A Case-Study on Variations Observed in Accelerometers Across Devices

TL;DR

This study investigates the variability in accelerometer data across different wearable devices, highlighting that device specifications significantly influence data discrepancies, which are important for accurate data interpretation.

Contribution

The paper provides a detailed analysis of accelerometer variations across devices, emphasizing the impact of device specifications on data consistency.

Findings

Significant variation exists between devices with different specs.

Variability persists even when recording the same activity with the same person.

Sensor data variance should be expected regardless of activity or user.

Abstract

Every year we grow more dependent on wearable devices to gather personalized data, such as our movements, heart rate, respiration, etc. To capture this data, devices contain sensors, such as accelerometers and gyroscopes, that are able to measure changes in their surroundings and pass along the information for better informed decisions. Although these sensors should behave similarly in different devices, that is not always the case. In this case study, we analyze accelerometers from three different devices recording the same actions with an aim to determine whether the discrepancies are due to variability within or between devices. We found the most significant variation between devices with different specifications, such as sensitivity and sampling frequency. Nevertheless, variance in the data should be assumed, even if data is gathered from the same person, activity, and type of sensor.