Running Algorithm Using Device Accelerometer

Project #747 on iSENSEProject.org

This project is set up to collect data from a mobile device that is recording accelerometer data and performing some calculations to collect a weighted moving average of a person in motion.

Data uploaded here will be studied in attempt to determine how one can tell if a user is standing, walking, running, or shaking the device strictly from accelerometer data.

See the publication of which my algorithm is based on here: http://www2.fiit.stuba.sk/~bielik/publ/abstracts/2012/tomlein-et-al-healthinf2012.pdf

---------

Right now, there seems to be a correlation to the WMA (weighted moving average) to the type of activity, regardless of whether the user's phone is in the user's hand or pocket.  Given below are WMA ranges and activity associated with them:

0 - 0.1: standing
0.1 - 0.7: walking
0.7 - 2.5: running
0.5 - 2.5, >2.5: shaking device (depending on proximity of local peaks)

Due to the overlap of running with shaking the device, a distance will have to be measured between successive min and max local peaks to determine if local-min to local-max peak appears to be a viable step or too quick (thus being a shake).