Incremental Deinterleaving of Radar Emitters

Abstract

In this paper, we present a new paradigm for the deinterleaving of radar signals in electronic support systems. Traditional systems observe the electromagnetic spectrum for a longer period of time and perform the deinterleaving operation batch-wise. This paper presents a new approach, where the deinterleaving is performed after each detected radar pulse, which we call pulse-to-pulse incremental deinterleaving. It enables minimum reaction time of the system for the detection of radar emitters. For that purpose the paper presents three incremental clustering algorithms based on DBSCAN, Adaptive Resonance Theory and the Leader approach. The algorithms are evaluated using simulated PDW data of realistic complex radar scenes. The results show, that incremental approaches can achieve a similar accuracy as non-incremental methods, but have much faster reaction times to the occurrence of new emitters.