aviation infrastructure but remain vulnerable to radio frequency interference (RFI), which may cause flight delays, operational inefficiency, and critical safety risks. The Automatic Dependent Surveillance-Broadcast (ADS-B) system, reliant on GNSS for precise positional data, also suffers performance degradation under interference. This study proposes a two-stage framework for GNSS interference detection using ADS-B data. First, an enhanced feature engineering pipeline integrates Akima interpolation to reconstruct incomplete trajectories and extracts spatiotemporal patterns through sliding-window statistical features. Second, an unsupervised Long Short-Term Memory (LSTM) autoencoder addresses data incompleteness and label scarcity by learning latent representations of normal operations. Finally, grid-based spatial feature extraction combined with a Gradient Boosting Decision Tree (GBDT) model enables spatial interpolation to generate interference risk maps across target airspace. The proposed framework provides significant support for interference source localization and enhances airspace risk management, thereby helping to reduce safety risks in civil aviation under GNSS-compromised environments.