Abstract:

This paper considers parameter estimation of a hybrid sinusoidal frequency modulated (FM) and polynomial phase signal (PPS) from a limited number of samples. We first show limitations of an existing method, the high-order ambiguity function (HAF), and then propose a new method by adopting the high-order phase function which was originally designed for the pure PPS. The proposed method estimates parameters of interest from peak locations in the time-frequency rate domain, which are less perturbed by the noise than peak values used by the HAF-based method. Numerical evaluation shows, the proposed method can handle the hybrid FM-PPS signal with low sinusoidal frequency and improve estimation accuracy in terms of mean squared error for several orders of magnitude.