Frequency tracking-adaptive filtering vs. time-frequency distributions

Frequency tracking using traditional least-mean-square (LMS) adaptive algorithms and time-frequency distributions (TFD) are compared. Two problems arise when time-frequency distributions are used for frequency tracking: the presence of cross terms and sidelobes. Methods to adequately suppress the cross terms and sidelobes for frequency tracking purposes are discussed. The important parameters for both methods are presented, and the results demonstrate, in most cases, the superior performance of TFDs versus traditional LMS adaptive algorithms