Hello Aya,

There are a few possible ways of doing it in Matlab.

To make sure we are on the same page: you want to test if there is a significant difference in means of two signals (time series represented as vectors) ?

Assuming so and If the signal values conform to normal distribution you can compute a parametric test using the following function (with settings depending on the nature of your data, look up documentation): ttest2 (two-sample t-test; part of statistical toolbox), e.g.

[h,p,ci,stats] = ttest2(sig1,sig2); with 2 signal vectors

You can also do a non-parametric test (again stat toolbox required), such as Kolmogorov-Smirnov:  kstest2 with the similar syntax as for t-test .

I hope it makes sense?

Iwo

Hello Iwo,

Thank you a lot for your answer,

Actually I was using the ttest2 but my signals haven't a normal distribution.

So I will try the kstest2 :) !

For non-normal data you can test the statistical significance using the Wilcoxon rank sum test:

p = ranksum(x,y)

or the Kruskal-Wallis test:

p = kruskalwallis(x)

Thank you Frederik

You could try this ...

Conference Paper A New Statistical Measure of Signal Similarity

Thank you all :)