P. Ubolkosold, G.F. Tchere, S. Knedlik, and O. Loffeld (Germany)
Frequency offset, nonlinear least-squares, extended Kalman filter, Cramer-Rao bound.
In this paper, the problem of extracting frequency offset from single frequency signal with a slowly time-varying amplitude is addressed. Herein two nonlinear frequency offset estimators are presented. The first relies on an ex tended Kalman filter (EKF) approach which is suitable for recursive or on-line processing. The second is derived us ing nonlinear least-squares (NLS) principle which is suit able for batch or off-line processing. The performances, mean-squared error (MSE), of these two estimators are compared to the Cramer-Rao-Bound (CRB). The simula tion results show that EKF has the same performance as NLS at high signal-to-noise ratio (SNR) with less compu tational complexity.
Important Links:
Go Back
