Using ML for Automated RF Side-Channel Attack Classification

This is a cool example of an ML project accessible to hackers.

Furthermore, it offers the way to combine SDR and ML for automated, almost industrialized attacks. Now you have a repeatable pipeline that works for many models of target device.

A fun read! Instead of actually designing the pipeline for detection one can just throw up a ML model (or many) and see what happens very quickly. The virtue signaling was pretty annoying though.

Interesting project and clear, logical write-up. Shows their practical approach to a real-world problem and (once again) demonstrates how "secure" devices often actually aren't. Even employing low-cost ML to augment the human endeavor... well done.