A new method for parameter estimation of high-order polynomial-phase signals

- A new method for parameter estimation of high-order polynomial phase signals (PPS) is proposed.
- The method reduces the order the PPS order to one by using two kinds of operators. Then the highest-order phase parameter is estimated by fast Fourier transform and one dimensional search.
- The sequence and the number of times of the two kinds of operators depend on the PPS order.
- Lower-order phase parameters are estimated by repeatedly performing the same procedure on dechirped signal.

Parameter estimation of a high-order polynomial phase signal (PPS) is considered in this paper. We propose a method to estimate phase parameters more efficiently and accurately. In the proposed method, we define an operator referred to as non-uniform sampled reducing-order operator (NURO) to reduce the order of a polynomial phase by half, when the order of PPS is even. By combined using NURO and phase differentiation (PD) operators, the PPS order is reduced to one, i.e., the PPS degenerates into a complex sinusoid. Then, the parameter estimation can be done by jointly using fast Fourier transform (FFT) and one-dimensional search. Compared with the traditional methods, the reducing-order procedure in the proposed method has lower-order nonlinearities. Simulation results show that the proposed method outperforms the hybrid CPF-HAF and HAF in both mean square error (MSE) and the threshold when the PPS order is higher than 5.