ACF-based classification of phase modulated waveforms

Classification of radar waveform phase modulation based on a sequence of observations is a simple problem complicated by a number of nuisance parameters. Without prior knowledge of waveform carrier frequency, time offset, amplitude, initial phase, and bandwidth, application of a matched filter classifier is not achievable. One must instead rely on waveform features that are invariant to these parameters. This paper presents an invariant feature set for waveform classification. Testing of these features on simulated radar waveforms has illustrated the desired invariance properties and robustness to low signal-to-noise ratio.