UAV Swarm Automatic Modulation Recognition with Multiple Interferences.

The complex factors such as Doppler frequency shift, multipath interference and atmospheric noise in the channel of unmanned aerial vehicle (UAV) swarm pose a huge challenge to the automatic modulation recognition. In response to the issue, this paper proposes an automatic modulation recognition method which is based on high-order cumulants and cyclic spectrum for UAV swarm communication signals. Firstly, a UAV swarm channel model interfered with multipath effect and Alpha noise is established. Secondly, the generalized cyclic spectral values and high-order cumulants are calculated from the received signals and extracted to comprise the sample of the UAV swarm communication signals. Finally, the samples are standardized and fed into a deep sparse autoencoder neural network (DSAEN) to achieve recognition of seven modulation types. The simulation results show that the proposed method improves recognition accuracy to over 97.5% in line of sight (LOS) scenes when Mixed Signal Noise Ratio (MSNR) is 0dB, showing excellent recognition performance in Low MSNR and high robustness to different LOS scenes. The combination of high-order cumulants and generalized cyclic spectral features greatly improves the automatic modulation recognition performance. The proposed method has significant value for automatic modulation recognition of UAV swarm communication signals.