Turning Tangent Empirical Mode Decomposition: A Framework for Mono- and Multivariate Signals

Area of Science:

• Signal Processing
• Data Analysis
• Algorithm Development

Background:

• Empirical Mode Decomposition (EMD) is a vital tool for analyzing non-linear and non-stationary signals.
• Existing EMD methods can be computationally intensive and complex, particularly for multivariate data.
• There is a need for more efficient and simpler signal decomposition techniques.

Purpose of the Study:

• To introduce a novel Empirical Mode Decomposition (EMD) algorithm, termed 2T-EMD.
• To demonstrate the computational lightness and algorithmic simplicity of 2T-EMD compared to existing methods.
• To validate the performance of 2T-EMD on simulated and real-world data.

Main Methods:

• The 2T-EMD algorithm redefines the signal mean envelope using new characteristic points.
• This novel approach enables the decomposition of multivariate signals without requiring signal projection.
• The method was compared against classical techniques using simulated mono- and multivariate signals.

Main Results:

• 2T-EMD demonstrated computational lightness and algorithmic simplicity.
• The algorithm successfully decomposed simulated mono- and multivariate signals, including noisy data like fractional Gaussian noise.
• The effectiveness of 2T-EMD was further validated through an application to real-life electroencephalogram (EEG) data.

Conclusions:

• 2T-EMD presents a computationally efficient and simple alternative for signal decomposition.
• The algorithm is suitable for both mono- and multivariate signal analysis, including noisy and real-world datasets.
• 2T-EMD shows promise for applications in various scientific and engineering fields requiring signal analysis.