Advancements in Motion Artifact Reduction for Wearable PPG Sensors

Introduction

Photoplethysmography (PPG) has become ubiquitous in consumer wearables, yet its reliability during physical activity remains a significant hurdle. The fundamental issue lies in the overlap between the frequency spectrum of motion artifacts (0.1–10 Hz) and the physiological cardiac signal (0.5–3 Hz).

The Noise Floor Problem

In ambulatory settings, the signal-to-noise ratio (SNR) can drop below -10dB. Under these conditions, identifying the systolic peak becomes statistically random without a secondary reference sensor.

Algorithmic Approaches

We are observing a paradigm shift towards Generative Adversarial Networks (GANs) for signal reconstruction. These models reconstruct the correct clean signal guided by physiological constraints of the cardiac cycle.

Future Directions

The integration of multi-wavelength sensors allows for depth-resolved artifact subtraction. By comparing the AC/DC ratios across different penetration depths, we can spatially filter out surface-level motion noise.