Implementing Fixed-Width Sliding Windowing technique on sensor signal

Question

I am trying to replicate the signal pre processing done on this Data Set: http://archive.ics.uci.edu/ml/datasets/Human+Activity+Recognition+Using+Smartphones#

"The sensor signals (accelerometer and gyroscope) were pre-processed by applying noise filters and then sampled in fixed-width sliding windows of 2.56 sec and 50% overlap (128 readings/window)"

I am trying to perform all my functionalities on Matlab and am stuck in trying to create a fixed width sliding window on my acceleration signals.

My main area is AI classification and have no background knowledge on signal preprocessing etc.. I've been trying to research etc for a long time but I cannot understand what I need to do to my signal.

I have signals which are produced at a 51.2 Sampling Frequency. Any help please on how I can generate a sampling window like the one done on the linked data set?

Answer 1

You should calculate window length in number of samples and step=window-overlap and use this:

https://stackoverflow.com/a/44190634/3344428 (you can clear unnecessary code for std, skewness etc.)

You should clarify window length, because you mentioned "128 readings/window", BUT if you use 2.56 sec window on 51.2 Hz signal, then window = 2.56*51.2 = 131 samples.