Spectrum Analysis for Modulation Classification in Aeronautical Wireless Communication Systems

The degeneration of signal time-frequency characteristics has impeded the performance of correlation algorithms. To address this issue, a generative adversarial network (GAN)-based recognition framework is proposed for aeronautical wireless communication systems. It consists of GAN and different dimensional of recognition networks. Firstly, we transform the sampling signals into time-frequency (TF) maps using a short-time Fourier transform (STFT), which shows an apparent signal frequency variation with time. And then, we design an improved GAN, transferring the TF maps affected by the multipath effect into pure maps, to weaken the interference of channels. Next, we put forward the two-dimensional (2D) recognition networks to extract signal time-frequency characteristics, and a deep long short-term memory (LSTM) network was introduced to obtain the time correlation from the TF maps. The experimental results show that the performance of the proposed GAN-based recognition framework is superior to that of conventional algorithms, especially performing in aeronautical multipath wireless channels. When the channel parameters change rapidly, the recognition rate of the proposed algorithm is more than 95.0%.