AF-MNS: A Novel AM-FM Based Measure of Non-Stationarity

Robust signal processing in any application depends on the representation and the understanding of the signals in hand. There is a plethora of signal processing methods, wherein choosing a method over others would depend on the attributes of the signals and their intended application. As of now, onl...

Full description

Saved in:
Bibliographic Details
Published in:IEEE communications letters 2021-03, Vol.25 (3), p.990-994
Main Authors: Singh, Pushpendra, Singhal, Amit, Fatimah, Binish, Gupta, Anubha, Joshi, Shiv Dutt
Format: Article
Language:English
Subjects:
AM-FM representation
Amplitude modulation
Amplitudes
Audio frequencies
Electroencephalography
Fourier decomposition method (FDM)
Frequency modulation
Hilbert space
Hilbert transform (HT)
Hilbert transformation
instantaneous amplitude (IA)
instantaneous frequency (IF)
measure of non-stationarity (MNS)
Modulation
Pattern classification
Representations
Signal classification
Signal processing
Signal representation
Citations: Items that this one cites
Items that cite this one
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Robust signal processing in any application depends on the representation and the understanding of the signals in hand. There is a plethora of signal processing methods, wherein choosing a method over others would depend on the attributes of the signals and their intended application. As of now, only spectral stationarity is discussed for the deterministic signals. This letter presents a novel amplitude-modulation-frequency-modulation (AM-FM) based measure of non-stationarity of the deterministic signals by extracting AM-FM components using the Fourier decomposition method (FDM). FDM generates analytic signal representation, where the real and imaginary parts are the Hilbert transform pairs. Simple and intuitive measures of amplitude and frequency non-stationarity are presented to quantitatively assess the extent of non-stationarity. Both these measures are zero for stationary signals, while at least one of them is positive for the non-stationary signals. Important observations regarding the stationarity of a signal are made with the help of some examples. The utility of the concept is demonstrated via an application of EEG signal classification.
ISSN:1089-7798
1558-2558
DOI:10.1109/LCOMM.2020.3041722