# Correlating accelerometer data

I am getting a signal from a device. This signal contains a non-random noise signal in it. I want to remove that noise from the signal. This noise is correlated to the motion the device is having. So now I've attached a accelerometer to the device to measure the motion produced by the device. I'm getting the data of X, Y, Z-axis from 3-axis accelerometer.

Now here is the process I thought to remove the noise.First get a motion signal from the 3 axis data which correlate to the motion noise present in the main signal. Then pass the main signal and the correlated noise signal through a LMS filter to get a filtered data. I've already designed a LMS filter but I'm stuck at how to get the motion signal(from the data of the 3-axis from accelerometer) which will correlate to the motion noise present in the main signal.

Can anyone help me with this plz??

• do you know how the noise is correlated to the motion? does it depend on acceleration, orientation, velocity, etc? – geometrikal Jul 8 '14 at 14:18
• For motion signal did you try magnitude of 3-axis accelerometer data i.e. sqrt(X^2 + Y^2 + Z^2) as feed to the LMS filter? What does the LMS filter produce? – Naveen Apr 13 '15 at 3:23

## 1 Answer

Motion can be described using displacement with time, velocity or acceleration. To get these quantities from the accelerometer data the following link may help. http://www.chrobotics.com/library/accel-position-velocity

• Looks like gyroscope would be needed to define the rotation matrix for conversion of accelerometer data from inertial plane to body plane. Is there any way to get a approximate estimate of the rotation matrix using only the accelerometer data? – Naveen May 4 '15 at 19:36