UAV Swarm Type Recognition Algorithm Based on Dual-Domain Fusion Features and Feature Selection

At present, the UAV type recognition algorithms in the literature only realize the identification of a single UAV type through the signal characteristics of the communication domain or radar domain, and there are problems such as low recognition accuracy. Aiming at the above problems, this paper proposes a UAV swarm type recognition algorithm based on the fusion characteristics of communication signals and radar signals. First, the high-order cumulant and instantaneous feature statistics of the swarm communication signal are extracted, and the radar track features are fused to construct the UAV swarm feature matrix; secondly, an improved feature selection algorithm-Secondary Screening of Neighbourhood Components Analysis (SSNCA) is proposed to reduce the dimensionality of the fusion feature matrix; finally, a Sparse Autoencoder Network (SAEN) is used for swarm type identification. The simulation results show that the algorithm significantly reduces the dimension of the swarm feature matrix (only 27% of the original matrix dimension); at the same time, when the Signal-to-Noise Ratio (SNR) is 0 dB, the correct rate of identifying five swarm types can reach 88%.