Detecting Radio Frequency Interference With Signal Fingerprints Classification

Satellite networks are essential worldwide and are often the only means of communication in remote regions. Because these networks rely on wireless communication, they are vulnerable to harmful interference signals, making it necessary to identify the source of interference. Geolocation is the main technique used for this purpose. However, geolocation may indicate an area containing multiple ground stations, which makes it difficult to determine the actual source of harmful interference. This work proposes a method to reduce the number of candidate stations identified by geolocation by applying classification models to radio frequency fingerprint features extracted from the signals. Experimental results using real data, consisting of 64,800 signal instances from six ground stations, show that the proposed method achieves an accuracy above 98%.