Reduced CW measurement for Drift Detection

Experiment:

-The dataset was captured using the Chipwhisperer CW308, with the target device being STM32F3 with an ARM Cortex-M4

-The target device performed an AES-128 encryption while we measured the leakage traces

-The experiment lasted for approximately 6.5 days 

-The data is organized in parts of 100k traces each. Each 100k-sized part was captured in approx. 38 minutes. Each trace has 5k time samples (features). The original experiment has a total of 254 parts of 100k traces each. 

-Every 100k-trace data part is called tracesi.mat and comes together with labeli.mat, for indexes i = 1, 2, ..., 254

-The labeli.mat is the value of a single sboxoutput of AES-128 i.e. the label ranges in the set {0,1,...,255}. We assume that successfully recovering the sboxoutput implies successfully recovering the respective key byte of AES-128.

-To limit the amout of data to process, we limited the uploaded tracesi.mat to certain subsets of the original experiment (denotes as Sets 1, 2, 3). The sets are described here:

Set 1: tracesi.mat and labeli.mat with index i = 0 until 10 
Set 1 contains the data that is close to the beginning of the the experiment
(approximately the first 7 hours of the experiment)

Set 2: tracesi.mat and labeli.mat with index i = 245 until 254
Set 2 contains the data that is close to the end of the the experiment
(approximately the last 6.3 hours of the experiment)


Set 3: tracesi.mat  and labeli.mat with index i = 0, 15, 30, 45, ...., 240 = 0:15:240
Set 3 contains data through the whole 6.5-day experiment in steps of approximately 10 hours. That is, traces0.mat and label0.mat are observed at hour 0. Then, traces15.mat and label15.mat are observed at hour 10. Then, traces30.mat and label30.mat are observed at hour 20 etc.


The full 6.5-day experiment that contains all tracesi.mat and all labelsi.mat for index i = 0,1,2,...,254 is available here (210 GBytes):

https://doi.org/10.21942/uva.24949077