fr4nky4ng / CDAE-Towards-Empowering-Denoising-in-SCA

CDAE-Towards-Empowering-Denoising-in-SCA

This repository contains codes and data from the paper entitled "CDAE: Towards Empowering Denoising in Side-Channel Analysis".

Discriptions

The following jupyter notebooks record the experiment details completely:

• cdae_aes_gpu.ipynb : records experiment on AES_GPU, the dataset is in AES_GPU,

• cdae_dpav2.ipynb : records experiment on DPAv2, the dataset can be generated by DPAv2_generate.ipynb,

• cdae_ascad.ipynb : records experiment on ASCAD, you may download the dataset yourself.

The following python scripts are neccessary for running the notebooks:

• TemplateAttacks.py: a Template Attacks class for profiling and attack,

• Evaluation.py: computes Guessing Entropy and Success Rate,

• PoiSelection.py: select POI for Template Attacks,

• utlis.py: some useful gadgets.

The folders contain the figures, results, traces and neural network weights:

• fig: all figures in the paper

• Traces: raw traces and denoised traces,

Datasets

All datasets are obtained from publicity databases:

• AES_GPU: https://github.com/DAE-Towards-Empowering-Denoising-in-SCA/AES_GPU

• DPAv2: http://www.dpacontest.org/v2/download.php

• ASCAD: https://github.com/ANSSI-FR/ASCAD