Analysis of Error-Agnostic Time- and Frequency-Domain Features Extracted From Measurements of 3-D Accelerometer Sensors

This paper investigates the expressive power of several time- and frequency-domain features extracted from 3-D accelerometer sensors. The raw data represent movements of humans and cars. The aim is to obtain a quantitative as well as a qualitative expression of the uncertainty associated with random...

Full description

Saved in:
Bibliographic Details
Published in:IEEE systems journal 2010-03, Vol.4 (1), p.26-33
Main Authors: Dargie, Waltenegus, Denko, Mieso K
Format: Article
Language:English
Subjects:
Accelerometer sensors
Accelerometers
Calibration
Correlation analysis
Data mining
Error analysis
Errors
Feature extraction
Frequency domain analysis
Frequency measurement
frequency-domain features
Humans
measurement errors
Movements
Placement
Raw
Sensor phenomena and characterization
Sensors
Three dimensional
time-domain features
Uncertainty
Wireless sensor networks
Citations: Items that this one cites
Items that cite this one
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:This paper investigates the expressive power of several time- and frequency-domain features extracted from 3-D accelerometer sensors. The raw data represent movements of humans and cars. The aim is to obtain a quantitative as well as a qualitative expression of the uncertainty associated with random placement of sensors in wireless sensor networks. Random placement causes calibration, location and orientation errors to occur. Different type of movements are considered-slow and fast movements; horizontal, vertical, and lateral movements; smooth and jerky movements, etc. Particular attention is given to the analysis of the existence of correlation between sets of raw data which should represent similar or correlated movements. The investigation demonstrates that while frequency-domain features are generally robust, there are also computationally less intensive time-domain features which have low to moderate uncertainty. Moreover, features extracted from slow movements are generally error prone, regardless of their specific domain.
ISSN:1932-8184
1937-9234
DOI:10.1109/JSYST.2009.2039735