Denoising Physiological Signals: Assessment, Techniques, and Wearable Implementation

This paper presents research on denoising physiological signals, focusing on the challenges posed by motion artifacts during physical activity, especially for individuals with disabilities. To do that, we conducted a review of the computational cost of various denoising techniques and analyzed the necessity of their use and the feasibility of their implementation in wearable systems. A database of physiological signals during various exercises has been collected to assess the impact of movement on signal quality. We found that not all signals require denoising, as skin temperature measurements are largely immune to artifacts, while EEG and ECG signals are significantly affected by movement. Among the activities tested, jumping caused the most contamination. A test is run to determine the power consumption for different complexity levels of the denoising technique and sampling rates. The aim is to identify efficient denoising methods suitable for implementation on low-power wearable devices used in monitoring physical activity.