QML-RANSAC: PPS and FM signals estimation in heavy noise environments

• The QML-RANSAC is proposed for polynomial phase signals parameters estimation.
• Random sampling is employed in getting parameter estimates in each iteration..
• In each iteration obtained results are compared with current estimates using maximum likelihood inspired function.
• Obtained results are excellent surpassing current state-of-the-art techniques in the field.

The QML-RANSAC estimator is proposed. It combines the quasi-maximum likelihood (QML) estimator with the random sample consensus (RANSAC). This technique can with reasonable calculation complexity work for lower the signal-to-noise ratio (SNR) than existing parametric estimators of polynomial phase signals (PPS) and nonparametric estimators of FM signals, i.e., it achieves lower SNR threshold than the state-of-the-art techniques in the field. Obtained results are better for about 3 dB with respect to the QML in term of the SNR threshold without increasing the mean squared error (MSE) above the threshold. The proposed estimator is tested on the PPS as a parametric estimator and for general FM signal estimation as a nonparametric estimator. An extension of the algorithm is proposed for multicomponent signals, as well.