Refinement in the estimation of multicomponent polynomial-phase signals

In this paper, we consider precise parameter estimation of multicomponent polynomial-phase signals (mc-PPSs). Two estimation refinement methods are proposed and both are initialized by coarse estimates provided by any technique for the mc-PPS estimation. The first method refines components, one by one, by estimating them once again, now from the received signal with all the components suppressed, except for the considered one. In the second method, the recently proposed estimation refinement technique for single PPS based on the filtering and phase unwrapping has been extended to the mc-PPSs. The first method approaches the Cramér-Rao lower bound (CRLB) within a few dBs, whereas the second one practically reaches the CRLB. Both methods are characterized by significantly lower complexity than computationally intensive nonlinear least squares (NLS) methods. Simulations confirm the effectiveness of the proposed methods.