The application of data-driven TF analysis methods in LFM signal parameter estimation

Based on Empirical wavelet transform (EWT) and sparse time-frequency analysis method, two new LFM parameter estimation methods are proposed. EWT method builds adaptive wavelets and decomposes the LFM signal into different modes, which can be processed through energy-oriented principal component extraction (EPCE) method to estimate the parameters of LFM signal. The second method tries to find the sparsest representation of multi-scale data within dictionary consisting of AM-FM intrinsic mode functions (IMF) through solving nonlinear L1 optimization problem. Comparisons are made with EEMD based EPCE method to show the usefulness of these two methods.