Composite signal detection using multisynchrosqueezing wavelet transform

It is important to study recognition and recovery of effective signal components for information extraction of composite signals. However, owing to the limitation of time-frequency (TF) resolution, composite signal detection has always been a difficult problem, particularly for signals with crossover frequencies, namely crossover signals. This study proposes a new approach to solve signal estimation based on optimization of TF distribution. Firstly, based on the theory of multisynchrosqueezing transform (MSST), a high-order multisynchrosqueezing wavelet transform (HMSWT) that can accurately estimate the nonlinear instantaneous frequency (IF) is studied. On this basis, wavelet coefficient reassignment is extended to two dimensions by introducing the chirp rate, and a high-order multisynchrosqueezing chirplet transform (HMSCT) is proposed to realize cross-component parameter estimation. This study realizes the theoretical expansion of MSST and performs multiple reassignments on a three-dimensional time-frequency-chirp rate (TFC) framework for the first time. Numerical experiments on typical multi-component FM signals show that the proposed method can provide a more accurate TF representation and high-precision component separation.