Spectrum Waterfall Completion in Jamming Enviroment: A General Adversarial Networks Method

It is very important for completing spectrum data in the wireless communication system. The traditional completion method mainly considered the missing completion operation of sparse data. To solve this problem, we proposed an efficient algorithm based on generative adversarial network (GAN), which can automatically mine the relationship of the data and complete the missing data accurately. The proposed method uses the sensing spectrum waterfall as the input sample of the network structure. We adopt a pre-classification module and generation module based on GAN to generate incomplete spectrum data in multiple jamming patterns. Simulation results show that the proposed algorithm can effectively complete the missing data, and its performance is better than that of the data completion algorithm without a clssifier.