ChipWhisperer TVLA is a python project that aims to make running Test Vector Leakage Assessments (TVLA) easier, a testing methodology for evaluating the resistance of a microcontroller to side-channel power leakage.
cwtvla focuses on AES (128 and 256) and supports the following TVLA tests:
-
Non-specific tests
- Fixed vs. Random Text
- Semifixed vs. Random Text
- Fixed vs. Random Key
-
Specific tests
- Random Vs. Random Text
cwtvla supports arbitrary leakage points for Random Vs. Random evaluation
for both value/distance of bits/bytes.
Make sure sphinx is installed and navigate to docs/, then run make html.
ktp = cwtvla.FixedVRandomText()
key, text = ktp.next_group_A()
waveA = capture_trace(scope, target, text, key)
key, text = ktp.next_group_B()
waveB = capture_trace(scope, target, text, key)For acquiring Random Vs. Random data, use the Fixed vs. Random Text group B and save both the trace waves and plaintext. A single set of Random Vs. Random data can be used for all Random Vs. Random analysis.
You can store your trace data however you like, but analysis functions expect numpy arrays.
Once you have your data, you can also use cwtvla for analysis of your data.
For non-specific tests, a simple:
t_val = cwtvla.t_test(group1, group2)
fail_points = cwtvla.check_t_test(t_val)
if len(fail_points) > 0:
print("Test failed at: {}".format(fail_points))will suffice.
Random Vs. Random analysis is more complicated and requires you to create a function describing where you want to evaluate leakage at. Some basic leakage models, such as the SBox output and the distance between the input and output of a round, are provided. You can also modify which rounds, bytes, and bits/vals are tested.
cwtvla also has a module to take care of setup and integrate with different ChipWhisperer
scopes and targets.
A basic showcase is available in the examples/ directory.