A GNSS Composite Interference Recognition Method Based on YOLOv5

In complex electromagnetic environments, composite interference consisting of various types of interference has become a major threat affecting GNSS signals. Existing interference recognition technologies typically target single types of interference and have lower recognition rates and real-time performance for composite interference. This paper proposes a GNSS composite interference recognition method based on the YOLOv5 (You Only Look Once version 5) algorithm, aiming to quickly and accurately identify the types of interference in GNSS signals. The method first models the GNSS interference signals and uses smoothed pseudo Wigner-Ville distribution (SPWVD) for time-frequency transformation, generating time-frequency images of the interference. These images are preprocessed and used as a training set input to train the YOLO model. The trained neural network then detects the composite interference images. Utilizing YOLO's rapid detection capabilities allows for real-time identification and classification of composite interference. Experimental results show that the average recognition accuracy of this method for GNSS composite interference reaches $\mathbf{9 2 . 6 \%}$.