Department of Computer Science and Technology

Security Group

Grizzly: power-analysis traces for an 8-bit load instruction

The Grizzly dataset contains recordings of the power-supply current of the 8-bit CPU Atmel XMEGA 256 A3U, an easily available microcontroller without side-channel countermeasures. It was prepared by Omar Choudary in August 2013 and forms the basis for the comparison of different template attacks described in these papers:

The MATLAB scripts for reproducing some of the results in our papers are available on github.

Please consult the file readme.txt file for more details on the scripts and the datasets.

Data from different devices

Recordings from all four target AVR XMEGA 256 devices are available below.

Device File
Alpha e2_bat_fb_alpha_raw_s_0_3071.raw.gz
Beta e2_bat_fb_beta_raw_s_0_3071.raw.gz
Beta Bis (a second aquisition campaign on device Beta) e2_bat_fb_bis_beta_raw_s_0_3071.raw.gz
Gamma e2_bat_fb_gamma_raw_s_0_3071.raw.gz
Delta e2_bat_fb_delta_raw_s_0_3071.raw.gz

You need to uncompress these gzip files, e.g. with gunzip.

Panda dataset

In order to test stochastic-based template attacks with 16-bit data, we have used the following dataset, acquired from the Beta device:

Note: it has about 10GB, compressed. Please see the readme file associated with the code and the paper below for more details about this dataset.

The code also contains scripts to use stochastic-based template attacks both with the Grizzly and Panda datasets. These should allow you to reproduce some of our results from this paper:

If you find our code or data useful please cite the relevant papers in your work.