Quasi ML algorithm for 2-D PPS estimation

The 2-D quasi-maximum likelihood algorithm for estimation of 2-D polynomial phase signals (2-D PPSs) is proposed. Estimation of all phase parameters is performed in an efficient manner using search only over a set of the window widths in the 2-D short-time Fourier transform (STFT). The mean squared error is on the Cramer-Rao lower bound that is big advantage with respect to all existing phase differentiation techniques even for low SNR. The proposed technique consists of two basic stages: rough estimation that is performed using 2-D STFT and fine stage with dechirping, downsampling, filtering, and polynomial interpolation. Obtained results are excellent even for 2-D PPSs of high-order or in the case when modulation is nonpolynomial.