Apologies for the poorly formulated question -- I've been away from DSP for a while.
Let us say I have amplitude vs time data for N signals. where amplitude is -127 to 127. For each signal I have say 900 to 1100 time samples.
I have a priori knowledge that the N signals can be grouped into n clusters where n << N.
I can take a guess at n. So for example if N=1000 I can guess that n=10.
If a signal belongs to a cluster -- then the sum of square error of the amplitude/time between the signal and the mean of the aplitude/time of the cluster is less than a equivalent measurement between all other such clusters.
The problem is that all of these signals are not aligned at time. So there is uncertainty when the first measurement was taken. The uncertainty for any signal is much less than the size of the signal. So say I have 1000 time measurements for a signal then the first measurement is at most 10 units of time delayed from zero.
Question: Given the above how to I cluster the signals given such phase distortions.
Thanks.
