Spectrum monitoring of radio digital video broadcasting based on an improved generative adversarial network

Due to the inherent broadcast nature of wireless communication systems, instances of radio jamming are common, such as natural interference and man-made interference, resulting in increasing demands for radio monitoring. A spectrum monitoring method based on a generative adversarial network model that is one of the most promising approaches of learning any kind of data distribution using unsupervised learning was proposed in this paper for the detection of anomaly spectrum with impulse noise. To validate the performance of the proposed model, both the simulated data set and the measured data set of radio digital video broadcasting were used to train and test the model. Experiments on the two data sets reached a consistent conclusion: as long as the energy of the interference is greater than a certain threshold, the detection accuracy increases with the increase of the interference power and pulse width. Compared with the existing anomaly detection models, our model was faster and more stable.