Missing sample recovery for wireless inertial sensor-based human movement acquisition

Kyoung Jae Kim, Vibhor R Agrawal, Ignacio Gaunaurd, Robert Gailey, Christopher Bennett

Research output: Contribution to journalArticle

8 Scopus citations


This paper presents a novel, practical, and effective routine to reconstruct missing samples from a time-domain sequence of wirelessly transmitted IMU data during high-level mobility activities. Our work extends previous approaches involving empirical mode decomposition (EMD)-based and auto-regressive (AR) model-based interpolation algorithms in two aspects: 1) we utilized a modified sifting process for signal decomposition into a set of intrinsic mode functions with missing samples, and 2) we expand previous AR modeling for recovery of audio signals to exploit the quasi-periodic characteristics of lower-limb movement during the modified Edgren side step test. To verify the improvements provided by the proposed extensions, a comparison study of traditional interpolation methods, such as cubic spline interpolation, AR model-based interpolations, and EMD-based interpolation is also made via simulation with real inertial signals recorded during high-speed movement. The evaluation was based on two performance criteria: Euclidian distance and Pearson correlation coefficient between the original signal and the reconstructed signal. The experimental results show that the proposed method improves upon traditional interpolation methods used in recovering missing samples.

Original languageEnglish (US)
Article number7419258
Pages (from-to)1191-1198
Number of pages8
JournalIEEE Transactions on Neural Systems and Rehabilitation Engineering
Issue number11
StatePublished - Nov 1 2016



  • Autoregressive (AR) model
  • empirical mode decomposition (EMD)
  • high-level mobility
  • inertial measurement unit (IMU)
  • missing sample recovery

ASJC Scopus subject areas

  • Neuroscience(all)
  • Biomedical Engineering
  • Computer Science Applications

Cite this