ADS-B Signal Recognition Method Based On Entropy Feature Fusion

Aiming at the problem that traditional single signal feature extraction algorithms cannot fully describe signal features, this paper proposes an ADS-B signal feature extraction and recognition method based on fusion entropy features. This article uses the ADS-B signal data set collected in a real environment as the original signal. There are a total of 198 airplanes in the data set, each with 200-600 samples, 3000 sampling points for each sample, and a sampling frequency of 50MHz. Eight planes out of 198 planes are randomly selected, 20 samples are randomly selected for each plane, and the singular spectrum entropy, wavelet energy spectrum entropy and Renyi entropy of the samples are extracted. The singular spectrum entropy and wavelet energy entropy, singular spectrum entropy and Renyi entropy are feature fusion respectively, and the random forest model is used as the classifier to classify the ADS-B signal. The recognition rate can reach 97.5%, which verifies the feasibility of the method.