Modeling of Noisy Acceleration Signals From Quasi-Periodic Movements for Drift-Free Position Estimation.

Area of Science:

• Biomechanics
• Signal Processing
• Wearable Technology

Background:

• Noisy acceleration signals from body movements complicate accurate position estimation.
• Existing methods struggle with time-variant harmonic structures and low signal-to-noise ratios.

Purpose of the Study:

• To present a novel data-driven approach for drift-free position estimation from noisy acceleration signals.
• To accurately describe time-variant harmonic structures in single-channel acceleration data.
• To enable reliable position estimation in low signal-to-noise ratio conditions.

Main Methods:

• Short-time modeling of acceleration dynamics (harmonic amplitudes and phases).
• Analytical integration for short-time position calculation.
• Overlap-add recombination for full-length position synthesis.

Main Results:

• The proposed method demonstrated superior performance compared to state-of-the-art techniques.
• Outperformed reference methods in Euclidean error, RMSE, correlation coefficient, and HNR.
• Effectively suppresses non-movement-related acceleration components, eliminating low-frequency artifacts.

Conclusions:

• The developed method provides drift-free position estimation, even with noisy acceleration signals.
• It is highly effective in handling quasi-periodic small-amplitude body movements.
• Potential applications include elderly frailty assessment and clinical gait analysis in aging and rehabilitation.