How to make a plot the correlation coefficient VS powertrace



Can you send an example of the type of trace you mean? Do you mean something like this (from the XMEGA Attack paper)?:

If so the current SW doesn’t do that, I’d have to check how to manually perform that. It’s actually something I’ve been meaning to add in the back of my mind, so if other people want it can speed up my effort…




Sure thing - give me a few days. As mentioned I haven’t actually done that yet but it’s on my todo as well, since I can use that myself. Having some external pressure is always good to know there is interest!


FYI: the git version of the CW software now includes this feature, which means it will be in the final 0.09 release. Here’s an example:




The code is at …, I don’t know if it’s much help as it’s all very specific to both the CW stats format + Python. But perhaps you can unwind enough to make it useful for you!




Might be able to reduce further - I don’t know off-hand what is typical on the SAKURA-X, I haven’t used it. But it’s at least in the ‘ballpark’ for a standard CPA style attack.


@Coflynn @aliakbarp i am not coder and i am only familiar with matlab code, in addition i am facing same problem that i can’t plot number of traces vs correlation coefficients and also don’t know how to do PGE.

If you can help me in that.

Kindly suggest me Matlab/C solution.

Thank you.



The ChipWhisperer software can do Traces vs. Correlation coefficient (its one of the windows that comes up already). You could look at the code in the software if it helps?