Noise assisted multivariate empirical mode decomposition applied to Doppler radar data

Abstract

The operation of the noise-assisted multivariate empirical mode decomposition (NA-MEMD) algorithm, which represents a breakthrough in data-adaptive analysis, is illustrated for the time-frequency analysis of Doppler radar signals. The performance of the NA-MEMD is here compared to the continuous wavelet transform, for both synthetic and real-world data applications, showing the advantage of the noise-assisted concept in terms of sparse time-frequency localization. For the considered application, we show how the approach gives a clear and natural interpretation of Doppler radar phenomena and enables the accurate tracking of object speeds.

Flow rate measurement of oil-gas-water wavy flow through a combined electrical and ultrasonic sensor

“…It is the extension of standard EMD [47] for multivariate data and the improvement of standard MEMD for avoid mode mixing. It uses extra independent white noise channels as references in the timefrequency domain, and adds them into the original multivariate signal to enable a more accurate time frequency decomposition [48]. Through a sifting process, NA-MEMD algorithm can produce the same number of intrinsic mode functions (IMFs) for all channels based on the regular MEMD procedures [49], and these IMFs are aligned in the corresponding frequency subbands.…”

Noise-Assisted Multivariate Empirical Mode Decomposition Based Emotion Recognition

“…In order to overcome these disadvantages a new approach called noise-assisted multivariate empirical mode decomposition (NA-MEMD) has recently been developed [24]. As an important step in data-adaptive analysis, the applications of NA-MEMD have so far been utilized resulting in positive outcomes, which are based on time-frequency axes with Doppler radar signals computer simulations and motor image EEG data from the BCI competition IV data set in time-frequency analysis of neuronal populations with instantaneous resolution phase synchronization using EEG-based prediction of epileptic seizures, in lung-heart sound discrimination, multichannel EMG signals, and rejecting the unwanted noise contained within the VLF-EM (very low-frequency electromagnetic method) data, which produced NA-MEMD [25][26][27][28][29][30][31]. Additionally, as an improved noise assisted method for multivariate signals decomposition, partial noise assisted multivariate EMD is proposed by Huang et al [32].…”

Bayesian Nonparametric Spectral Estimation

“…The need for frequency representation arises naturally in a number of disciplines such as natural sound processing [1,2], astrophysics [3], biomedical engineering [4] and Doppler-radar data analysis [5]. When the signal of interest is known without uncertainty, the frequency representation can be obtained by means of the Fourier transform [6].…”