Multivariate Fast Iterative Filtering for the decomposition of nonstationary signals

TL;DR

This paper introduces Multivariate Fast Iterative Filtering (MvFIF), a new algorithm for decomposing nonstationary multivariate signals, demonstrating its convergence, robustness, and superior performance compared to existing methods.

Contribution

The paper presents MvFIF, a novel, fast, and reliable multivariate signal decomposition method with proven convergence and superior performance over existing techniques.

Findings

Successfully separates multivariate modulated oscillations

Produces a quasi-dyadic filterbank for white noise

Robust to noise perturbations even with many channels

Abstract

In this work, we present a new technique for the decomposition of multivariate data, which we call Multivariate Fast Iterative Filtering (MvFIF) algorithm. We study its properties, proving rigorously that it converges in finite time when applied to the decomposition of any kind of multivariate signal. We test MvFIF performance using a wide variety of artificial and real multivariate signals, showing its ability to: separate multivariate modulated oscillations; align frequencies along different channels; produce a quasi--dyadic filterbank when decomposing white Gaussian noise; decompose the signal in a quasi--orthogonal set of components; being robust to noise perturbation, even when the number of channels is increased considerably. Finally, we compare its performance with the one of the main methods developed so far in the literature, proving that MvFIF produces, without any a priori assumption on the signal under investigation and in a fast and reliable manner, a uniquely defined decomposition of any multivariate signal.