Improved algorithm for de-interleaving radar signals with overlapping features in the dynamically varying electromagnetic environment

As an indispensable part of electronic support measure, the de-interleaving technique is used to separate interleaved radar pulse streams. At present, clustering based on radar features is one of the most effective de-interleaving methods. In a dynamically varying electromagnetic environment, the features of intercepted radar pulses overlap each other. Compared with other de-interleaving algorithms, the fuzzy adaptive resonance theory (Fuzzy ART) has obvious advantages in classifying such radar features. However, it still faces several problems: (i) the vigilance parameter used to regulate de-interleaving performance is difficult to reach its optimal value and (ii) since the unified discrimination threshold is selected for different regions, the algorithm suffers from category proliferation problem. This study addresses these problems by constructing a new vigilance model to replace the unified vigilance parameter and introducing a dual-vigilance mechanism to ART-based de-interleaving systems. It demonstrates this idea in the context of Fuzzy ART, presented as Fuzzy ART based on a 3D fuzzy model with two vigilance thresholds (2VT-3DFA). 2VT-3DFA suppressed the excessive proliferation of categories, and its clustering quality was 20% higher than that of conventional algorithms in dynamically varying signal environment.