I am running AES 256 on CW305 (this effect is also taking place in factory AES 128). The higher frequency i use, I see a higher droop in the power trace like the following picture. What might be the reason? At 100MHz i do not see distinct clear peaks like lower frequency trace (each peak has other smaller peaks with it).
Following are two traces collected from oscilloscope, top one is running at 100MHz, bottom one is running at 20MHz. Scaling is different
This is just the nature of measuring power via a shunt. Prior to the capture, the target is essentially idle and consuming minimal power. When the target then starts the AES encryption, all that logic becomes active and causes a sudden change in the target’s power requirements. The faster the clock, the more drastic the sudden change in power requirement.
As a result, the AES rounds may be harder to see, but they are still there. Remember that side-channel attacks don’t need accurate absolute power measurements; they rely on relative power differences (which are still being captured, in spite of that “droop”).
