Compressive Sensing for IMU-based Gait Activity Assessment Using Discrete Wavelet Transform

Rapid increase in the volume of data generated from the embedded devices and sensors must be compressed with optimum compression ratio without losing information for sharing the resources utilizing low bandwidth and storing in low spaces. Specialized signal compression mechanisms exist for many biomedical signals like electrocardiograms. However, studies rarely exist in the state of the art to compress gait signals captured using Inertial Measurement Units. Among the works that exist, the average compression ratio is about 5.5. We propose a wavelet decomposition-based gait signal compression method for inertial measurement units. We experimented with gait activity data obtained from 10 human subjects with informed consent. From this experiment, we have achieved a state-of-the-art compression ratio of up to 7.4 with a percentage root mean square difference of 0.22. This signal compression method can be of immense importance in edge and fog computing environments as data transmission can be performed with minimal bandwidth consumption and can work in resource-constrained environments.