Comparisons of parametric methods for polynomial phase signal parameter estimation

In this paper, parametric methods for parameter estimation of polynomial phase signals, such as the suboptimal likelihood estimation and the maximum likelihood estimation, are discussed and compared. Comparisons are given in three aspects: the variances of the estimates, the output SNR, and the effect of the error propagation and cross terms. It shows that the suboptimal likelihood estimation methods are more efficient than the maximum likelihood estimation. However, the performances achieved from the suboptimal methods degrade rapidly when the SNR is below certain threshold, while the maximum likelihood estimation is free from the threshold effect. Moreover, unlike the suboptimal likelihood estimation, the maximum likelihood estimation method does not suffer from the error propagation and the cross terms due to its linearity.