Skip to content

Latest commit

 

History

History
32 lines (23 loc) · 1.04 KB

Readme.md

File metadata and controls

32 lines (23 loc) · 1.04 KB

GPU Side Channel Attack

This repository implements a side-channel attack on Jetson Nano that allows the attacker to do website fingerprinting on the victims browser and track users behavior.

This is a proof-of-concept from the following paper - Side Channel Attacks on GPUs

We use the edge device Jetson Nano to demonstrate the attack.

Get Started

Install CUDA

Use the following steps to Install CUDA on Jetson Nano

Run the code

# Build the file
mkdir build
make

# Run the code
make run

# You should now see the performance metrics being returned every second
# These metrics can be used to infer behavior of the user
# Check the folder `analysis/` for more information

Get list of available performance metrics

cd /usr/local/cuda/extras/CUPTI/samples/cupti_query
sudo ./cupti_query -device 0 -getmetrics