SCAAML - Side Channel Attacks Assisted by Machine Learning

Background

SCAAML showcases the use of Machine Learning assist in Side-channel attacks against cryptography hardware, in an automated fashion.

SCAAML paper:

Training Models

See ./notebooks/EndToEndAttack.ipynb for an example of training a model to learn to discern the key byte from the trace.

Running Attacks

See ./notebooks/AttackOnly.ipynb for an example of training a model to learn to discern the key byte from the trace.

Results

In short, this can be trained to learn to discern the correct byte for the TinyAES implementation with 100% accuracy, given a small handful of traces measured from the chip while using the same key.

More comprehensive details coming soon.