Comprehensive Review of Pulse Repetitions Interval (PRI) Classification Schemes

This research is focused on comprehensive analysis of the Pulse Repetition Interval (PRI) Classification schemes; applied to the intercepted pulses by an Electronic Intelligence (ELINT) receiver / Radar Warning Receiver. The intercepted pulse train is interleaved in time; these pulses are first pre-sorted on the basis of some fundamental parameters of the captured pulses termed as Pulse Descriptor Words (PDWs); then de-interleaved on the basis of Statistical Time of Arrival (TOA) analysis. The first step after de-interleaving and association of PDWs to candidate emitters is the PRI classification, which then serves as one of the most important parameters for emitter identification. There are a total of six different PRI types (e.g. Constant, Staggered, Jittered, Sliding, Dwell & Switch and Sinusoidal). The PRI classification schemes found in literature are neither comprehensive (i.e. able to differentiate between all six types) nor computationally efficient as most of them employ machine learning techniques. A PRI classification technique was proposed by the authors, which employed feature extraction out of these pulse trains followed by simple threshold based hierarchical classification on top of the feature set to classify all six types of PRIs. A comparison of this scheme with all other schemes not only proves this scheme superior to other schemes in most of the test cases but also proves it to be computationally efficient due to its simplistic approach. A key parameter used for comparison is PRI Classification rate; determined on the basis of synthetic data generated in MATLAB. Robustness of these algorithms were also tested against impurities like Missing pulses, Spurious pulses and Incidental jitter which are inherent part of the ELINT receiver / RWR.