Multi-functional radar signal sorting based on weighted undirected graph features

Spaceborne passive reconnaissance system is the main technical path for radar radiation source perception. In modern complex electromagnetic environment, traditional radar signal sorting (RSS) methods face "batch increasing" and "batch decreasing" challenges from complex parameter variation pulses of multi-function radars (MFRs). To overcome the shortcomings above, this paper proposes an RSS method based on weighted undirected graph features, which constructs the interleaved pulse sequences into a complex network by using the sliding window nearest neighbor linking method. Then, this proposed method combines the label propagation algorithm and adaptive density peak clustering method to realize RSS of MFRs. Experiments based on simulated MFR dataset demonstrate that this method outperforms the existing RSS methods.