The Synchrosqueezing algorithm: a robust analysis tool for signals with time-varying spectrum

We analyze the Synchrosqueezing transform, a consistent and invertible time-frequency analysis tool that can identify and extract oscillating components (of time-varying frequency and amplitude) from regularly sampled time series. We first describe a fast algorithm implementing the transform. Second, we show Synchrosqueezing is robust to bounded perturbations of the signal. This stability property extends the applicability of Synchrosqueezing to the analysis of nonuniformly sampled and noisy time series, which are ubiquitous in engineering and the natural sciences. Numerical simulations and examples in paleoclimatology show that Synchrosqueezing provides a natural way to analyze and filter a variety of signals.

• Type: Journal
• Year: 2011
• Where: CORR
• Authors: Eugene Brevdo, Neven S. Fuckar, Gaurav Thakur, Hau-Tieng Wu
• Keywords: CORR 2011, Education, Noisy Time Series, Time Series, Time-frequency Analysis Tool