A Novel Instantaneous Phase Detection Approach and Its Application in SSVEP-Based Brain-Computer Interfaces

Area of Science:

• Signal Processing
• Biomedical Engineering
• Machine Learning

Background:

• Accurate phase estimation is crucial for analyzing complex signals.
• Conventional methods like Discrete Fourier Transform (DFT) suffer from spectral leakage and require corrections.
• Existing phase estimators for brain-computer interfaces (BCIs) have limitations regarding sampling rates and stimulus frequencies.

Purpose of the Study:

• To introduce a novel phase estimator based on fully-traversed DFT.
• To demonstrate the merits of direct phase extraction and suppressed spectral leakage.
• To validate the estimator's performance in noisy multi-tone signals and SSVEP BCIs.

Main Methods:

• Development of a phase estimator utilizing all possible truncated DFT spectra (fully-traversed DFT).
• Theoretical proof showing the estimator adheres to a 2-parameter joint estimation model.
• Numerical simulations and real-world data analysis using SSVEP BCI data.

Main Results:

• The proposed estimator achieves direct phase extraction without correction.
• It effectively suppresses spectral leakage, outperforming conventional DFT.
• Demonstrated superior classification of SSVEP phases in BCI applications compared to traditional DFT.
• Proven compliance with a 2-parameter joint estimation model.

Conclusions:

• The fully-traversed DFT phase estimator offers significant advantages in accuracy and efficiency.
• It provides a more robust method for analyzing noisy multi-tone signals.
• The estimator's flexibility in handling sampling and stimulus frequencies broadens its applicability in phase-coded SSVEP BCIs.