A Novel Cluster-Sorting Method for Radar Signals Based on Graph Autoencoder
Yongchun Li, Ling Yang, Yifang Zhang, Chunxia Chen, Tao Leí, Zhibin Yu · 2021 International Conference on Control, Automation and Information Sciences (ICCAIS) · 2021
Aiming at the problems of traditional clustering algorithms that require manual screening of features, only learn the attribute information of the input signal, and have poor results in radar signal sorting. In this paper, a novel cluster-sorting method for radar signals based on Graph Autoencoder (GAE) is proposed. The key of this method is to obtain the adjacency matrix of the radar signal data by using the K nearest neighbor (KNN) algorithm, and then obtain the graph structure representation of the radar signals. The low-dimensional vector embedding is obtained after Graph Autoencoder learns the relevant information of the graph. Next the K-Means method is applied to the clustering of low-dimensional vector embedding, and the cluster-sorting of radar signals are completed. Finally, the proposed algorithm is compared with five typical clustering algorithms. The experimental results show that the proposed algorithm is better than the typical method in radar signal cluster-sorting, and the signal sorting accuracy is higher 12.5%.