Pulse Sequence Segmentation for Multifunction Radar Using Multiple Reconnaissance Receivers

Multifunction radar (MFR) can simultaneously execute multiple different tasks on the radar timeline, and has the characteristics of flexible working mode, instantaneous switching of beam direction, and adaptive signal modulation and parameters. From the perspective of the electronic reconnaissance, pulse sequence segmentation for MFR, followed by clustering segments into waveform units or radar words, is the primary task of recognizing MFR's behavior or state. This paper develops two unsupervised pulse sequence segmentation methods using the stepped pulse amplitude sequences intercepted by multiple reconnaissance receivers from different observation angles, and both of them have two core steps: sliding window based hypothesis test excluding the case that pulses in the sliding window belong to the same waveform unit, and adjacent points mergence further merging false switching points into true switching points. Computer simulations verify that the proposed methods could implement pulse sequence segmentation without any prior information, and exhibits superior performance than the competing methods in harsh conditions, such as extremely low signal to noise ratio, high rate of missing or spurious pulses.