Breaking with CPA on secAES masked ATmega

Embedded Security Side Channel Power Analysis
1

What I need to do for breaking this type of trace in AES. I don’t know how much samples I’ve needed to put because I tried to test some scales of samples (100, 316, 1000, 3160, 10000, 24400) and anyone break this AES. So what I need to do for break this trace. Extract from secAES version 1

bokeh_plot (3)
bokeh_plot (3)800×600 58 KB

2

I haven’t dealt with masking much, but I believe you typically need much more complicated attacks (higher order CPA/DPA attacks, the paper I listed earlier uses a neural network). I did find https://github.com/SideChannelMarvels/Daredevil from some searching, so you may want to check that out.

Alex

3

Thanks Alex, you are help me so much my friend :blush: